Tue, 26 Feb 2002
Michael K. Brewer 

Anorad Electronics Broadcast Noise
==================================

  When I arrived at the IOTA site, I found that both Eltrol cards had
an oscillation at about 1 Mhz on the output lines to the motors.  The
amplitude of the oscillation varied from zero to about 10 Volts peak
to peak depending upon where the cart was in relation to the permanent
magnets embedded in the track. The oscillation was radiating and was
easily detectable by tuning a radio to 1000 Khz AM.  Since I could see
the oscillation throughout the Anorad system, I was not sure where it
was coming from.  I tried to kill it by putting a 0.5 uf capacitor
between the negative input and output of the op amp driving the output
stage of the motor amplifier on the C phase of the SD2 Eltrol card.
This killed the 1 Mhz oscillation but left a lower level (about 2
volts peak to peak) 4 Mhz oscillation.  Since there was no way that a
4 Mhz oscillation could get through the op amp when a 1 Mhz
oscillation couldn't, I decided that the ouput stage itself must be
the cause of the instability.  Wes had given me the number of Richard
Yaros, the person at Eltrol who designed the card, so I called him for
advice.  He agreed that it was probably the output stage that was
oscillating due to too much capacitance in the long (about 25 feet)
lines from the drivers to the motor.  He suggested that I increase the
capacitance from base to ground on the output stage from 180 pf to not
more than 0.01 uf.  He also didn't much like the capacitor I had put
across the op amp driving the output stage as it would lower the step
response of the system.  I saw that the newer (SD1) Eltrol had 1000 pf
capacitors from base to ground on the output stage instead of 180 pf
and was still oscillating so I went right to 0.01 uf on all 6 output
stages (A, B and C on both cards).  I also removed the 0.5 uf
capacitor that I had added previously.  The result was that both the 1
Mhz and 4 Mhz oscillations dissappeared and were replaced with a new
oscillation that retained the same amplitude (about 2 volts peak to
peak) but changed frequency from about 500 Khz to 250 Khz depending
upon where the cart was located relative to the track magnets. This
oscillation wasn't detectable on the radio, so we tried it with the
PICNIC camera and found that there was no longer any noise correlated
with the movement of SD2 or of SD1 when SD1 was moving toward home.
When SD1 was moving away from home, we could still see a slight (3-4
ADU) increase in noise.  I called Richard Yaros back and he now
decided that it would be OK to put small capacitors across the op amps
driving the output stages after all.  He suggested 100 pf on phases A
and B and 20 pf on phase C.  I tried this and it worked fine for SD1,
completely eliminating all oscillations.  Unfortunately, on SD2, while
the 500 Khz oscillation went away, the 1 Mhz oscillation returned.  I
slowly reduced the values of the capacitors on SD2 and finally
eliminated all oscillation with values of 20 pf on phases A and B and
8.2 pf on phase C.  This final step had no further effect on the noise
detected in the PICNIC camera.