Detecting open gate on SCR switch

[Paul E. Schoen]

I am testing my smart SCR firing circuit with a simple transformer load and
a lamp on its output. The test circuit is about as follows, except I have a
50 VA isolation transformer supplying the ACHI and ACLO. (My first ASCII
diagram!)

CB
ACHI--[1A]---- SCR1 >|--------|
| | | |-------
--|< SCR2 ---- )|( |
120V )|( 16V O Lamp 28V
)|( |
ACLO--------------------------| |-------

I was able to program the PIC to turn the SCRs on and off, adjust phase
angle to dim the lamp, and check for primary voltage across the SCRs before
allowing initiation.

Then, I wanted to check for an open gate, which I figured would cause a
large DC current in one direction, and would probably trip the 1 amp
circuit breaker. I assumed I would read a large voltage across the SCR that
was not triggered, and that I could use this to cause the firing circuit to
shut down.

Much to my surprise, when I removed either of the gate wires, the circuit
seemed to perform almost the same as when both were connected. I did read
about 2 VDC across the SCRs, and the AC voltage (at about 50% modulation)
changed from about 64 VAC to 66 VAC, when one gate was disconnected.

I am thinking that possibly the ungated SCR is being triggered by dV/dT
from an inductive kick from the transformer, and possibly it is related to
the high impedance of the source and the load. I will need to look at the
waveforms with a scope, and also try a lower impedance source. It may also
help to put a snubber across the SCRs.

Your thoughts will be appreciated, and I will post my additional test
results.

Thanks,

Paul

 

[Luhan]

I am testing my smart SCR firing circuit with a simple transformer load and
a lamp on its output. The test circuit is about as follows, except I have a
50 VA isolation transformer supplying the ACHI and ACLO. (My first ASCII
diagram!)

CB
ACHI--[1A]---- SCR1 >|--------|
| | | |-------
--|< SCR2 ---- )|( |
120V )|( 16V O Lamp 28V
)|( |
ACLO--------------------------| |-------

I was able to program the PIC to turn the SCRs on and off, adjust phase
angle to dim the lamp, and check for primary voltage across the SCRs before
allowing initiation.

Then, I wanted to check for an open gate, which I figured would cause a
large DC current in one direction, and would probably trip the 1 amp
circuit breaker. I assumed I would read a large voltage across the SCR that
was not triggered, and that I could use this to cause the firing circuit to
shut down.

Much to my surprise, when I removed either of the gate wires, the circuit
seemed to perform almost the same as when both were connected. I did read
about 2 VDC across the SCRs, and the AC voltage (at about 50% modulation)
changed from about 64 VAC to 66 VAC, when one gate was disconnected.

I am thinking that possibly the ungated SCR is being triggered by dV/dT
from an inductive kick from the transformer, and possibly it is related to
the high impedance of the source and the load. I will need to look at the
waveforms with a scope, and also try a lower impedance source. It may also
help to put a snubber across the SCRs.

Your thoughts will be appreciated, and I will post my additional test
results.

Thanks,

Paul

If I understand the situation correctly, you are floating an SCR input.
I suspect this makes it react to internal leakage of the high voltage
AC. Try connecting a resistor from the input to bottom of the SCR.

Luhan

 

[Paul E. Schoen]

I am testing my smart SCR firing circuit with a simple transformer load
and
a lamp on its output. The test circuit is about as follows, except I
have a
50 VA isolation transformer supplying the ACHI and ACLO. (My first ASCII
diagram!)

CB
ACHI--[1A]-+-- SCR1 >|--+-----+
| | | +-------
+-|< SCR2 ---+ )|( |
120V )|( 16V O Lamp 28V
)|( |
ACLO--------------------------+ +-------

I was able to program the PIC to turn the SCRs on and off, adjust phase
angle to dim the lamp, and check for primary voltage across the SCRs
before
allowing initiation.

Then, I wanted to check for an open gate, which I figured would cause a
large DC current in one direction, and would probably trip the 1 amp
circuit breaker. I assumed I would read a large voltage across the SCR
that
was not triggered, and that I could use this to cause the firing circuit
to
shut down.

Much to my surprise, when I removed either of the gate wires, the
circuit
seemed to perform almost the same as when both were connected. I did
read
about 2 VDC across the SCRs, and the AC voltage (at about 50%
modulation)
changed from about 64 VAC to 66 VAC, when one gate was disconnected.

I am thinking that possibly the ungated SCR is being triggered by dV/dT
from an inductive kick from the transformer, and possibly it is related
to
the high impedance of the source and the load. I will need to look at
the
waveforms with a scope, and also try a lower impedance source. It may
also
help to put a snubber across the SCRs.

Your thoughts will be appreciated, and I will post my additional test
results.

Thanks,

Paul

If I understand the situation correctly, you are floating an SCR input.
I suspect this makes it react to internal leakage of the high voltage
AC. Try connecting a resistor from the input to bottom of the SCR.

Luhan

The SCR I am using is a dual AC switch rated at 90 amps and 600 V. The
gates have internal shunt resistors of about 20 ohms, so removing a gate
connection does not make it especially sensitive. The same thing occurred
when I removed the gate drive optoisolator.

I thought perhaps it was because I drive both gates at the same time, even
for 180 degrees conduction angle. I changed the software so that both gates
are turned off when there is zero volts across the SCR, and only the gate
for the SCR with positive voltage on the anode is turned on. In this case,
I saw about ten cycles of very distorted asymmetrical waveform, which
caused a fairly large current flow that would have tripped the breaker
eventually, but then the waveform stabilized with just a bit of delay after
the zero crossing in one direction.

Finally, I completely removed one of the SCRs from the circuit, (cathode
and gate, anode internally connected to other cathode), and now the
waveform starts cleanly. After about three cycles, there is a little
turn-on delay glitch in one direction. It appears that the gate of the SCR
which is connected is triggered just about constantly.

So, the SCR is acting a bit like a triac, where a gate signal causes
current flow in either direction. However, the triac requires the gate
signal to be in the same polarity as the applied voltage.

A
|
|
\
->|
Q1 |-----+
/-| |
/ |
| /--+
| |/
G----+---| Q2
| |->
+-+ \
|R| \
+-+ |
|------ K

Looking at the equivalent circuit, what happens when the anode (A) is
negative WRT the cathode (K), with R in this case 20 ohms, and Q2 is turned
ON with a voltage from G to K? The only current path I see is through the
resistor, Q1C-E (or Q2B-C), and then reverse through Q1B-E. Possibly
forward biasing Q2 causes the B-E junction of Q1 to break down at a low
voltage. I have not seen much information on SCR characteristics in the
reverse quadrant with gate current applied, but I saw one spec that only
allowed about 5 volts reverse under that condition. When both SCRs are
conducting, it is not a problem.

I'll have to do some more thinking and experimenting. Things may be
different when a full current load is applied, but I hope to understand
what is happening before I do that. Thanks for any additional thoughts.

Paul

 

[Paul E. Schoen]

I now have the SCR circuit working properly, although I still do not know
exactly why an SCR can conduct on both directions with a positive gate
current.

To detect an inoperative gate, I simply check for the presence of any
voltage of either polarity across the SCR pair after the trigger signal has
been applied. I had to wait a couple hundred microseconds to allow for
propagation through the optocoupler, but it now shuts down within one cycle
if a gate is disconnected.

The fun stuff will begin soon. This circuit will need to be tested under
full power on a breaker test set. When we have had problems in the past
with bad SCR gate drives or connections, the input current draw was often
high enough to trip the main breaker to our building. We have 208 VAC at
200 amperes. We use a step-up autotransformer to get 480 VAC for the test
set, at about 90 amperes. When the SCR malfunctions, I figure that
sometimes we get well over 2000 amps at the 208 VAC mains, and the breaker
trips in about one cycle. So we will see if this circuit is fast enough to
beat the breaker.

Paul