PDA

View Full Version : In progress - 8020, linear rails, ball screws, steppers



stevespo
11-17-2006, 11:52 PM
After many months of reading the forum and scribbling on graph paper, I have finally put chopsaw to aluminum and started building. My basic design goals were to have a solid benchtop (moving gantry) machine that I can use for woodworking - with enough accuracy to (hopefully) cut pearl for inlay. My budget ruled out commercial units, so I chose the DIY route. Also I guess I was looking for a challenging project that could bring together a bunch of interests.

I accumulated a lot of great materials off of eBay and through the Zone classifieds.

PC 890 (new)
Parker Actuator 6.5" travel (Z axis)
Lots of 8020, connectors, etc
33" THK SR15 rigid rails
39.5" IKO SR15 rigid rails
2 38.5" rolled ballscrews with preloaded nuts and bearings
Honeywell 914/924CE switches for home switches
Freud Diablo 10" non-ferrous table/chopsaw blade
Xylotex 3 Axis with 480oz steppers

Months ago, I purchased the Xylotex 3 axis controller kit and installed it in a box with PS, etc. I was able to test out the actuator and ballscrews by driving them from my laptop with Mach2. The accuracy blew me away, and made me very hopeful. While I was playing with the electronics I went through several iterations of design and did some mockups with the 8020.

This past Monday I finally decided it was time to stop planning. I was nervous about cutting the aluminum, but it's actually very easy with my 10" chopsaw and contractor's saw. Using a spray can of Tap-Magic to keep things lubed.

I've been able to put in 4-5 hours each day this week and have made some good progress. I'll post some pictures on a sort of day-by-day basis.

So, Monday & Tuesday got the basic framework together for the base, measuring 37" wide by 34" deep. 8020 extrusions and some 1/2" extruded bar stock (4 1/2" wide). The side rails are 3"x3" and the rest is 1.5"x3" on it's side. My drill press is lousy, so getting the holes lined up took some time. A small mill with a nice sliding table would have saved me many hours.

stevespo
11-18-2006, 12:02 AM
Made some mounting brackets out of 1/2" stock and the gantry uprights out of 1530 stock and some 1515 for 45 degree bracing. Cut the 3/4" holes with a Forstner bit and bolted everything together. Took some fussing to get everything square and plumb, but pleased with the results.

With a temporary bar clamped between the uprights, the "gantry" just sails on those rails. As best I can measure, they are within .002" of being parallel and very square. Given that I'm assembling this on a fairly flat MDF tabletop and not a surface plate, I'm going to get things as close as I can and then move onto the next step.

Steve

stevespo
11-18-2006, 12:29 AM
Keeping busy. Made a sturdy arm to hang the Z-axis on. This is 2 pieces of 1530 (1.5"x3.0") stock bolted together on edge, and then bolted to the uprights. It is very rigid. Added the lower part of the gantry using a piece of 3075 (.75" x 3"). I can't seem to get this to rack, so I feel like I've made a good choice by going with the 8020 and using the larger extrusions.

The rails are attached with long strips of steel, that were pre-drilled and tapped for 4mm socket head screws on 60mm centers. Very handy. I knew I wouldn't be able to accurately drill/tap those holes, so this helped me out tremendously. Thank you 8020 Surplus. All my 8020 supplies were purchased through the outlet and very easy dealing with them.

I then made a bracket to mount the actuator, and attached the Z-axis. I debated about how to handle this detail. I wanted to have a decent amount of clearance under the upright, but didn't want to sacrifice too much rigidity. I placed the bottom of the rail 6.75" above the table. This will be 6" above the MDF workboard. Then factoring in clamping/fixturing, router chuck, etc - I should still have decent clearance. The nice thing about this design is that if I run into problems, I can lower the entire arm by unbolting and sliding it down the upright.

I was a little hestitant about mounting the actuator this way - so that the stepper is moving the router, and the weight of the actuator and itself. In testing things out, the stepper had no problem with the weight, and I felt it would be the stiffest way to mount the router. I could hang the router off of the moving table and keep the actuator/motor fixed, but will reserve that in case my first choice doesn't work out. Would appreciate any ideas on that.

Now I need to get the 2nd ballscrew mounted and attach the moving parts to the drive system. I can move the Z up and down and I'm anxious to get all 3 axis going. Not sure if I'll have any time this weekend, but I'm hopeful.

Steve

GaryCorlew
11-18-2006, 07:28 AM
This is a great looking build, I made mine from 8020 extrusions as well

stevespo
11-18-2006, 10:53 PM
Less than a year ago, I knew almost nothing about CNC. Now I feel like I'm getting close to having something really useful and accurate in my shop.

Seeing other people's success with the 8020 really inspired me to give it a shot. I hope the combination of choices I have made (and the attention I've given to the design) will pay off. I'd like to hear any feedback people have, because I'm sure I've done some things in a less than ideal way.

This forum has been invaluable in helping to educate me. I've definitely been inspired by all the great work people are doing and their willingness to share that experience. I'll keep the photos coming and hope to eventually post a short video when I'm making some cuts. I know I will encounter some problems before that happens and I look forward to getting more help along the way.

Steve

epineh
11-19-2006, 06:29 AM
Nice work Steve, keep the pics coming.

Russell.

stevespo
11-20-2006, 08:03 PM
Progress sure does slow down when you have to stop and think about things. Needed to finalize how I was going to attach the drive mechanism to the gantry. Managed to make 2 flanges which thread onto the ball nuts. 7/8" Forstner bit and a 15/16-16 tap worked very nicely.

The ballscrew and bearing blocks attach to a piece of 1530 that runs down the center of the frame. I can pull this out quickly and work on things. You will notice the high precision Baltic birch end blocks - don't laugh. They're temporary, just to get things together and tested. I plan make "real" ones out of aluminum (maybe delrin?) when the mill is functioning. If I can't do it myself, I'll have a local machine shop do it.

Photos include closeups of each end block (front and back). Another shot of the back of the machine, showing the stepper in place. I'm able to move the gantry back and forth quite easily. Speed right now is limited to 54 ipm. I'm running on a 400Mhz laptop (7 years old), and I have problems if I try and go any faster. I have an old 800Mhz desktop (AMD) that I plan to setup and use. One of my neighbors was tossing this out and all it needed was a new hard drive. I hope that's fast enough. If not, I will have to scrounge something else. What burns me is I have a fast new laptop with no parallel port! From what I've read, the PCMCIA -> parallel adapters won't do the trick.

Tomorrow I will attach the last ballscrew and I should be able to drive all 3 axis independantly and simultaneously (PC willing). Then I just need to make one last mounting plate for the router and take it for a spin.

Steve

Mike F
11-21-2006, 03:31 PM
This is one, good looking machine. That 8020 really does make a professional job. Of course, it does take a modicum of genius to put it together properly too :) :) and it looks like you're doing a cracking job.

Please do keep the pics coming as a lot of would be builders get lots of ideas from them rather than the text. Once they are hooked by the looks, they will start reading, and this one's a looker.

Mike

stevespo
11-21-2006, 08:09 PM
Thanks for the comments. The 8020 does make for a clean looking machine, IMO. I'm a long way from being a genius with it, but happy with the results. It's a wonderful tool for prototyping - I hope it will hold up for this use. I'm sure a bit of Lock-Tite wouldn't hurt.

Today I needed to fabricate a couple of brackets and a mounting plate (nema-23) for the X axis. I oriented things so that you face the long (X) axis when standing in front. I guess this isn't typical, but it seemed to make more sense for my workspace and the way I hope to use it.

Couple of pictures of the X-axis ballscrew, mounting brackets and then a few shots of the results of my "Sharpie" testing. Just kept drawing a circle and a square (offset by .5").

At first, I was encountering some hesitation and lost steps when moving the gantry, but backed off the speed and it performed very well. Not sure if it's my slow computer, my Mach2 setup, my controller/motors, etc. I need to go over my settings again and read the manual.

I can run the gantry at 45 ipm reliably, but faster caused problems. Definitely no problems with racking/alignment. I can chuck the ballscrew in my portable drill (lowest torque setting) and fly the entire gantry around. Smooth and very fast at over 180 ipm.

Interestingly, I can move the upper part (what the Z moves on) at 190 ipm using the stepper. Moving a lot less mass around, but both parts are just silky smooth on the rails. I will keep trying to figure that one out and will get the faster computer setup ASAP. I don't know what to expect as far as performance goes with the Xylotex board and the 480 oz. steppers - any ideas?

Next step is to mount the router, home/limit switches, etc. I'm confident that I have everything lined up well and I'm ready to tackle some routing... after the Thanksgiving holiday. Hoped to be further along, but still happy with the progress.

Steve

stevespo
11-27-2006, 11:29 PM
Success! I made another mounting plate and attached the PC890 to the Z axis. After some "air routing" I decided to cut for real and the results were great. Only tried the simple circle/square, but they were just about dead on (within .001").

Circle is 2" diam, and 4" square. Cut to .375" deep with a .250" upcut bit in 3 passes. I've been playing with the Mach 3 motor tuning and have things running smoothly at 45 ipm, cutting around 30 ipm. I hope to improve this, but will require some playing around with the controller and software.

I still need to wire up the home/limit switches, and deal with dust collection. I might need to counter-balance the weight of the Z actuator and router. I seemed to notice some slipping on the Z, but can't say for sure.

I'm eager to start cutting some more complex shapes. So far, my expectations have been exceeded with the accuracy. I'd like to improve the speed, but it seems like a usable setup as-is. I'd appreciate any ideas as to what to expect with 5 tpi ball screws, the Xylotex and 425oz/in steppers.

Steve

stevespo
11-28-2006, 10:40 PM
Spent the day cutting parts and getting some practice. I'm amazed at the quality of cuts that I am getting. When I tried to cut one of the sample PhotoVCarve pictures, I definitely saw slipping on the Z-axis. I had suspected it before, but it was very minor. After a few minutes of VCarve, it was very noticable.

So, I read up on counterbalancing and decided to go with a gas strut. I happened to have 2 struts from a pickup cab's rear window, each rated at 24 lbs. My Z mechanism and router weigh 28 lbs. So, I attached one strut with the hope that my 425 oz/in stepper can maintain the 4 lbs or so differential. I think this will work ok. You can see the strut towards the right hand side of the photo, attached between the base plate and the upper part of the Z actuator.

I'm going to need to upgrade my computer at some point. It runs Mach 3 ok, but could definitely be faster. The real issue is when I want to edit a drawing w/BobCAD and then use Mach 3 to cut the part. Pretty slow trying to multitask. Maybe just stick a new motherboard and more memory into it - or look for a cheapo on sale.

Steve

robe_uk
11-30-2006, 04:00 AM
Hi Steve
Great looking machine you have there, I always like the look of a 8020 machine.

I am sure i have seen somewhere a link to an interface which will connect to a computer usb and allow connection to the commond stepper drivers, am at work at the mo so will try and dig it out later, details are sketchy as i never paid to much attention at the time, maybe someone else has seen it?

robe_uk
11-30-2006, 09:34 AM
Hi again
here are two links for usb to parrallel port, i dont know if they are any use to you as i havent looked into them in any detail.
http://usbcnc.com/
http://www.cncdudez.com/usb-par.html

Good luck with the rest of your build.

Rob

ViperTX
11-30-2006, 10:07 AM
Steve, it's absolutely amazing what you've been able to complete with a minimum amount of tools....see now you'll be the inspiration to other dreamers!

Great Job!

WilliamD
11-30-2006, 12:00 PM
Very impressive! I really love the professional look of the 80/20 machines also. With those linear rails and ballscrews, I bet you do have some pretty good accuracy! ;) Very nice job drilling the holes with the drill press. Just based on the pictures, I would swear you have a mill setup somewhere. Congratulations!

stevespo
11-30-2006, 08:26 PM
Thanks for the feedback. I appreciate you guys taking the time to look. Just me, my chopsaw and drill press. Doing all that work on my drill press was a pain - especially since the amount of runout is pretty extreme. Took me a long time, but came out fairly accurately.

One "trick" is that I did CAD drawings of the parts, printed them out and then spray mounted them to the aluminum plate. Then I could use a center punch and start without having to measure anything. If I had really thought things through, I should have bought a small mill or at least a sliding table.

I think I discovered why my Z was misbehaving. Nothing to do with the weight or mechanicals. I was losing steps during aggressive movement because my Xylotex controller was overheating and causing "thermal shutdowns". This is mentioned in the documentation. I dialed back the voltage (from 3.6V to 3V), added another fan and it is working beautifully. Just did one of the PhotoVCarve demos and it still tracked perfectly after 12,000 instructions.

I am still seeing some intermittent hesitation on the X & Y. Doesn't seem to be losing steps, but it balks occasionally. Seems to be speed related. Some speeds just appear to work more smoothly. All this seems kind of like voodoo to me. Not sure if it's electrical in nature, perhaps software, or even my computer not being fast enough. Sometimes it's flawless, other times a bit jumpy. Again, not losing steps - just not smooth - like I'd expect.

I'll keep playing detective, but the accuracy and cut quality is fantastic.

Steve

stevespo
12-03-2006, 10:52 AM
Installed my home switches (Honeywell 914CE & 924CE). They're doing double duty as limits as well. Installed a dial indicator to make sure that when I say go "home" it returns to 0 - and it really does.

I know a lot of people skip home/limits, but (for me) it has made using the machine much easier and more enjoyable. Being able to say "go home" and have everything return to the same location is a HUGE benefit for me. Even more so when I stop playing and start doing some real work on it.

To set the Z height, I borrowed a trick from a friend. I am using a .250" thick bar of aluminum like a home switch. I followed a diagram for a typical NO switch. The router is attached to ground, the bar gets 5V via a 10K pullup. I am using pin 10 (defined as Z home in Mach 3). When the circuit is closed, the signal goes low, and the Z is "home". I have Mach adjust for the thickness of the bar.

Another photo is of the "Big Ben" demo in PhotoVCarve. This was cut with a .125" ballnose cutter, shallow depth, 30/15 ipm. It may not look like much, but this was over 10,000 gcode instructions.

The last photo is some gold mother of pearl I cut and inlaid into ebony. My 8 month old daughter's name. The pearl letters are ~.5" high, cut slowly @ 5/3 ipm with a 1/32" carbide upcut. The pearl is .060" thick and cut in 4 passes. The ebony was cut with the same bit, but at 10/6 ipm. I'm sure this is pretty conservative. I undersized the letters by .001" around the perimeter, and they slipped in perfectly. The fit is amazing.

Friends keep asking me "how did you learn to build this machine? Did you read a book?" I tell them I read every DIY post on CNCzone and tried to make some good decisions. Without this type of community and resource, no way I could have done this. What a thrill to see is in action.

Steve

blau_schuh
12-05-2006, 08:49 PM
Steve, great job on this machine. If my 8020 frame looks half as good as yours I'll be ecstatic. The rest looks spot on as well.

eman5oh
12-05-2006, 09:06 PM
Steve, I see what look like rubber caps or wipers on your ball nuts. where did you get these? I would like to find some wipers for my 5/8 5 tpi ball screws, if anyone else knows where you can get some that would fit that size screw let me know. Nice looking machine btw.

stevespo
12-05-2006, 09:30 PM
The brush wipers came with the ballscrews/nuts (I don't even know what brand they are). They're not labled, but look like Thomson or Nook. One end came with a retainer, and on the other I attached them to my homemade flange. Google "ball nut brush wipers" and you'll turn up some sources.

On another note, I cut some aluminum today, and it worked beautifully. Light, slow cuts with a shot of cutting fluid. Standard 1/4" carbide upcut router bit for woodworking running at 23K RPM. The finish was amazing. Not something I'm going to do very often, but good to know that I can do some light duty milling.

Steve

blau_schuh
12-05-2006, 10:45 PM
Steve, another question/comment. What is the reasoning behind putting your rails on the sides of the 8020 rather than laying them on top?

Wouldn't the stress from the gantry on the linear slides be less?

stevespo
12-06-2006, 10:23 AM
I can't say I put a lot of thought into their location. I chose the sides, because it made installation and alignment pretty easy. It seems like other builders do this as well so I figured I couldn't go too wrong.

I considered the inner sides, because it would keep them away from dust/etc, but it seemed to make assembly and alignment harder. On the outside, I can disconnect the lower strut and slide the entire gantry off, if I needed to move it.

Putting the rails on top makes a lot of sense, but I didn't want to "box in" my work area. This way I can put slightly over sized pieces on the top, although I'm still constrained by the gantry uprights. Most of the "on the top" designs seem to use dual screws to drive the gantry, and I wanted to stick with one screw underneath the table, because that's what I had access to.

Again, not for any particular reason - it's just what I envisioned when I started to design the machine.

Steve

bill south
12-06-2006, 10:47 AM
.

One "trick" is that I did CAD drawings of the parts, printed them out and then spray mounted them to the aluminum plate. Then I could use a center punch and start without having to measure anything. If I had really thought things through, I should have bought a small mill or at least a sliding table.

Steve

Steve;
This work is fantastic and looks very professional. Would you consider sharing the cad files you prepared and shoot, while I'm asking, a materials list?? It would save folks a bunch of time.
If you are considering selling them, let me know the price and I may be interested. As you know, even though the design is simple (in a very smart way), it will still take hours of fabricating/designing to get where you have already been. Good job!
Bill
:)

Bowman
12-06-2006, 11:24 AM
Very nice looking build I am sure you will get lots of fun from it, well done.

As for your Xylotex board I have mine set on my Sherline mill to 3.3V/3.4V just below that 3.6 max to leave a little overhead so as not to max it out. Also have it well cooled with a fan blowing in and one out moving air constantly across the board. I haven't run it for hours on end but I never have had a heat issue with those settings.

As for your 425 oz. steppers what are the specs regarding voltage and have you seen the speed/torque curve they produce? Are you microstepping the motors and what diameter are your screws? Remember that with an increase in RPMs the stepper loses torque.

I plan to use the 282 oz. steppers on my build with my xylotex board to see if I can get satisfactory results before spending the $$ for Geckos and NEMA 34 motors.

Also when you cut Aluminum take into account the chip load per tooth at the feedrate/rpm your running. If your spinning a 4 flute cutter at 12000 rpm and not moving very fast the cutter will just continue to recut the chips and make them weld to the cutter. I have never cut Alum on a DIY router but if you think its stiff enough try cutting some Alum at a depth of say 25/33% of the cutters diameter and slowly work up the feed speed. With a cutter rotating at 12000 rpm it needs to move pretty quick to maintain the correct chip load per tooth and not damage the cutter.


Bowman

cncfoam
12-08-2006, 10:30 PM
Guys,

Here is the link to my build with 8020. http://www.cnczone.com/forums/showthread.php?t=16818

I tried to detail as much as possible. So between this build ( nice job, by the way) and mine, those of you looking to build an 8020 build, there is some nice info.

also this build is a good build with 8020. http://www.cnczone.com/forums/showthread.php?t=20656

EZ

stevespo
12-09-2006, 12:59 PM
Bill, I will post my drawings when I get a chance to clean them up and make sure they're accurate. The useful ones are the mounting plates and those should work with few mods. I could also provide a basic cut list for the 8020, but those dimensions could be changed to suit the builder, based on rails, screws, etc.

Bowman, I do not have a torque graph for the 425oz/in steppers, but they are from Xylotex. I am also running at 3.4V, 1/8 microstepping. The screws are 5tpi rolled ballscrews, 5/8" diameter. I can run them pretty fast, but they stall and lose steps. Around 42 IPM (210 RPM), they have been rock solid. I'm a little suprised that I can't produce more speed with a reasonable amount of torque.

Like I had mentioned, my meager cordless drill on it's weakest torque setting can toss both axis around at 200 IPM. My 8 mo old daughter can make that drill stall, so I think my motor tuning is also suspect. It's possible that for my setup, the 282's would have been a better choice, but I haven't seen the curves to compare. Maybe just operator error.

EZ, both of these 8020 builds inspired me. Great looking machines and lots of good detail to get people like me started. Thanks!

Steve

Mr.Chips
12-09-2006, 01:42 PM
Spent the day cutting parts and getting some practice. I'm amazed at the quality of cuts that I am getting. When I tried to cut one of the sample PhotoVCarve pictures, I definitely saw slipping on the Z-axis. I had suspected it before, but it was very minor. After a few minutes of VCarve, it was very noticable.

So, I read up on counterbalancing and decided to go with a gas strut. I happened to have 2 struts from a pickup cab's rear window, each rated at 24 lbs. My Z mechanism and router weigh 28 lbs. So, I attached one strut with the hope that my 425 oz/in stepper can maintain the 4 lbs or so differential. I think this will work ok. You can see the strut towards the right hand side of the photo, attached between the base plate and the upper part of the Z actuator.

I'm going to need to upgrade my computer at some point. It runs Mach 3 ok, but could definitely be faster. The real issue is when I want to edit a drawing w/BobCAD and then use Mach 3 to cut the part. Pretty slow trying to multitask. Maybe just stick a new motherboard and more memory into it - or look for a cheapo on sale.

Steve

Hello Steve,

The "Z Slipping" issue, is this something that everyone has and just doesn't mention?

Or Is this something that is just on your machine?

Why is it slipping? You would think with 425 oz/in steppers that would not be an issue. Could you discuss this a little?

Did you make or purchase your "Z" Axis? Like the bellows cover.

Like the looks of your 8020 construction, a very clean build

Hager.

stevespo
12-09-2006, 02:39 PM
Hi Hager, I did solve my "Z slipping" problem. For some people, this problem is weight related and is solved by using a counterbalance. That was my first thought, and I spent a good bit of time adding struts only to find that it made no difference. The 425 oz/in stepper is a good match for my Z axis, and I've removed my struts.

In my case, it was unrelated to the weight of the router/Z. My controller was overheating, causing "thermal shutdowns" (ie. lost steps). This was only during extended, aggressive Z movements. I backed off the voltage, added a big fan, and the tracking on all 3 axis is perfect. Positioning and repeatability are just great.

My Z axis is a Parker actuator purchased used on eBay. THK brand appear much more frequently, but the Parker is a very nice unit. All aluminum construction, 6" wide, 6.5" travel. 5/8" 5 TPI screw, on 15mm rails. Even came with optical limit switches. A perfect find for my CNC, and it made the build a lot easier!

Steve

Mr.Chips
12-09-2006, 04:05 PM
In my case, it was unrelated to the weight of the router/Z. My controller was overheating, causing "thermal shutdowns" (ie. lost steps). This was only during extended, aggressive Z movements. I backed off the voltage, added a big fan, and the tracking on all 3 axis is perfect. Positioning and repeatability are just great.

Steve

Yep, that sounds like the problem. So do you think you need the piston "Z" help after making these adjustments?

It probally doesn't hurt though, it is better to have it as netural lift force as possible.

Hager

crocky
02-21-2007, 08:38 PM
Bill, I will post my drawings when I get a chance to clean them up and make sure they're accurate. The useful ones are the mounting plates and those should work with few mods. I could also provide a basic cut list for the 8020, but those dimensions could be changed to suit the builder, based on rails, screws, etc.

Steve

Hi Steve,

I am just about to begin building a machine like yours :) and some cad drawings would really be very useful (even unedited would be fine). Just something to get started with.

How is the machine going now?

Bob

stevespo
02-22-2007, 12:58 AM
The machine is working out better than I hoped. It's definitely exceeded my expectations in terms of accuracy and utility. I think the 8020 is a great platform for this kind of project. I'm using it almost daily and have been doing some very precise inlay with great results - not to mention making other parts, jigs, fixtures, etc.

The speed is also decent. With the Xylotex setup I'm getting about 90 ipm "blended" movements. So 90 ipm on one axis or 90 ipm total split 2 or 3 ways. My PS doesn't have enough power to drive the system faster than that. I generally am using small cutters, so 60 ipm has been a good upper limit. Lost steps have not been a problem.

I'll get the drawings together as soon as possible. Should I just post them here as attachments? Maybe just a single .zip archive? It's really just the mounting plates, but it might be helpful. I'll also make a quick cutting list so you can get an idea of what materials were used where. If you go through the early photos you can see how the machine went together and I'll elaborate wherever you'd find it helpful.

Steve

Bowman
02-22-2007, 02:50 PM
Great job. Post the drawings anyway you like a zip works good for one fail swoop at it. I am really leaning towards the 8020 as a base frame and possibly uprights althought I am tempted to go plate for those 2 pieces. My Y is already made from a nice piece of extruded aluminum with plenty of t-slots for goodies and my linear bearings are mounted perfectly with a 32" ballscrew. This is attached to another slim extrusion w/ t-slots perfect for mounting the adjustable height router mounts in the future.

Working on the Z axis mount now that I have the materials and thinking about the X axis setup as I am still up in the air about its configuration. The rails I have are half inch diameter doubles with a solid quarter inch plate seperating the two round rails. The bearings wear closed for top and bottom but I cut them in half to make them open. Just need to come up with a good way to seat the unused round bottom into a half inch slot the length of the rails which are 70" long. I am hoping not to get much deflection and if I do I will try adjustable screw stop mounts on the flats. With the long length I need a good support underneth so I don't get vertical flexing.

Any ideas from anyone on a good solution to this issue. I was thinking MDF slot cut to a half at .375" or so deep. Painted and sealed to try and avoid distortion over time but not sure if it will hold up down south here in the 90% humidity of summer. I could go AL stips with slots just the added expense and fab work since I can cut the long MDF strips by hand with the router myself.

Again sweet rig and nice results I am sure nothing but more great things to come from it all. I can't wait to get cranked up on mine again.

btw- about how much did the basic frame materials and mounting accessories cost you if you don't mind me asking? Ball park is cool I know my lengths/requirements will differ somewhat.


Bo

crocky
02-22-2007, 04:56 PM
Hi Steve,

Just zip them up and put them here in your log. It will be good for anyone else that wants them too :)

I am a little bit disabled, nothing on the right side works and I am determined to have a CNC Router and 80/20 is definitely the way for me given that I only have the left hand working. I already have the xylotec 3 axis ready to go kit and the router so I am moving.

I have got onto 8020 Australia and they will cut a kit for me, I just have to give them the dimensions. I am going for 900mm x 700mm x 200mm so it should be pretty easy to do. I will order linear bearings and rails from here in Melbourne and the ballscrews will probably come from the States.

Glad to hear that it is working so well and daily use is even better :)

Is there anything that you would change now that you have been using it for a while?

Regards,
Bob

PS. Just ordered the bearing and rails :)

stevespo
08-07-2007, 11:13 AM
I apologize for the long delay in posting these files. I wish I had complete drawings to pass along, but I only did a few parts to help me during the build. These mounting brackets might be helpful for people using a similar design. They are in BobCAD .cad format, but I believe I can export them in other formats if necessary. Just let me know.

The key things to note about these parts are that they're built around the 8020 15 series (1.5"). So the larger holes are spaced on 1.5", 3.0", etc. The smaller metric holes are generally used for mounting THK style 15mm rails. The 10-24 cap screws are for motors, router, etc.

Steve

stevespo
08-07-2007, 11:35 AM
I have been using this machine very steadily (almost daily), for the past 8 months. Not in a production setting - but I have been doing a lot of prototyping, inlay, parts fabrication, playing around, etc.

My overall feeling is that I am VERY happy with the machine. It has exceeded my expectations for a DIY unit. If I were starting over, what would I advise people?

Take extra care during assembly to get everything super square and parallel. A big surface plate and milling machine sure would be handy, but you can still do ok with basic tools and a large workbench. I have gone back and done a little tweaking and the 8020 design allows for this.

Match your electronics and motors to your expectations. I am happy with my Xylotex setup, but there are times when I'd like a bit more speed (who wouldn't). This is a big, heavy machine and I still cut reliably at 42-60 IPM, which is great for small scale work on a benchtop machine. I recommend this setup, but don't expect blazing performance due to the limited power supply and power handling.

If I were starting over, I would probably consider some larger steppers and Geckos - but my setup works so well right now that improving it is a low priority. Servos are always another possibility.

I'd really like a spindle capable of lower speeds for cutting non-ferrous metals and plastics. The PC 890 has been very solid and accurate, but it appears I need to replace the bearings already. I am not seeing ANY runout, just a lot of heat buildup in the shaft and collet even after short time periods. This doesn't seem too unreasonable after several hundreds of hours of use.

Workholding? A vacuum table would be awesome. I have made a lot of small HDPE vacuum fixtures, but a large table would be a big time saver. I have gone through many rolls of double-stick tape. T-slots would be useful, but I generally screws down my fixtures and use 1/4" pins for positioning them on the table.

Again, the accuracy is just great. Here is a small inlay I did for a friend. This is a .023" cutter in abalone. Moving about 6 IPM and taking very light cuts at .015" depth. The other is an example of milling aluminum slots in 8020. As you can see, very clean and chatter free.

Steve

stevespo
08-22-2007, 10:55 PM
Ok, two weeks ago I said that improving my electronics was a "low priority". Then I discovered that Gecko was having a sale, and that changed things a little bit. I had already decided to clean up my wiring and add a proper e-stop and relay control, so I decided to bite the bullet and swap out the Xylotex for (3) Gecko G203V drivers.

I had a great phone conversation with Mariss last week and he felt that I could achieve the speeds I desired (120 IPM rapids) with the current 24V/5A power supply and my existing 425 oz/in steppers. He felt that my current top speeds (60 IPM rapids) were limited by midband resonance and that the G203Vs would solve it. Knowing that I wasn't going to have to replace everything definitely helped my decision.

Today I started my rewiring, and how does it work? Amazing! I'm absolutely floored by what I'm seeing. The motors are silent. They are smooth and fast. My heavy gantry accelerates effortlessly. The full testing may have to wait for the weekend, but I'm seeing *250 IPM rapids* with absolutely no lost steps. Incredible. The motors stay cool. The drivers stay cool (set at roughly 2.8A). It's like a new robot. Now the limiting factor is the 45K pulses/second that Mach generates.

So, this is not meant as a slam on Xylotex. It has been a solid performer for me and it was a great way to get up and running. For the type of work I'm doing, it will be very helpful to have more speed. My hat's off the the folks at Gecko. Stay tuned.

Mr.Chips
08-23-2007, 01:36 AM
Hello Steve,

I'm glad to hear that everything is running smoothly with you and that upgrading to Geckos has been such a good experience.

I have a stack of 80/20, linerear bearings and ball screws just waiting to get my first machine running and proved. Have been plauged with problems, on a trip to Austin TX I took my control box, steppers and PC to a fellow CNCZone user and he found that I wasn't getting a signal through my optically protected breakout board, have ordered replacement parts so in a week or so I'll try again. He also found some settings that were in error in my Mach 3 setup.

Keep thoes pictures coming.

Hager

eman5oh
08-23-2007, 08:15 AM
The Gecko's may cost considerable more than a hobby set up, but you get a considerable performance gain. I have used other hobby based systems in the past and now only use Gecko's as the performance is well worth the extra cost.

sandiegocadcam
08-26-2007, 11:20 PM
hello stevespo.....i sent you a PM. let me know what you can do.

thanks,

sandiegocadcam
08-26-2007, 11:21 PM
sorry..double post.

stevespo
08-28-2007, 09:22 PM
I have rewired my drivers, although it's still not quite as neat as I'd like - it's a big improvement over my first try. I moved everything from a small metal box to a drawer under my router - as well as putting the PC and vacuum pump close by. Very convenient. I still need to add a drawer front and a door (and some ventilation), but it's getting there.

The picture shows my small power supply, Geckos, and 2 boards from Hubbard CNC. I'm using the relay/breakout board as well as a small board to simplify using limit/home switches. The relay board makes adding an eStop very easy, and eventually I will have it controlling my router and dust collection as well.

The wires and bar of aluminum are how I zero my Z axis. Just a simple NO switch. I clamp the ground to the router and +5VDC goes to the .250" bar. When they make contact, the switch is closed and I automatically offset by .250" to give me 0.

As far as performance goes - although I can generate 250 IPM rapids, they aren't reliable. I wish I understood why, because the router seems to fly around effortlessly, and then will seem to stall out at random moments. Perhaps more time spent playing with the motor tuning will help here.

Even at 120 IPM I occasionally lose steps, so it's not as big an improvement as I initially thought. However, I can run very reliably at 90 IPM, so a nice 50% gain so far. I am going to upgrade my 24V/5A power supply and hopefully that will provide more power for the Geckos and steppers. Last resort is trying some larger NEMA 34 steppers. It's so much nicer running jobs at 90 IPM, I can't wait to get a bit more speed out of the system.

Steve

stevespo
09-09-2007, 11:27 PM
Upgraded to a 72V/12A power supply, rearranged wiring to avoid possible interference, played around with dampers and resonance supression - still no solution, but... the problem might be a real simple one.

My speed issues appear to be related to my "slow" (800 MHz) CPU speed. I literally grabbed it off the curb on trash day, so you can't expect too much! Too slow to maintain the timing necessary to drive my motors reliably above 90-120 IPM. At least that's my current theory...

http://www.cnczone.com/forums/showthread.php?t=43258

Steve

Rhodan
10-17-2007, 09:50 AM
Hi Steve,

Fine looking machine! I've just been looking around for some 80/20 suppliers in my area as I've got some ideas on making a larger second machine (bought my first one, a KT2514 from K2).

About your lost steps, it may not be a speed issue directly. I had some problems with my motors stalling intermittently (infrequent) and it turned out to be the acceleration setting in Mach 3. Accel was set to 70 which appears to be right on the very edge of my system's capabilities. After some wild-goosechase time, I dropped the accelleration to 60 and have had absolutely no problems since. I could probably work my way back and squeeze out a bit more but I really don't see a lot of difference overall and I get less shaking this way (kerk buts will probably last longer too :)

Just a shot in the dark but hey, can't hurt to check.

Rhodan
10-17-2007, 02:28 PM
Argh... Now this got me thinking and I started messing with my setup again. I'd forgotten about the mach timing setting and when I checked it was at the slowest setting (25kHz?). Since I'm running a 3.2gig P4 I changed to the fastest setting and started playing around. Now I also had just updated Mach3 to the newest release (downloaded two days ago) so I don't know whether the timing or the update had the most effect.

First thing I did was change the velocity from 70 to 135 (fastest it would let me select). Wow, does that ever move! Then I changed accelleration (it was set at 6, not 60) up to 12, 24, 60, then 100. I couldn't get the motors to stall at all but man does 100 accel bang the machine around... I dropped that to 30 just for my own comfort!

I haven't cut anythng big yet but I did do a little aluminium engraving job as a test and had no lost steps. I'll be doing a 14" diameter VCarved sign later today that's fairly involved and has some fairly heavy (for me) cutting with the V-Bit.

So, with the lowest timing setting I had a hard time with an accelleration over 6, now it'll do 100 (didn't try higher). Maybe if I had all rigid guides and ballscrews instead of plastic kerk nuts I'd brave that ;)

MRM RCModels
10-21-2007, 06:20 PM
Stevespo,

Great work! I've decided to use a 8020 set up to ease the alignment of my THK rails. My only question is what size screws (M4, M5?) did you use to mount the THK rails on the 8020s? Also, what 8020 connectors (standard T-pin?) did you use to mount the rails?

Thanks

MRM

stevespo
10-22-2007, 01:13 PM
Rhodan,

Thanks for the information. I am still running my old, slow, 800Mhz AMD machine at 25K pulses/second. My acceleration is set between 25-40 and my velocity is 90 (max). This is a comfortable speed given my current setup. Once I get a faster PC for the shop, I'm sure that I can bump up velocity significantly. Running in CV mode gives slightly better performance, and I'm sure that higher accelerations would also.

MRM,

The THK-style rails are a great match for the 8020. I used 15mm rails on both axis, and was lucky enough to find 1 meter lengths of pre-drilled nut stock intended for those rails (ebay!). I believe it was 4mm holes on 60mm centers, but please double check. 8020 will custom drill/tap those nuts to any spec you need. Part # 8900 or 8900-36, p99.

One thing that would be very handy is to mill a very shallow 15mm slot directly above the extruded t-slot you're using for fastening. This would allow you to register the rails more easily, and quickly get them parallel and planar. This isn't necessary (I didn't do it), but something I thought of later to make assembly easier and more accurate.

Steve

Matt29c
11-06-2007, 10:19 AM
Hey, About how much did that cost you? and how sturdy is it? Looks very professional. and

stevespo
11-07-2007, 06:47 PM
The approximate cost, so far, is about $2500. Checking my notes and invoices, that's about as close as I can calculate right now. It will probably continue to go up over time as I make improvements.

That is for the aluminum plate, 8020 frame, motors, drivers, ballscrews, linear rails, Z actuator, router, fasteners, etc. Most of these items came off of eBay and 'zone classifieds. Trying to spec everything out as new would cost more - the actuator and rails are very pricey brand new.

Because I was kind of 'designing on the fly', I ended up purchasing a lot of extra hardware, and materials - so I have spent quite a bit more on the project. I'm slowly selling that stuff off and incorporating it into other projects - so it's not a loss by any means. Just be prepared to spend more if you're doing your own design.

There are also a lot of tools that I did not include in that price. Things like metal cutting blades, taps, endmills, etc. Unless you already have a metal shop, this stuff adds up pretty quickly. But again, it can be used on other things down the road. Luckily I already had a full woodworking shop. Tablesaw, chopsaw, bandsaws, drill press, etc. A milling machine would have been great!

The machine is very sturdy. If you look at the basic structure, the upper arm of the gantry is (2) 40" long pieces of 1530 (1.5"x3" cross section) bolted edge to edge. The uprights are also 1530 with 1515 reinforcements. This is sliding on 15mm rails attached to (2) 3' long sections of 3030 (3"x3"). The worktable has an additional 3 pieces of 1530 to support the MDF and workpieces. The front/back and most mounting plates are 1/2" aluminum plate. The Parker actuator is a very solid piece of equipment.

Even with a decent amount of mass in these components, the rails allow everything to glide with just fingertip pressure. I'm very happy with the structure of the machine. In some respects, this machine is probably overbuilt - but I wanted something that I wouldn't quickly outgrow. I'm quite certain I could put bigger motors on it, and/or a VFD spindle and have few/no problems.

It doesn't compete with the steel frame machines for stiffness, but it does what I need it to do.

Steve

blumpie
11-27-2007, 10:11 AM
Steve,

If your still looking for a computer, I can probably help you out. I am sitting on a pile of them now. Perhaps a trade for some of your spare parts?

pm if you like.

stevespo
05-04-2009, 10:57 AM
I guess the DIY machines are never really "finished", they just keep evolving.

I fiinally got around to installing my new 2.2kW spindle, Igus chain, rewiring my control box, and installing a new computer (2.8GHz P4). With my Gecko 203v and 72V/12A power supply this is providing 150 IPM rapids and 120 IPM cutting speeds.

The spindle has been a nice addition. I like the ER20 collet and it's much quieter than my PC 890. Required adding some filter caps to compensate for the extra "noise" from the spindle as well as running all my cables in the Igus chain. Still waiting for some double shielded 16-4 spindle wire.

I looked around for a suitable (cheap, 18"x24") metal box and ended up just making a plywood box with an aluminum backplate. If I find a proper enclosure, I can just relocate the plate. It's very nice having the spindle on/off and vacuum pump under relay control. I'm going to add another relay for dust collection and possibly my spindle water cooler.

Future changes? Definitely a new breakout board for optoisolation, filtering and spindle speed control. Possibly new screws or even servos for some additional speed and acceleration. That's fairly low on the priority list, but if I can scrounge some decent motors...

Steve

stevespo
05-20-2009, 09:07 PM
Now that I'm running my machine at higher speeds, I've realized that my "temporary" plywood bearing blocks are becoming a weak link. Time to make some new ones, using my new spindle.

I'm going to stick with the 30mm 7200B angular contact bearings and a simple skate bearing on the end. I'll preload the AC bearings, lock those on the machined end of the screw, lock the bearings in my blocks and then bolt on the assembly with standard 8020 brackets.

The aluminum milled beautifully. I ran a .250" carbide upcut at 24K RPM, 90 IPM rough, 60 IPM finish, with a shallow .010" depth of cut. I sprayed a shot of WD-40 every so often, especially before the finish pass. The quality was excellent. The pockets were within .001" and I achieved a very nice fit. Firm hand pressure can seat the bearings.

Here's a few pictures and a short video:

<object width="425" height="344"><param name="movie" value="http://www.youtube.com/v/5oydlX8t660&hl=en&fs=1"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/5oydlX8t660&hl=en&fs=1" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="425" height="344"></embed></object>

Steve

stevespo
05-20-2009, 09:12 PM
Chrome failed to upload pictures. 2nd try.

mike2768
05-21-2009, 10:38 AM
WOW great build for sure. I plan on doing a build some. I have a X2 mini mill that i converted to CNC now but it's way to small for what i would lke to cut. I to think the 8020 is a great loking machine. I seen the video of you cutting alum. that answers my question if the 8020 is ridgit enough. Again what a great machine. I plan on using the spindle from my X2 mini mill for the new machine i build

Is there anyone with 8020 plans for building a machine like you built

Again great build,
Mike

stevespo
05-21-2009, 03:05 PM
Mike,

There are a bunch of 8020 build threads, and I'm sure you could take one of the designs and easily modify it. You can get a lot of good ideas by ordering the 8020 catalog from the website. There is also a free set of plans available that works with Ahren's linear motion components. The machine looks very good and there are videos of it doing a bunch of cuts.

http://www.cncrouterparts.com/
http://www.cnczone.com/forums/showthread.php?t=73670

One difference with my machine is that my gantry side supports are two pieces of 1530 arranged in an "L" shape, with a piece of 1515 bolted in at 45 degrees. Then the mounting bracket for the side rails is a .500" piece of 6061 - so that entire unit is VERY solid. The 15mm linear rails are also pretty solid. I don't know how well the 8020 would work in terms of a full time milling machine, but in terms of handling a high speed router and taking light passes in aluminum - it is definitely stiff/strong enough.

The X2 spindle running at a few thousand RPM is going to need to move a lot slower than a high speed spindle. I've been toying with the idea of an 8020 milling machine, strictly for light duty milling work on aluminum, pearl, plastic. My machine does well with those materials, but I'd like a T slot table or something less sensitive to liquids and moisture than my MDF spoilboard.

Steve

contactirfu
08-05-2009, 11:32 AM
Steve - just wanted to know what software you used for programming - how is that the bit comes up and goes back again!

Thanks in Advance!

R/
IRfan

stevespo
08-05-2009, 02:03 PM
That toolpath was generated by BobCAD v23. For a simple, circular pocket, the tool should NOT return to "safe Z" after each pass. This is very inefficient, but that is the path that was generated.

Other types of (complex) pockets seem to generate the exact OPPOSITE behavior. The tool NEVER returns to "safe Z", instead it travels around a circuitous path to avoid running into islands. This is also terrible behavior because it adds significant time to these operations. Not to mention that if you specifiy more than a 50&#37; stepover, it misses portions of the pocket which you have to clean up manually.

I've posted this behavior on the BobCAD support site, and didn't get much in the way of a response. I believe they want us to upgrade to the "high speed pocketing" as that may fix some of these issues. I seriously doubt it.

To my mind, these are pretty basic operations that worked just fine in v20, and have degraded in v23. Yes, the CAM tree is a nice feature and there are many more types of pocketing operations, but they have slid backwards on some basic functionality, in both the CAD and CAM portions of the product.

Yes, there are workarounds (like using a 50% stepover for everything), but that isn't how the software is supposed to work! I'll get off my soapbox now...

Steve

contactirfu
08-06-2009, 12:33 PM
Thank you steve -

sonicwonder2000
09-06-2009, 10:40 PM
I just went through this whole thread and this was a mind-blowing build! Incredible machine - seems like you are getting some equally impresive performance out of it. What did you use to go about checking the calibration of your machine? Did you chuck a dial indicator?? Specifically, how did you make sure your axis were at right angles to each other - or did the aluminum extrusion pretty much eliminate that as being an issue?

stevespo
09-06-2009, 11:58 PM
I've been really happy with the accuracy and performance. Eventually I may upgrade the motors, but I'm getting reasonable speeds for the work I do. The 8020 is very easy to work with, but it doesn't guarantee that everything is square. Even with a carefully tuned chopsaw/tablesaw, my base and uprights started out a little out of square.

It took several attempts with a 12" square (Starrett), and feeler gauges to get things square. That is a pretty good way to make sure that you're within .001" (or so) over a 12" length. The 8020 can be loosened and shifted slightly until you have the accuracy you're after. Keep some shim stock handy. Tap, tap with the rubber mallet - then tighten things back up.

To verify things, it's a matter of making test cuts. Is a 2" square, really 2.000"? Is a 6" square, really 6.000? Do the diagonals match? Give or take +/- .001?

The big question (for me) was with a 12" square, how to you check a 24"x32" table? To check the overall "squareness" of my X/Y plane, I drilled a hole in each corner of the table and then measured the distance of those diagonals. Make a small adjustment, plug the holes, re-drill, then re-measured. I quit fussing around when the measurements were within .003" of each other over a 40" span. That's close enough for my needs!

I squared the Z axis to the spoilboard, using my 12" square. You could also chuck a test indicator and rotate the spindle. Then I surfaced the table and still had some ridges left where the router bit was leaving overlapping marks. I rechecked things, added a small shim, and then resurfaced the board again. This time it was completely flat. No detectable ridges or defects on the surface. So, this is definitely an iterative process. The 8020 makes this type of tweaking very straightforward.

Other important tools - a good dial indicator, dial caliper, a quality 24" straightedge. I never skimp on measuring tools because they're the basis for any type of accurate work. The squares have to be square and the straightedges are guaranteed straight.

It all depends on your goals for the machine, and the types of work you intend to do. No sense in going overboard, if you don't have to.

Steve

MikeyC38
11-16-2009, 08:51 AM
Hi Steve

I've really enjoyed reading the posts on your build of a CNC machine using Aluminium extrusions. I too have come to similar design decisions as yours, as in the UK it is cheaper to build in ali than wood! My design is loosely based on David Steele's Solsylva 25" x 25" plans (scaled up to metric equivalent of 37" x 25") which I can HIGHLY recommend for anyone wanting to start out on this path. Also the help and encouragement from the regular posters to this forum. The advice I would give to any one starting out is to READ,READ and READ again. Build up your knowledge and understanding through reading the posts and assessing other peoples design. Then design and paper check before you start buying and building. The old adage "Measure twice, cut once" is so true. So I'm doing my design in 3D CAD to check fit and clearances, then start buying the bits. My estimated cost is £1100 including all hardware, Mach 3 but minus computer.

Regards

Mike

stevespo
11-16-2009, 11:04 AM
Thanks Mike!

Good luck with your build and don't hesitate to ask questions along the way. I totally agree with your advice to do the homework, study the existing, successful designs, etc.

My machine is still evolving as I try different types of work. The structure - frame, rails, screws have all held up really well. Now that I'm trying to do heavier 3D work at faster speeds, I'm (again) realizing the need for some different motors. Perhaps more powerful steppers, or it might be time for some type of servo setup. DC/brush, DC/brushless, AC? I don't know. More things to think about.

All in all, the machine has been a workhorse, I just keep asking it to do more and more.

Steve

stevespo
12-30-2009, 04:25 PM
After much reading and talking with friends, I've decided to go with a closed loop servo system. Now that I'm finally tackling some fairly heavy 3d carving, the steppers are losing steps unless I really nurse back the feedrates, and that then creates other problems. Premature wear, scorched parts, and long cutting times to mention a few.

So, what type of motors and how large? Based on my machine and requirements, we did some calculations and came up with a figure of 59 oz-in of torque to accelerate the loads and maintain cuts at reasonable speeds. I recently re-ran the numbers and came up with 90 oz-in, so it's hard to know for sure. The number I had come up with a long time ago looking at the torque curves for my motors was 100 oz-in, so they're all in the same ballpark.

The Gecko 320 drives with a 50 oz-in Nema 23 servo seemed like a good option, especially with peak torque in the neighborhood of 350 oz-in. Even so, I wasn't convinced and bought a set to play around with. I found I could fault them fairly easily, so my confidence wasn't real high with this solution. Really, they should probably be reduced 2:1 or 2.5:1 to give me enough continuous torque and drop the RPMs.

So what's next?

I eventually decided to go with a DC brushless servo setup and Granite VSD-E drives from ZealCNC (http://www.zealcnc.com). They offer a 70mm 300W motor that provides 150 oz-in continuous from 0-3000 RPM and over 3x that in terms of peak. My 72VDC PSU is a perfect match. So, this is a motor that I can direct drive and feel confident that I won't run out of power. It also has really nice quality encoders, and the Granite drives have a lot of really great features that the Geckos are lacking.

So, I purchased those from Jason (thanks so much!) and quickly got them running on the bench. Very easy to setup and configure. They (motors and drives) are a wonderfully HIGH quality product, and the form factor is just a little bit bigger than my NEMA 23s, but 3x the power. Really perfect. Loads of power, incredibly quiet and smooth. I'm just blown away and feel that I've got the perfect match for my machine.

I'm still waiting for some cabling and small parts, but I milled some motor adapters the other day out of some 3/8" 6061 aluminum. These will fit right onto my existing NEMA 23 motor mounts and bridge the difference from 57mm to 70mm. I also have new couplers to handle the 14mm shafts on the motor.

So, it might still be some days (or weeks) before I get around to installing them for real, but I am really looking forward to it. Stay tuned.

Steve

MRM RCModels
12-30-2009, 04:56 PM
Nice servos! I have the same stepper set up as you. At what IPM where you starting to lose steps? Did you try gearing the steppers down? I'm afraid I might run into the same problem on my big machine.

Thx

stevespo
12-30-2009, 05:16 PM
With the 425 oz-in steppers and G203V drivers (72VDC PSU) I'm pretty good in the XY plane up to 120-180 IPM. Above 180 IPM and I can stop movement with modest hand pressure. This is 900 RPM with my 5 TPI ballscrews. These particular motors have a pretty lousy torque curve as the RPMs increase.

The other issue is the Z axis where I start to lose steps cutting above 60 IPM (300 RPM). I've gas strut counterbalanced the weight of the actuator and spindle, but it didn't really seem to help. I've realized there could be other issues at work here, possibly a slipping shaft coupler.

So, I considered gearing the steppers down to reduce RPM and impove the position on the torque curve, but I really wanted to close that loop and use the feedback to know that I was really positioned where I thought I was.

I also considered larger/different steppers, but same thing - no feedback loop. I know there are no guarantees even with servos, but I'm going to feel much move confident with my new setup. Its also something that could be moved to a larger machine if I ever decide to build a new one.

Lots of potential options: new motors, screws, timing belts, etc. Decided to go with the option that I felt had the most potential for success - brushless DC servos. We'll see.

Steve

ger21
12-30-2009, 06:57 PM
You really should be getting better performance than that. I get similar speeds with a Xylotex, 24V, 250 oz motors and acme screws. 55ipm on the Z axis with 1/2-10, and 150-175ipm on the X and Y (respectively) with 1/2-8 2 start. I was 3D carving the other day along the Y axis at 170ipm. The speed is limited on steep areas by the Z axis, but on flatter areas it does get up near 170ipm. If I had a double shaft motor on the Z axis, I could put a damper on it and go even faster, probably at least 80ipm.

And I've never had the machine lose a single step, and I run it pretty close to the edge.

stevespo
12-30-2009, 07:51 PM
Gerry, yeah - I've been thinking the same thing for 3 years!

I can get much faster rapids, but there's no torque available to do anything. The mechanicals are silky smooth. Light finger pressure and even the ~100lb gantry just glides. I can connect my 9.6V cordless on the lowest torque setting and it moves with ease at 250 IPM. Everything is lubed and well maintained.

I've never been able to definitively figure out what the issue is, but I have to believe it's my stepper motors. These particular motors have a really lousy RPM/torque curve. They're very high inductance (8mH) and low current (2.5A/phase). They could actually be run at higher voltage, perhaps 90VDC.

I'm sure that there are much better NEMA 23 stepper options out there. That would have been my cheapest option, but a relatively small incremental improvement. I'm ready to make the move to servos.

Steve

rc-monster mike
12-30-2009, 11:22 PM
Awesome machine, Steve. Just read the whole thread, as I am considering building or buying a cnc router myself. This is one of my favorites. I just figured I would pipe in on the stepper/servo thing here. There are a couple scenarios that I have seen that may apply to you(and may help).

While it is true that the 425 oz/in steppers need lots of voltage to make speed and have a poor speed/torque curve, I still think you may be battling a computer issue that may go away when you upgrade the computer. This is just a hunch, but a few years ago I had missing step issues with an old laptop. My buddy had the same laptop and didn't have the issues, so I initially didn't think anything of it. I drove myself crazy for well over a year with the setup - upgraded the power supply, new steppers, etc - couldn't get over 500 or so rpms out of any steppers(including the 425s at one time) reliably. Well, one day the laptop died. THANK GOD! When I put a cheapo desktop computer to the system, everything changed for the good. The same system that would skip and miss steps on the old computer(and would fault out the servos) at anything above 500 rpms will now run the steppers at 2000+ rpms no sweat(I run them at 1750 rpms for rapids every day - 350 ipm but can get 450+ ipm on my 5mm pitch ballscrews without missing steps). The only time it misses steps is when I screw up via something stupid - we won't go there.

Another issue I have seen with laptops and newer desktops is low voltage from the parallel port, which is less obvious, but results in position errors over time. I use a voltage boost on my parallel port, which fixed this issue. I have 3 gecko stepper systems - one using G201 drives, one using G210 drives and one using G203s. The 210s work perfectly with a buffered signal(buffering in the BOB), but without the buffered signal, they slowly lost position. The 201s seem to not exibit the same issue. It has been a few years, but IIRC, it has something to do with a pulse rate or something(talked to Gecko, but haven't had any issue in a few years, so the details are rusty)? Sherline mode can sometimes fix this issue.

Moral of all this is that you may want to look at a few things aside from simply swapping to servos - the computer for one. If the computer is the root of your troubles, the servo system won't meet a reasonable expectation any more than your stepper system does. My 425s would easily run 1000+ rpms reliably after the computer upgrade(on +/-44 volts), but couldn't manage over 300rpms reliably before.

stevespo
12-31-2009, 12:40 AM
Hi Mike,

I appreciate the advice. You are right about the computer. For a long time I was using a very old computer and it was definitely the weak link. Since then, I've used two "new" computers with fast clock speeds and high pulse rates, but I'd still run into problems at speeds higher than 180 IPM.

You guys have got me thinking that there must be something else going on. In the past, I could ALWAYS stop my X axis from moving at around 180 IPM. This is using hand pressure and leaning into the moving axis pretty hard.

I just went to the shop and started running through my same tests. 120 IPM, 150 IPM, 180 IPM, 200 IPM, etc, etc. At 200 IPM I would expect to stop the axis and lose steps - but I didn't! I kept going higher, 250 IPM, 280 IPM, 300 IPM. I still couldn't stop it. I kept increasing the acceleration as well. No problem! I'm leaning hard into this with 200lbs behind it and it's just pushing me out of the way.

What is going on?!

For one thing, I am using a new (old) 1.7Ghz computer, but I know I ran these tests on my 2.8Ghz computer and it was failing around 200 IPM. So, it's probably not computer speed. The faster computer eventually had a parallel port problem, so I retired it. It's possible that it was the pulse train. Still I don't know.

Then I remembered something that I did a few weeks back. I noticed that my motor power cable was hanging oddly and when I inspected it one of the wires was severely strained, almost to the point of breaking. In fact, several strands were completely broken. I replaced the connector, introduced some strain relief and didn't think too much about it. This was the power cable to the X axis. I wonder if that bad wire was causing problems? If it's ability to carry current was compromised, that could be the culprit!

So, I kept running my "stress" test, and finally could stop the X axis from moving at 360 IPM at an acceleration of 60 in/sec/sec. That's pretty brisk. I still could not stop it at 300 IPM, so something has definitely changed since the last time I ran this test.

Ok, then why has my Z axis been acting up and losing position during fast (120 IPM), heavy cutting? It may have the hardest job of hauling that spindle up and down, but with my counterbalancing in effect, it should actually be pretty neutral.

If the X is pushing strong at 1500 RPM, the Z's 600 RPM should still be well within the decent part of the torque curve. I think the answer here might be as simple as a slipping motor coupler. I torqued it down quite well tonight and it zips up and down quite happily at 180 IPM. I'm going to run my carving test again tomorrow at high speed to see what happens.

The good news is that the rest of the machine was perfectly happy at those high speeds. No issues with the linear rails, vibration, or the ballscrews. I'm definitely going forward with my DC servo upgrade. Even if my stepper "problems" have all been my own doing, I am still going to feel more comfortable with encoder feedback, and huge torque available across a very broad RPM range (0-3000). I've gotten all my moneys worth out of those steppers and drives and don't mind taking the next step forward.

I'm going to learn a few things in the process and that's always a good thing.

Steve

rc-monster mike
12-31-2009, 01:13 AM
Your 28lb z axis shouldn't be problematic at all. I am thinking the coupler or something similar is(was) the real issue. I have a +/-50lb z that can run up and down at 400 ipm without missing steps - I have a 20lb air cylinder on it and run it at 300 or 350 ipm(not sure which) just "because", but it isn't needed(I peck drill aluminum a lot, so the extra 20lb "push" keeps my mind at ease). I also mill aluminum with high helix, and don't want the z to be pulled own into the material(or vise). I think your 100lb gantry accelerating and decelerating is far more load than the 28lb z axis.
IIRC, the problem with the 210 drivers was related to mach3 - the driver wanted a certain pulse width or something that Mach couldn't deliver - sorry, but I don't remember the specifics now - it has been a couple years without issues. The buffered BOB fixed that issue(cleaner pulse stream delivered to the Geckos or something). The computer was the 1st(and far worse) problem, though. Glad you got that sorted out on your end.
I look forward to seeing the servo plan come together. The machine looks terrific and deserves the proper 'motivation'. I am curious how long brushes last on these little DC servos. I haven't given servo much thought since the computer issue went away and 350ipm became easy and reliable. The steppers are brushless, so they should last 'almost' forever. :)

stevespo
12-31-2009, 01:36 AM
The servos that I'm going to be using are brushless DC servos, so they should also last 'almost' forever. I've also got a 24lb gas cylinder on my Z, so it should be almost neutral. The issue has to be a slipping shaft coupler. More info on that tomorrow.

Here's a video of the machine moving around at 360 IPM, with some 514 IPM 'blended' diagonal moves. Same old steppers as before, completely different performance.

I'm going to be cutting some parts tomorrow so we'll see how it goes.

Steve

<object width="425" height="344"><param name="movie" value="http://www.youtube.com/v/nx8oEiYkVGI&hl=en_US&fs=1&rel=0"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/nx8oEiYkVGI&hl=en_US&fs=1&rel=0" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="425" height="344"></embed></object>

ger21
12-31-2009, 08:38 AM
I was going to question the 60Khz kernel speed, but did the math and see that you need it to get the 360ipm.

Running at 60Khz definitely needs a fast PC with a good pulse stream, so a slower PC could have a lot to do with the slower speeds.

stevespo
12-31-2009, 08:57 AM
Yes, 360 IPM is right at the edge of what 60KHz can provide. The rapids were quite stable at that speed, although I wouldn't cut at that speed. I actually bumped it up to 75KHz and got above 400 IPM, but I think it's too much for the PC. The PC was only generating about 72KHz.

My last PC was considerably faster, but the parallel port had some issues with low voltage, etc. All along it could have been the PC (again!).

One nice thing about the Granite drives is that there is infinitely configurable step scaling. So, even though the encoders provide 4000 CPR, I can scale it back to 2000 CPR to match what the steppers are doing, or even go down to 1000 CPR and take a big load off the PC. That's still very high resolution.

Steve

ger21
12-31-2009, 09:15 AM
Do you have any issues with the screws whipping, or are they short enough to avoid that?

Drakkn
12-31-2009, 09:28 AM
Hi Steve

I've really enjoyed reading the posts on your build of a CNC machine using Aluminium extrusions. I too have come to similar design decisions as yours, as in the UK it is cheaper to build in ali than wood! My design is loosely based on David Steele's Solsylva 25" x 25" plans (scaled up to metric equivalent of 37" x 25") which I can HIGHLY recommend for anyone wanting to start out on this path. Also the help and encouragement from the regular posters to this forum. The advice I would give to any one starting out is to READ,READ and READ again. Build up your knowledge and understanding through reading the posts and assessing other peoples design. Then design and paper check before you start buying and building. The old adage "Measure twice, cut once" is so true. So I'm doing my design in 3D CAD to check fit and clearances, then start buying the bits. My estimated cost is £1100 including all hardware, Mach 3 but minus computer.



Mike

Hi

So where in the uk can you get 8020 extrusions?

Regards

Tony

stevespo
12-31-2009, 09:35 AM
No issues with screw whip even at 400 IPM (2000 RPM). My X screw is 40" long and the Y is about 32". They're both .625" diameter and .2" pitch rolled ballscrews.

I went back and re-read my thread and I find it hysterical that I'm still talking about PC speed as being the limiting factor after 3 years. I've been trying to solve performance issues this whole time and it might have been as simple as a computer swap.

New builders take note of that one - follow the Mach recommendations. Get a (reasonably) decent desktop computer and keep it super clean if you're going for higher feed rates and pulse streams.

Steve

stevespo
12-31-2009, 10:22 PM
Ran a couple small sheets of plywood through the machine today and I increased my cutting speed from 120 IPM to 180 IPM and then 200 IPM. No problems at all. Running the 1/8" Onsrud 60-102 at 18K RPM and it just breezed through the work. I'm finally starting to reach the recommended chipload of .012" - .018".

No issues, no lost steps. I may run some more parts tomorrow, or try something more challenging - like some aggressive 3d carving.

So, now one batch of parts (36pc) is taking me 8 minutes to cut. It's a complex little 2d shape. Just a couple of weeks ago, it was 50 minutes, then new tooling brought that down to 15 minutes, and now at 200 IPM feed rate (240 IPM rapid) it's down to 8 minutes. What a difference good tooling can make. Super clean cuts, no fuzz, zero cleanup.

I just love it when something is working well. Now it's almost time to start messing around with it again and see what happens...

Steve

mike2768
01-01-2010, 09:33 AM
I to am having some speed trouble. My computer is an older one and i can only get at best 120 IPM (at 35,000 kernal) thats going up to the next kernal setting in mach Higher then 25,000. I tryed to go higher and i get the jerking of the steppers. I guess it's time for a newer PC.

Do you have any idea what to look for in a PC when paying one for fast one. Any info would be great.


No issues with screw whip even at 400 IPM (2000 RPM). My X screw is 40" long and the Y is about 32". They're both .625" diameter and .2" pitch rolled ballscrews.

I went back and re-read my thread and I find it hysterical that I'm still talking about PC speed as being the limiting factor after 3 years. I've been trying to solve performance issues this whole time and it might have been as simple as a computer swap.

New builders take note of that one - follow the Mach recommendations. Get a (reasonably) decent desktop computer and keep it super clean if you're going for higher feed rates and pulse streams.

Steve

MikeyC38
01-01-2010, 09:39 AM
Hi

So where in the uk can you get 8020 extrusions?

Regards

Tony

Hi Tony,

I will be using Bosch profiles from Machine Building Systems (http://www.mbsitem.co.uk/) Not the cheapest but the stiffest I've found in the UK (4.2kg/m) The Bosch profiles (Profile 8) are here http://catalog.item.info/Onlinekatalog/web/EN/artikeluebersicht/Construction_Profiles_8_1000021846. The standard 2.0 kg/m profiles are not stiff enough in the y-axis (80mm) plane.

Building using this heavier profile reduces the maximum deflection of the beam to 0.06mm for a point load of 120N.

Now if I could only source in the UK some reasonably priced multi-start trapezoidal leadscrews and antibacklash nuts!

Regards

Mike

rc-monster mike
01-01-2010, 10:04 AM
The servos that I'm going to be using are brushless DC servos, so they should also last 'almost' forever. I've also got a 24lb gas cylinder on my Z, so it should be almost neutral. The issue has to be a slipping shaft coupler. More info on that tomorrow.

Here's a video of the machine moving around at 360 IPM, with some 514 IPM 'blended' diagonal moves. Same old steppers as before, completely different performance.

I'm going to be cutting some parts tomorrow so we'll see how it goes.

Steve

<object width="425" height="344"><param name="movie"


The machine should fly along nicely with the servos. Also, I wonder if your acceleration setting vs gas cylinder is possibly causing an issue on your z axis? The gas cylinder will act to slow an aggressive acceleration - this is part of the reason I run my z axis slower than the x and y on my mill - I "timed" the speed of the cylinder I am using(fully compress and let it go and track how long it takes to open vs the distance) and this was(is) the limiting factor for my Z axis speed. My 20lb spring times out to +/- 300ipm, but there is a notable acceleration observed, so max speed is a little faster. For the relatively short z travel on a router, acceleration is more useful than top speed, as it only moves over a short distance.

stevespo
01-01-2010, 10:54 AM
Do you have any idea what to look for in a PC when paying one for fast one. Any info would be great.

Mike, machsupport.com says 1Ghz (P4 or better), 512MB RAM and a "non-integrated" video card. Also a parallel port if you're going that route. I don't know if form factor plays a role, but I had better luck with a micro-ATX or ATX sized machines. People use them, but I'd avoid laptops for many reasons.

Grab a 2.0-2.8Ghz P4 with 1GB and you should be all set to run 75KHz or better. The other thing is to keep is absolutely as "clean" as possible. Avoid virus protection, complex apps, auto-updaters, anything that might cause interruptions in the pulse stream.

rc-monster-mike - Thanks for the tip on the gas strut. I was wondering about the same thing. I'm running the Z at 90-120 IPM or so and it seems quite fast and stable. I really don't feel the need for more speed on the Z, just reliability. I'll keep experimenting and see what happens. My problem was the deep 3d cuts with a 1/2" ballnose (in hardwood), so one easy solution is to rough it out and take lighter cuts in multiple passes at higher speeds.

Steve

stevespo
01-03-2010, 09:38 PM
Carved two guitar (tele) necks today. I swapped out the coupler on the Z axis, the one that I thought could be slipping. The new coupler appears to be rock solid.

Just for fun, I decided to double the speeds that I had been using. I went with 180 IPM for feedrate and 90 IPM on the Z feedrate. My max speeds configured in Mach are 240 IPM (XY) and 120 IPM (Z). I believe my accelerations are 40-50 IPM/M.

The first neck in rock maple, carved perfectly. I profiled, roughed and then finish passed everything in 12 minutes. No issues whatsoever. At 16000 RPM, no burn marks, the finish quality was excellent. I was really excited.

The second neck, I used a piece of Goncalo Alves I had lying around. It's a little denser than maple, and I thought it might look nice. The profile and roughing went great. As soon as it started the rapid up/down movements for the finish pass, I noticed that it was already out of position. It was cutting deeper than it was supposed to, and it's hard to know why.

Potentially it could be losing steps on the movement upward, or it could be that the 1/2" ball nose bit was pulling the spindle downward. That seems unlikely because so little of the cutter is engaged. I'm removing .125" depth and .020" stepover, so I can't imagine there is enough force present to pull the Z down. The tool is new and quite sharp.

It's more likely that the movement upward is fighting with the gas strut and I could try lowering my acceleration and see if that helps (or just remove the strut). Right now, that's too fast for that particular toolpath in that wood. I may try another one tomorrow, but it still seems like the (stepper) Z movements are going to be speed limited.

Side Note: Although I specified 90 IPM for my Z movements, because I'm actually making XZ or YZ arc movements, it appears to use the faster feedrate (180 IPM). I'm sure that Mach is enforcing my max Z speed (of 120 IPM), but the blended speed is somewhat higher and I'm sure I'd have better luck if brought the overall speed back down.

All of this is interesting (to me anyway), but somewhat irrelevant because I hope to have those stepper motors replaced by my DC servos in the next week or two. Still just waiting on some cables that USPS seems to have lost during the holiday rush.

Steve

stevespo
03-15-2010, 01:38 AM
Well, after a surprisingly busy winter and a long lapse in improvements I have finally finished my new electronics cabinet and attached the BLDC servos to the machine.

"Holy (insert here)!". It's actually quite frightening to see my gantry accelerate effortlessly and then move at 600 IPM across the (short) 2' span. Basically 2 seconds. No need for that kind of speed, but it is fun to see that it's possible. I can't see running it any faster than 400 IPM rapid, which is more than adequate and extremely smooth.

I have not done any kind of tuning yet, just using the settings I arrived at when I was configuring things on the bench. I can't budge the machine when it's at a standstill. I can even begin to slow it down when it's moving at 400+ IPM. So, far my little experiment is looking very fruitful.

I still have to hook up one set of limits and do some tuning. I'll post some pictures and videos when I get a chance. Hopefully tomorrow or soon afterwards. Can't wait to start carving some parts and see how it performs.

Steve

MikeyC38
03-15-2010, 09:27 AM
Great to hear about the performance improvement you are getting from the servodrives. For me, this justifies the use of 8020 profile to build the machine becuase you can make this type of upgrade and get serious improvement in accuracy and speed of milling because the basic machine stiffness will allow these improvements to be seen.

Well done steve!

MikeyC38

Jason3
03-15-2010, 05:19 PM
Excellent - great work! I hope the upgrade lives up to expectations :) Looking forward to the video footage now!

Best regards,

Jason

ger21
03-15-2010, 07:07 PM
What's the conversion rate from US dollars to NZ dollars? 'Cause that looks like a pricey upgrade, looking at Jason's site. :)

Jason3
03-15-2010, 07:57 PM
What's the conversion rate from US dollars to NZ dollars? 'Cause that looks like a pricey upgrade, looking at Jason's site. :)

It had better perform then!

Exchange rate is about NZ$1.00 = US$0.70 today. Always worth asking for a quote, I can't update the website every time our wobbly dollar falls over :)

Best regards,

Jason

stevespo
03-15-2010, 08:42 PM
Well, price is always a big consideration, but IMO the overall value for the setup is VERY high. Here's the thinking that went into the decision.


I wanted to feel absolutely sure that this system would meet/exceed my needs, and I have no doubts that it will. I'm certain of that already, just seeing it power that machine around effortlessly.
I didn't want to under size things again and have to start over. I just don't have the time or energy to keep iterating.
It's actually sized such that it can easily move up to a much larger machine and still have fantastic performance. That wouldn't be the case with most other options.


I carefully considered the Gecko 320/340 route with the 50 oz-in NEMA 23 servos, but the only way it was going to work well was to gear it down 2:1 or even 3:1. Sacrifice top speed for torque and I felt that the acceleration would also suffer. The individual components may be cheaper (perhaps 1/2 price) but then you add in a new PSU, the time, effort and money to make the reduction mechanisms, etc. The NEMA 34s are a possibility but just too big and bulky for my machine.

I also wanted to direct drive and I wanted a minimum of 100 oz/in all the way up to 3000 RPM. The only way to achieve that seemed to be BLDC or AC servos. These particular motors will do 150 oz/in continuous and peak at 3-4x that. 300W is a lot of power to throw at a benchtop router, but like I mentioned, my long term goal is something larger and even more capable.

So, the pricing was roughly what a "plug and play" Gecko system would run from many vendors (Logitrol, etc). It's less than 1/2 the price of what Techno charges for their servo system. So, at least for me, it was the right fit in terms of price/performance.

Plus, Jason is absolutely fantastic when it comes to sales/service as well as designing and sizing these systems. The machines he builds are fantastic and the components he uses are top notch. I just felt like I couldn't go wrong.

No knock on the stepper based systems. My machine was actually moving quite well and performing adequately. I certainly could have done a whole new stepper setup for less money, but I really wanted to close the loop and gain some experience with servos as well.

Steve

stevespo
03-20-2010, 01:15 AM
Here's a video of the new motors in action. Just doing some different moves at 450 IPM. It takes a true CNC junkie to find this exciting. :)

Steve

<object width="425" height="344"><param name="movie" value="http://www.youtube.com/v/F2KZlWUpD-I&hl=en_US&fs=1&rel=0"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/F2KZlWUpD-I&hl=en_US&fs=1&rel=0" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="425" height="344"></embed></object>

stevespo
03-20-2010, 01:24 AM
And a couple of photos of the new motors, electronics, motor mounts, coupling, etc.

Steve

ger21
03-20-2010, 10:45 AM
I sure wish I hadn't had to sell the 5 400 watt brushless servos that I had. :(


Just doing some different moves at 450 IPM. It takes a true CNC junkie to find this exciting.

Very nice. But now that we're cutting at 800ipm at my new job, 450 is now very boring. ;)

stevespo
03-20-2010, 03:13 PM
Yeah, I know what you mean. The 450 just seems so tame. It's easy to get spoiled, and then want more. :)

You start these projects and think "all I need is 60 IPM", then it's 120 IPM, 240 IPM, etc. I'm pretty certain this is the end of the line for this particular machine (speed wise).

I can declare it "finished". (ok, at least for the time being).

Steve

ger21
03-20-2010, 03:27 PM
When I started mine 6 years ago :rolleyes: I was hoping for 100, and get 150-180ipm. Not nearly enough. :) Next one should rapid at 600-800. :wee: But that probably won't be nearly enough. ;)

stevespo
03-20-2010, 03:38 PM
It's great to hear I'm not the only one afflicted with NFSS (need-for-speed-syndrome).
Hmnn... I guess if I swapped out my .2" pitch screws for .5", then I could...

Steve

stevespo
03-22-2010, 06:25 PM
Short video showing the smooth action of the linear rails.
THK on the Y, and IKO on the X, both 15mm.

<object width="480" height="385"><param name="movie" value="http://www.youtube.com/v/muTaVJxW7Ew&hl=en_US&fs=1&"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/muTaVJxW7Ew&hl=en_US&fs=1&" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="480" height="385"></embed></object>

Steve

diyengineer
03-23-2010, 06:12 PM
You have a beautiful machine!

diyengineer
03-23-2010, 06:15 PM
How do the direct drive servos work out for you?

http://www.cnczone.com/forums/showthread.php?p=753304#post753304

Please shoot me your 2cents since your machine looks soo good. I want to direct drive steppers/servos but cant really decide. your opinion would mean a lot!
Thanks!

Jason3
03-23-2010, 06:23 PM
Steve - what a tidy installation! Nice work on the electronics enclosure - looks great. That sure is smooth - you've clearly aligned everything just so :)

Best regards,

Jason

stevespo
03-24-2010, 04:19 PM
Gerry, you asked about screw whip a while back. I guess I just hadn't noticed it before, but I am seeing it now. Ever since I started running my machine faster, I had noticed a funny noise, but I was thinking it was related to my AC bearings.

After doing some tests, it's definitely a vibration/resonance with the screws at higher speeds. On the X screw (length 40") it is noticeable above 250 IPM, although it seems to increase/decrease as I fuss around with my end blocks and play with the alignment.

On the Y screw (length 32"), there is a similar noise that is pretty consistent across the speed range. It seems to increase above 300 IPM, which is in line with the screws critical speed. I don't see any obvious whipping, but the noise is present.

Part of the problem might be the way I have my end blocks mounted on the 8020 rails. They are not rigidly/directly attached to the 8020. Instead they are mounted to L brackets, on one side only. So, some of the noise might be vibration transferring through the blocks, bracket and into the 8020.

I'll have to play around to learn more. Perhaps make up a new mounting bracket and make sure I have my alignment nailed. The machine seems to run just fine at faster speeds, but it makes more noise. At 250 IPM and below, it's much quieter.

Steve

Dylwad
03-25-2010, 02:07 AM
Angular contact bearings and some tension on the leadscrews will raise their critical speed, We ran a 6 foot 1/2-8 2 start at 400 IPM without whip after putting some tension on it.

If you have inch ballscrews angular contact bearings require a bushing of some sort, I have only been able to find metric bearings. I use 15x35x11mm bearings http://www.vxb.com/page/bearings/PROD/ANGULAR_CONTACT/Kit1086 on 1/2 inch with custom bushings. They have them shielded and sealed as well, but are a bit wider. They take the axial and thrust loads great.

Think of it as a guitar string, the more tension, the higher resonant frequency. You dont want to go overboard and introduce friction, but its easy to adjust to match your required rapid speeds.

Great machine BTW, you sure packed alot of power into a small machine!

Dylan

stevespo
03-25-2010, 10:01 AM
Dylan, I'm sure your right about that. At this point I have no way to tension my ball screws because I'm using a fixed/simple bearing arrangement. The fixed end is (2) preloaded 7200B AC bearings (30x10x9mm) and the simple end is just a standard ball bearing. Short of re-machining the ends and using different bearing blocks, any ideas for me to try?

Steve

Dylwad
03-25-2010, 01:31 PM
I would have to see pictures of the ends of the ballscrews. Keeping the fixed ends like they are and doing an AC on the simple end and drilling and tapping the end of the ballscrews comes to mind, using a short SHCS and washer to tighten up and tension the rod. Dont know how easy it is to drill and tap your ballscrews, Might need to be annealed if the ends are more than surface hardened.

Its tricky with your bearing block setup as far as I can tell, tension might bend the angle brackets used to bolt them to the 8020, might have to make the bearing blocks wider for through bolting to the 8020, or support both sides of them with angle brackets, or cap them from the sides.

stevespo
03-25-2010, 03:15 PM
Dylan, I understand what you're describing. It's going to be difficult/impossible with my current end block. I'll monkey around a bit more with alignment and then just run it below the critical speed. That's certainly decent performance.

Perhaps on a future build, I'll use a fixed AC bearing on both ends, as that clearly provides the best support and tensioning options.

Steve

stevespo
04-09-2010, 11:55 PM
A little more monkeying around. I added (2) bolts to the bracket where the X axis ball nut attaches. This does seem to stabilize the screw a bit better and there is less resonance at higher speeds.

I also rearranged my Z axis. I originally attached it so that the table was fixed to the gantry and the entire actuator moved up and down (25lbs + spindle + motor). I thought this would be the stiffest possible arrangement. The by-product of this design was that my 7+" of travel was turned into 4.5" because of the position where the router was mounted. Most of the range of motion was on the downward side, which is not really where I need it.

So, I flipped it around and modified the mounting plates. Pretty easy. Now I have a good 6" of clearance over the table, and the full range of motion on the Z. I can bring a tool way up and over a workpiece without any interference. Another advantage is now the motor is just moving the spindle up and down and not the entire actuator.

Did I lose any stiffness? Hard to say just yet because I haven't done any heavy duty carving since the mod. I could definitely imagine a heavier/stiffer mounting plate which could help (if necessary).

Where it's positioned in the picture is for the smallest/shortest bit I'd ever use, so the spindle would be under very little stress. With larger bits, I'd slide the spindle up and that would help with the "lever" effect on the mounting bracket. I think it will be ok, and I really need the extra travel.

Any comments or thoughts on this mod?

Steve

Jason3
04-10-2010, 05:53 AM
Nice change. I think it's better around this way - it really doesn't look like you're compromising very much, if at all, for the extra travel. It would be interesting to see whether you can detect any difference at all in the cut quality.

Less mass moving on the Z can't be bad :)

Best regards,

Jason

Dylwad
04-10-2010, 06:06 PM
I think it looks great, If anything you could add a gusset between the top of the spindle mount and the Z plate on both sides to tie things together, should be easy to do. Spreading the spindle mount plates apart to the extremes could only help too. Looks easy to slide the spindle up on the mount to stiffen things up for heavy cuts.

mhasting2004
04-15-2010, 11:52 AM
Nice build Steve

I'm at that stepper/servo crossroad myselfand am happy to see it working so well for you. My build is a bit bigger ...http://www.cnczone.com/forums/showthread.php?t=53824&page=7 (http://www.cnczone.com/forums/showthread.php?t=53824&page=7) ... and made with junk thrown from work so no design has gone into the choice of drives or motors its just what I happen to have. This means I've saved alot in build costs and I feel justified to spend wisely on servos to enhance the machines capabilities.

One question I have is are you using a smooth stepper? I see no mention in the thread and wonder if it wouldn't help with your pc clock issues?

I have no vested interest but I have one and am driving my machine from a laptop at the moment.

Cheers

Mark

stevespo
04-15-2010, 11:08 PM
Mark,

Great looking machine you've got there. The 300W Zeal servos should be a great fit for your needs. Not quite as large as a NEMA 34, but bigger than a NEMA 23. I'm sure you could adapt the 70mm motors to your existing mounts and they should have loads of power for your application.

I'm still using the parallel port, so no smooth stepper. My PC issues seem to have gone away with my most recent machine, a Dell Dimension 4400 running at 1.7GHz. I have my steps/inch turned down to 2000 (previously 10,000 with the steppers), so the load on the machine is fairly low. Even a 25K Mach kernel is fine. No issues with the pulse train, even at speeds of 600 IPM (3000 RPM).

No doubt people are doing great things with steppers, but if you want a closed loop system, servos are the way to go. The Granite drives provide a lot of flexibility with tuning and setup, but I really haven't done much other than use the default settings for the motors and verify that the PID settings are reasonable. It was easy to setup and configure and I'm thrilled with the performance.

Steve

Dylwad
04-16-2010, 01:45 AM
Steve,

I'm using the same Dimension 4400, I Highly recommend a CPU upgrade, the 1.7's were better than the 1.6's, more from the 1.6 having a maze for a cache, but for $10-$15 for a 2.4Ghz + CPU its worth the upgrade, and its as simple as plugging it in.

http://www.geeks.com/details.asp?invtid=P42800D478-533&cat=CPU will de-rate to 2.08Ghz after installation due to the 400Mhz front side bus on the 4400 motherboard.

My 4400 tripled its speed and then some, I had the CPU laying around from another build.

Dont know if you know this, but some 4400's had problems with service pack 2, the fix is a newer hard drive believe it or not. I used a 120GB WD out of a "dead" DVR. SP3 runs great on it now, wouldnt load windows at all with the stock 40GB HD with SP2.

Dylan

stevespo
04-16-2010, 02:06 PM
Dylan,

I can never remember if it's 1.6 or 1.7, so I double checked - and it's a 1.6Ghz Dimension 4400. So, the CPU has never really been a "problem" for me, but it wouldn't hurt to do the upgrade either.

I actually did replace the hard drive when I did my clean XP install. Pretty certain it auto-updated itself to SP3, but I'll check when I'm back in the shop. Thanks for the tips. This Dell has been super reliable for me, so why not give it a little boost.

Steve

stevespo
04-27-2010, 10:42 AM
Installed the 2.8GHz P4 (actually clocking at 2.10 Ghz) and the benchmarks are telling me it's running 25-40% faster for compute intensive applications. Pretty good for $15 + s/h.

Steve

RicknBeachcrest
12-23-2011, 12:26 PM
Hi Steve,

It's been a while since this thread has had any postings. I'm in the process of building an 8020 CNC router. I have read your build thread several times and each time see something I hadn't seen before. Your thread is one of two that I am using for guidance in my build. I'm getting real close to making some decisions on the drive screw for my machine. I am planning on using ball screws for my X and Y axis.

I have a few of questions, if you don't mind.

1) You used 30mm 7200B angular contact bearings on your ball screws. From my understanding these are not sealed. How have these worked out for you? The sealed bearings are a little over 3 mm wider (5200-2RS Double-row Angular Contact Ball Bearing).
5200-2RS Bearing Angular Contact Sealed 10x30x14.3 Ball Bearings (http://www.vxb.com/page/bearings/PROD/Kit7279)

2) What was the thickness of the aluminum that you used for each half of the bearing blocks?

3) My biggest problem (that I can see now) is finding a source for machining the ends of the ball screws. Did your have to adjust the length of yours?

4) I hope this isn't too dumb of a question. My understanding is that there are two AC bearings in each bearing block and the nut on the end of the ball screw squeezes these to AC bearing together. Is this right?

Thanks for your help.
Rick

stevespo
12-23-2011, 12:57 PM
Hi Rick, this thread has been quiet, but not the machine! I'm still using it regularly to make parts for myself and some long term customers.


1) You used 30mm 7200B angular contact bearings on your ball screws. From my understanding these are not sealed. How have these worked out for you? The sealed bearings are a little over 3 mm wider (5200-2RS Double-row Angular Contact Ball Bearing).
5200-2RS Bearing Angular Contact Sealed 10x30x14.3 Ball Bearings (http://www.vxb.com/page/bearings/PROD/Kit7279)


The 7200B bearings have worked out very well. They are not sealed, but in my case, one set is under the table and the other is in back of the upper gantry and well protected from dust. It would not be hard to add some type of dust cover to the end blocks, or you could certainly go with sealed bearings and play it safe.

It looks like the 5200B are a double row AC bearing and (probably?) don't require a matched set with a manual preload. If that is the case, then the 5200Bs have two advantages over the 7200B and seem like a great choice.


2) What was the thickness of the aluminum that you used for each half of the bearing blocks?


It was .500" 6061 aluminum and I milled the parts with the CNC.


3) My biggest problem (that I can see now) is finding a source for machining the ends of the ball screws. Did your have to adjust the length of yours?


I had a local machine shop shorten (and turn the end of) one of my screws. The screws that I had were hardened steel and required some finesse to turn. I don't think it was anything exotic or difficult, but the machinist was a little surprised at how tough they were (which is a good thing!)


4) I hope this isn't too dumb of a question. My understanding is that there are two AC bearings in each bearing block and the nut on the end of the ball screw squeezes these to AC bearing together. Is this right?


Not dumb at all. With the 7200B, they are used as a pair within the bearing blocks. I'm trying to remember how they went together. I believe the inner raceways are clamped tightly together on the screw shaft (with the lock nut) and the preload was managed with some metal shim stock to allow them to turn freely (as a unit), but resist the axial forces that the screw might encounter. I may have some photos or a description somewhere to refresh my memory.

In theory, that wouldn't be necessary with something like the 5200B - but then you also can't control the preload because it is going to be a function of the dual row unit. I might actually have some of those around for another project, so I will look at my supplies and see if I can tell you anything more about that tonight or tomorrow.

Just found your build thread - looking forward to checking it out.

Steve

RicknBeachcrest
12-23-2011, 10:52 PM
Steve,
Thanks for the quick reply. As far as my CNC build, I truly am standing on the shoulders of those who have come before me. I have absolutely no machining experience, but do a lot of wood working. This is my driving motivation, also the personal satisfaction of building such a machine. I am very grateful to all on this forum who have taken the time to describe (in copious detail) their builds.

My present task has been building a bench for the CNC router. I need to get the machine together so as to determine how exactly things will fit together. Currently I just have an assortment of sub-assemblies sitting on the floor. It will be exciting to see it in a more 3 dimensional form.

Rick

stevespo
12-24-2011, 05:42 PM
Hi Rick, just pulled out my "box of extra CNC stuff" and I do have some extra 7200Bs and a pair of 5200Bs as well. I'm pretty certain I read about them after I used my 7200s and I thought it might be a reasonable replacement as the OD is the same (30mm). I remember there was some debate as to whether they were "true" AC bearings or not. I can't say. They do appear to be quite solid and I don't detect any side play while handling them.

The 7200s have a lot of play (individually) and the key is using them in a matched pair. If you put the logo sides together and clamp the inner raceway, the slop disappears and you can tune the preload with shims. Precision ground AC bearings are all ready to go, but the low cost version requires some trial and error. With the bearings clamped on the screw, you can measure the axial play and adjust the load.

One set felt just fine "as-is" and the other one used a very small shim to get the same type of feel/resistance. I'm definitely NOT an expert, but relied on several other threads here that went through the process.

Steve

stevespo
02-08-2012, 10:01 AM
Here are the files from post #34 converted to IGS format. Just the mounting plates and other misc small parts.

Steve

RicknBeachcrest
02-24-2012, 12:18 PM
Keeping busy.

The rails are attached with long strips of steel, that were pre-drilled and tapped for 4mm socket head screws on 60mm centers. Very handy. I knew I wouldn't be able to accurately drill/tap those holes, so this helped me out tremendously. Thank you 8020 Surplus. All my 8020 supplies were purchased through the outlet and very easy dealing with them.

Steve

Hi Steve,
I think I need to use some of these long strips of steel you used in your buld. I have been using 4mm T-nuts and it has been giving me some problems. The T-nuts have a corrugated seat, so when they are mounted they bit into the extrusion. Now this would be good (I guess), if I didn't need to fine tune the placement of the rails.

So any idea where I could find some of these long strips of steel you used?

Thanks.
Rick

PS. I checked 8020 surplus, but didn't find any there. Maybe I didn't look in the right place.

stevespo
02-24-2012, 12:36 PM
It's in the PDF catalog. I think the part number is 8900-36 for the 36" lengths of undrilled/untapped 15 series nut stock. I sometimes see it on eBay in the 8020 surplus store, which is where I found mine. Just search for "8020 8900" and you should find it there. They will drill/tap it for an additional charge.

Steve

RicknBeachcrest
02-24-2012, 02:04 PM
Thanks Steve,
I emailed them to see what they would charge to drill and tap the holes for four 25" bars.
Rick

RicknBeachcrest
02-29-2012, 11:31 PM
Hi Steve,

I am in the process of deciding on ball screws. Are you satisfied with the lead(.2") on your screws?

I was thinking that lead is a good compromise between speed and accuracy. But then again, I know next to nothing about it. My machine will only have a cutting space of about 18" square, so there really isn't a whole bunch of real estate to cover.

Also what kind of couplers are you using?

Thanks.
Rick

stevespo
03-01-2012, 12:33 AM
Rick, the .2"/turn lead has been good, but I (personally) would go with something coarser next time around. For my machine, with a 25x32" work envelope, a .5"/turn lead (or probably 10mm, 12mm, 15mm, metric) would be a very good fit.

You're correct that is is a bit of a trade off in terms of speed/accuracy. With steppers you may lose some positional accuracy because you're dividing each screw revolution into 200 steps. Fewer revolutions per inch means less granularity with motion.

My servos have 1000 CPR (quad) encoders, which I am only using in 200 step/turn mode, so I could definitely use a coarser screw and adjust the step count to keep the same accuracy I'm used to. The accuracy is a function of the encoder and not the motor.

The downside of my current screws is top speed. Theoretically, I could drive them at 3000 RPM (600 IPM), but in reality I start to see resonance (whipping) at more like 300 IPM. Do I need to move that fast? No, not really - but it's kind of sad not to be taking full advantage of the great motors and electronics I have!

For a smaller machine, with 18x18" of travel, screw whip is not going to be an issue even at very high speeds. You can calculate the max speed with a few variables.

So, for your machine - I would probably go with the .2"/turn screws, which are a good balance. If you're using brushed DC servos, it's a good choice because you get increased mechanical advantage with the finer pitch and you can run the servos at very high RPM.

If you are using steppers, the max speed is dependent on the motors, electronics and screws. You may hit your max RPM with the steppers and .2"/lead, but it's (probably) going to be over 250 IPM. Pretty fast. If you really need more speed than that and want to keep the RPMs down into the higher torque range of a stepper, then a coarser screw would also work well, but you trade off some resolution.

Oh yes - I was using the aluminum helical couplers and then switched to the Lovejoy style with the urethane spiders when I went to the DC servos. I couldn't go from the 16mm motor shaft to the .25" screw end with anything else.

Steve

RicknBeachcrest
03-01-2012, 04:12 PM
Thanks again for your great insight.

One of the first purchases with this project was the steppers and controller. I am going to use the g540 with 280 oz steppers sold by Gecko.
G723-280-4 Stepper Motor (http://www.geckodrive.com/g723-280-p-45.html)

Now my next learning exercise is matching up AC bearings to the fixed end of the ball screw. I thought that the end machining was pretty standardized. From my looking at the various sites it seems to vary from manufacturer to manufacturer.

stevespo
03-01-2012, 04:27 PM
That is a versatile motor/driver package. You could go with fine or coarse screws and still have a nice balance. If it's important to have 1000 steps/inch resolution, then the .2" (or 5mm) is a great choice. If you wanted a bit more speed and lower RPM, then a .5" (or 10mm) pitch would be great. That combo has a nice torque curve and you shouldn't run out of headroom regardless.

I would not fabricate my own end blocks and AC bearings the next time around. I'd order a matched set of blocks, screws and anti-backlash ball nuts and focus on building and using the machine. Of course, there is some entertainment and educational value in doing that stuff yourself. The cost savings (IMO) is minimal when you consider the time and effort, sourcing parts, etc.

Steve

RicknBeachcrest
03-02-2012, 05:54 PM
Thanks for the info on the steppers. I looked, but could not find a torque curve for this stepper. I kind of planned from way back to go with a .2" (5mm) lead. (I was originally looking at Roton threads which were just .2 lead for .5/8") But it makes sense to consider a 10mm lead. With a 5mm lead that would give the resolution to about .001" per step. Looking at it this way seems to be a little over kill on resolution.

I'll be using this machine 99% of time for wood working projects and .001" resolution per step seems a little much. So if I decrease the resolution by 1/2 (10mm lead) that would still give me a resolution of .002" per step, which I think is more than enough for my needs.

Am I figuring this right?

End block fabrication:
I was going to ask your opinion on this. I think I will plan on using the matched set of blocks, unless this just won't fit. I built the router mount which was a good learning experience, but I really was wondering about the wisdom of this decision, since I could have purchased one from K2CNC for about $68.
Thanks for all your help.
Rick

stevespo
03-02-2012, 06:43 PM
I haven't seen a torque curve for that particular motor. Gecko might have one if you called them up. My impression (from looking at Keling specs, etc) was that the 270 oz-in and 380 oz-in motors had more available torque at higher RPM than my old (high impedance, triple-stack) 425 oz-in. Those *definitely* would have benefited from the .4"/10mm pitch screws, even running at 72VDC.

It's all a function of driver/motor/PSU and it's sometimes hard to know until you have it wired up and working. That is when theory becomes reality.

The limiting factor (in terms of accuracy) has probably less to do with the screw pitch (step count, etc) and more with the limitations of rolled ball screws (probably +/- .002" per foot), the structure of the machine, type of linear rails, runout on the spindle (router, lam trimmer), etc. You are right that .001" or .002" is plenty for woodworking. We're not doing aerospace work here.

I think the .4"/10mm pitch will keep the RPMs down, keep the pulse train down, give you very smooth, fast motion and probably keep the motors working in the strongest part of their torque curve. You really can't go wrong either way. It's more a matter of which way you want to hedge.

The end block fabrication not a big deal, but I used my CNC to do the work. So, you'd have to cobble something together (like I did) out of wood, plastic, etc and get the machine working. Then make the new blocks and swap everything out, reinstall, realign. That is the PITA. Optionally, you could have someone else machine them, but then you're probably better off just buying a matching set and having everything work.

The standard blocks don't often mount cleanly on the 8020, so some small adapter plates in aluminum might be needed. But that's just cutting /drilling and the accuracy doesn't have to be real high. Just my opinion, as always.

I'm happy to take this discussion over to your thread, I've been subscribed for a while.

Steve

gera229
07-23-2012, 10:35 PM
I haven't seen a torque curve for that particular motor. Gecko might have one if you called them up. My impression (from looking at Keling specs, etc) was that the 270 oz-in and 380 oz-in motors had more available torque at higher RPM than my old (high impedance, triple-stack) 425 oz-in. Those *definitely* would have benefited from the .4"/10mm pitch screws, even running at 72VDC.


Did you supply the 425oz-in with enough power?
It also could have been that the inductance of it was higher than the other stepper motors.
I read that higher inductance means it loses torque quicker throughout the RPM range.
Lower inductance would be better.

From post #120:


You're correct that is is a bit of a trade off in terms of speed/accuracy. With steppers you may lose some positional accuracy because you're dividing each screw revolution into 200 steps. Fewer revolutions per inch means less granularity with motion.

My servos have 1000 CPR (quad) encoders, which I am only using in 200 step/turn mode, so I could definitely use a coarser screw and adjust the step count to keep the same accuracy I'm used to. The accuracy is a function of the encoder and not the motor.


By positional accuracy are you referring to resolution?
Or positional recognition?

As for the different size screws and keeping the accuracy that you're used to if upgrading to a larger pitch screw:
Have you considered micro-stepping with stepper motors?

Encoders for servos may have the ability to produce more steps per turn than micro-stepping for steppers, but you aren't using it that high anyway. And while the resolution will be high at a high amount of steps/turn, the speed would certainly lack. Who doesn't want speed? I most definitely do :D.

I am still new to CNC, but just pointing some things out.

Also in this thread I've heard you talking about how servos are closed-loop.
What exactly does closed-loop mean?

stevespo
07-24-2012, 12:02 PM
Did you supply the 425oz-in with enough power?
It also could have been that the inductance of it was higher than the other stepper motors.
I read that higher inductance means it loses torque quicker throughout the RPM range.
Lower inductance would be better.


Yes to all of the above. Wired bipolar parallel, the inductance is 6.8mH, so the ideal voltage would be more like 32*sqrt(6.8) = 83V. I was using an unregulated 72VDC supply at 10A, so adequate voltage and current were not the issue. Sure, I could have gone a little higher on the voltage, but I was pretty close to the upper limit with 72VDC (they already ran quite hot).



By positional accuracy are you referring to resolution?
Or positional recognition?

As for the different size screws and keeping the accuracy that you're used to if upgrading to a larger pitch screw:
Have you considered micro-stepping with stepper motors?

Encoders for servos may have the ability to produce more steps per turn than micro-stepping for steppers, but you aren't using it that high anyway. And while the resolution will be high at a high amount of steps/turn, the speed would certainly lack. Who doesn't want speed? I most definitely do :D.

I am still new to CNC, but just pointing some things out.

Also in this thread I've heard you talking about how servos are closed-loop.
What exactly does closed-loop mean?

I'm talking about resolution. Microstepping is helpful for smooth motion and fighting resonance, but it won't necessarily offer greater precision. There are MANY CNCzone discussions on the topic and I don't have the details committed to memory. Assuming greater than 200 step/rev accuracy for a stepper motor is problematic and I would leave the explanation to the experts.

I used 10:1 microstepping for years and it's a very good arrangement, provided you have matched your accuracy needs to the motors and screws. My issue with these particular steppers was more to do with max RPM and top speed than accuracy or resolution. They are fine motors, but would have benefited from a lower TPI screw to provide more speed (at the cost of resolution).

Closed-loop means that the encoder's positional output is fed back into the controller to create a feedback loop. If an external force causes the motor to lose it's position, the controller will sense this and adjust based on what the encoder is telling it. A PID algorithm takes the inputs and computes the movement require to maintain position or move to a new location.

Consider my 300W BLDC servo motors which provide consistent torque across the 0-3000 RPM range, with peak current of 20A. Another league of motor/controller entirely. The configurable quad encoders can provide whatever resolution I want from 1-4000 CPR.

My limitation now is ball screw resonance and the ability of Mach to generate a fast enough pulse train. That's one reason for running them at 200 steps/rev like the steppers.

It's a VERY nice setup and has been absolutely accurate and problem free. Given the amount of time I spent troubleshooting and upgrading my stepper setup, I wish I had gone to servos much sooner, but it's all part of learning.

gera229
07-24-2012, 04:00 PM
I want to try servos now, but my budget is just too low for servos and my setup would be mainly for woodworking so servos might be overkill.

Still, the positional feedback you describe there just adds on to the accuracy and the resolution of the quad encoders. Plus eliminates the worrying of lost steps and guarantees accuracy.

I supposed I will learn along the way too. For me right now about 100 ipm is fine and that's faster than an old MacroMill CNC I used with 16 ipm max which is far too slow and rather limited on the size of material it can cut. *Might want more speed later*
I also used a plasma cutter with large servos on the X and Y axis (I think they were servos because they were cylinder shaped and quiet) and a smaller stepper (Might have been a small servo, but it was square shaped like a stepper) on the Z axis and there was a setting in its CAM program of up to 1000 ipm. Now that was amazing.

Servos are much quiet than steppers, and that's another great thing about them. They don't produce that high pitched noise that steppers do. Is that correct for your set-up?

stevespo
07-25-2012, 12:10 PM
I want to try servos now, but my budget is just too low for servos and my setup would be mainly for woodworking so servos might be overkill.

Start with what meets your needs and fits your budget. DIY machines can be tweaked and upgraded because you'll know every nut, bolt, screw, wire, etc. Focus on a very solid frame and accurate mechanical parts and the motors and electronics become a "bolt on" upgrade later.

Still, the positional feedback you describe there just adds on to the accuracy and the resolution of the quad encoders. Plus eliminates the worrying of lost steps and guarantees accuracy.

Yes, it's nice - but more costly. Most brushed DC servos are geared down with a belt/pulley system to increase torque and keep the RPMs high. Extra complexity and cost there too. My BLDC (brushless DC) motors and Granite drivers would be considered on the high-end of the DIY market. There are lots of stepper combinations that will work very well.

I supposed I will learn along the way too. For me right now about 100 ipm is fine and that's faster than an old MacroMill CNC I used with 16 ipm max which is far too slow and rather limited on the size of material it can cut. *Might want more speed later*

That's what they all say!

I also used a plasma cutter with large servos on the X and Y axis (I think they were servos because they were cylinder shaped and quiet) and a smaller stepper (Might have been a small servo, but it was square shaped like a stepper) on the Z axis and there was a setting in its CAM program of up to 1000 ipm. Now that was amazing.

Servos are much quiet than steppers, and that's another great thing about them. They don't produce that high pitched noise that steppers do. Is that correct for your set-up?

Yes, the servos are very quiet - essentially silent. I never minded the stepper noise and it's not noticeable over the sound of a router/spindle and a dust collector in the shop. You can get round steppers and square servos. Many of the BLDC and AC servos are in a square package that look like steppers. The encoder on the end is the visual clue, but sometimes it's integrated and not obvious.

Have fun!