Set up using AMC AB15A100 drives, brushed DC motor Prototrak Plus with Encoders

I think I'd pick the 1.6uF one.

However I question if your switch settings are what you thought they were for the 1.267uF image. (Are you sure, maybe that was actually the 1.737uF setting?)

Anyway, it isn't just the overshoot at the top of the cruse plateau you should be concerned with. The overshoot at the end of the move when stopping is just as important.

I think it is time to try adding some P in Linuxcnc and see how small you can make the following error. (As the f-error gets smaller, you'll probably need to increase the gain for the f-error trace so you can see what is going on.) But keep the jog speed below what ever speed makes it jerk.

(Yeash, 20ipm seems very slow to me. I'm used to working with big wood working machines where cutting feeds are often 200ipm or more, and positioning feeds can be well over 1000. but I'll admit those are fast machines.)

Todd, thanks you for a 2nd opinion, I am complete good with trying some "P" & more (depending on how I make out with P) with the current settings. I will monitor the motor temp, because we are at the edge of the range with pot2. Later this weekend I will try the 1.6uF setting if necessary. As far as the potential mix up of 1.267 & 1.737 setting, if I revisit & easy to try, I will redo. I have read & will read again your recent post so I am shore where & what in the trace to keep eye on. If I read your suggestion correctly it will become necessary to mix units on the hal trace the f-error trace?

Many thanks

Todd, thank you! Attached are just 2 pic of the trace, notice the f-error units (in white) are different 5m/div than others at 100/div. I am going to need to study all the pic I have to understand how I am doing with tuning. Does the traces attached appear moving in the correct direction to you? Notice 1 of pics I adjusted FF1 from 1.0 to .96. How much can FF1 tolerate from 1.0? Feed rate is tweaked also. Read the file name closely, pot2 is back to 11.5 in some, because the motor had some vibrations that would not go away.

Many thanks!

Because F-error is not in units of velocity (it is units of position) and the fact that you're not really comparing it to the other traces, it's gain scale in Halscope is mostly irrelevant, other than knowing it to calculate what the actual following error is. In your first image the peak f-error is about 0.027" in the 2nd its a bit more than 0.015". So getting better. Keep adding P until you start to see instability, then back off a bit.

Todd, it is not acceptable to slow the feed rate, just to get f-error down, correct? I need to keep feed rate in the 30 ipm range or higher if it will move quicker.

Thanks

What the PID in Linuxcnc does is look at the feedback from the encoder and adjusts the velocity command to the servo drive to try to minimize the following error. The P term adds a value to the velocity command proportional to the size of the following error. So as the error grows the amount added increases. Increasing P should decrease the following error at all speeds, until it becomes large enough to cause instability. But the amount of the effect may be different at different speeds or unstable at one speed and not another.

The FF1 term adds a value to the command proportional to the commanded velocity, so if set to =1 it will add the amount of the velocity command. (This is why when FF1=1 and all other PID terms are zero, it simply passes the velocity command through open loop.)

PS: The problem is because of your drives jerk problem at higher commanded speeds, If P sees too large of an error it could cause a velocity command to the drive large enough to trigger the jerk problem. Then every thing may fall apart.

I think I'd work at tuning P at 20ipm for now. Get that as good as you can. Then try increasing the velocity to find the limit where things fall apart.

There is still something seriously wrong causing your jerk problem at higher speed commands. Hopefully the AMC tech will get back to you with some better ideas of what the problem could be. But properly diagnosing this may not be possible without a real o-scope.

Todd, this pic is on the verge of the motor starting to jerk. How much should I be thinking about backing off of "P"? P is not so large of a number.

Many thanks!

To honor A-M-C's suggestion it was to get a power supply to match the motor requirements. Am I clear in my thinking that if I can find a power supply 60 VDC; 7 Amp Min (what else should I be thinking about? wattage, 110v input); what I have seen toroid wound style of transformer in the power supply? I may be in for new power supply. Welcome a second opinion.

At this point your system is not capable of 30ipm velocity commands. I know in open loop you had it up to nearly 32ipm, but the servo needs some headroom to work. For now stick to 20ipm jog commands.

I thought that this once was a working system, and you are using the original motors and power supply? It should still be capable of moving the motors at the original speeds, unless there is something wrong with it. And since you don't seem to be seeing significant voltage drop under load, and the motor is only getting up to about half it's rated speed for the voltage available. It doesn't seem to me that the power supply is what is limiting you right now. (but I could be wrong, I don't really have any experience spec-ing a servo power supply.)

(Dumb question have you checked your servo motors brushes to see if they are in good shape?)