A paucity of evidence proves challenging to an engineer investigating a servo drive gone a bit haywire
By William Ketel, Contributing Writer
The circumstances leading to my involvement in this case were a bit unusual, to say the least. I was called to service a machine that I had never seen and knew nothing about, in a place that I had never been, by people whom I did not know.
It seems that I had been recommended by a co-worker from our previous place of employment, Sun Electric’s Industrial Systems Division. He was a chap whom I did recognize but had not really worked with since he was a machine builder and I was a controls engineer. The businessman who had called me had an “alley shop” that you really did have to enter from an alley!
The alley shop had sold a life test machine to a truck parts OEM development laboratory, and now it had developed problems. So I went to the laboratory, located on a beautiful engineering campus, and located the individual who had called and asked for service on the machine. I asked the operator to explain the problem. He told me: “Every once in a while the thing would not cycle right.” I asked about any possible drawings or information on the machine and learned that there were none.
I examined the machine, which was being used to life test prototype window regulator mechanisms. They were quite expensive, since they were made in a machine shop one at a time. The tester consisted of a servo motor and amplifier package set up as a variable speed drive. There was a shaft position encoder on the motor output, driving an electronic up-down counter with multiple preset outputs.
The test consisted of driving the input shaft of the window regulator one direction at a fairly high speed, then at a preset number of counts, slowing down and driving the regulator shaft to the end of travel at a preset count value, where the torque limit on the servo drive would sense the load increase and stop the drive. There would be a very short delay, and then the system would reverse and drive in the opposite direction, with a similar slow approach in the last very short distance to the end of travel.
The failure mode was in not slowing for the approach, but rather banging into the stop at full speed, which made the test non-valid. Not surprisingly, it also often broke the test parts. The problem in servicing was that this would not happen very often, and almost never while I was watching. The machine failed this way once and then worked perfectly for the next three hours as I watched.
No one was willing to pay me to sit around all day and watch the machine work perfectly, so I left, and tried to get more information from the alley shop. It seems that the circuit was created by a panel builder who then moved to Texas before producing any drawings.
Three days later I was called by the engineer at the lab. He informed me that the tester was malfunctioning a bit more frequently this time, and he wanted it fixed. When I arrived, the system was running, and it did malfunction several times. In addition, this engineer (whom I had met on my previous trip), pointed out that the tester was simply not changing to the low-speed setting when it malfunctioned.
So now I studied the circuit by tracing out the wires a bit. Suddenly the problem became clear and obvious: The servo speed was set with a potentiometer supplied with the servo amplifier. Ten volts DC was applied to the outside terminals on the high-speed potentiometer, and the wiper supplied the speed set-point voltage to the servo amplifier.
The counter operated an Allen Bradley 700P series relay, which, when the contacts closed, placed another potentiometer wired as a rheostat, between the wiper terminal and the supply common, which would load the set-point voltage down to the low speed value. The contacts in that relay were rated 10 amps and 600 volts.
The signal that was being switched was about 2 volts, and the current was about 1 milliamp. The result was that the self-cleaning function of the contacts did not work, resulting in an occasional high-resistance closure. The quick cure was replacing the power contact with a reed contact module.
That was the end of the problem, the customer was happy and I got paid.
Contributing Writer William Ketel is a hands-on electrical engineer who enjoys troubleshooting and diagnostics and Ham radio,(extra class). His industrial machine projects range from an evaporator valve calibrator to a brake drum inspection machine to crash sled controls, and a package to calibrate developmental crash sensors.
Tweet This
12 
