Re: [CAD_CAM_EDM_DRO] Major encoder trouble, need advice.

Henrik Olsson wrote:

> I agree, it 'has' to be the encoder, but of course USDigital
> disagrees.... Yes, all measurements on the encoder has been done with it
> connected TO and powered BY the UHU drive. I can't see the encoder
> count but I can see if it detects invalid counts and I have tried, with
> the motor power supply turned off, spinning the motor shaft "by hand"
> from very slow to as fast as I can and I don't see any reports of
> invalid transitions then. I'll try to rig something up so I can spin it
> with a cordless drill or whatever.
>
>

Ahh, that's very interesting, but I don't think speed has anything to do
with it. I think it is noise.

> After solving another issue with the drive I tried it with the AMT
> encoder and 15m of unshielded cable lying in a ring right on top of the
> motor and drive and run it for over 90 minutes without trouble. So if it
> is noice - which it indeed looks like - the USDigital encoder must be
> overly sensitive.
>
> USDigital recomends the 26C32 receiver while the HP-UHU has the
> SN75115... big difference?
>
>

VERY unlikely. An awfully complicated chip, with strobe, differential
output and response time control. Maybe there would be some benefit in
slowing it down with the response time control pin to filter out
extremely short pulses.
All my encoder counters have digital filtering in the FPGA to ignore
transient invalid transitions.

> I'll definetly try with a 4.7uF at the encoder. Also, on the drive PCB
> the low voltage GND and the motor supply GND is separated but connected
> to each other at one single point. I wonder if cutting this trace and
> connecting the -ive of both supplies outside the board will help or make
> it worse...
>
>

Without seeing the exact details, I can't be sure. The IR FET drivers
have some serious limits on the excursions of the low-side FET source
pin to the logic ground. Generally, they can't tolerate more than about
-5 V on the source compared to logic gnd. So, you usually want to keep
this connection between logic and power ground quite short. But, there
are these other concerns that may argue in the other direction.


Jon