Engineers are baffled by a sub-transmission line for an electric
company losing power at dinner time
I was working for a small distributed control system company selling
into the electric utility market. One of the voltage sensors we worked
with puts out a current related to the voltage on the distribution
line. The lines were typically between 7KV and 96KV.
The current source was extremely high compliance, and we designed a
product that could be powered by the sensors. Each sensor could supply
20W. We would double the signal to provide +/- 150V to run the system,
and the circuit was designed to always collect full sine waves. If the
voltage on the top cap exceeded 150V, it would skip power collection
and just short the current to GND. That allowed the system to be
powered by 120 and not collect power from the sensors.
One system in the Chicago area kept losing power between 5 and 7 pm.
Worse, these systems were controlling sections on the 96KV
sub-transmission line around Chicago and were blacking out an entire
neighborhood due to a low-voltage reading from the sensors. We had them
power the control from 120 and the problem persisted. The control came
back and we could find no problem. Worse yet, the new control they had
installed had the same problem.
There were only two hardware engineers in the company, and we were
both sent out to the customer site to see what was going on. The first
two nights the problem did not occur and all signals looked normal. The
third night was particularly cold, and the system started to misbehave.
I had a Tektronix 3K series 4channel scope and could see the voltage
sensors start to collect power in ½ cycles. The system sensed
positive going zero crossings to turn on or off the shorting circuit
for power collection.
The line had some big noise spikes that were causing the power
collection to shut down after only ½ cycle. This caused the
bottom cap to have very low voltage and the system would switch to
battery power. It also placed a DC bias on the sensor, causing it to
send less current and cause a low-voltage trip turning out the lights
in the neighborhood. I disabled switch operation, quickly moved one
probe to the 120VAC input and saw the top of the sine wave missing,
(Our units did not have PFC.) When the diodes in our supply began to
conduct, the voltage dropped. As power usage increased in the
neighborhood, the line voltage dropped and eventually the system would
start to collect power from the sensors.
The line the control system was sensing was driven by a transformer
designed to cancel third harmonics. I told the local linemen, their
supervisor, and his supervisor (our customer) that I could tell by the
waveforms that the system was sensing this special type of signal, and
that they must have a loose connection at the far end of the line
supplying 120 to the control.
They demanded to know how I could tell that. I told them that the
third harmonics showing up on the control power indicated high
impedance in the system especially considering that the source was a
transformer designed to cancel third harmonics. They told me
that there was more than one loose nut around, demanding to know how
I could tell all that by looking at the waveforms. I told them I could
prove it using the scope and they said welcomed me to go ahead and show them. Using the scope I
ran an FFT of the third harmonic canceling waveform and told them the
spectrum would be a mirror image for the signal on the 120VAC coming
I saved the first FFT plot to memory 1 and displayed it. I then
proceeded to run an FFT on the 120, and it was almost exactly the
mirror image of the saved plot. They sent a truck to check the other
end of the line, and we never heard about that installation again.
Author Robert Godes is CTO of Brillouin Energy Corp.