SCR theory question

[russrmartin]

Hey guys. I have been asked by our techs to help explain this, and I am at a loss at the moment. We have heaters that are controlled by SCRs. Today they had a fire because a wire had the insulation crack and create a short to the metal clad covering. The tech investigating came to me and said the following:

When there is no load on the SCR output, he measured 230V, phase to phase. This was measured when the controlling source of the SCR was off. Meaning, that load or no load, the heaters should not be on. With the last statement still being true, he said that when a heater was connected to the output of the SCR, the voltage measurement then went to zero. Meaning, that when the heater is hooked up, and the SCR control is off, there is no voltage, as we would like. His theory was that the heater lead shorted to ground, severing the heater circuit. This opened the circuit on the SCR output. Knowing that with the SCR turned off, and no load connected, he measured 230volts, he assumes that once the heater circuit was broken, the SCR output went high. This caused a short to ground, which initiated the flame. He said he checked 4 of our units, and all 4 behaved the same. Can anyone explain to me how or why an SCR will pass voltage to it's output when the control side is not high(active, on) etc. This is beyond my knowledge, and because of the safety ramifications, this will need to be explained. Thanks.

Russ

 

[TConnolly]

Its called leakage current. All solid state devices have a leakage current. This leakage current is a micro-ampere leakage, but with no load connected the potential will rise to line voltage. With a modern DMM you can see this voltage as the DMM imposes no load on the circuit being measured. If your tech would run and get an old Simpson analog volt meter which imposes a load on the circuit, he would measure 0 volts.

Not only will you find the same phenomen on SCRs, but you also find it on solid state PLC outputs, motors soft starters, and even motor drives. When designing SCR controlled heater circuits I usually place a isolation contactor upstream of the SCR. I do this mainly not because of leakage, but because an SCR almost always fails in the ON state and should this happen there must be a way to shut down the heaters in an overtemperature condition.

 

[russrmartin]

Thanks Alaric. Most of our SCRs have contactors directly upstream, but for some reason, that is not the case. Am I correct in understanding that this leakage current could not have been the cause of the fire? Would there be enough potential there with the leakage current to cause an arc large enough to start a fire?

 

[CroCop]

Great post.
I learned this in school quite a while back, and Alaric is spot on.
I tried telling a fellow tech about SCR leakage on a DC drive card, and he wouldn't have any of it. Said that there was no energy in the circuit, because the motor wasn't turning. We put $20 on it, and I got $20 and a hell of a laugh watching my buddy get bit by 80 Vdc.

The contactor ahead of the SCR is the perfect solution, and if you've got 4 SCRs, if they all come from the same feed, you can put the contact in line with all 4.

You can use a 10K pot in parallel with your DMM to shunt the voltage, and you'll read the on/off state of the SCR.

 

[CroCop]

The typical leakage is in the micro amp range. I wouldn't think this would cause a fire under normal circumstances.

 

[russrmartin]

Thanks guys. You've been a great help.

Russ

 

[Lancie1]

Russ,

The "leakage current" is a normal phenomonon for SCR's and most likely did not cause your fire. I would check to see that the wires are sized correctly for the load current, and that the insulation is not getting hot due to heat from the heater. Otherwise, you have experienced the Law of Entrophy, which says that all things generally go downhill.

 

[keithkyll]

The leakage current is caused by the 'Snubber' circuit inside the SCR controller. The Snubber is a resistor and cap connected in series across the SCR. It bypasses noise. Without it, noise spikes can be high enough to cause the SCR to self-trigger. The leakage is typically 2-4 ma.
Use a single phase heater circuit example to explain the fire.
208 volts, 1 amp, 208 ohms, 208 watts. If this heater shorted to ground exactly in the middle, each half of the heater now has 120 volts at 104 ohms. Each half goes to 138 watts, and draws 1.15 amps.

If the same heater shorts to ground 1/4 of the way, the short side has 120 volts, 52 ohms, 2.3 amps. This 25% section of the heater normally has 52 watts, but is now running at 276 watts! Instead of a pretty orange, it will be a bright white. A white filament is over 2000 degrees C (light bulb filament). In air, the filament melts, spraying hot metal everywhere.
I'll let you finish the story about how the fire started.
This example was with a 200 watt heater. Yours is much higher. Note that the SCR controller also took a surge of several times it's rating. Hopefully, the fuse blew in time to protect the SCR.

 

[SimonGoldsworthy]

Is this a 3-phase system ? I've had problems with 3-phase SCR controlled heaters where the neutral wire was not the same size as the main conductors and suffered melt down. (Once you use SCR's the the load will not be 'balanced' due to the harmonic currents that flow in certain configurations)

 

[russrmartin]

Thanks for the additional info, but the case here is that the heater shorted and blew. The operators shut the machine off, including the heaters. While they were waiting for things to cool down, supposed quite awhile after the initial short, this is when something ignited a fire. This is the trick to the explanation.

 

[TConnolly]

Russ, it could be possible that when the heater blew it damaged one or more of the SCR semiconductors. These nearly always fail in a shorted condition. A short would not be distinguishable between leakage and a short with a DMM when the heater is disconnected from the load terminals.

Disconnect the SCR modules from the power buss in the controller (some disassembly probably required) and place your meter on diode check. Measure forwards and backwards continuity through each SCR. If you read a short in BOTH directions, the SCR semiconductor junction has been damaged and the SCR module will need to be replaced.

If you are using hokey puck style SCRs they will need to be clamped snugly in a vise or C-clamp - pucks require clamping in a heat sink and incorporate a clamp force activated physical switch.

 

[Wino]

In addition, if a three phase SCR bridge has a bad snubber (for example, an open resistor), then the voltage on the output side goes to approximately L-N minus the resistance of the good snubber resistor. For example, in a typical 480V three phase bridge, a bad snubber will typically cause about 150VAC on the DC bus.

This is "real" voltage, and will bite you hard if you touch it.

There are lots of things that could cause the fire while cooling. The least of which is dripping insulation that eventually puddles enough to get next to something that could ignite it. Once ignited, there's no way to tell how hot the hot thing was, or whether the insulation burned then dripped, or dripped then burned.

 

[Terry Woods]

"...but the case here is that the heater shorted and blew. The operators shut the machine off, including the heaters. While they were waiting for things to cool down, supposed quite awhile after the initial short, this is when something ignited a fire. This is the trick to the explanation."

I don't believe that the fire was an electrical issue at all.

Apparently, they "heard" the heater short and blow. (Some smells wrong here.)

Since the system was shut down... don't ya think, maybe, that the fire was caused by excessive local heat build-up? I should expect that if they shut the system down then any ventilation would have been shutdown as well. Without ventilation, any heat remaining in the immediate area might tend to... collect, and finally, go critical?

I suggest that the process be fitted out with, at least, a "Potential-Fire Button". This would cut power to everything, except it would leave on a minimum ventilation system to remove the heat.

I don't think that leaving the entire ventilation system on would be good... that might tend to "fan the embers into flame". However, there should be enough ventilation to prevent collecting heat from going critical.

There might then also be a "Fire-Button" which would cause a deluge system to go on. This would minimize the damage.

 

[russrmartin]

Thanks for all the interest and discussion. I'm not fully up to date, because this wasn't a responsibility of mine, but I have learned few things here, which is what I was after. I do know that the SCR controlling the heater in question is fine, and that it probably has been deemed not the cause of the fire, but I have not heard what they conclude had started it. Either way, it sounds like power contactors should be installed in front of the SCRs. It may have been something like Terry suggested, as I was told it was just a tiny flame up. Thanks again for all the discussion.

Russ