Mesa 7i96s encoder input sanity check

I've hooked up a new higher-resolution encoder to my 7i96s and am getting a problem.

Encoder.04.velocity-rpm drops to zero, and encoder.04.count stops incrementing at about 900-1000rpm.  Velocity and counts are good below that speed.

Encoder is a US Digital H6 with the EM2 module. 10000 lines/rev, so 40k/ppr post-quad.  Differential line driver output, 5v.  Belt driven 1:1 from spindle.

US Digital specs indicate the encoder should be good for 720khz.  The output frequency should be 166khz at 1000rpm, (500khz at 3000rpm max spindle speed).

Is my math wrong?  Am I pushing the signal too fast for the 7i96s?

Is your encoder cable shielded and connected to ground on both ends?

I would probably switch to a 1024 lines/rev encoder for a spindle.

As a workaround you can try to disable the encoder filter. If it counts higher with the filter disabled, that might mean you have EMI issues at higher frequencies.

Example:
setp hm2_7i95.0.encoder.04.filter 0

filter

If set to True (the default), the quadrature counter needs 15 sample clocks to register a change on any of the three input lines (any pulse shorter than this is rejected as noise). If set to False, the quadrature counter needs only 3 clocks to register a change. The default encoder sample clock runs at approximately 25 to 33 MHz but can be changed globally with the sample-frequency or muxed-sample-frequency pin.

It's shielded and grounded at both ends.

I have a 600/line encoder but since this (eventually) going to be a c-axis as well, I was hoping higher-res feedback would help axis positioning.

EDIT - just turned the filter off and it stopped losing counts.  Now I need to see why I'm getting random velocity jumps - maybe need an input filter on the drive.

Thank you.

Don't ground at both ends, just ground the shield at the 7i96s. since the encoder is plastic it's probably better to ground the shield at the 7i96. Noise shouldnt really be that big of a deal, that's the whole point of differential signalling. Is this running from a VFD?

That encoder looks a little light duty to be belt driven, belts put radial load on the shaft. might want to support the shaft with a bearing if possible. I'd be curious to see this on a scope to see if it's actually noise or dropped pulses

blazini36 wrote:

Don't ground at both ends, just ground the shield at the 7i96s. since the encoder is plastic it's probably better to ground the shield at the 7i96. Noise shouldnt really be that big of a deal, that's the whole point of differential signalling. Is this running from a VFD?

That encoder looks a little light duty to be belt driven, belts put radial load on the shaft. might want to support the shaft with a bearing if possible. I'd be curious to see this on a scope to see if it's actually noise or dropped pulses
 

You're right - it's grounded at the Mesa only.  I'm not sure how the shield is connected internally to the encoder read-head... but it's obviously not grounded through the plastic case.

It's belt driven with 2GT timing belt - very little belt tension on it.  The shaft is mounted on double ball bearings internally, and I don't think I'm exceeding the max radial load, but time will tell of course.  I should have mounted the pulley on independent bearings and connected with a little flex coupler... but that would have meant actual work.

I tried capturing the signal with halscope, but I think it's moving too fast.  Really choppy signals on the display and I don't have a proper scope to capture the signals.

The RPM jumping was due to the programmable pull-up/down resistors in the drive.  Had the PWM+ input pulled up.  Once I set it to pull-down it's behaving very nicely.  Very steady RPM, within +/- 15RPM of commanded, and that's probably due to my belt ratio estimate being slightly off.

blazini36 wrote:

 Is this running from a VFD?

 

Forgot to mention... it's a Copley XSL-230-40 driving a re-purposed BLDC.  Motor is one of those 2-3kw (depending on who you believe) BLDC's that came on various Syil and similar mills a number of years ago.

I yanked off the halls and stuck a commutation encoder on it, then lied to the Copley about how many amps it'll handle.  I'll probably smoke it, but I've got a big Yaskawa with the same flange ready to go when that happens.

I did something similar with a little Chinese mill with a 1hp BLDC. I had it running off an 8i20, with an encoder on the shaft. Issue was the software BLDC component was a too slow to run it in velocity mode. It acted like a rubber band then the cutter hit the work

Is this machine in the shop with the plasma? I have a usb scope I tote around with my laptop for service calls, I could doublecheck the encoder signal for ya. You did set the jumpers on the 7i96s for differential right? Maybe it was just the filter slowing things down too much for a high res encoder but the card can handle something like 10mhz encoders.

Jumpers are set for differential.  And I ran it up to 2500rpm just now and the encoder/Mesa behaved fine with the filters off.

This particular problem-child is at my house, not the plasma shop, but if you do make it over this way (and aren't wiped out) you're welcome to visit my Pit of Despair.  I'm only 10 minutes from the 'big' shop.

Well that was definitely a good call by 0x2102, I wouldnt have even thought about the filter. Ive only seen it cause problems on a low res encoder running too slow to be counted, then again that's probably the only case where you need it on.