Smoothstepper pros and cons please?

[Ranscon]

I have a diy (mdf with steel reinforcement) router machine for wood routing. The cutting area is 20" x 40". I'm running Mach 3, with 425 oz/in nema 23 steppers, Keling 4030 drivers, C-10 bob, 1/2-10 Acme lead screws, and a 24v power supply.

My machine runs smoothly at 35 ipm. When I try to increase this in motor tuning the motors "whine" and stall and the gantry stops moving or "slips". There is not a lot of friction in the machine as it moves quite freely when I hook up my 14v cordless drill to the leadscrews.

Does anybody have any ideas as to some good starting settings for velocity and accel? I would like to get this to 80-100 ipm.

I was told that a Smoothstepper might accomplish this by changing the connection from parallel port to usb. Is this correct? I would love to hear some feedback on Smoothstepper, good and bad. Any other ideas? How do some of these machines with smaller steppers possibly run at 200-300 ipm and more?

Thanks, Jeff

[stevespo]

Jeff, I have a similarly sized machine (25"x32", 8020 aluminum), with similar motors. 5 TPI ballscrews, linear rails - very smooth and fast (250 IPM) with my 9.6V cordless.

However, with my 425 oz/in NEMA 23 steppers I hit a wall with my rapids around 100 IPM. I can cut very reliably at 90 IPM, which is ok, but I can't do much safely above that.

One theory is that I have a slow computer (800Mhz), and that Mach is not running as fast as it should, or that the timing might be off. I have a 1.6GHz machine I need to test this theory against, but I believe this is not the problem.

Second theory is that this particular motor is only effective until about 500 RPM (100 IPM for me). I've seen references to this in several places and the torque charts seem to bear this out. The torque is about 50% at 500 RPM and drops to about 17% at 1000 RPM. Some of the smaller steppers (like the 270 oz/in size) actually produce more torque at 500 and 1000 RPM than the larger 425 oz/in do. The smaller motor has a much flatter, more linear graph.

@ 500 RPM, 269 oz/in = 1.3 Nm, 425 oz/in = 1.1 Nm
@ 1000 RPM 269 oz/in = .9 Nm, 425 oz/in = .45 Nm

Right now, I'm really leaning towards theory #2, but it's hard to say. Initially, I thought the problem could be fixed with higher voltage. I was running a Xylotex 24VDC setup and then switched to Gecko G203Vs and a 72VDC, with no real change in performance. That was an expensive experiment. For me, it's not voltage/current/mid-band resonance - it's something else.

In your case, the 1/2-10 lead screws aren't doing you any favors, since you're going to need 1000 RPM to hit the 100 IPM mark and the motors may not have the juice to get you there. I'm skeptical that a Smoothstepper is going to help you out here, but I'm definitely no expert.

If my computer upgrade fails to get me anywhere, I'm considering swapping in a different motor(s), a different screw (2 TPI), trying a timing belt reducer, or if all else fails - maybe it's time for a Smoothstepper.

If I upgrade my X/Y to a NEMA 34 640 oz/in, I should have loads of torque:
@500 RPM, 640 oz/in = 3.6 Nm
@1000 RPM 640 oz/in = 2.3 Nm

This motor has a very flat, linear torque curve and much more available torque than what we're running now. I have to believe this is the proper solution. Again, there may be something else going on here, but this is what I am working on at the moment.

Please keep this thread updated with your progress and I'll do the same.

Steve - sorry for the long-winded response

[Torchhead]

Okay here's some Physics to munch on. Steppers will only spin so fast at a given load. (torque-RPM curves). The upper end of most with decent drivers and DC motor voltage is about 800 RPM, give or take a hundred RPM. What that means is if your system can provide the number of step per second to spin the motor at it's full RPM NOTHING in the pulse side is going to make it spin any faster. So having a Pulse card that can do 4 million steps per second will not help. If you run the math, 800 RPM is 13.33 Rev/sec. With microstepping drives you need to furnish 26,666 steps per second. MACH 3 with a parallel port will easily do 45,000 steps per second on a 1.8GHZ computer. About the only performance increase from a smooth stepper would be from a smoother pulse stream to the drives. That might buy you 8 to 10% more upper speed.

Your problem is mechanical. A 5:1 screw will never let you get over 120 to 150 IPM even if there were minimal friction. If you have acme leadscrews you are losing part of your torque there. A better match would have been a "2 start" screw (effective 2.5 TPI). You also need to make sure nothing is binding or takes a lot of torque to move. The motors will stall sooner the more torque that is required.

The stepper effect of losing torque with RPM says you need to match the need for speed with the proper gearing. You can actually benefit from a gear/belt ratio that lets the motor operate in the first 2/3's of it range. What you are seeing on the graphs is the faster tail-off of torque with the bigger motor. The fact remains at lower cutting speeds the larger motor will have almost double the torque. My advice is to re-gear to take advantage of the bigger motors better low end torque. A 2:1 belt increase will give you 150 IPM speeds and still let you operate down in the better part of the motors torque curve.

In any design it's a balance between torque and speed. You can trade one for the other. The resolution drops with increased speed as well. Usually on leadscrew router designs that is not the limiting factor.

Often the problem is perceived the opposite way. Most Builders think bigger is better so bigger motors should work better ....right?. Well yes, but it depends. With the same gearing they might not!

One thing that is interesting though. The doubling of voltage should give you almost double the RPM. You should have seen a noticable difference. On a small table top router we use in production we switched from a Xylotex and 24V supply to a G250 based unit (one of our BladeRunner's) and 48V. We say more than a 2X increase in overall speeds. Best we could get with the Xylotex was about 60 IPM now we can rapid at 150. Something just don't sound right when you said there was no difference. The 24VDC would not have let you get anywhere near the upper limit of your motor RPM.

Did you change the step polarity in MACH for 203V's? It needs to be "POS" (red X) on all the Step signals. Having it wrong will cause the motors to run rough and slow and stall too soon.


TOM CAUDLE
www.CandCNC.com
Totally Modular

[stevespo]

Tom, thanks very much for the help and advice. The math/physics makes very good sense and I should have paid more attention to RPM/torque/speed with picking my motors. I thought I'd be happy with 5TPI and 60 IPM, but once you start running long jobs, you (naturally) want more speed.

http://www.cnczone.com/forums/showthread.php?t=27527

Looking back over my build log, I did indeed get a boost from the Geckos, but not so much from the 72V power supply. I had been running pretty reliably at 60 IPM with the Xylotex and once I installed the Geckos that had moved to 90 IPM reliably (still with 24V). When I made the move to the 72V/12A supply, I didn't get ANY additional performance so I figured that the bottleneck was elsewhere. It could still be the computer, or possibly motor tuning as well. I clearly need to do more experimenting.

With the 5 TPI screw at 450 RPM, I'm getting 90 IPM cut speeds very reliably. Never lost a step. As I approach 120 rapids, it becomes very unreliable, so maybe my cutoff is 600 RPM. At 10x microstepping, that is 2000 steps/rev, and requires 20K PPS from Mach to maintain. My machine is a very old 800Mhz AMD and might be the weak link. I'm going to swap that out when I get a chance. If the 5:1 screws are going to limit me to 120-150 IPM I then need to make a decision to reduce RPM via a timing belt, new screws, or new motors.

My machine is certainly very solid, very accurate, but I'd love to run certain jobs more quickly. I feel like there should be some basic things I can do to wring a bit more performance out of my setup w/o the hassle and expense of moving to servos.

Thanks again,

Steve

[Torchhead]

The PC is one of your bottlenecks. We have found that a 1.8GHZ is good and a 2/4GHZ works up to 65,000 steps per second. There are several places on ebay to buy used pentium 4 boxes with plenty of RAM and a built in parallel port.

One such store is AssetRecovery. There are others. Lots of used 17" LCD monitors are starting to show up as well. You can put together a nice controller PC for about 150 to 175.00 with monitor.

I would expect 600 RPM if there were lots of friction. We have a router table with even larger steppers (740 oz-in) and a 5 TPI leadscrew. We get safe speeds of 150 IPM and can run it to 180. It has NEVER lost steps running at 150. The leadscrews are acme with anti-backlash nuts.

The polarity of the STEP signals in MACH to the 203V's is REALLY important. With 72Volts and 203's you should be able to hit 800 RPM on the motors with a moderate load.

It may be your PC just won't hack the pulses needed. I used a 900 mhz AMD for 2 years with MACH2 on a servo Plasma table and it ran fine. When I went to MACH3 it would buck and snort at anything above about 100 IPM. I upgraded to a 2.4 AMD MB and it returned to normal.

TOM CAUDLE
www.CandCNC.com
Totally Modular CNC Electronics

[stevespo]

Tom, I had a few minutes today to run some tests. I made the polarity switch, and was running 150 IPM rapids on both the X and Y axis. Individual straight line moves, combined X/Y angular movements, and even arcs. No problems at all. 150 IPM = 750 RPM. Fast and very smooth, no sign of any problems.

X velocity 150/acceleration 22
Y velocity 150/acceleration 16

My gantry is fairly heavy (8020), but silky smooth. With the ballscrew disengaged, light one finger pressure will move it. My 9.6V cordless on the lowest torque setting (1-15max) will move it at 250 IPM. So, binding/friction should not be an issue.

The really funky thing is that my tests never seemed repeatable. Tomorrow I could sit down and run the same exact programs and it may start failing at 120 IPM. It's just very hard to pinpoint. At least that has how things were in the past.

Maybe the polarity setting was the culprit. A new computer is certain to help too. I have a 1.6Ghz P4 sitting idle next to me, I just have to reformat/reinstall and hook it up. Maybe I should spring for a used 2.4Ghz+ and take the computer out of the equation.

Thanks again, this is very helpful.

Steve

[Torchhead]

So how fast do you want it to go? It sounds like the speeds are close to what the current configuration can produce. You might eek a little more out with a faster computer but 900 RPM is close to wide open. You have to start looking at servo's to get faster performance AND have adequate torque for acceleration. Given a choice between velocity and accleration (they always are inverse of each other) pick acceleration.

TOM CAUDLE
www.CandCNC.com

[stevespo]

If I can get these speeds reliably with the setup I have, then I'll be very happy. I just picked up a 2.4Ghz P4 so I can at least make sure that CPU speed isn't the problem. If I continue having problems above 100 IPM, it's time for a more drastic change (ie. servos).

Steve

[semi-lucid]

Tom

Could you expand the discussion to include servo motors.

If I want to achieve high resolution with steppers, I have to gear down to the point that one step equals the desired resolution. If I'm limited to 800 rpm, I get slow rapids.

Most steppers are 1.8 degree's per step, what about the ones that claim .9 degree's per step? Does torque suffer from the smaller step?

Some servo motors claim rated speeds of 3000 rpm. Are these speeds realistic?

Servo's also claim up to 4000 encoder counts per revolution. Can these motors actually be controlled down to 1/4000 of a revolution?

Thanks.

[stevespo]

The resolution of the steppers is a function of the motor resolution, the screw pitch and the driver's capability. On my setup, I have 200 steps/rev (1.8 degree), 5 TPI and 10x microstepping drivers. 200 x 5 x 10 = 10,000 microsteps/inch. That's a lot!

I can put a dial indicator against any axis and watch it tick off the tenths, so the accuracy is there. Given my rolled ballscrews, and the kind of work that I do, this is serious overkill. .0005" - .001" resolution would be plenty. I could definitely gear these up to 2:1 or 2.5:1 and gain a big boost in RPM without a significant loss in accuracy.

I have no experience with servos, just what I have read. Their rated speed should be realistic, but people typically gear them down so that (during rapids) they are operating at 75-80% of their max speed. This reduction also gives them a torque boost.

I don't know the answer to your last question. If the encoder has that kind of resolution (1/4000) it's a question of whether the driver and motor can actually take advantage of it. Multiply this by your screw pitch and this will be your effective resolution per inch. To really take advantage of this would require high precision ground ballscrews, double anti-backlash nuts, and a really fast pulse stream to get decent high end speed. Based on my understanding, anyway.

Steve

[semi-lucid]

Steve

I have seen mention of micro-stepping, half-stepping and what not, but I sort of disregarded it, assuming that there would be a loss of torque.

I need to go find a definitive article to read. I was leaning in the stepper direction, before I read that 800 rpm stuff.

Regards

[stevespo]

No, I don't believe there is any loss of torque/smoothness with microstepping, but I've never done any quantitative measurements. Can't say for sure. I've found the accuracy/repeatability to be outstanding.

It all depends on your application. If there is a fairly hard limit around 800 RPM, then you need to choose your screws carefully to achieve the accuracy and performance you're looking for. I'm quite merrily rolling at 150 IPM (750 RPM), and the limitation right now is my 25k pulse stream from Mach3. Can I go much higher with these motors, and no gearing? Doubtful.

For many applications, people gravitate towards servos due to their higher RPM, continuous torque, closed-loop feedback, etc. I've never heard of anyone regretting their servo investment, but there is a bit more upfront cost and effort. If I was starting from scratch, that is probably what I would do.

Personally, I've got too much time and money invested in my stepper setup to just toss it and start over. Besides, it's been ultra-reliable and easy to work with. Any problems I've had have just been part of the learning curve, trying to make sense of all the parts and how they interact.

For a few bucks, I can build a timing-belt reducer and drop my motor RPMs and increase my screw speed. Bolt it on and see what happens. Much cheaper than $1000 (or more) to swap them out for servos (which is always an option at some point). Of course, at some point you say "enough is enough" and either accept the limits of your machine or start over.

What kind of machine are you looking to build?

Steve