Hurco BMC20 Dynomotion Retrofit

[TomKerekes]

Hi Troy,

So would speed mode be more likely to have larger position errors at higher speed? Im just thinking that it would need some error to drive output high just to cruise at a high speed without any load on it. Kinda sounds like a poor setup to me. Or maybe integral gain would correct for this in practice.

Yes Integral gain should do this. See:

Dynomotion

Your overshoot and lack of regenerative resistor could be related.

I agree. Without the resistor the drive may not have anywhere to put the energy besides its capacitors (which may be insufficient) so the drive doesn't have the capability to stop quickly.

But it may also be a tuning issue. If there is a negative error (ie lagging during acceleration) then later there will be a positive error (ie overshoot). The Integrator forces a balance between positive and negative errors. This is a consequence of having an average error of zero.

Higher gains (ie P and I) should reduce the initial negative errors and/or correct them along the motion to bring the average error back to zero before the deceleration.

As a last resort FF might be used to reduce the initial lagging.

Your small D Gain is probably not doing anything. Usually it needs to be a few times bigger than P to have an effect.

The Filters weren't shown so we don't know what they are.

We can't really see very well from your plots. At that scale the error appears to be nearly zero. Either zoom in, plot error, or better yet include the raw data so we can plot as we wish.

HTH
Regards

[Need TECH Help!]

Hi Tom,
If i adjust the I gain to .002 the axis becomes more unstable. I have not applied any filters besides the Drives Constant Torque Filter.
Here is the text data created by plot.

Troy

[Need TECH Help!]

So i done a quick test on the X axis using DMM drives as Step/Dir input and set Kmotion as StepDirCL using KFLOP JP5. Here is plot and data file from Step/Response screen. What would cause the Output to be the shape it is? Or is it normal? As axis movement was smooth and there was no overshoot when checking with an indicator. As a reminder, my setup has Servo motor directly coupled to ballscrew, no gear reduction or pulley reduction.

[TomKerekes]

Hi Troy,

The Step/Dir with the drive closing the loop is much better ~6X better (180 counts vs 30 counts of error). Even with much higher Acceleration. There is still some overshoot during deceleration.

This should indicate a regenerative braking resistor is not currently required to get good results. So I'm not sure why you can't achieve the same result in Velocity mode. It could be that the Drive velocity mode is doing something funny ie delayed or "smoothed"? You might try tuning it again using higher D gain. And with higher D gain which tends to make the output "spikey" a 2nd order low pass filter (1000Hz Q=1.4) might be used in the last filter.

BTW what is the resolution of your system?

Regards

[Need TECH Help!]

Hi Tom,

It could be that the Drive velocity mode is doing something funny ie delayed or "smoothed"?

This is what iam suspecting as its like iam fighting something else when tunning with analog. DMM told me when i was using there software for tunning that there "adaptive tuning" is always running in the background?? I emailed them to see if this can be turned off for trouble shooting.

It seems like when i tried a higher D value that output line looked like "fur" Will try it with the filter you described. Can i just simply change this on the filter screen and filter will be automatically applied when i download channel to KFLOP from the Config/Flash screen?And then use the Step/Response screen?

As for resolution, Ballscrews are 10mm/.3937 pitch and encoders are 16 bit. There is also gearing configuring in the drive.

Not to jump ahead to far, but if i go with step/direction, i will lose my other 8 differential inputs on Kanalog. Was going to use this for Z index. Can the Z index marks on encoder be connected to Konnect or Kanalog JP12 using an 200ohm resistor to bring down 5V signal?


Thanks,
Troy

[Need TECH Help!]

Ok. DMM got back with me about the auto tune in there drives. Said it can not be disabled, but it does not have any effect at all when in Analog, Speed Servo mode, only Main Gain and Speed Gain setting are in effect. And to check wiring connection at Kanalog. So i did a quick test by removing cable's braided shield drain (cable from drive encoder output to Kanalog.) from Kanalog GND and am getting a much different response from servo plus much quieter when sitting still and moving. There drive manual says the ground of encoder output connecter,d-sub, is at the shell of connector and to connect it to the controller ground. Was thinking of drain is why i connected the 2. So should i just leave off the drain at each end and connect a insulated wire from shell to Kanalog Ground?

Thanks,
Troy

[mmurray70]

Hi Tom,

It could be that the Drive velocity mode is doing something funny ie delayed or "smoothed"?

This is what iam suspecting as its like iam fighting something else when tunning with analog. DMM told me when i was using there software for tunning that there "adaptive tuning" is always running in the background?? I emailed them to see if this can be turned off for trouble shooting.

It seems like when i tried a higher D value that output line looked like "fur" Will try it with the filter you described. Can i just simply change this on the filter screen and filter will be automatically applied when i download channel to KFLOP from the Config/Flash screen?And then use the Step/Response screen?

As for resolution, Ballscrews are 10mm/.3937 pitch and encoders are 16 bit. There is also gearing configuring in the drive.

Not to jump ahead to far, but if i go with step/direction, i will lose my other 8 differential inputs on Kanalog. Was going to use this for Z index. Can the Z index marks on encoder be connected to Konnect or Kanalog JP12 using an 200ohm resistor to bring down 5V signal?


Thanks,
Troy

What multiplier is used in the drive? 16 bit encoder with 10mm screws would be 0.000006" per pulse so if you had 30 counts error thats less then 2 tenths without multiplier. I think id be pretty happy with that!

How small an error would you guys hope for with a system with that resolution?

[Need TECH Help!]

Hi mmurray,
I may end up with just using the Step/dir output as iam still having troubles trying to tune analog. And yes iam happy with the small error in Step/Dir mode . But was hoping to get Analog working so i could keep my other 8 differential inputs.
I took another look at manual and it states....
"Electronic Gearing ( GEAR_NUM Parameter )
Gear number (Gear_Num) is set from 500 to 16,384, default value is 4,096. Gear number provides an
electrical gear ratio: 4096 / Gear_Num, from 0.25 ~ 8.192. For example, if Gear number = 4,096, then
16,384 input counts from pulse will turn motor exactly one revolution. If Gear number = 500, 2,000 pulses
will turn motor one revolution. Gear number parameter is only applicable to position servo mode.

So looks that parameter doesnt apply when in analog Speed Servo mode. But the encoder output is scalable. ...

"This real time emulated encoder output is scalable using the LINE_NUM parameter set in the DMMDRV program. The pulse output per
revolution is set according to: Pulse output per revolution = LINE_NUM* 4

For example, if LINE_NUM = 2,000 then 8,000 pulses will be output per motor revolution. The Z pulse is output once per motor revolution.
LINE_NUM can be set from 500 to 2,048."

Dont know if this matters or how i should use it in analog speed servo mode. Manual is lacking.


Troy

[TomKerekes]

Hi Troy,

So i did a quick test by removing cable's braided shield drain (cable from drive encoder output to Kanalog.) from Kanalog GND and am getting a much different response from servo plus much quieter when sitting still and moving. There drive manual says the ground of encoder output connecter,d-sub, is at the shell of connector and to connect it to the controller ground. Was thinking of drain is why i connected the 2. So should i just leave off the drain at each end and connect a insulated wire from shell to Kanalog Ground?

I don't really follow this. Strange disconnecting a shield would have an effect on servo performance. I'd probably need to see a diagram to help. It didn't look to me like you were losing/gaining encoder counts before due to noise. I suppose connecting the two grounds through the shields could somehow introduce noise into the analog signals to the drive.

I'm hoping you can get the analog mode to work. It should work as well.

"Electronic Gearing ( GEAR_NUM Parameter )
Gear number (Gear_Num) is set from 500 to 16,384, default value is 4,096. Gear number provides an
electrical gear ratio: 4096 / Gear_Num, from 0.25 ~ 8.192. For example, if Gear number = 4,096, then
16,384 input counts from pulse will turn motor exactly one revolution. If Gear number = 500, 2,000 pulses
will turn motor one revolution. Gear number parameter is only applicable to position servo mode.

So looks that parameter doesnt apply when in analog Speed Servo mode. But the encoder output is scalable. ...

"This real time emulated encoder output is scalable using the LINE_NUM parameter set in the DMMDRV program. The pulse output per
revolution is set according to: Pulse output per revolution = LINE_NUM* 4

For example, if LINE_NUM = 2,000 then 8,000 pulses will be output per motor revolution. The Z pulse is output once per motor revolution.
LINE_NUM can be set from 500 to 2,048."

I'm afraid I don't follow that either. I don't see where you tell us your settings. But the simplest thing to do to be sure is to just verify the result by making a big move, measure the distance, check the change in the number of encoder counts, and divide. If you have different resolutions in the different modes we won't be comparing velocity, acceleration, errors, as apples to apples.

Regards

[Need TECH Help!]

Hi Tom,
Iam going to rework encoder cable so that one of the extra unused wires inside of cable will go to shell of drive encoder output and the other end to Kanalog ground then remove drain from drive end and connect it to Kanalog ground.
The Electronic Gearing and scale able encoder output reply was aimed toward mmurry asking about what electronic gearing i had. What i posted was from manual and i did not know how the LINE_NUMBER for scaling encoder output would effect analog tuning in Kmotion. I posted a screen shot a few posts back of the DMM tuning software that shows the current settings of drive. I have not changed any gearing for either step and direction or analog. I left them both as default which is 500.

Thanks,
Troy

[Need TECH Help!]

Well, i think iam just going to lose the the other differential inputs and go Step/Direction. I cant get anything constant when tuning analog. And between the lacking manual and DMM rep telling me that certain drive parameters dont apply when in Analog Speed servo mode and in fact they do, and not getting around this learning curve of analog, i feel like iam running in circles. Not to mention the slightest change in settings and i get a run away condition. Here is the latest plot. Getting around .02 or .03" overshoot. Its easily visible by watching motor.

[mmurray70]

Troy, Looks like the line number determines how many pulses per rev are passed to Kanalogs encoder inputs. Your setting of 500 x 4 would give 2000 counts per rev which is very low. Even the max line number of 2048 would give a maximium number of counts of only 8192 which is still kinda low considering your servos have 16 bit or 65,536 count per rev encoders.

I spoke with DMM last year about my lathe (that still hasnt be retrofitted) and they mentioned the high resoloution encoders were very important for performance. And it looks like if you use analog control you are stuck using a fraction of the encoder counts? Thats disappointing. Might be worth bumping it up to the maximum anyway. You will have to adjust tunning again after obviously, probably start from scratch again.

Are you using any filters or just playing with PID settings? Again, i didnt do the tuning on my machine so im not an expert, but briefly played with step response after the fact and the PID settings my machine runs with are completely unstable without filters applied so filters are definitely important. One more thing you might want to try is more I gain but add a lower integrator max limit to possibly help keep it stable. Im guessing that more I gain is needed for performance but you may have to tweak other things to make it stable.

[Need TECH Help!]

Troy, Looks like the line number determines how many pulses per rev are passed to Kanalogs encoder inputs. Your setting of 500 x 4 would give 2000 counts per rev which is very low. Even the max line number of 2048 would give a maximium number of counts of only 8192 which is still kinda low considering your servos have 16 bit or 65,536 count per rev encoders.

I spoke with DMM last year about my lathe (that still hasnt be retrofitted) and they mentioned the high resoloution encoders were very important for performance. And it looks like if you use analog control you are stuck using a fraction of the encoder counts? Thats disappointing. Might be worth bumping it up to the maximum anyway. You will have to adjust tunning again after obviously, probably start from scratch again.

Now that you explained it,that sounds like a "rip off". Will try setting to max.

Are you using any filters or just playing with PID settings? Again, i didnt do the tuning on my machine so im not an expert, but briefly played with step response after the fact and the PID settings my machine runs with are completely unstable without filters applied so filters are definitely important. One more thing you might want to try is more I gain but add a lower integrator max limit to possibly help keep it stable. Im guessing that more I gain is needed for performance but you may have to tweak other things to make it stable.

Was only adjusting PID. Will setting filters on filter screen automatically download to motor channel and will be applied to Step Response screen? Or?

[TomKerekes]

Hi Troy,

Was only adjusting PID. Will setting filters on filter screen automatically download to motor channel and will be applied to Step Response screen? Or?

yes it will. Select "Low Pass 2nd", Freq=1000, Q=1.4 then push "Compute" to determine the Z Domain Coefficients. After that any "Move" on the Step Response Screen First Downloads the Config/Step Response/Filters Screen Parameters then performs the Move.

The Low Pass filter should allow higher D gains without the Output being so "Fuzzy" as you described. The "Fuzziness" is caused by spikes in the output caused by step changes in the encoder position. Plotting the derivative (slope) of an input like a staircase results in spikes, For example with a D Gain of 25 when the Encoder Input suddenly (over the 90us sample time) changes by 1 count the output will spike up 25 DAC counts for 90us. The Low pass filter will smooth out the spike to have less amplitude but over longer time.

Measurement of current velocity is very crucial to good servo control. The higher the resolution of the encoder the better current velocity can be determined as the "steps" are smaller and there are more of them over a time interval.

With the Drive in Velocity Mode the Drive should be computing Velocity (hopefully using the full 16-bits of resolution?) and using that to maintain and correct velocity. So D Gain shouldn't theoretically be needed in KFLOP's Position loop. But depending on the characteristics of the Drive it may be helpful to add it.

Regards

[Need TECH Help!]

So I set line number to 2048 and added filter in kmotion. And slowly started adding I in kmotion until I got to .0009 Response was getting better with less overshoot so added a little more I to get a total of 1. Did test move of 10000 like before and servo jerked hard and started to runaway. So I rebooted drive, put I value back to. 0009 and tested again... servo jerked hard again and started to runaway. I am done with analog for these drive s . Thinking of trying to get my money back from DMM, also been waiting for almost 2 weeks for them to send me my 3rd drive. And still have not got back with me on resistors. Totally bad experience with DMM all around.

[Need TECH Help!]

Made a typo, I adjusted I for a total of .001 not 1.

[TomKerekes]

Hi Troy,

Sorry to hear your frustration. If you wish to pursue analog mode further (either Velocity mode or Torque mode) let us know.

Regarding the jerk/runaway: looking at the Step Response Plot would give clues to what went wrong (what was Output to the Drive? what did the Drive do based on Encoder Positions?).

Also setting a smaller Max Following Error should disable in the event of a Jerk or Runaway for whatever reason.

Regards

[m_c]

Troy, Looks like the line number determines how many pulses per rev are passed to Kanalogs encoder inputs. Your setting of 500 x 4 would give 2000 counts per rev which is very low. Even the max line number of 2048 would give a maximium number of counts of only 8192 which is still kinda low considering your servos have 16 bit or 65,536 count per rev encoders.

I spoke with DMM last year about my lathe (that still hasnt be retrofitted) and they mentioned the high resoloution encoders were very important for performance. And it looks like if you use analog control you are stuck using a fraction of the encoder counts? Thats disappointing. Might be worth bumping it up to the maximum anyway. You will have to adjust tunning again after obviously, probably start from scratch again.

Are you using any filters or just playing with PID settings? Again, i didnt do the tuning on my machine so im not an expert, but briefly played with step response after the fact and the PID settings my machine runs with are completely unstable without filters applied so filters are definitely important. One more thing you might want to try is more I gain but add a lower integrator max limit to possibly help keep it stable. Im guessing that more I gain is needed for performance but you may have to tweak other things to make it stable.

Even though the encoders are 16bit, the drive won't use that level of resolution at speed.
16bit encoders typically use serial comms, and having found a datasheet for a random 16 bit rotary encoder (http://www.beisensors.com/pdfs/chm5-...encoder_en.pdf) using SSI, the maximum clock frequency is 166KHz, it takes 21 cycles for each position read, so 166'000 / 21 gives a maximum of 7904 reads per second. Now unless you're only doing 60RPM, that means the drive is not getting the motor position updated as fast as a conventional incremental encoder, and the drive itself is using internal gearing to generate the emulated incremental output.

And to compare that to the KFlop internal servo loop, it updates every 90uS, or at a rate of ~11KHz, which is 40% faster than the above encoder example, however the KFlop essentially works in a similar way. Every 90uS, it reads the encoder position from an internal encoder counting register, compares to the previous position, and uses that to decide what needs changed/moved/updated.

The major advantage of Absolute encoders, is you don't need to search for the index pulse to establish the shaft position, plus you don't need to worry about losing/gaining pulses, as serial comms will either work, or not work. And even if you do lose a position or two, with good servo drive coding, things should be recoverable with no noticeable problem.

[Need TECH Help!]

Hi Tom,
I am not going any further with analog right now. Need to see if i can get predictability out of these drives with Step/Dir first, as i might just call the whole thing a loss and source some other drives and servos. DMM said they cant refund me and they think that the issue is my wiring.As i told them that when i adjust the Torque Constant Filter or On Position Range (DMM software) that motor will growl and run rough. Even thow they said these 2 settings dont apply to analog mode of drive.?? Dont know...maybe, But, i get the same growling issue when using there RS232 cable and no other I/O connection to drive. Its like these drives have a really fine line when tuning before they jump off the deep end.

There was other items i asked about also, but like every other email i sent them it is only have answered at best. They also sent me a wiring diagram and it was how i was already connected to Kanalog. Trying to get my money back on the 3rd replacement drive i bought and was sent the wrong model, and still have the wrong model almost 2 weeks later. What a mess all around with DMM TECH.
Plus they mentioned that there drives are used with Kanalog all the time???
Sorry for the Rant. This will be my first and last rant.

After i make new cables for Step/Dir i will report back how the drives are tuning.
It would be the cats meow if Dynomotion built there own drives and servos.

Thanks again,
Troy

[Need TECH Help!]

Here is plot of Y axis in Step Direction mode. Move is about 5.0". I have a filter applied. But filter didnt seem to change anything.

Troy