Major encoder trouble, need advice.

Hi guys,
I have a problem I've been working on for several weeks now and I need any
help I can get. I appologise in advance long post but I'll try to cover what
I've done and what I've tried. But first a bit of background....

I'm aiming to use the new HP-UHU drive to run my motors and after quite a
bit of struggle with various issues I now have three drives that seems to
work pretty good. The problem is with the encoder signal.... The drive has a
counter that increments for every invalid transition it sees (or think it
sees) on the encoder input. Invalid transitions can come from overspeeding
(the CPU can't keep up with the frequency) or from noice.



Now, I've tested the drives and motors with a 3600 lines encoder, (14400
counts / rev) and it works fine up to around 541rpm with that encoder then
it starts to report invalid transitions. I knew the bandwidth was limited
from the get-go so that's fine - now I know where the limit is and that the
*drive* AND the differential interface (based on SN75115) on the drive
works. (Initially it didn't work due to a problem with the drive but now it
does.)

Motors are rated at 2000rpm so I settled for USDigital E7P, 625 lines with
differential output. With my mechanics this will give me 0.001mm theoretical
resolution and still stay inside the envelope of both the drive and Mach3.




OK, finally to the point. With the E7P in place I can not get it to run
reliably at ANY speed. The setup is the same as what was working with the
3600 lines encoder, same powersupply, same motorcable, same
control-computer, same motor just another encoder and the same grounding
scheme. The motorcable is screened and screen connected to motor supply
-ive. Encoder cable is screened twisted pair (only 300mm long for now) with
screen connected to logic level GND.

Each differential pair is terminated, where entering the drive, with a
120ohm resisitor in series with 4.7nF capacitor as per USDigital
instructions. It was impossible to get it to work reliably so since the
encoder was new I turned towards the drive which has had it's issues. I
scoped around, swapped the differential reciever IC, swapped drives, and
generally tried whatever trick I could come up with but nothing helped.

I've tried with and without the encoder cable shield grounded, with and
without termination, with only the 120ohm resistors as termination. I've
tried bypassing the differential reciever circuit using the A and B channels
only. I've tried running the A and B channels to homebrew differential
transmitter and running THAT to drive. NOTHING seemed to help.




Then I mounted an AMT encoder (you know those capacitive ones with
adjustable resolution that Digi-key sells) my second motor and gave it a go.
NO PROBLEMS, cycled thru my testprogram for 45 minutes without trouble so
the drive IS INDEED OK.

I contacted USDigital and they suggested that I change encoder to a new one,
might have damaged the disc while installing it.... Since trying everything
else I gave it go. I VERY carefully installed the new encoder (electronics
AND disc) with the supplied mounting tools to get correct distance etc and
checked runout of disc - it was 0.02mm, can't be much better than that.

100% sure of having fixed the problem I gave it go. Couldn't even jog for a
couple of seconds without getting into trouble. I even lowered the motor
supply voltage from 130V to 30V and run the motor 50revs at 20rpm and it
STILL acted up. I can't imagine it being noice then....

I'm getting desperate here. USDigital says it's not the encoder since I
changed that. I say it's not the drive since it works with another encoder
and it's not the motor since it used to work before mounting the E7P. But
what the heck is it??

I've checked the output of the encoder and it looks fine. Nice sharp square
pulses, 50% duty cycle, 90deg shift. Each complementary pair is always the
inverse of each other etc. I don't own a logic analyzer so it's REALLY hard
to detect if a single pulse "dissapears".

Again, sorry for this really long post, even so I probably forgot to mention
something. I would really appreciate any pointers you might have.

Sincerely,
/Henrik.