Hello All,

Another long time lurker turned builder. I've read just about every post over the last couple of months (and some many more times than once) and have found the Zone to be an absolutely fantastic resource and wonderful community. Thanks to all the contributors.

Now on to the deep end of the pool into which I have jumped. Against all good advice I built big. Hopefully the attached pic works. My plan was to 1) CAD up my design 2) build the mechanicals 3) go for help. So now I'm at 3.

A bit of background to the project for everybody:

Construction: 7Ply plywood for moving gantry parts. Cold rolled steel tubing for all bearing surfaces. All bearing blocks are skate bearings in a 120' configuration... so 6 bearings per block. Dimensions X: 7ft, Y5ft, Z18"... workable area is 4ftx5.5ftx15". Table base is made of mdf with I joist stringers. Spindle is a 2-1/4hp makita router.

Purpose: above and beyond the fun of tools - I'm hoping to have a machine that is capable of producing 3D wooden scultpure pieces... hence the large Zaxis (Also a faroff dream to use it to produce a classic wooden boat)

Results to Date: well, I'm reasonably happy with the mechanicals, all the axii move very nicely with very little finger pressure. Its a lot bigger than I envisioned... but this wouldn't be the first project that got out of hand. I've put in 1/2" 13tpi plain old threaded rod on the Z and Y axis... this leads to some questions below. The rigidity seems very good given the materials, even with the Z fully extended it takes a good shoulder to deflect the spindle. (Sorry for the scientific measure)

Design goals: This prototype will hopefully work and perform to reasonable accuracy (don't really need absolute accuracy so long as relative accuracy is there for sculpting) and at a reasonable speed. I hope to be able to cut finishing passes at 100"/min. Eventually I'd like to evolve to proper linear slides, aluminum framing, etc etc. At that point I'd like to be in the 200-300 "/min category.

Questions Questions

Q1: Is this monster going to move... and what will it take. My uniformed guess/design notion was something like the camtronics 600 oz/in package or what DaSigntist is using on his machine. Is this pie in the sky given my 100"/min wish, and 200-300 "/min dream?

Q2: What are my options for drive mechanism?

The threaded rod on the Y and Z works reasonabley well - I can turn the Y to 500 rpm no problem... at 1000 rpm (using my drill to power the rod) the whip is nasty. I'm experimenting with tensioning the rod between the end supports although I think my rod has some bow in it from the store. So at 500 rpm I'm around 40"/min which is going to kill me on the finishing passes. I'll post another msg to this thread with a few thoughts I had plus picture. Basically I'd like a reasonably quick system that is fairly cheap... I'm going to test it on the X axis. I'm considering a timing belt set up, or a chain drive setup (cheapest), or a wire rope drive setup. I can always fall back to cheap threaded rod and low "/min performance. As a target, I'd like to get the X drive set up for 150 bucks.


So that's where I'm at to date. Thanks all for the inspiration. All comments/criticism/suggestions most welcomed. Oh yeah, I'll be completely disassembling what you see in the pic (for painting/alignment) so feel free to offer suggestions that take me back to square one.

Thanks all.

AJ

Q1: Is this monster going to move... and what will it take. My uniformed guess/design notion was something like the camtronics 600 oz/in package or what DaSigntist is using on his machine. Is this pie in the sky given my 100"/min wish, and 200-300 "/min dream?

You could get by will much smaller servo motors. I run my all aluminum router on 430 oz/in motors and I know it would do just as well with the "Cheap Servo" setup using the 360 oz/in ones. I can get 150 IPM or more with these motors and this is using a 2:1 ratio belt drive setup.

Q2: What are my options for drive mechanism?

To help with the speed get the .200 pitch ball screw stock from reid tool or McMaster Carr.

As a follow up to the above intro - for the X axis I'm considering these options


1) Cheap threaded rod, low rpms, slow speed :(
2) Either, timing belt, chain drive, wire rope set up on each side of the X, driven by a common shaft which is suitably reduced to the motor.

re #2. to date it seems that the timing belt setup (with cogs) is the most expensive. I've looked at McMaster carr and don't see myself getting anywhere close to my 150$ target. Any other resources/materials to consider? That's what brought me to the chain drive... a lightweight chain at $1 per foot is in the budget and all sorts of cogs are available/cheap. My only reservation is what about backlash? The wire rope was another idea... if at the drive end it took 2 or 3 raps around the drive pulley I think no slippage would occur. this would be by far the cheapest. All ideas at this point.... please comment on hypotheses/experience with any/all. The picture below shows roughly what I'm thinking (I hope!). In any setup, I'd have the idler pully adjustable to tension the belt/wire/chain.

If any of these work I may retrofit to the Y axis too.


Thanks
AJ

PS: if the underlying theme of cheap hasn't come through yet... well yes, I'm another dreamer that wants to build performance without paying for it! :D

You might want to look at multiple start acme rod. I got 1/2-8 2 start (4 turns / inch) from MSC for about $22 for 6 ft. This will get you 100 ipm at 400 rpm. It will also work a lot better than plain threaded rod, and it's pretty cheap. For about twice the money, you can get 1/2-10 5 start (2 turns per inch), which will double your speed, but you'll probably need more powerful motors. If you really want to go over 100 ipm, you're probably going to have to start spending some money.

Gerry

yes. generally speed costs money. The chain will not be slop free. The cable idea has posibilities. If it's cheap you could try it on your long axis and let us know. The problem tho is you would still need belts to reduce the "spool" speed. If the spool is 3" in diameter, then it's ~10" per revolution, which means it will need to spin ~10 rpm for 100 imp. A 1" diameter spool will need to spin 30 rpm. This goes for your chain drive as well.

Multistart acme may be your best bet.

Eric

I am going to test out the wire rope... just bought a few pieces to put it together, then will share the results.


What do you think of this auction item on ebay

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=2579389161&category=25276&rd=1

The 2 servos with gear boxes are interesting although I am fairly in the dark on what's required as yet. The seller doesn't know the oz/in... does the wattage (170w)mean anything to anyone out there - what is that equal to in oz/in? He guessed they were 100 oz/in... but they look pretty big to be that light.... what do I know though - at this point, not much!

The thing that attracted me was the 25:1 reduction gear boxes... this would bring my rpm's down to a usable level with the wire rope or timing belt idea.

I am assuming I could then go the gecko route for the rest. If I win this auction, am I painting myself into a corner for a Gecko setup (re. max amps, voltage etc)?

Thanks

-AJ

cheapskate is my middle name
I'm in the process of building a machine and am going to use a garage door opener threaded drive, it uses 3 start threads, and am now attempting to refine the thread cutting op

Yes you will need geckos (or similar $ ) to drive those servos. They should work tho. Ask the ebay fellow about backlash in the gearboxes.

BTW I meant to mention before that I like your router. Very nice job.

Eric

Thanks all for the quick feedback.

re. the gearbox/servo setup I'm considering off ebay... the fellow said

>>>>>>
"I don’t have the oz/in but there is a general conservative rule of thumb:
Oz/in = (Watts * 1351) / RPM. This gives us 88.3 oz/in. This is a very conservative equation. I think the torque is well over 100 oz/in, as I can’t stop the motors with my hand."
>>>>>

I'm still dithering as to whether this will be strong enough to move the X and Y axis. Can anyone confirm this rule of thumb eqn.

The motors are Super L - Sanyo Denki L720-039E84, 170W, 32V, 7.9A, 2600 RPM with 250 Pulse Encoders. I'm hoping they are stronger than 100 oz/in.... does anyone else know how to convert Watts to oz/in? or know where I can dredge up a spec sheet for this motor?

Originally posted by AJ_Mac2001
The motors are Super L - Sanyo Denki L720-039E84, 170W, 32V, 7.9A, 2600 RPM with 250 Pulse Encoders. I'm hoping they are stronger than 100 oz/in.... does anyone else know how to convert Watts to oz/in? or know where I can dredge up a spec sheet for this motor?

Watts [W] is the unit for power, while oz*in is a unit of torque. They are related as Power [W] = angular velocity [rad/s] * Torque [Nm]. (1 Nm = about 141.69 oz*in.)

Now, if the 2600 RPM spec is the top speed at which the motor can still supply full torque, then the torque of this motor is 170 [W]/(2600 [RPM]/60*2*pi) [rad/s] = 0.624 Nm or about 88.4 oz*in.

My guess is that this is a continous rating, and that you might be able to drive the motor at up to 3 times this torque for shorter moments. (However you should really try to find a data sheet to confirm this since you might demagnetize the stator magnets if you supply too much current to the motor.)

DO NOT try to run the motor without some kind of current limit - if you were to just connect 32V at the motor terminals you would probably burn out the windings. If you want to spin the motors to see that they are ok, use a single D cell battery.

// Arvid

Thanks Arvidb for the science .... nice to start to put some comprehension to all this new info. I am quite electronically ignorant but hopefully improving with everybody's help.

Sounds like they are quite low power then? I have no relative feel for what 100 oz-in is... can someone equate that to an everyday appliance. Any guessestimate as to what a 14V cordless drill would rate at in oz-in? My drill spins the axii with no problem... but I suspect it is much more powerful that 100 oz-in.

Re. my wire rope trial. Seems to work adequately. some pics attached. i tried wrapping the wire around the drive pulley 2 times, but it tries to climb over itself and snags. Therefore, I had to stay with a half wrap. (if anybody knows of a source for small wire rope pulleys that have a helix carved into them like a winch reel, that would solve that problem). But with only a half wrap, and then taking up tension with the turnbuckle it is very rigid and slippage doesn't seem to be an issue as I couldn't manually make it slip. The only issue now is that at the required tension it takes a fairly strong set of fingers to turn the drive shaft.... back to my problem of not knowing what torque is enough torque.

I've attached the image of the ebay item I'm considering to drive the shaft. Any comments would be most helpful as to its suitability. Sanyo Denki Servo L720-039E84, 170W, 32V, 7.9A, 2600 RPM attached to a 25:1 gearbox.

here's the wire rope trial pics.

and one more.

remember that 100 oz in in the force it can hold at 1" from the center of the shaft. That's 6.5 lbs. It's a lot. If it's a 100 at rated speed that's huge, as a stepper is rated as holding torque, and goes down a ton at any kind of speed.

My motors are ~125 watts and run my machine at 150" per minute. At 100" per minute I need my whole shoulder on the gantry to make it stall.

Eric

Maybe I am not clear on something. Are you going to run lead screws with a very low TPI, so that the cable will only move a bit, or are you going to eliminate the turnbucle in the actual application? I think it will be difficult to join the ends of the cable without a bump or something that wont like wrapping around the pulleys. I like the idea, if the cable is wrapped around the pulleys it will give more grip too. Why not use timing belts?
co

The turnbuckle could be located at the bottom, with the axis fastened to the top cable. The turnbuckle wont need to go over the pulley.

Instead of pulleys, you will need spools, which will have a flat surface, and not a vee shape. the cable will wrap a few times around the flat spool no problem. You will need a gearbox.

Repeatability will probably become a problem over time.

If you're spending the bucks on a gecko system and servo's with gearboxes, you will want a more positive drive. The belt drive is the way to go.

The KISS rule states you should go with a leadscrew. Cheap and proven, perhaps slower unless you go with a .2 lead or better.

Eric

Just for your information, a Porter Cable 14.4V drill has a max torque of about 4960 oz-in of torque, according to their website (310 in-lbs).

Gerry

Before you bid on the ebay motors you might want to check this thread out.

http://www.cnczone.com/showthread.php?s=&threadid=2082&highlight=cheap+servos

I picked up a set of these, was under $40 for 3. Just have to get some encoders. I think there around $30ea.

-Wes

Yes... it seems there is no cheap way out. I may go with threaded rod to get going for cheap, and then upgrade as my confidence level rises.

re. the turnbuckle idea, the turnbuckle itself would be fastened to the bearing block so only the wire would be turning around the pulley. The pulley in turn would be driven on an axle that is attached to a suitable reduction gearbox. Inb this manner a common axle can drive both sides of the X axis.

Thanks Balsaman for clarifying the right terminolgoy for the pulleys that won't bind up with mutliple wraps. (aka spools)Any ideas out there for cheap off the shelf spools?? Nothing in the McMastercarr lineup.

Right now I'm thinking of getting things moving on the X axis, albeit at 50 ipm, with threaded rod - although I can't just yet let go of the wire rope idea... if I find some spools I may play with it a bit more. It seems the timing belt, set up in the same way as the wire rope, will be my goal as budget permits. I think this will be more economic than 2x 78" lead screws with the necessary tackle to sync them together.

Back to the hardware stores to peruse for inspiration..

-AJ

Originally posted by balsaman

Instead of pulleys, you will need spools, which will have a flat surface, and not a vee shape. the cable will wrap a few times around the flat spool no problem. You will need a gearbox.

Repeatability will probably become a problem over time.


Eric

Balsaman, can you expand a bit on why you see repeatability being a problem.

Not that I'm dearly attached to the wire rope, but I can't see around it at the moment. My guess at this point is that if I can find a spool then slippage can reduced to zero with the use of a couple of wraps on the drive spool. The other variances I see entering the system would be backlash between the motor and spool drive shaft, and possible stretch in the cable. The backlash I can live with, (and probably will be inversley proportional to the money I spend on the reduction system) and I think the stretch would be minimal and easily taken up with periodic retensioning of the turnbuckle.

??? not meaning to question anyone's advice... your experience is far deeper than mine... just inquisitive... can't help but ask why

ciao,
AJ

Rack and Pinion might work. With a single motor attached to the gantry turning a geared way the hell down shaft with a pinion gear on each end (bearing supported at the motor and pinion ends) turning against a rack on each side. The Rack and pinion would be a little more expensive than a leadscrew (Not ballscrew) but would be fast as hell even geared down and could be done with a single motor.

With that kind of gantry design you almost have to go with dual drive of some kind even if one side is just a slave.

I can enlighten you on the wire rope idea. Tried it on my table to try and eliminate having two motors. On my table the motor is driving a leadscrew on the left side. Leadscrew is coupled to gantry and rope is coupled to gantry(on left side). Rope transmits movement to the other side via Bishop Wisecarver Vee bearings where it is then coupled to the right side of ganrty. It will work if you don't care about super precise cuts. The end driven by the rope has a little "spring" in it. Of course your design is a little different than mine so it may work better. I was not satisfied because it slowed my rapids from 375 to about >200 ipm. The rope causes friction/resistance as it wraps around the bearings. I am looking for less resistance on my gantry.

I can see timing belts in my tables future if I can't solve the problem soon. Anyone need 1/2" 5 start leadscrew????

Inventit,

Thanks for the pic, nice to see it working in your application. I'd hope that having a "twin" setup driven by a common drive shaft my reduce the spring. My biggest reason for wanting to make this work is cost (naturally). Wire rope is 20 cents per foot whereas the cheapest timing belt I've seen is 4$/foot. Also wire rope is locally availalbe whereas I'd be ordering from south of the line for the timing belt/cogs which adds extra time/cost/complications. Given a 7 foot X axis, I need 14 feet per side to make a loop, times 2. That's starting to add up... I feel if I can make the wire rope work reasonably well and can actually make something selleable as a result of the machine, then I can reinvest and move up the line to timing belts/low pitch lead screws. And if it works for X, then I'm bound to want to retrofit Y. I like the sound of your 200 ipm.

You said it slowed from 375ipm rapids... was this with only the one side of the gantry lead screw.... I take it the gantry skewed itself with the one screw and that's why you added the wire setup? what kind of oz-in torque are you using and what size/weight gantry is that pushing?

thanks for the help!

AJ

I think the slippage will be minimal but cumulative. A cnc needs a positive drive system so a thousand minute errors don't add up to a big error.

Eric

Thinking about the wire rope idea, there is considerable problem created by the rope wrapping and moving laterally across the face of the spool.

The only method I can think of that would solve such a problem is to use two spools, one 4 groove driver and one three groove idler, for example.

Both spools have circular grooves to maximize traction. The axis of the idler spool is tilted relative to the driver, so that the rope coming off of groove 1 of the driver, wraps around idler groove1 by half a wrap and feeds back onto the driver on driver groove2. The rope wraps 1/2 turn around the driver again and comes onto the idler groove 2, etc.

The rope from the gantry enters the driver and leaves the driver on the same side, but on opposite ends of the driver spool, after it has made the multiple wraps around the two spools.

Both ends of the cable are anchored to the gantry, with suitable tensioner, and the cable passes around one single groove idler on the far end of the rail. This would serve to keep the tension on the cable the same in both directions (because it is sort of an endless loop), and could be run nice and tight because the cable would essentially be pulling against itself, with the gantry being dragged along for the ride.

I am sure you could get an automotive timing belt, one of the long ones that goes around everything in the motor.(I think GM used these) and some scrap pulleys from a junk yard. Put a wheel for tension and away you will go. You can probably get the pulleys for next to nothing, and the belt might cost 30 bucks new, or get one from the scrap yard too.

Just a thought. Why not use two belts, one going to one screw and the other to the second screw. This would lead to more complexity since each belt would need to be tensioned separately and the two belts would not be in the same plane so the lengths/positions of the lead screws would not be identical. However you do get both screws moving together and all the other good things that come from a single-motor design.

Just my 2 cents,
bb99

Ajmack,

I really like this guys whole machine, really clean design. In particular his drive system for dual leadscrews seems really good.
This Thread by rmtucker: (http://www.cnczone.com/showthread.php?s=&threadid=1355)
He shows a good picture of it in in message #16.

Chris

cbcnc - I agree!.. I certainly aspire to RMTucker's level of workmanship, it's very very professional looking. I hope that I am beginning one of a succession of machines.... and that the next generation can be more robust with aluminum parts, high quality slides, etc etc.


bb99 - I think I will try and design from a single motor per axis. I suscribe to your line about all good things coming from a single motor :)

HuFlungDung - I'm not yet finished with the wire rope. I will flog this horse one more time. I've now got all the pieces together to try a real size trial. I am going to try and prevent the wire from tracking across/off the spool, with the use of some bearings as guides onto the spool. Just can't beat the cost. I think for about 70$ I can have the X axis drive sorted out (back to the reduction gearbox). Course if it doesn't work, that's 70$ into the experience account.

But now I must go Christmas tree hunting - So long and best of the season to everybody.

Hey AJ, really nice design.

Any of the suggestions above are pretty feasable and potentially accurate, it is mostly decision/value time on your hardware. One of my favorite products that I produce is a sign that reads " quality, price or speed,,,, pick any two". This transfers into cnc machining better than I wish it did.

If you are going to be doing sculpture ( I am assuming wood here ) don't anticipate running at very high feed rates on your finish pass anyway. But do pay attention to your z axis. The constant reversal up and down will act like a relentless jackhammer. There is also a common tendacy to install conservative drive specs on the z-axis because most routers are intended for 2-1/2 d work. The hypothetical feed rate on your x/y moves will be all but stopped by the z axis if you choke it down in any manner. Due to the strains put on it, I would highly suggest a ballscrew for that axis at least.
Depending on your work and the size of your sculpture, be prepared for several times you anticipated machining time on those final cuts.

Best of luck,
Phill Pittman

www.masterwerkes.com
digicarve@verizon.net

Thanks Carver,

You assumed right on wood being the subject of my sculpting. Your points as to speed / axis drive design are well taken.

What kind of speed can you do a finish pass at if you are basically doing a 2-1/2d toolpath? I would like to think I could power this homebrew effort to do roughout cutting at 50-100ipm (1/2 end mill - with 1/4" stepover, 1/2" stepdown) and then do my final finishing at 150+ipm... is this another dream? In terms of the finish pass, the bulk of my work would be completed with spiral type tool path (ie very gradual changes in the Z as the x & y are brought to finished dimensions)... is this what people refer to as 2-1/2 d? As opposed to an xy raster with radical Z axis changes?

What power spindle do you use? I'm hoping my 2-1/2HP makita will gobble up softwoods like hemlock/fir/pine/etc... if I get into hardwoods maybe that will slow things down. Any suggestions as to what speeds, rate of removal I can expect with this as a spindle.

Thanks,
AJ

OK... I'll finally concede on the wire rope. Spent the better part of a day trying to make it work but to no avail. I thought some bearings that guided the wire onto the spool would help tracking - but nosireee, the wire likes to translate across the spool and right off the other side. Haarumph! Looks like I am due a few "Itolduso's" - and its time to pay for performance.

OK, I'm over that now. So onto looking at a Belt drive or dual lead screws. The debate is now whether I go with two motors slaved together driving two screws, 1 motor belted to two screws, or one motor through a reduction gearbox driving a shaft connected to two timing belt setups. Have to run the numbers on these to see what looks most economical.

Ciao,
AJ

What kind of speed can you do a finish pass at if you are basically doing a 2-1/2d toolpath? I would like to think I could power this homebrew effort to do roughout cutting at 50-100ipm (1/2 end mill - with 1/4" stepover, 1/2" stepdown) and then do my final finishing at 150+ipm... is this another dream? In terms of the finish pass, the bulk of my work would be completed with spiral type tool path (ie very gradual changes in the Z as the x & y are brought to finished dimensions)... is this what people refer to as 2-1/2 d? As opposed to an xy raster with radical Z axis changes?

You are dreaming about cutting at over 150 IPM for the finishing cuts on your home machine. I run my 3D carving tool paths using a Pentium III 500 Mhz machine and Mach 2. My machine can move a linear move at close to 200 IPM but the P.C. can't deliver enough steps on all three axes to go faster than 50 IPM for my finish cuts. I can tell mach 2 to go faster but it will not do it.

Remember that pulse rate is GOD for speed on your (servo) machine. You can only go as fast as the steps can be sent to the drives.

Regarding dual servos for the X axis, Mach 2 has some built in logic to accomodate this...

First reference from Mach 2:

Second reference from Mach 2...

Hi Jeff,

Your post Re. pulserates, PC's, and maximum feedrates on 3D finish toolpaths was a real eye opener.

Given what's available today at the hobbiest level/budget - what is the constraint to how fast you can machine a complex 3D toolpath. I don't really understand it all yet, but I take from your last post that the PC is the weakest link in terms of computing/processing the code fast enough. I've read the Mach 1 and 2 can process at 25 kHz and 45 kHz respectively. Can a newer PC match this rate of processing? Where is the bottlekneck liable to occur given Mach2 software, a newer PC, gecko driver setup, and suitable servo/encoder setup?

Or I guess the simpler way to ask this is where would you spend money to make your machine go faster? and how much faster do you think you could make it go?

Thanks for the reality check!

AJ

Beerfizz,

Thanks for the Mach2 excerpts. So far its sounds like it would cost about the same to drive the X-axis using: 1) two motors, 2 leadscrews, and 2drivers VERSUS 2) two lead screws, mechanical syncing between screws [timing belt], and one motor/driver.

Thks,
AJ

From my readings throughout this site, I have found that many have had problems with using two slaved motors. Racking accumulates unless the system is constantly refreshed/reset. Maybe with servos this is minimal. I believe that two shafts with one common motor with a timing belt and pulleys is best though the cost in time and money may be same. Worth the time though to set it up as such.