Trinamic Discussion Groups

[shorton]

Hi guys:

This problem appears to be somewhere between TMCL, or a a general module issue. Wasn't sure which sub-forum to stick it. Feel free to move as appropriate..

I am using the 110 module with a Linengineering NEMA17 .9deg (400 step/rev) motor. I have also tried a 1.8 deg with the same result/problem.

My application is a simple linear slide. Moves a load from point A to point B and back with a very low duty cycle (minutes between moves). The movement is about 8"/200mm. With my pulley, this works out to about 4 turns of the motor over ~2 seconds. The load/torque is very small and much less than the motors rating. It takes a lot of extra external force to hold it and prevent it from moving (stall the motor). I am using a 24v 2A powersupply. I have the motor max current set to 1180 per Linengineering's recomendations for my motor.

I have tried a lot of combinations of speed and accelleration and I get nice, smooth quiet movement in the fast constant velocity part of my motion. At my vendors recommendation, I tried setting the fastdecay to
-1, 0, 2048, 1024, I could see no difference with any setting of that parameter.

In the Vel/Acc settings that seem to work best, I get very quick (almost instant) accelleration to the target velocity. A very sharp start, which is fine. It's a noisy start as it goes through the slower speeds on it's way to finaly velocity, but good enough (on start acceleration).

But it decellerates differently. The decelleration is more visible as it's a seemingly a good bit slower (to the eye). My device loudly "growls" (eeeerrrrroooowwww - technical term <smile>) to a stop with a slower decelleration (about .5 sec). The decelleration is very noisy but what's worse it the deceleration has a coupe very "rough" spots. At some settings I can see/hear a clear pair of closley spaced "bumps" as it goes through the deceleration curve (eeeerrrrroooowBwBwww). Sometimes those "bumps" are strong enough that the motor loses some steps.

Of course I checked the mechainsm in case it might be the cause or a contributor. The mechanism is a ball bearing based linear slide with negligible friction. It functions from end-to-end perfectly smooth with no change or anything to cause the rough decelleration. I can change my travel by 1/2, basically stopping the device mid stream, the "bumps" follow to the new location, more evidence it is not the mechanics causing them. Also, the motor behaves identically with no load at all.

For what it's worth it seems to "bump" on decelleration worse in one direction than the other (i.e. from left rotation to 0 vs from right to 0). But it's hard to quantify. We have changed motors and have the same result.

Is there something I can do to get the decelleration as quick as the acceleration, or mainly, to make the deceleration not so rough?

My pdiv and rdiv were at the defaults (3/7). Per the other post about slow deceleration relative to accel., I have tried -1 on rdiv and changed the accel accordingly, but that does not fix it. i've tried a variety of settings on acceleration, too.

Let me know if there is any other info I can give that will help. I'm anxious to go to production with this device but I can't release it with a rough motion.

What I need more than anything is smooth accurate motion, quick is a plus but I have a lot of leeway there. Quieter woud be good, too.

Thanks,
Scott

[Olav Kahlbaum (TRINAMIC)]

Hi Scott,
please give us more details about all your parameter settings, so that we can reproduce the behaviour here. We can then better give recommendations for your parameter settings so that your problem can be avoided.

Generally, with some low acceleration values, the motor decellerates slower than it accellerates. Here it mostly helps to change the accelleration value just a little bit (increase or decrease by 10 or less). This behaviour is also likely to happen when moving short ramps where the maximum positioning speed (parameter #4) can not be reached. In such cases it can be worth a try to reduce this parameter to a value that can be reached.

Regards
Olav

[shorton]

Olav:

I've worked on this a good deal this weekend. I can't seem to find a place where I get smooth acc/deceleration and reliable, consistent positioning. I have reset the module to defaults using the onboard jumper, and I have loaded the latest firmware from the website.

As I described originally, I'm just moving from one position, back to another, a very simple device.

I'm seeing a significant "bump" in the motion at one end of my travel. I've ruled out the mechanicals by trying differnet start-stop points, the bump follows to the new stop position.

Today, I ran the motor alone, mechanically disconnected and you can visually see the "bump" by watching the pulley. I have seen the pulley slightly reverse direction for a few degrees before it continues it's decelleration. Finger pressure on the pulley and you can feel the bump as well.

My test program is very simple. ROrate until a limit switch is triggered, set position counter to known value (i.e. 0), then move back and forth a specific distance with a 2 second pause between direction changes.

One set of params that causes the "bump" are:

Speed 700
Accel 85
pdiv 3
rdiv 8
uSteps 4
max current = 1180
MoveDistance=26208

The bump occurs worse and the far end (i.e. position =26208) than the start end (pos=0). It is also not consistent. Occasionally it won't "bump" but it does it enough to make my device completely unreliable in positioning accuracy.

When I get this rough deceleration in my device it sets up a huge vibration that causes the motor to stall at worst, lose steps at best. When the device is sitting on my bench, the stopping motion "bump" will cause things to rattle/vibrate on my bench. Not a very scientific measurment, but I mention it so show it is there.

I tried 2 different motors, they behaved the same. The motors are from Linengineering, one of your US distributors. The motors showing the behavior are NEMA 17, 400uS/rev motors (0.9 deg).

I had tried a great many accel/speed combinations. I did find the device worked a little better but still not reliable at some higher accelerations. In the 250 range (all else the same), I get a "musical" decelleration. Does not happen on accel, only decel. From 258 and up, the decel does not cause the musical sequential tones. And for much lower values (<100 IIRC) it won't do it however it's deceleration is very noisy with a grinding "growl" sound as it stops.

Hopefully you can duplicate and help me find some parameters that will work for me.

Thanks,
Scott

[shorton]

Andrew, thanks for the input. But I think I've addressed all your suggestions already, as much as possible.

1. For resetting due to many experiments, I reset the whole module back to factory default using the jumper on the module procedure.

2. As for choosing a lower top speed, this won't help becasue the place it happens it at a very low speed. To low to use, so I must go through it. I can't go that slow. The anomaly happens right before the stop, and immediately after takeoff.

3. I understand about friction and iniertia. I don't have much friction, but I have a good bit of inertia (11lbf load). But, I ahd already considered these and so I did try to disconnect the motor, ( as I described in previous post above), and the behavior persists with nothin connected to teh motor but an empty timing pulley.

4. The bump (jump back) observerd happens on decel and accel, but oddly enough, only manifests itself mainly at one end of the travel. SO, figuring I was stopping on some partial microstep, I even attempted to use a total travel that was in multiples of 16 to avoid any microstep step problems.




I have hooked up a larger ("double stack") motor that is a 1.8 deg motor and it seems to eb working better. But it's size is a problem, I'd much rather get the smaller motor to work properly.

[Olav Kahlbaum (TRINAMIC)]

Hi guys,
I have tried it out with Scott's settings. In deed the TMC428 (this is the chip that controls all the ramps, as ramp generation is done in hardware) handles decelleration a little bit different than the acceleration phase. You will see this especially when you are using a low acceleration parameter and pulse_div/ramp_div values that differ very much from each other.

So I have also tried using the following settings:
pulse_div = 4
ramp_div = 7
acceleration = 85
max_speed = 1400

I think that accelleration and decelleration are more equal then. When changing the pulse_div, ramp_div and microstep resolution values please bear in mind to set the acceleration as the last parameter always, after you have set the other parameters.

But I also think that Scott's problem is mainly a problem of resonance. You are right, the TMC428 (that is used on nearly all the TMCM modules) can only do contiguous ramps. The only way to overcome this would be to use a TMCL program which does some polling on the speed and bypasses the reonance speed ranges (but this is hard to do, as Andrew already mentioned). But in most cases you will find suitable parameters so that resonace will not become a big problem.
Finding the resonance speeds with the help of the StallGuard profiler is also a very good idea.
Please also bear in mind that when moving only short distances the maximum positing speed might not be reached.

[Olav Kahlbaum (TRINAMIC)]

Hi Andrew,
the easiest way for "turning off" accelleration and decellaration is to set the accelleration parameter to its maximum value (2047) and the ramp_div parameter to 0.

[shorton]

Andrew:

Very interesting, but a lot to get a handle on your meaning. I do not understand some of it, but will try to study it.

[shorton]

Andrew:

As I continue to test I have found that no matter what I use for current, if I use your p correction AND I use the minimum speed, that the deceleration is MUCH smoother. It does take both of them. I still get the chirp/growl blip on start. If I had a way to get the same jump at start that would be nice.

Also, as hard as I try if I lower the current to anything less than the max the mfgr said I coudl use withthe motor (1170), it is much too easy to stall. The "stick my finger in it" test reveals a mouse could make it stall. At the higher currents is ls less susceptable.

Over the next few days I will replace this 400us/rev motor with an alternate "high torque" doublestack 200us/rev motor and try this same battery of tests. It may cure the issue as it will have more torque at the lower currents.

As I asked in the recent reply, I was a little fuzzy on yur calculation of the value params 137 and 146. For a given set of "regular" params, can you clarify how to calculate them?

If there is a way to get a similar skipping of the resonant speeds on start that would be great, too. You have mentioned it and I discussed it above, but I do not see how to do it, yet.

Getting close. Calling it a day.

Best, Scott

[lungfish]

Andrew, Shorton!

I got a twelve axis automaton featuring 4 TMCM303 here and have run into each and every problem Shorton describes.

My machine includes an arm/table positioning unit (like a harddisk), the arm can only move a short way, which prevents extensive stall guard frequency tests. Alas, this axis gives me a bad time.

Other axis include ball bearing / thread rod linear units.

I hope one of these days I will (with your help) derive a solution which gets rid of melodical deceleration, slow speed crawling into the target position, resonance.

Since my machine features different axis types, I get the feeling that you two are discussing the basic issues one will encounter whenever using step motor drives.

The most interesting part is to determine which issues are originated by TMC and which are intrinsic to step motor designs.

Please keep enlightening us!

[Olav Kahlbaum (TRINAMIC)]

Concerning motor current: It is often a little bit confusing, but most motor manufacturers give the maximum RMS current for their motors, while manufacturers of stepper motor drivers (like us) give the maximum peak current for their drives. SAP 6 and SAP 7 set peak currents. The formula is: RMS=Peak/sqrt(2).
So, when the motor manufacturer says that the maximum (RMS) current for his motor is e.g. 1A, you can set SAP 6 to 1400 or even 1500 (1A*sqrt(2)=1A*1.414 which is approximately 1400).

[shorton]

Andrew:

I am trying to run the calculation and usage of "p" in your examples to ground. I am using the following sample code. It has all the parameters that have been involved, and I believe I have them being set in the proper important order:

[shorton]

As I see it from Andrews help and my experiments, I see that for acceleration I have to find the best copromise. I can:

a) start abrubtly, no acceleration. The motor and mechanics I'm using will do it, but this produces a lot of impact/shock load on my device, the load but most importantly for my application, the mounts. Noise from 0 to v_max is minimal.

b) start with steep acceleration, not much different that A above in that I stillget significant shock on the system from the abrubt start. More, but not much acceleration noise

c) start with a medium acceleration ramp. Smoother takeoff impact wise, but noiser as it goes through the slower more resonant frequencies. I can fiddle with pulse_div/v_max and ramp_div/a_max to try to find combinations that skip the most resonance points but this hasn't helped much.

d) adjust currents in all of above. I found a low currents I have beautiful, silky smooth, quiet takeoff's but they lack enough torque to prevet a stall in anything other than perfect setup conditions.

Anyway, I think I can work with that to find the best compromise for acceleration.

One thing that is not clear to me is the parameters for the acceleration threshold. Does this parameter (#136) apply different current reductions to specific moves requested with specific acceleration values, the value of which willbe either above or below the threshold for the whole move, OR does it apply to different velocities passed as part of a single acceleration move? I have tried adding a maximum current reduction (7) to both high and low params (#144/145) but I can see no affect whatsoever.

Now, as we have determined and has been verified, decelleration does not behave the same as acceleration. The ramp is different. Even when I find smooth(er) acceleration setups, I still get a growl to stop.

So far my only solution as been to apply a v_min (#130) setting to the setup. This can cure the noise, but it produces an abrubt stop introducing some shock load. Like acceleration, I need to reduce the shock load and the noise. That is the next hurdle.

One thing I considered was reducing the max current but only during the deceleration. This might be possible using Andrews code in the thread "changing speeds on the fly" only instead of changing speeds, I'd change currents. Feasable? Stopping ought to be easier than starting and resulting in less current requirement, less niose?

Thanks, Scott

[shorton]

[B. Dwersteg (TRINAMIC)]

Dear Shorton an Andrew,
Olav and me have followed this discussion in many points, albeit not giving much input. From my reading of the last posts, I still wonder why your application loses steps. I see you have tried out "SAP 203,-1" for mixed decay on? Please retry with the latest version of the module firmware / with your actual double stack motor. The next question is: Motor coil resistance, and supply voltage, and at which RPM you first loose steps, to get an impression, if there is an electrical problem. The text reads like mechanical vibrations are leading to a high torque ripple from your dynamic load, thus stalling the motor. To get a real impression, the best would be: Can you make a short movie with a digital camera which shows the problems? If possible, please mail to Andrew and to us (please refer to our contact address given on the Trinamic website).
Thank you!

[shorton]

Thanks. I am not missing steps anymore, as far as I can tell. I judge that from the repeatability of the movement. The motor that I originally started with did as described, that is, had a more specific point where it would skip steps. I set that motor aside for another day. The motor I am using now is a higher rated torque model, with 200us/rev and it does not have a severe resonance point like the original motor. My apologies for changing horses in the middle of the race if it added any confusion about my solution. I do still get a rougher sounding decelleration which is what I'm trying to address now. I'm away form the lab most of this week but will try to make a clip of the motion and post for your guys to download asap. Thanks again, Scott