Re: Fw: Stg2, 2- problems

[Jon Elson <jmelson-at-artsci.wustl.edu>]

Fred Proctor wrote:

>
> > Encoder 1,3 and I presume 5 and 7 gives out, flipping values between
> > the exact counts and counts * 65537. So a movement of 0,001mm gives
> > out the value of 1 and 65537, a movement of 0,002 gives out 2 and
> > 131074... This can be seen on the original STG-module and with
> > stg2diag. The latter gives whit every flip a linefeed so the values
> > can be easily read. The value 65537 can be interesting; as 2 to the
> > power of 16 + 1 = 65537. I don't know of this problem is only
> > encountered with my board but I will ask STG if they know this problem
> > yet.
>
> I've never seen this. You need to be sure that the power supply that
> powers the encoders shares its ground with the Servo To Go digital
> ground. Otherwise, ground loop problems will cause noise and
> unpredictable feedback.

This isn't an encoder problem.  It is clearly the LSB is getting into the
wrong byte.  Are some shifts done to assemble the bytes (it seems
like it is 2 16-bit words from the encoder counter chips)?  It sounds
like the bit count for the shift is wrong, causing the LSB to get wrapped
around to bit 16.  Possibly the byte ordering is wrong when the board
is addressed at some particular I/O location.  One other possibility is
that this user entered something like 0x201 for his I/O address of the
board, when it is really at 0x200, causing all the words to be scrambled.
That is beginning to sound like the first place to look.

Jon