When is PID too high??

[Lasershop]

dspMC / Mach3 conversion on a V2XT. Tuning the axis motors (servo) and the machine wouldnt do the same thing twice. Started out at "P" 4000 and "D" an times 25 factor which is 100000. Motors run very smooth but going back to zero is all over the place. All ways and screws are in very good condition. Lube pump is working. I have the P now at 25000 and the D times 25. Seems alot more accurate within 2 or 3 counts. Scale has always been 100. What do I do, just keep going? Or am I way out of typical range as it is. Stock SEM motors and 250 encoders. Anything less than 2500 P wont move the motors. Your thoughts??
RonO

Also, motors have a tach but I have the amps in torque mode right now. Still learning

[H.O]

Hi,
I don't have any experience with the DSPMC or your particular setup but you don't say anything about the I-term... If you end up short or slightly overshooting targetposition (and not getting back) you probably need to increase I for it too move that last bit.

[Lasershop]

The "I" has been at 0. What do you suggest? Increase "I" one at a time or more?
Thanks, RonO

[Lasershop]

That was just weird.. with the "I" at only 1 the motor would go like 20 counts very slowly after a stop.

[H.O]

Are you saying it stopped at "target" and then moved 20 counts "away"? If so, that sounds strange. If it stopped "short" and then moved 20 counts "towards" the target it's working just as planned but you need to increase it further.

The I-term looks at the error over time, if a discrepancy between target position and actual position exists the I-term keeps adding more and more "effort" (torque command in your case) untill there no longer is an error. Too much I can cause large overshoots or even oscillations so be carefull, you might also need to increase the D further to dampen it when you're "getting close".

[Lasershop]

Originally Posted by H.O Are you saying it stopped at "target" and then moved 20 counts "away"? If so, that sounds strange. If it stopped "short" and then moved 20 counts "towards" the target it's working just as planned but you need to increase it further. The I-term looks at the error over time, if a discrepancy between target position and actual position exists the I-term keeps adding more and more "effort" (torque command in your case) untill there no longer is an error. Too much I can cause large overshoots or even oscillations so be carefull, you might also need to increase the D further to dampen it when you're "getting close".

Came to a near stop and "continued" very slowly 20 or so counts.

[electric2u]

can someone tell me if the moters on this web page good or bad? http://www.probotix.com/index.php?vi...&product_id=33

i'm looking at using them on a x 36" y 80" z 6" made out of MDF i well be working on doors and other things i think of to do with it.

or what size do you think they may work on?


thank you

[TOTALLYRC]

Originally Posted by Lasershop dspMC / Mach3 conversion on a V2XT. Tuning the axis motors (servo) and the machine wouldnt do the same thing twice. Started out at "P" 4000 and "D" an times 25 factor which is 100000. Motors run very smooth but going back to zero is all over the place. All ways and screws are in very good condition. Lube pump is working. I have the P now at 25000 and the D times 25. Seems alot more accurate within 2 or 3 counts. Scale has always been 100. What do I do, just keep going? Or am I way out of typical range as it is. Stock SEM motors and 250 encoders. Anything less than 2500 P wont move the motors. Your thoughts?? RonO Also, motors have a tach but I have the amps in torque mode right now. Still learning

Use of the tach in velocity mode is supposed to give better low speed response.

You say that returning to zero is all over the place, what does that mean exactly and what are you using for a deadband? I typically use 3 or 4 but that is me.

Different encoder counts will affect how the pid reacts. When I was setting up a new encoder to replace a bad one I had it set for 48 counts and with the existing pid settings the axis would bounce all over the place when jogging. When I say bounce, I mean bounce to the tune of a half inch or so. Real freaky to watch.

I currently have 500 on the x and z and 1024 on the Y. Since the dspmc will handle much higher than that I am planing on turning the encoders all the way to their max of 2048 (8192 in quad) and retuning. This will give ~81920 steps per inch on my machine.

Originally Posted by Lasershop The "I" has been at 0. What do you suggest? Increase "I" one at a time or more? Thanks, RonO

Remember the scale factor of 100 which means that your I term of 1 is actually 100.

Originally Posted by Lasershop That was just weird.. with the "I" at only 1 the motor would go like 20 counts very slowly after a stop.

Same thing I found when setting mine up. The I term seems to slow the settling time of the servo loop.

Originally Posted by Lasershop Came to a near stop and "continued" very slowly 20 or so counts.

I have not yet tried it but you could try using an I term of .5 and see how that works.

I have spoken with other people who are using pid terms much higher than yours.

I would say that as long as it works there is no right or wrong. I provided the numbers in the manual because like you I found that very low numbers result in no axis movement at all. The provided numbers give a place to start that will provide movement and get you on your way to tuning the motors.

Using your numbers, how does it look in axis works?

Mike

[H.O]

Came to a near stop and "continued" very slowly 20 or so counts.

And after moving those 20 or so counts it was closer to the target position than after the the initial near stop, yes? If so, give it some more I and see if it get's there faster.

If the tuning software provides a way to plot the response, which I believe it does, please post a screenshot.

[Lasershop]

I bench tuned the Servo Dynamics amps per the manual and double checked before I installed them. Same amps that were in the machine when it ran under the stock Bridgeport motion. I made my PID adjustments using Axis Works and can get the motors to run smooth and kept adjusting until I could move forward and back 6 inches (100 ipm) and always land at 0. That took a couple days. Machine is rated at 250ipm.

Under Mach 3 I can jog just fine but "return to zero" is random at best. Different velocities, different accelerations. "All over the place" means it never lands at zero and doesn't do the same thing twice. I'm getting closer with the high P and D numbers. I can now keep the return to zero within 3 or 4 counts. Is that as good as I can expect? I just don't know if there is a ceiling on PID numbers to where there is no longer any affect. I started as per the dspmc manual with P at 4000 and D at 100000. I at 0.

I'm now (on my X axis) P at 25000 and D at 625000. Scale always 100. Its much better than before retuning to zero within 1 to 2 counts. Going to recheck all my grounds this morning and make sure they are all done right. I haven't got my post software set up yet but I worry about a program with hundreds of moves and all the accumulative error.

[Lasershop]

Originally Posted by H.O And after moving those 20 or so counts it was closer to the target position than after the the initial near stop, yes? If so, give it some more I and see if it get's there faster. If the tuning software provides a way to plot the response, which I believe it does, please post a screenshot.

Ill get you that screen shot this morning sometime.
Thanks, RonO

[Lasershop]

Originally Posted by TOTALLYRC I currently have 500 on the x and z and 1024 on the Y. Since the dspmc will handle much higher than that I am planing on turning the encoders all the way to their max of 2048 (8192 in quad) and retuning. This will give ~81920 steps per inch on my machine. Mike

What do you mean "turning the encoders all the way to their max of 2048"?
and "500 on the x and z and 1024 on the Y"? Are those PID settings? Thats almost my original question whether there is a max setting.