1
\$\begingroup\$

I am trying to read a resolver sensor attached to a stepper motor through a long cable (500 meters). More specifically with a resolver to digital converter.

However I get an error respect to the actual position like in the figure, error that is not present using a short cable instead of the long one.Figure

In the right figure, the error in degrees is represented against turns of the motor (1 turn=360 degrees). It is clear how the error is repetitive. In fact the error is very close to this one at different rotor speeds, and with different excitation signal (7.5 kHz) peak to peak voltage values.

Assuming that in the cable there is coupling between the sine and cosine signals from the secondaries, I can reach only an expression explaining a second harmonic error with the motor position, but not the first harmonic. And the error fits quite well with a 1st + 2nd harmonic of the motor position, as shown in the following figure: enter image description here

Any idea of what effect might be producing this kind of error?

\$\endgroup\$
  • \$\begingroup\$ I'm not sure I understand what error is happening on the longer cable. Do you have a picture of "short cable" error versus "long cable" error? Is the cable fully differential for both sin and cos waveforms? \$\endgroup\$ – Andy aka Mar 16 '15 at 13:50
  • \$\begingroup\$ The position read with the resolver and the resolver to digital converter is near zero compared to the position read with an encoder when a short cable (standard case) is used to connect the converter and the resolver. The 3 lines (excitation, sine and cosine) are twisted pairs but in the converter side the grounds are shorted together, as close to the converter as possible (in the converter manual is indicated to be done this way). I can take a picture but I don't think it will provide more info (maybe I did not understand correctly, but if it could be useful let me know). \$\endgroup\$ – rpicatoste Mar 16 '15 at 13:57
  • \$\begingroup\$ What about reading the position using a short cable, and using some suitable digital medium to transfer it 500m? \$\endgroup\$ – DThought Mar 16 '15 at 14:09
  • \$\begingroup\$ Unfortunately this is not possible, because the motor is in a high radiation area (limitations to place electronics). \$\endgroup\$ – rpicatoste Mar 16 '15 at 14:11
1
\$\begingroup\$

The presence of harmonics suggests simply that the waveform is being distorted. 500 meters each way is a long run, and even more so for 7.5 KHz. Assuming that the excitation is a nice clean, low-distortion sine wave, have you measured the amplitude of the converter inputs? I'd guess you need some sort of receiver amplifier to get the signals up to specified amplitude. If you already have such amplifiers, check to see if they are producing the distortion.

The obvious tool for this job is a spectrum analyzer. You do not state the harmonic distortion levels required to get your (excellent) error fit. If it's a few per cent, you probably can't see it on an oscilloscope.

So. Levels first. Actually, you need to check the amplitude levels at the resolver first, and adjust the driver amplitude to get the correct level. Then look at the converter and amplify as necessary. And note that if you've adjusted the drive level already to get proper converter level, you may be overdriving the resolver, and producing distortion that way.

You say the lines are shielded. Are they individually shielded or are they a multipair cable with a common shield? And while it's probably not your problem, be aware that in some cases where pickup is a problem, individual shields with an overall shield has been used.

There is an obvious problem with my advice so far, but I don't know how to get around it. Your cable runs are already in place, and due to radiation concerns you probably can't get to the resolver to measure its levels. In that case, you're just going to have to try blind experimenting with drive levels, monitoring the distortion levels of the returns with a spectrum analyzer.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.