[Wavelength]

I have a pair of Yaskawa SGD7S-1R6A00A002 drives, SGM7A-02A7A21 motors (Sigma 7 series, 200W). At the moment I am bench testing a new setup, checking control-board to drive setup. The problem I am having is that as soon as the servo motor is enabled, the drives are emitting extremely annoying audible noise, at ~10.6kHz, and additionally creating substantial EMI, such that surrounding equipment (control board, laptop) functions poorly, if at all.

Details:
- The problem is only when the servo motor is enabled. Servodrive on, but motor not enabled there's no issue.
- If the servo is enabled but the motor isn't plugged in (drive doesn't seem to know this as long as encoder is hooked up), there's no noise or feedback.
- The motor power wiring is to spec, as far as I can tell (4-conductor 20AWG wire, shielded. Motor grounded, wire shield grounded).
- Motor power wiring is as short as I can make it (about 75cm). Originally was wired to my whole spool, ~4meters; reducing to 75cm seemed to make no difference.
- No tuning parameters seem to have any effect on this. The highest I can even set a notch filter is 5kHz.
- Motors seem to move fine, no noise/vibration. Movement or load have no discernable impact on the audible or EMI noise.
- Both drives do this, both motors do this.
- Drives are powered by 230V single phase, provided by step-up transformer from 115V mains (230V single phase is allowed for these drives, and the parameter for it has been set)
- I acquired both drives and motors used. Claimed to be off working machines, do seem to be in good shape.

Really hoping someone here can point me in the right direction with this issue! This is my first foray into servos, so I'll take no offense to any suggestion; there's every possibility that I've overlooked something basic. However I'm beginning to worry that this is just the way it is with these drives, which would render them very expensive, unusable hunks of metal.

[EinarH]

Did you find the cause for this? I have the exact same problem. Used sgmcs 02B and 05B motors and a supposedly new/unused SGD7W-2R8A20A700.

~10.5 kHz noise only when motors are active. It is very annoying indeed.

[Wavelength]

Nope, I have not figured it out, and it is really damping my enthusiasm on otherwise very nice gear.

I did put a better quality line filter on the power line, and that helped dramatically with the EMI issues, but did absolutely nothing for the 10.6kHz whine.

I've found nothing to improve the audible noise. I have only one theory even, which is that perhaps the Servopack capacitors needs replacement. The Servopacks do track capacitor lifetime, and both are basically at 100%. I find that suspicious in a used pack, so I wonder if someone reset the time on them? However, I'm not willing to break open the pack and replace the capacitors without good reason to believe that that's the issue, and so far I've found nothing to elevate this beyond pure speculation. If your packs are new and still making the noise, that seems to be a point against this theory.

Since I will have water available to the machine already, I'm considering bolting the packs to a coldplate and putting them in a soundproofed enclosure. It feels ridiculous, but I don't see any other way to stand being in the room with them for any period of time without ear protection. And to think, one of the major reasons I went to servos was to avoid the noise of steppers.

[joeavaerage]

Hi,
I reckon you are going up the garden path.

I suspect the servos are dithering.

Lets say the commanded position is 123456 but the actual position (encoder) is 123457. The servo will try to drive that one step backwards, but it overshoots to 123455,
so now it tries to drives forward and overshoots to 123457....and so on. and it will dither back and forth at a high frequency and yet by so little amount you cannot detect
any movement, ie dither.

Most servos have a parameter that defines a 'Zero Error Widow', that is a parameter usually in encoder counts that if the servo is within the window, ie within a few counts, it will stop
trying to get even closer. It prevents dither. You'll have to read the manual and find the parameter in Yaskawa's terminology that means the same thing and try widening that window.

Another possibility is to reduce the integral gain a little and probably also the differential gain by about the same percentage. It could be the the servo is oscillating, again by such a small
amount you cannot detect movement, but none the less its oscillating. Excessive integral gain will do that.

Craig

[Wavelength]

Interesting; that is not something that it ever would have occurred to me could happen at such high frequencies. Pretty impressive, actually.

Appreciate the suggestion, Craig! I'll take a look at this when I have the chance to hook up the laptop and tuning software. Fingers crossed.

[Wavelength]

I have increased the error window to the point of significant slop before the servo tries to correct, and gains down so low the servo would be unusable. Neither had any effect on the 10.6khz noise.

I will again note that movement or load have no discernable impact on the noise. While I still don't claim any expertise in servos, I don't see how dithering or oscillation noise wouldn't be reactive in some way to what the servo is doing?

[pippin88]

What's the bandwidth of the servo drive control loop?

I would guess if it was dithering causing the noise, the bandwidth would have to be equal or higher than the sound frequency??
I could be completely wrong.

Most servos don't have 10kHz control loop bandwidth.
Delta servos are 2.5-3.1kHz (current generation, depending on model).