[phantomcow2]

I am doing a final physics project on resonance of stepper motors, going to be taking many trials with a program and finding out how much inaccuracy is caused by resonance. That's a very general overview. But, some of the drives I will be using has midband resonance protection. I need to give some explanation of how this actually works. Does anybody in here know?
If so, would you mind explaining how this is electronically suppressed? Or link me to an explanation already existing?
Thanks

[kreutz]

I am doing a final physics project on resonance of stepper motors, going to be taking many trials with a program and finding out how much inaccuracy is caused by resonance. That's a very general overview. But, some of the drives I will be using has midband resonance protection. I need to give some explanation of how this actually works. Does anybody in here know?
If so, would you mind explaining how this is electronically suppressed? Or link me to an explanation already existing?
Thanks

Look at this post:

http://www.cnczone.com/forums/showpo...&postcount=181

Many techniques are proprietary, in general terms, as Mariss point out, you introduce a compensating phase shift (+/- 60 degrees maximum) on the incoming step pulses in response to the phase difference between the input step frequency and the real rotor frequency. The rotor frequency could be measured by means of a encoder, BEMF zero crossing, or use the synchronous detection feature of the output switching measured on the current sense resistors and extracting the error information by means of a low pass filter followed by DC suppression (actually a bandpass filter).

I can't give more information on the procedure because it is a patented subject, but you can find out for yourself...

[phantomcow2]

Ugh, so getting information won't be easy. I've got a lot to learn about this stuff. Luckily I have 'till June to do it.
It's a fascinating subject. I've got some Pacific Scientific drivers that make these steppers sound like servo's, and then some basic hobbyist drives that make steppers sound like a chattering mill when the steppers are at low speed

[Madclicker]

Look at this post:

http://www.cnczone.com/forums/showpo...&postcount=181

Many techniques are proprietary, in general terms, as Mariss point out, you introduce a compensating phase shift (+/- 60 degrees maximum) on the incoming step pulses in response to the phase difference between the input step frequency and the real rotor frequency.

No, what he said was that they use a second order lead compensator made from opamps. That would add 180 degrees of phase lead, ideally with the zeros of the compensator placed very close to the offending poles of the system.

The rotor frequency could be measured by means of a encoder, BEMF zero crossing, or use the synchronous detection feature of the output switching measured on the current sense resistors and extracting the error information by means of a low pass filter followed by DC suppression (actually a bandpass filter).

All you need is feedback from the sensing resistor.

I can't give more information on the procedure because it is a patented subject, but you can find out for yourself...

Patents are public record and can be discussed freely.


I don't know for a fact, but I suspect that the new gecko drives are going to be compensated digitally instead of using opamps. That how I'm going to do it.

[phantomcow2]

So, midband resonance is caused by the actual rotor position being out of phase with the theoretical? I need to wiki some new electrical terms in my vocabulary now, thanks for the help.

[Madclicker]

So, midband resonance is caused by the actual rotor position being out of phase with the theoretical? I need to wiki some new electrical terms in my vocabulary now, thanks for the help.

No, the resonance is a symptom of the system that is passing through step frequencies where the system phase response is greater than 180 degrees.

The motor has 2 close in poles. Here is how they are modeled:

DC motor system model

The motor has 180 degrees of phase lag. Not a stability problem. The problem comes with the drive. It adds another system pole and 90 degrees of phase lag. This makes it possible for the open loop phase response to be greater than 180 degrees....an unstable system. The growling of mid-band resonance is just the physical manifestation of the instability. The real problem is that the current through the motor is being "choked off". That's what causes the stalling.

Forget the electronics for now. To write a paper about this you'll need to explain it at the system level. There are several links on the above page that will get you started.

[Madclicker]

Also, that link to the motor model is for an armature controlled DC motor. I haven't found the model work-up for a stepper yet.

[Madclicker]

Here's some work at Berkeley:

http://www.cs.berkeley.edu/~jfc/yzhu...287/node4.html

[kreutz]

Take a look at this:

[phantomcow2]

Thanks for hte links and comments folks. I've got a ton to learn by June. I don't totally understand resonance in a stepper motor, at least I don't think so. That's alright, my teacher told me we should be learning something of interest for this project.

[Madclicker]

Take a look at this:

Thanks for the link. I'm collecting them on this subject. I want to design a poor mans gecko drive.

[kreutz]

A little more information:

http://www.euclidres.com/apps/steppe...r/stepper.html

[Madclicker]

A little more information:

http://www.euclidres.com/apps/steppe...r/stepper.html

I saw and bookmarked that page in my journey. Nice pics of the stepper actually ringing.

I've had a tough time getting any kind of a model for a stepper motor, but I think I finally did today. The pdf that actually helped me was right here on the zone:

http://www.cnczone.com/forums/attach...8&d=1171941849

The missing piece I didn't have in the model was the damping factor of the 2nd order resonant mechanical pairs. It says here that .02 is typical. That's really low and would explain the stability problem. Other places I've found said that the resonance occurs anywhere from 70 to 200 steps/sec depending on the motor. That would also be hz and I can get radians from that. That gives me the complete model for the resonant mechanical pole pair except for the gain. The electrical pole is at R/L.

I'm going to continue this quest on my build log. Good luck with your paper and PM me if I can help.

[epineh]

How does midband resonance suppression work?

I suppose magic isn't a good enough explanation?...damn, I got nothin'

Russell.

[Madclicker]

Feedback control systems are a little magical. Understanding how they work requires a background in linear systems theory. The prerequisites for the first controls class I took was basically 3 years of engineering school.

In a nutshell, here's how it's done:

1. Model the system. (previous post)

2. Design a compensator. In this case it will be a second order lead compensator with a single pole far enough out to drag the motor electrical pole with it.

3. Implement it. In hardware, you can use opamps to implement the differential equations. In software or firmware you can use the equivalent difference equations.