Maker Pro
Maker Pro

Post-mortem of SCR trigger board failure

P

P E Schoen

Jan 1, 1970
0
Yesterday I was at the facility where I do design work for high current
circuit breaker test sets, and they had experienced a rather dramatic
failure of an SCR controller board I had designed some time ago. The SCRs
are connected to a 480 VAC mains supply, and the output is a multiple
section high current transformer with one section connected to a variable
autotransformer (Powerstat), and the other sections are either shorted or
connected directly to the SCR output. The secondary consists of a heavy
copper bus bar rated at about 6000 amps continuous and able to produce
currents approaching 100,000 amps into a short.

In this case, they had installed a retrofit which replaced the
instrumentation and the unreliable SCR trigger can that was in the unit.
Part of the testing involves initiating the unit with an open circuit and
reading output voltage with various settings of the variable autotransformer
(vernier), and coarse taps from 1 to 9. Maximum output voltage is about 12
VAC. During this test, high surge currents were observed on the input from
the AC line, and it would trip a 30 amp service breaker on tap 2 where the
second section is connected to the output of the SCRs. The waveform
indicated high peaks that looked like saturation. We have experienced
similar problems with this type test set before, and it has seemed to be
remanent magnetism from short AC pulses with unequal positive and negative
half-cycles, causing a net DC component. And when the unit is connected to a
higher rated mains supply, it usually survives the high initial surge
(several thousand amps) and then works OK. It is particularly bad with an
open circuit secondary. With a load, the waveform looks good, and the
controller is designed to initiate at about 70 degrees phase angle, and
provides 5 full cycles so no significant DC component is present.

However, in this case, when they connected the test set to the 400 ampere
480 VAC mains supply, it caused a severe overcurrent event which resulted in
the SCR pucks shorted and the tracks blown off the bottom of the trigger
board and a plating of vaporized metal on the surface. The gate supplies are
totally separate and isolated by means of split bobbin transformers and
3500V rated optoisolators. These components did not appear to be damaged,
and the only visible damage was the vaporized tracks on the bottom of the
board. So I think something caused a high voltage to be imposed on the gate
circuit itself, which connects to the gates of the SCRs through twisted
pairs of probably #18AWG wires. There is no fusing on these wires, and it
appears that the PCB tracks functioned as a fuse.

After replacing the SCRs and the controller board, they had resumed testing,
but connected the power section to a low-current isolated variable supply.
We observed similar high current spikes as we attempted open circuit
testing. Then they connected a shunt as an output load, and after one or two
pulses with some high current surges, the output current looked normal and
there were only occasional and brief high current transients on the input
which resolved after about 20 cycles, and was indicative of DC offset and
remanent magnetism. So it appears that drawing current from the transformer
in a controlled manner demagnetized it.

My opinion is that the SCRs may have been previously damaged and weakened,
and when they encountered the high current spike from the remanent
magnetism, one of them failed partially open, and briefly imposed much of
the 480V mains voltage across the gate circuit, causing it to vaporize, and
then finally failed as a short circuit, which removed the high power from
the gate circuit and then simply fed the transformer primary with normal
voltage and current, although it may also be that the 400A main breaker
tripped.

This is the first time I know of that an SCR board has failed in this
manner, and I wanted to know if anyone else has experienced anything like
this. In a post a few years back I brought up the possibility of fusing the
gate circuit, and I have found some fast-acting SMT fuses that are rated at
500 VAC, so perhaps they would be a good safety measure. It may also be
necessary to add a TVS zener across the gate circuit downstream of the fuses
to clamp the voltage and provide a low impedance current path for the fuses
to act on.

Any thoughts?

Thanks,

Paul
 
Top