Initially, when I thought I was going to use the VSD-A from Granite Devices, I
fabricated some brackets and extensionshafts so I could
mount some really nice SCANCON encoders I had to the
motors. These encoder have differential linedriver and
3600 lines resolution, 14400 counts per revolution in
quadrature - perfect for the VSD-A.
This worked very nice
with the VSD-A but because of the limited acceleration I
switched to the HP-UHU drive which I found, during
testing, had a max encoder count frequency of ~130kHz.
With 14400 counts rev that would have allowed me a
maximum speed of 541rpm, or rougly 1/4 of the motors
rated speed - not very good.
I decided to get new
encoders, I settled for the USDigital E7P because of
three main reasons. The E7P is available with
differential output, it's small enough to fit inside the
already fabricated covers and it was availble with 625
lines of resolution which would give me a nice round
figure of 1000 steps/mm on the X- and Y-axis, 2000
steps/mm on the Z-axis and stay well below the 130kHz
frequency limit on the HP-UHU. (2000rpm * 2500 / 60 =
83.3kHz). Can't go wrong with that can you?
Well, to make a very
long story a whole lot shorter the USDigital E7P encoder
simply refused to work reliably with my setup. The UHU
servo chip has an internal counter that keeps track of
any invalid transistions on the encoder input. For
example the the state of the A- and B inputs can't go
from 0 0 to 1 1 with out "going thru" either 1 0 or 0 1.
If that happens it means that we longer know for sure
where the motor is. This can happen for a few reasons.
The frequency (ie the motorspeed) is to high so the chip
can't keep up or it could be noise coming into the
system causing false readings. I knew it wasn't a speed
issue because I had verified the max frequency when
testing with the SCANCON encoders. Besides it didn't
matter one bit if I tested at 1000rpm or 10rpm it still
lost position. So it had to be a noise problem then.
During the course of
several weeks I tried every trick in the book, and some
that are not in the book to get these encoders to work.
I tried with long and short cables, twisted and
untwisted pairs, shielded and unshielded, many different
grounding strategies regarding the shields and the
common point for the motor power supply and the low
voltage supply of the drive. I tried with different
voltages for the motorpowersupply, anything from 130VDC
down to 40VDC.
I tried feeding the
encoder from a battery powered supply, I tried bypassing
the differential receiver in the drive, using only the
A- and B-singals (not the complementary ones). I tried
with a homebrew linedriver circuit, I switched motors
and I switched drives - no matter what I did I got the
same results. Drive complains about invalid transitions
on the encoder inputs with loss of position as result -
confident it was a problem with the actual encoders I
tried two other encoders I had around. One Hengstler and
one AMT102, both worked beautifully although the
Hengstler had 3600 lines so the speed had to be held
down. I tried a few of the different grounding and
shielding strategies I had tried with the E7P and it all
worked, even with unshielded cable. Yet nothing worked
with USDigital E7P encoders.
During all this
USDigital claimed that there had to be something wrong
with my setup since I could verify the integrity of
signals on the E7P with my scope - which I could. Rise
and falltimes looked OK, voltage levels OK, dutycycle
and phaseshift OK but never the less - when mounted on
the motor it simply DID NOT WORK.
After telling them that
my system worked with 2 other encoders but not with the
E7P they offered to take them back "for investigation"
but I had to pay for the shipping. At this point I was
so tired of it I decided to write it off as a lesson
ever buy USDigital crap again.
But now what, I still
needed enocders. I got pointed towards Renco Encoders
and found their R35i model. After E-mailing them it
turned out it was available with differential linedriver,
625 lines and it would fit inside the already fabricated covers.
After explaining I about my problems with 'another
manufacturers' encoders they assured me I would have no
such problems - guess what, they where right! The Renco
35i, the AMT103 the Hengstler AND the SCANCON all works
fine while the USDigital E7P doesn't. (Yes, I DID try more
than one E7P with the same results.)
The next photo shows
the final test setup with the Renco R35i encoder. The
encoder cable during this test was roughly 15m long, twisted pairs
as a test the shield/screen is left floating. The cable
was also wrapped around the running VFD you see on the
wall there. It was left like this, cycling thru a G-code
program for hours without any sign of trouble.