Hi Guys,

I frequened this forum a couple years ago and collected quite a few parts for my planned machine. I have the following:

Thompson linear rails:
• 48" X 1" supported X
• 24" X 3/4" unsupported Y
• 18" X 3/4" supported Z

Other parts:
• 3 servo motors 360 oz/in with encoders
• Lots of plate aluminum and some extruded
• Computer
• Router

I'm tired of seeing all this equipment sitting in a pile, I either have to finish this project I started over a year ago or sell the parts. I'm intimidated by the power supply, drive boards, mounting the servos in the best configuration, and finding the correct linear screws. I'm no electrician but can assemble from plans and solder quite well. I'm suffering from information overload and procrastination. Could someone please show me the light? I'm trying to figure out how much it will cost me to get all the parts needed. From what I can remember I will need the following:

• 3 linear screws
• power supply
• drive boards
• software

I'd really appreciate pointers, suggestions, condolences... whatever you can offer that might get me out of this funk. I'd prefer to finish the project rather than sell the parts. Of course I'm kinda tight on funds as well so any cost savings ideas would be wonderful. I was considering Gecko drives back when I started collecting parts. Is there a less expensive alternative? I'm really trying to keep the costs down to finish but understand that might be futile.

Thanks.

Darren

Hi Daren,

I can't help with the linear Screws but here's a Link to Phils Electronics Site.

http://pminmo.com/

Its great for the electronics do it yourselfer.

Googling Turbocnc and Mach3 will point you to trial programs that can be used for setup and even small jobs.

Hope this helps.

Thanks for the link, unfortunately I didn't see anything for servo drivers :( Did I miss a link?

Darren,

Hi! I know how you feel! Where to start? It can make your head hurt. I think you can get a lot of help here, but it might be best if you can offer up a little more information. Do you know the specifications on your motors and encoders? That may point to certain drives. What kind of machining are you wanting to do? Metalworking? Wood? That will determine how much "oomph" you need. And is this to be a hobby machine, or a real shop tool? Knowing that, you can better pick out a drivetrain. I say drivetrain, as you might use belt drive or rack and pinion versus screws. Also, what about budget and your skills with building? That may also steer you one way or the other. In a nutshell, the more information you can provide the better. Look forward to hearing back from you...don't give up!

Sincerely

Evodyne

Ok, more info:

The servo's are Litton Clifton Precision Model# JDH-2250-BX-1C they are 360 oz/in and the encoders are US Digital E2-250-25-G, they have Ground, Index, Ch.A, +5v, and Ch.B pins. Please let me know if you need more info, I'll try and look it up on-line.

This will be a serious hobby machine. I will be working with wood mostly but would like the ability to cut aluminum, speed on aluminum isn't terribly important as long as it isn't prohibitively slow. I'm looking forward to any advice you can provide. I'm quite capable as a woodworker, have been remodeling homes and building furniture for many years. I'm mechanically inclined but don't know much about CNC to start with. I work in computers for a living so I'm adept with computers/software.

Budget, well... I'd like to keep it under $500 to complete it but I realize I might be dreaming. I'm just looking for the best bang for the buck to complete my machine. I'm a real DIY guy, I've assembled my own high end audio amplifiers and speakers from kits so I can follow schematics and am good with a soldering iron. I just don't know how to design stuff like power supplies etc. I'm great at duplicating proven designs with instruction.

I'd like to think I can build pretty much anything with proper guidance. :)

Sorry Darren,
I didn't notice the servos.
Alan is developing a servo controller but I don't know how advanced he is.

http://www.fromorbit.com/projects/picservo/

Cheers

Ok, more info:

The servo's are Litton Clifton Precision Model# JDH-2250-BX-1C they are 360 oz/in and the encoders are US Digital E2-250-25-G, they have Ground, Index, Ch.A, +5v, and Ch.B pins.
Do you have a datasheet or any thing on the motors? Stall current, rated voltage, rated rpm, that sort of stuff?

This will be a serious hobby machine. I will be working with wood mostly but would like the ability to cut aluminum, speed on aluminum isn't terribly important as long as it isn't prohibitively slow. I'm looking forward to any advice you can provide. I'm quite capable as a woodworker, have been remodeling homes and building furniture for many years. I'm mechanically inclined but don't know much about CNC to start with. I work in computers for a living so I'm adept with computers/software.
Hey, I'm like you-except for the capable, mechanically inclined part. :D See, you've already go the upper hand! I'll fess up-I've yet to start building my machine-though I have nearly all the parts.

Here's a short blurb on power supplies, but my time is limited tonight. O.K, servos are going to give you more upper end power and speed than steppers, but to take advantage of it you need those specs (voltages, currents, etc.) to match them with the right power supply or supplies. You want a supply that has a voltage at or just below the motors rated voltage. You might see values for continuous torque and stall torque. Your 360 oz-in rating is one of those two. The continuous torque is what the motor can deliver all day long and not overheat. It will have a corresponding current that will produce that torque when supplied with the rated voltage. But a servo will try it's best to give you what you want: load it down and it will produce more and more torque. It will produce the most just as it stalls out from too much load. That value is your stall torque. You can't run that way too long, as you'll heat up quickly. But it gives you acceleration and the abiltiy to push hard (for a short time) when needed. But to get it your power supply needs to deliver a lot of current. How much? The value given by the stall current value.

Example: I've got some big servos that will continuously put out 300+ oz-in when powered by 60V and drawing a few amps. But they will deliver up to 1500 oz-in when fully loaded. My supply has to deliver a stall current of 45 amps though for me to get that. For one motor!

So, basically, you want to get enough current from a supply to meet the stall current rating. I could go on, but I need to get off line. Get a pencil and paper and rough out what one motor would require. Believe it or not, you don't need three times that for three motors-you are very unlikely to simultaneously stall all three. Three motors will need exactly the same voltage. How much more current hopefully someone else can tell you. I didn't go this route-I'm building with three separate supplies. Ugh! I really need to go. Maybe more later. Bye!

Lance

Hi Darren,

It looks like those servos are the same "cheap servos" that can be found here (http://cnczone.com/forums/showthread.php?t=2082). The ideal power supply for those is 36V @ 20 to 25A. You could construct a power supply fairly simply with a transformer, capacitor, and a rectifier. The 250 CPR encoders you have would work well with G320 Gecko Drive. Depending on the size and torque you want your router to have, you may want to go with a 4:1 or 2:1 belt reduction between the motor and screw (If you are going to use a lead screw), however, it all depends on what you are going to be cutting and the size. Also, you may want to consider 80/20 Aluminum Framing for the construction of your router. Hope this helps.

Yes, those are the servos. I bought them from Jeff Davis (HomeCNC) a while back. He modified them to accept the US Digital encoders and a NEMA 23 style mounting plate. He also tested them with a Gecko 320 drive with good results.

Here are the specs:
Use a 36VDC power supply.
No-load RPM at 36VDC = 3,130
(Ke = 11.5, KRPM = 36V / Ke)
80% of no-load RPM = 2,500 RPM
Leadscrew RPM @ 120 IPM, 5TPI = 600RPM
Reduction Ratio = 2,500 RPM / 600 RPM = 4.1666:1, Use 4:1
Kt = 1.351 Ke = 15.5 in-oz / amp
Maximum continuous current = 3 Amps
Maximum continuous torque = 3A * Kt = 46.5 in-oz
Torque on leadscrew = 46.5 in-oz * 4:1 reduction = 186 in-oz
Leadscrew thrust = pi/8 * TPI * in-oz = 365 lbs
Maximum speed at full load = 2500RPM /(ratio * TPI) = 125 IPM
Maximum speed, no-load = 3,130RPM / (ratio * TPI) = 156 IPM
Power delivered to load = RPM * in-oz / 1351 = 86 Watts

A link to the servos: http://www.automec-direct.netfirms.com/html/clifton_dc_motor.html

Here is a quote from Jeff in the other thread:


Yes, I finished my testing and I have the full specs on these motors.

They are the next size smaller than the ones I get new from Poly-sci. They are 360 oz/in peak motors at a peak of 20 amps. This means they have an 18 oz/in per amp rating.

Testing results:

Made a bar that had a hole 1" from the center and placed a wire hook there to attach to a bucket. The only thing I had to add for weight was 45 cal bullets. I had my Gecko drives current limited to around 10 amps. I kept adding weight to the bucket and lifting the bucket with the servo motor. I got to a point where the motor would just lift the weight past horizontal and then the drive would fault and the bar would fall down vertical. I weighed the bucket and it was close to 11 pounds. So it lifted 11 in/lbs with my current limited to around 10 amps.

This is very close to what my Poly-Sci specs say. If you do the math, the spec says 18 oz/in per amp. So 18 * 10 amp is 180 0z/in. Convert that to pounds would be 180 / 16 = 11.25 lbs. I'm a 1/4 pound off. Maybe I don't have the current limit screw set to exactly 10 amps

The perfect power supply for these would be a 36VDC supply @ 20 amps. You could leave the Gecko drives current limit screw all the way up to 20 amps. This way if you needed the current for one axis that is working hard it can take what ever is available from the power supply.

Darren,

Hey-looks like he layed it out pretty well for you. So if you went that route you know pretty much the screw and supply parameters. Roton (http://www.roton.com/web/index.jsp) has a nice site with screw information, you might take a peek. jjwl89 was right, the Geckodrive G320s should work fine. Mariss has some nice products and other members have told good stories about his support when they "break" one of his units.

Lance

I would start building and by the time to get to the electronics....you'll have more buck available....

Well, it isn't a matter of obtaining the money, I have a set budget I don't want to exceed. I'm hoping to gather the remaining parts for $500 or less. I don't want to start building until I have everything, I really want to sit down with all the parts and work my design from there.

I was hoping there might be a less expensive option than the Gecko drives. That would make a huge difference in price but I suppose I could keep an eye out for used Gecko's. I guess I also need to look into belt drive type mounting for the servos and screws.

I got this power supply schematic from JavaDog (JavaDog, let me know if you didn't want this posted and I'll remove it). He has the exact same servo's I bought from Jeff. I'm considering using this design for my power supply. I don't know much about designing power supplies but it looks like two of the Gecko's will be driven by one side of the power supply and a single by the other. I'm assuming the Z axis would be one that is doubled up because of the minimal load it would see most of the time.

Schematic (http://users.adelphia.net/~javamoose/CNC/Schematics_Rev2.jpg)

Can someone suggest a good place to find large caps like the ones called out in the diagram? I'm having trouble locating some that size. Also, could I stray a little either direction on size? How much?

Thanks for any info. I just got a bonus at work so I'm getting ready to purchase 3 Gecko G320's and need a power supply next. Then on to screws and reduction and mounting motors....

Darren

Darren,

The supply uses electrolytic capacitors. These capacitors have two ratings: a working voltage rating (given in volts) and a capacitance rating (given in microfarads). They also have a polarity-one terminal is the "positive" and the other is the "negative".

Your supply will output about 24 VAC * 1.414 or 34 VDC. A volt or two less actually due to diode voltage drops in the bridge rectifiers. The capacitors voltage rating MUST exceed the 34 volts. They can be rated much higher (i.e. a 70 volt cap. is fine), but it can not be rated lower. You should be able to use 36 or 40 volt capacitors. I'd shoot a little higher just to be on the safe side.

The overall value of capacitance adds up for capacitors in parallel. So each pair of 37,000 uF caps act as a single 74,000 uF capacitor. More capacitance is better, but you can actually use quite a bit less and do just fine-even a single 4,700 uF capacitor would work in a pinch. Larger ones give less AC ripple and store more energy that can be drawn on for a short time if needed.

One final note: DON'T wire the capacitors in backwards-i.e. the negaitive voltage to the positive capacitor terminal. Bad things (like a small explosion) can happen!

Have fun!

Lance

Darren, I use the Sprague electrolytics at mouser.com.

Darren,

Thought of this after I posted the last one...

Your motors give a peak torque of 360 oz-in. Your ke is 15.5 oz-in/Amp, so you peak amperage will be 23.2 Amps. You want the supply to be able to handle this at 34 volts. If it can't it will bog down under load and you won't see torque anywhere near the 360 oz-in rating.

The diode bridges are rated for 25 Amps. They will be fine-you wont be pulling this kind of current too often. You can always simply buy bridges with higher ratings. If you buy bridges, make sure their voltage rating (PIV or Peak Inverse Voltage rating) exceeds your 34 volts. A lot of times they are already rated for 400 or 600 volts, so no problem there.

Similarly the transformer and fuse current ratings should be bumped up as well. 34 VDC * 23 Amps is 782, say 800 Watts. For reference, 746 Watts = 1 HP.

Your transformer should then have a 24 volt secondary (the center tap isn't needed-you're not using it) that can handle 800 VA. Currently (no pun intended) it is set for 10A * 24 VAC or 240 VA.

On the primary side it would draw 800 VA / 110 VAC or 7.3 amps. The 10 Amp breaker/switch should be O.K.

The slo-blo fuses would need to have their trip rating bumped up. As they are now they would trip you out at 5A. At 15.5 oz-in/Amp the best you would get is about 78 oz-in of torque. Not quite 360 oz-in, huh?

I would want to be able to push the 360 oz-in of torque briefly, but blow a fuse if at that level for too long. Why? The motor is going to get very, very hot at this level of drive. Blow a fuse or a motor? So I think I'd try to get a slo-blo at or just under the 23 amp value.

Hope this makes sense.

Lance

Evodyne....I think you strayed on the servos...he claims they are only rated at 3 amps max.....

Lance,

Thanks for all the great information. It will take me a bit of time to digest what you said, I'll study the diagram and read your post more closely tonight and see if I understand exactly what you are saying. It makes sense to me initially but I will need to look it over more.

Ouch, that size cap is spendy... $30.84 each and I need 4 for this design. Anyone have suggestions for a less costly route? Sorry, I'm really quite green in this area. I can assemble it, just can't design it :)

http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=442920&e_categoryid=387&e_pcodeid=75007

I found these for half the price:

http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=442552&e_categoryid=387&e_pcodeid=75007

And these:
http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=442526&e_categoryid=387&e_pcodeid=75007


Good idea? Bad? Seems close to what I need.

Evodyne....I think you strayed on the servos...he claims they are only rated at 3 amps max.....

ViperTX....

Hi! The 3.0 Amp rating is for continuous running-running all day without overheating. He specifies the torque rating at 3.0 amps at like 45 oz-in, which makes sense as his ke rating is 15.5 oz-in/Amp. This is a different spec. than peak torque. If you look back he quotes "Jeff", who has the same motors, as saying the peak torque and current are 360 oz-in and 20 some amps. That's peak or stall current, when the rotor is nearly locked. In this state the full 34 volts of the supply sees only the winding resistance of 1.7 ohms, giving 20 amps. The motor in this case is dissipating all the power from the supply (800 Watts) as heat. Good for pushing through a tough spot-just not for long. This is the advantage of servos over steppers-plenty of reserve torque when needed. You don't have to build a supply to operate at this level, but it defeats the purpose of buying 360 oz-in motors if you don't.

Regards,

Lance :cheers:

Ouch, that size cap is spendy... $30.84 each and I need 4 for this design. Anyone have suggestions for a less costly route? Sorry, I'm really quite green in this area. I can assemble it, just can't design it :)

http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=442920&e_categoryid=387&e_pcodeid=75007

I found these for half the price:

http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=442552&e_categoryid=387&e_pcodeid=75007

And these:
http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=442526&e_categoryid=387&e_pcodeid=75007


Good idea? Bad? Seems close to what I need.

Darren,

Take a peek at the attached .pdf file off of the Geckodrive website. Look at the capacitor information in particular. Notice anything? Wow! Those capacitance rating CAN be a lot less! This implies you really only need two capacitors vs. four. I've built several supplies similar to these and used between 8,000 and 10,000 uF total (one cap. per transformer) and been happy with the results. I actually would get the 2nd or 3rd choices you listed-you dont need 75 volt caps. You could certainly use them, but way pay extra when the others are just fine? Have fun!

Lance

Sorry if I'm butting in here; I was asked to take a look.

Darren did a good job of summerizing the motor specs; that makes things easier.

The motor output power is 86 Watts while power in is 108 Watts (3A * 36V). A quick reality check shows an efficiency of 79% (86W / 108W) and meshes nicely with the theoretical expected efficiency of 80%. The data is good.

The quandry seems to be power supply design. Do you design to maximum limits or do you pick something smaller and more practical? Let me throw in this about engineering; it is the science of designing something that is adequate to the task and not a bit more. Airplanes wouldn't fly otherwise.

A Kt of 15.5 in-oz/amp requires 23A at 360 in-oz at stall yet the max continous torque is 45.6 in-oz requiring 3A. The ratio is 7.66:1 as is the price, size and weight of the supply. At one end you need an 828 Watt supply, the other a 108 Watt one. Big difference.

Let's look at the implications to help decide which end of the scale makes sense.

Let's start at the cheap end of the scale. Since a switching-type drive is being considered, we can think in terms of Watts. For a 23A stall at 36V means the motor has a resistance of 1.56 Ohms. 108 Watts into 1.56 Ohms is 8.3 amps for a peak torque of 129 in-oz. Not 360 but still pretty respectable. Switching drives can and do put more current into the motor than the supply current draw.

8.3A is 2.77 times more than the continuous rating of 3A and the motor will generate 7.65 times more heat than it can continuously dissipate (the I^2*R thing). Take the reciprocal and you get 13% (1/7.65) maximum duty-cycle for this torque. Not bad.

Now let's go to the Big Watts end of the scale and compare. 23A is 7.67 times 3A and the motor generates 59 times more heat than safe, giving a 1.7% duty cycle. The extra Watts you have paid for can only be used for 1.7% of the time! This is a bad utilization of an expensive resource. Kind of like only being able to drive your car 1 day every two months. It gets tedious to calculate but the net practical result is to shave a few milliseconds off of an average motion control move compared to the 108W supply.

Finally, how about a practical middle-ground solution. Most machines have 3 axis. You design for a 3 times 108 Watts to service them (324W). Assume (safely) only one motor accelerates maximally. 324 Watts into 1.56 Ohms is 14.4A and 223 in-oz. This is so close to the 360 in-oz number as to make the difference meaningless.

My recommendation? Design a supply to meet the max continuous current rating of the motor. Anything above that has rapidly diminishing returns in terms of performance. Supply capacitor size for three axis? 20,000uF for a 9A, 36VDC supply.

Again, sorry for butting in on a thread I haven't been following.

Mariss

Mariss,

You didn't butt in-I asked you to take a peek! Please, don't apologize when your insight is nothing but helpful. The whole point of the forums is to share our thoughts, ideas, and wisdom, right? Right! I'm glad you popped over to take a look-it will certainly help Darren and I see your point about the value vs. return. You do a nice job illustrating this with some simple math. Thanks! Still seems a shame though...kinda seems like buying a Hemi 'Cuda and then installing a single barrel carburator off of a Cub Cadet tractor. Bad analogy? Probably. Oh well...

Lance

Mariss,

Thanks for bringing your point of view to the thread. I'm happy you "butted in". I really don't know much about this sort of thing so all perspectives are good for me. I like your practical approach, especially since it saves me money. I just need to try and get a materials list and schematic drawn up somehow for you guys to check out.

So, you are recommending a single 22,000uF capacitor in a 36VDC power supply at 9A?

I hope I said that right :) I'm going to do some studying so I can communicate intelligently. I took electronics in High School but that was 16 years ago :(

Darren from my observations on this forum there are a few catagories of members here.

Those who do lots of drawings of thier machine design, and then dissapear never to be heard of again.

Those who collect parts for years, and never do any thing with them.

Those who just start buiding a machine, and eventually finish it.

A philosphy of mine is, to jump straight in, think on your feet, dont understand it all b4 you start, and edit your work as you go. I have a friend who was once toying with the idea, of building a machine. I advised him to just spend some money on parts, and then he would be committed. Well he took that advice and he now owns/built a fantastic CNC machine.

A successful man I once new said his philosophy was to "Bite off as much as he could, and start chewing like buggery".
Another member on this forum once said to me, and it really helped me, "If youve run out of money, or dont know what to do, just drill a hole." That sounds so simplistic, but it means there is always something you can do on your machine that doesnt cost money, and will get it one step further to been completed.

Just start.

ynneb,

Seems I'm well on my way then! I purchased probably a little more than half the mechanical parts about a year ago and got caught up in other things. I'm now staring at the pile of parts wondering where to begin again. I believe you have identified me in one of your descriptions. I suffer from "analysis paralysis" so to speak. I research things until I thoroughly confuse myself :)

I guess I'll need to start cutting some aluminum this weekend and see about getting this thing started. I get a bit discouraged when I think of the belt reduction etc... That's another part I'm not familiar with yet. I'm quite capable of building this, just need the knowledge to do it right. I appreciate your encouragement!

I guess I'll need to start cutting some aluminum this weekend

Great. Now for the bit that I underlined. "Think on your feet" You dont want to cut that alum wrong, so think a few steps ahead at how everything will fit together. But only a few steps ahead, not the entire machine construction ahead. Although a general idea of how the machine should be, would also be a good idea. But I expect you have a general idea already. (Does that sound confusing? )

I want to see some pictures of your cutting after this W end. OK ?

I want to see some pictures of your cutting after this W end. OK ?

Oh, sure... make me stick to what I say eh? :)

Darren,

I just noticed that your location is Auburn, WA. I work in Renton. I've been scrounging/gathering parts for a CNC router and so far I have aluminum extrusion, some small stepper motors, hobbycnc 4axis board, and just recently I've acquired several lab/bench variable power supplies. I just happen to have an extra Systron Donner HR40-10C (0-40VoltsDC/0-10Amps) that might work for you if you are interested in trading something or buying it. I'm still looking for linear guides and ACME or ballscrews.

I haven't even thought about cutting aluminum yet since I'm waiting to size the machine to whatever guides and screws I can get. Anyway, send me a PM if you are interested.

Later...

juzwuz

Darren,
Nah, didn't mind the posting at all. Since I haven't built my PSU yet - the info is good. Although, kinda annoying that this is coming up now and not in the 8 pages of my thread. :boxing:

Looking forward to how all this turns out...

Darren,
Nah, didn't mind the posting at all. Since I haven't built my PSU yet - the info is good. Although, kinda annoying that this is coming up now and not in the 8 pages of my thread. :boxing:

Looking forward to how all this turns out...


Yeah, unfortunately this is coming a little late to save you any money unless you are interested in selling any parts I suppose :) I like your design, you already have it so might as well build it eh? Maybe we could present our designs as a option 1 and 2 for a budget minded power supply.

Hey, have you figured out how you are mounting the servos? Are you going with belt driven reduction?

Hey, have you figured out how you are mounting the servos? Are you going with belt driven reduction?

Yup, belt-drive 2:1...

I think that is the last bit of info I need to figure out, I can't really design the table or gantry without knowing how I'll mount the servos and drive the screws.

Do you have a good thread for reference? Since we have similar setups I'd love to see what you came up with.

I think that is the last bit of info I need to figure out, I can't really design the table or gantry without knowing how I'll mount the servos and drive the screws.

Why not, I did! :idea: I kind just winged the design...

Do you have a good thread for reference? Since we have similar setups I'd love to see what you came up with.

Sure, my build-log is here (http://www.cnczone.com/forums/showthread.php?t=6452). :)

Thought I'd post an update and declare that I'm not one of those guys that fades away without a machine :)

Well, I am still collecting parts and thinking. I've been extremely busy with work and other things but the CNC router is still on my mind. Just made a purchase on Ebay of a new supported X axis. I'm still procrastinating the power supply, Gecko drives, and lead screws. I'm hoping to start back up soon. Wow, I've been at this for quite a while with no machine to show for it... I swear, I will build it!

I have all supported Thomson bearing slides now.
• I have 30" x axis, 48" y axis, and 18" z axis.
• 3 servo's as described earlier
• Dedicated computer with AutoCAD 6
• Plenty of 1/2" aluminum plate, angle, channel etc... to build everything
• router as a spindle

I still need:
• lead screws for all axis and associated couplings, bearings...
• Gecko's
• break out board?
• Power supply
• ????

I'm taking updated pic of my materials etc... I'm rethinking my initial design as well.

Here are my updated materials:

Y axis (48")
http://www.garagehobbies.com/images/cnc/yAxis.jpg

X axis (30")
http://www.garagehobbies.com/images/cnc/xAxis.jpg
http://www.garagehobbies.com/images/cnc/xAxis3.jpg

Z axis (18")
http://www.garagehobbies.com/images/cnc/zAxis.jpg

Misc Aluminum
http://www.garagehobbies.com/images/cnc/aluminumCChannel.jpg
http://www.garagehobbies.com/images/cnc/angle.jpg
http://www.garagehobbies.com/images/cnc/extruded4x7.jpg
http://www.garagehobbies.com/images/cnc/plate2.jpg
http://www.garagehobbies.com/images/cnc/plate.jpg
http://www.garagehobbies.com/images/cnc/tChannel.jpg
http://www.garagehobbies.com/images/cnc/scrapPile.jpg

Hey! Glad you're back at it! Sorry to say, you just missed the big Gecko drive sale. They were on sale for the 100+ unit price.

Just my luck... Oh well. I look forward to putting the rest of the puzzle together.

Here's a hint to help get you started. Stop buying stuff and start building it. You don't need to worry about the PSU or the drives until you have something mechanical that's close.

Eric

I'm assuming this is your first machine. If so, you have a couple of decisions to make:

1. How badly do you want to finish
2. What do you need the machine to do to be worthwhile to you
3. How much more money are you able or willing to spend

If you continue on the approach of high-end parts that you're on now, you'll easily spend close to $2k finishing this. The Geckos, supporting parts, PSU, etc will be >$700 most likely. Leadscrews will depend, but you could easily spend anywhere from a couple hundred on basic acme thread, to $1000+ on ball screws. And then there's all the little stuff which you never think about until the third trip to the hardware store/etc ends with a $100 receipt. I'd say you'd be very lucky to finish without spending another $1000.

If you don't care and you enjoy daydreaming about what you might build someday, then relax and keep buying parts. This is about enjoying yourself. But if you really, really want to finish a machine soon, then I'd suggest taking a set of plans like the JGRO that you can work the parts you have into, and go from there. If you're really into this, the first machine you build will lead you straight to the second.

Without plans and with all those high-quality parts, you've set a very high bar for yourself to get started. Stop working in steel and precision parts and switch to MDF and you won't worry about messing something up. Lay the servos aside and get some medium-sized steppers and a HobbyCNC board and wait until after it runs to decide if you really need the warp drive. Just because you've spent $X and taken 2+ years to get here doesn't mean you're any closer to being finished yet. Simplify, simplify, simplify!

Yes, this is my first machine and I do tend to want the best of everything when setting out on a project like this but I offset that with finding good deals. My priorities have changed a bit and money isn't as big an issue at this point. I don't need the machine, I'm building it solely for fun and to learn. I'll definitely get some use out of it for hobbies as well.

Spending $1,000 more to finish would be desireable. My intention is to be able to cut aluminum reliably with a decent amount of precision although I'll be cutting mdf etc... as well. I've probably only spent about $800 on materials so far. I've been waiting for good deals but at this point I have most of the parts so getting the right materials to finish is more important than cost.

I'm ok with acme screws unless there is some reason I shouldn't be. I do tend to succomb to upgradeitis. I'm a little hesitant to start putting things together without the lead screws in hand. I'd like to have them for mocking up etc... am I being too freaky?

Actually the one thing holding me back from starting to assemble is getting a proper base to build from. I'm not sure I have the appropriate materials to build a solid platform for the Y axis to mount on.

Darren,

8020 extrusion has been mentioned by others as an effective and easy way to build your frame, as well as your gantry. In a bit of shameless self promotion, I might suggest using the setup described on my website (www.cncrouterparts.com) for acme screws and end bearings, which just happens to work well with 8020. Cheap, easy to assemble, and adjustable. With the motors you have, high pitch ACME screws (which are far more efficient than standard 1/2-10), and anti-backlash nuts from dumpstercnc, you'll have a pretty solid machine without breaking the bank. You can upgrade to ball screws later if you feel the need.

Ahren
www.cncrouterparts.com

Thanks for the info Ahren, I had thought about 8020 for the base or table but haven't ever worked with it. I'd think I would need rather large extrusions to prevent trosional issues and racking etc... Am I overthinking the table?

Your site is a great resource.

For the main part of your table, a surplus steel table or even a bit of pallet rack might be a more economical way to go. However, for all of the parts you want to keep relatively straight (and adjustable!) 8020 is pretty great. It's easy to work with, and the t-slots give you lots of options for your design. I've made some progress on my router this week -- it's time for me to re-open my build log. I'm using some of the parts I sell on my website now. I'll post some more pictures this weekend, and will let you know when I do.

Thanks, I'll be anxious to see the pics.

$800 for all those parts isn't half bad. I wouldn't hesitate to start out with 1/2-10 acme though Ahren is right to say it's not the best. I suspect however that it is unlikely to be a limiting factor in your first iteration, and it's comparatively inexpensive to work with and easy to find. Good luck with your build!

Ok guys, I'm having a hard time judging what size acme rod I'll need. I don't know much about this part.

Earlier in the thread 5 tpi with a 4:1 gear reduction was suggested and appears to be a good recommendation. I'm looking at http://www.use-enco.com/CGI/INPDFF?PMPAGE=625&PMITEM=AB408-0220 and am trying to figure out which rod to buy. It doesn't go by tpi from what I can tell. Should I have a specific diameter for a 48" length? I want to get all axis purchased today if possible. How do I tell which rod is 5tpi??? Yup, I'm pretty dumb about this part but anxious to learn.


1. do I use different diameters for different axis?
2. should all axis have the same tpi?
3. help, I'm getting confused with all the info

Don't spin your wheels!

To a large extent the only thing that matters is the total drivetrain ratio between the motor shaft and the moving table. If you have a 4:1 drive on a 5:1 screw, you have effectively got the same thing as a motor direct-driving a 20:1 screw. Rack drives usually require a reduction gear since the drive mechanism itself does not provide reduction like a lead screw does.

On my machine a 125oz motor driving a 1/2-10 acme screw at 50ipm will generate enough force to push the machine over if you hold an axis steady. I would start with 1/2-10 screws with direct drive since this will be easier to get right. If it works well and you still need more, you can swap in 2-start 1/2-5 screws and/or bigger motors. 1/2-10 is dirt cheap and steppers go for very good prices on eBay so the upgrade path is very simple.

In terms of axes, all controllers (well, I know EMC and Mach, at least) will allow you to specify something like a "SCALE" for each axis. This is usually specified in the number of steps to move the axis one inch or one mm, depending on which unit you use. A lot of people will use steeper (fewer turns) on the XY where jog speed is important and steeper on the Z where you're moving a lot more weight. However, my impression is that most people over-specify their requirements by at least 200%--since the price gap between a 125oz and 200oz motor is sometimes just a few dollars, it does allow you to get away with more misalignment, etc.

Ok, I'm a little confused and a little better off than I was a few minutes ago after reading your post :D

1. So, 1/2 - 10 would be 1/2" diameter 10 tpi???
2. 2 start... I assume that means two threads start off? Do I need to consdier this when buying nuts?


I'm using servo motors as was trying to avoid direct connect coupling for ease of mounting etc. It was recommended that I use a 4:1 reduction so I figured I could kill two birds with one stone. Good idea? Bad?

Screws are typically specified by diameter, thread pitch, thread type, and number of "starts" or threads. A normal threaded rod has one continuous groove while a 2-start has two, and so on. You definitely need to keep this in mind when buying nuts! All other things being equal, a 2-start thread will move you twice as far per turn as a single start, and so on.

It is easy to get lost in the weeds here. Axis travel per "step" is still what governs your operating limits so a 2:1 reduction drive with a 2-start screw gets you to the same place as a direct-drive on a single-start. A 2-start screw is not "better" for all applications. Keep in mind that the screws that McMaster sells are used for a million applications other than CNC so there are a lot of other reasons to specify these.

I do not have a lot of experience with servos so you will have to ask someone else. Steppers generate more torque at lower speeds so reduction drives can actually work against you. I am not sure how a belt drive simplifies mounting (alignment is still going to matter, and you'll have more things to align) but if that's the only reason then you can do a 1:1 drive. However I think servos like higher speeds more than steppers so you want want or need the reduction. Look around elsewhere on here for more on that.

Your real limitation is the critical speed of the screw -- if you rotate faster than this, you will get whip. Nook industries has a critical speed calculator here. For reference, you will probably have "B" style end-fixity:

http://www.nookindustries.com/acme/AcmeCharts.cfm

I'll save you the trouble of reading it and tell you that for a 48" axis, you can go about 200 IPM with a 1/2-10 5 start, vs. 40 IPM for regular 1/2-10 1 start. Even going up to 1-5 (1" diameter, which quadruples your inertia, meaning your motors will have to work 4x harder) still only gets you to about 160 IPM. It's all about how fast you're spinning, and 1/2-10 5 start is tough to beat for price/performance, since you only have to spin it 1/5 as fast to get the same linear speed.

I highly recommend using the 1/2-10 5 start from mcmaster carr. A 6 foot piece is part number 98940A020. I'm using the 3 foot length pieces on my router, and they are awesome. Also, while it doesn't say the manufacturer on McMaster's site, you will most likely get the screw from Nook. This has advantages in that it is probably much straighter than what you'd get from Enco or some other company.

Other than that, might I suggest end bearings and motor mounts from www.cncrouterparts.com :)? Oh, and dumpster anti-backlash nuts: www.dumpstercnc.com.

Good luck,

Ahren
www.cncrouterparts.com

The reduction between the motor and the screw won't matter when calculating critical speed -- it's the speed of the screw that matters. It's probably a good idea to use reduction for your servos, since servos operate more efficiently at higher RPM's.

Ok, good info ahren. A couple of questions for ya...

1. If I went with a 5 start how does that affect my reduction? I'm understanding what you said but trying to get my ratio correct is becoming more complicated from what I can tell. So, my motor will be spinning faster but the screw will be counter acting that correct?

2. If I need a 4:1 reduction for my motor, how does this look at the screw?

1) Just the opposite. The multiple starts is in fact an "inverse" reduction -- your motor will be turning slower for the same linear speed, which requires more torque, not less (but this is fundamentally necessary to get around the whip problem). With the 5 start, you'll be moving 1/2" linearly for every rotation with a straight connection to your motor.

2) The 4:1 ratio for your motor isn't a hard and fast rule -- it's just to bump the RPM's to get your servo into a more efficient range, so what you really want to look at is RPM's. To get the same linear speed as you would have with 1/2-10 single start and a 4:1 reducer, you'd have to have a 20:1 reducer with the 5 start. However, one of the benefits of the 5 start IS higher speeds, so I don't know if you'd want to lose all of that.

If it were me, I'd go with a 5:1 ratio on the servo and the 5 start screw. This will mean that at 200 IPM, the screw will be rotating at 400 RPM (200"/min * 2 rotation/1"), and your motor will be spinning 2000 RPM, which is a respectable speed for a servo.

Ok, so in another thread it was pointed out that Roton has ball screws for a really good price $10.00 a foot for the following:

http://www.roton.com/Mating_Components.aspx?family=7059321

How would I determine what reduction etc... I'd need for something like this?

Ball screws have a pitch just like lead screws, which is given in inches / rotation: as an example, a pitch of 0.2' means 0.2" of linear travel per rotation, or 5 turns per inch. Use the nook table above to find a screw with the same diameter and pitch as your roton screw to get the critical speed. This will be in IPM -- divide this by the screw pitch to get critical RPM. Then, gear up until your servo is running at an acceptable speed. I suggest 2000 RPM for your servo, but some can go much faster -- check your motor specs. Basically, multiply your motor gearing ratio by the critical RPM to see how fast your motor will be running at max linear speed -- you want this to be towards the upper range of your servo speed.

Keep in mind that ball screws are usually hardened, and turning the ends down to fit in a bearing can be a bear -- you can't just slide them in the bearing like you can get away with in ACME. Plus, you'll need a preloaded nut or a double nut to remove backlash from the system. It can be done, and when it's done right, the results are great. However, it isn't as simple as the ACME system I proposed, and if it isn't done right, you'll have paid extra for a system that's kind of sloppy.

My $0.02,

Ahren
www.cncrouterparts.com

Good food for thought. ACME might just be the ticket. I have a friend that can turn the screws down for me but I'm not sure he can handle hardened steel.

I'll browse your website a bit and do some other research. One way or another I'll get something ordered.

Darren,
Another option if you have access to a lathe (or better yet, a friend with a lathe) is to make adapters for the ball screws. A clamping adapter might be easier to make than actually turning the screw, and this could let you adapt to my bearings (or someone else's). You would still have to deal with preload on the nut, but it might be another option for you.

Ahren
www.cncrouterparts.com

Darren,
Another option if you have access to a lathe (or better yet, a friend with a lathe) is to make adapters for the ball screws. A clamping adapter might be easier to make than actually turning the screw, and this could let you adapt to my bearings (or someone else's). You would still have to deal with preload on the nut, but it might be another option for you.

Ahren
www.cncrouterparts.com

I do indeed have a friend with a lathe. He said he could probably even turn down a hardened ballscrew. He said that hardening is usually easy to grind through then machine the rest of the way. I personally don't know much about it but he seems confident.

By clamping mechanism are you thinking something that would clamp around the threads that has the correct diameter end on it for the bearings or coupling?

If your friend can grind through the hardening, you probably don't need an adapter, but yes, I was thinking of something that slipped over the ballscrew threads, and then had a smaller diameter cut to fit in a bearing.

If your friend is cutting them down for you, buy your bearings first -- FYI angular contact bearings are usually metric. Once you have those, you'll want to cut a shoulder long enough to fit through two of them mounted back to back, and put a thread after that that you can use to generate a preload on the bearings. Good luck with either way you go.

Since this is your first build my advice is to go with something like 1/2-10 acme rod and delrin nuts. With a good flex coupling this type of system will tolerate quite a bit of slop and misalignment, which is more likely than not on a first build. No matter what I do I can't get the Y-axis on my larger machine totally straight, it's probably off by a total of between 1/16-1/32" but it has minimal backlash on that axis and can position with around .005" accuracy. If I need better I will use the machine to cut new parts. I don't know how much misalignment a ball screw will tolerate without binding but there's a good chance you will find out.

Also just to reinforce what ahren said, the real cost of ball screws is in the whole system, not just the screw itself. McMaster has 5/8" screw for about $15/foot. The nuts are about $60 each while the mounting blocks are $300! I'm sure you can find cheaper but bear in mind that with precision machine parts, precision gets exponentially more expensive. A cheap ball screw system may not in fact be more accurate than a good acme rod. Ball screws are not just used for accuracy, they're also used because they're good for moving heavy loads very quickly due to their high efficiency.

Thanks sansbury, I think I'll stick with ACME for my first shot as you suggest. I just need to figure out how many starts now and what reduction I'd need for each start. I'm trying to understand what others have said but I'm not quite getting it.

Here's the easiest way i can think of to break it down:

Diameter: the largest OD of the screw. Thicker screws can carry more load and resist whipping better. Whipping is a factor of the unsupported length of screw, its diameter, and how fast it's turning. At 48" you need to think about whipping. 1/2-10 going 2000rpm will definitely whip without intermediate supports. There are calculators out there you can use to determine what's acceptable.

Pitch: The number of turns per inch. Steeper pitch = more distance traveled per revolution and less mechanical advantage via reduction.

Starts: These act as a multiplier of the pitch. Standard screws are single start, so a 1/2-10 1-start will move the nut 1" for each ten turns. A 2-start will go 2", a 5-start will go 5". This also affects mechanical advantage; a 2-start screw will provide half the reduction of a 1-start screw with the same thread pitch. However, it will also move twice as far (fast) for the same number of revolutions (per minute).

Here is a thread talking about this for a larger machine than yours:
http://www.cnczone.com/forums/showthread.php?t=8138

As I mentioned earlier in this thread, I would get multi-start ACME screw -- you can get much faster linear speeds at the same screw RPM. If you're worried about resolution, gear down before the screw (which you wanted to do anyway to get your servos spinning faster). Per a previous post, I would go 1/2-10 5 start, with a 5:1 reducer between the motor and the screw.

Thanks again to both of you for being patient with me.

Ahren, sounds like a great recommendation and I have it noted from before and was seriously considering going that route. I am just trying to wrap my head around all of this so I understand everything completely. I really appreciate everyone's adivce on this. It has been a long time coming and I'm finally getting somewhere.

I have a friend that can turn the ends down on the Acme thread if I need. I just need to find good ratio reducer. Anyone have a preferred supplier? I'll be checking my post and checking the websites provided to me as well earlier in the thread.

The part I'm having trouble with is this:

It was recommended to me initially by the person I bought the servo's from to use a 4:1 reduction, that was considering a single start acme. So, if I went with the 5 start as recommended for reducing whip wouldn't that mean I would require a 20:1 reduction to get the same effect? I know a 20:1 reduction isn't really practical. Maybe I'm using incorrect math.

Darren

Your math is basically right, but whether it matters really depends. The question proceeds from the resolution you want the machine to provide. I would guess that .005-.01" is a very good goal to aim for, especially if you're cutting wood, which will probably grow or shrink by several times that with humidity changes.

Motor resolution is typically discussed in terms of angular position change, since the motor turns in a circle. Steppers, for instance, are specified as being able to step in units like 7.5 or 1.8 degrees per step. Servos are electromechanically different in design, but the principle of commanding the motor shaft to turn a specified number of degrees is the same.

Now, assuming you used 5-start 1/2-10 acme screws, we can do the math:
1. 10tpi / 5 starts = 2 turns per inch, or .5" per turn
2. (0.5" / .005") = 100 or 1/100th of a revolution to move .005"
3. there are 360 degrees in a circle so 1/100th of a revolution is 3.6deg of angular position change

For reference, a typical hobbyist stepper motor will step in increments of 1.8 degrees. Whether you need to use reduction on the servo is above my pay grade as I'm not familiar with how torque curves on them work, but given the power of your servos versus your application, I suspect you may be able to do a direct drive with good results. At the least, you can try it without tears since it involves all the same parts (fewer, actually) you'd need for a reduction drive. Both machines I've built have been taken apart, adjusted, reworked, and reassembled a dozen or more times prior to "real use." No matter what you do it's going to have to come apart a few times.

Excellent explanation. I really appreciate that. I think I can proceed now that I understand.

Thanks again!

Everything Sansbury posted is correct, but you also need to take into account the feedback system you are using on your servo motor, since you aren't using steppers. Your feedback system is probably a rotary encoder, so your true resolution will be determined by the number of counts on your encoder disk. 100 CPR (counts per revolution) is quite low -- it's easy to get encoders with 512 counts per revolution or more. To calculate your effective resolution, just take your pitch in inches and divide by the number of encoder counts per rev. So, for 5 start acme with 512 CPR encoders, you'd have 0.5"/512 = 0.000976", or about 0.001". Any gearing you use will make this number even smaller.

Now, you may be thinking "0.001, that's better than I need!" Well, maybe it is, but maybe it isn't. Since you've decided to take the plunge with servos, you'll have to deal with tuning them. Steppers are easy, since they have discrete positions that they either make it to, or not. Servos have a continuous motion that is regulated by the feedback loop set up in your controller. For a DC brushed motor (typical of hobby servos, but I'm not sure what you have), this basically means the controller checks the position of the encoder, and increases the voltage if the actual motion is falling behind the commanded motion, or reduces (even reverses) it if the actual motion is getting ahead.

Why is this important for you? Well, even if you tune your system perfectly, under load your motor is unlikely to be able to track exactly the commanded position you give it. It's not uncommon to be 10-20 encoder counts off, and this is considered a pretty good tuning job unless you have an advanced controller. With 0.001" resolution, a 20 count error means a 0.020" error in position. Not so great. However, with 5:1 gearing, and a 512 CPR revolution encoder, your effective resolution is now about 0.0002", so that 20 count error would only be about 0.004". Furthermore, you'll have the benefit of running your servos at higher speeds, where they are more efficient (opposite of stepper motors, that have maximum torque at lower speeds).

OK, once you digest all that, I can tell you about quadrature decoding for your encoders, which is a trick that increases your resolution by another factor of 4. But let's leave that for another post. What I've posted above should get you some more knowledge about what to do with your system, and continues to support your original 4:1 or 5:1 gearing ratio. Of course, you could also use single start ACME to lower your pitch, but we've already discussed the foibles of this approach in terms of critical rotation speed.

Ahren
www.cncrouterparts.com