I had a transistor blow up in an engine DME (engine computer). I am replacing the transistors (IGBTs - Triple Darlingtons - 14A rated), but I am trying to find the root cause. in this application, and a typical, setup, 12vdc goes to an ignition coil (low-moderate resistance value) and then to the Collector. The base to goes to the CPU pin and the emitter goes to ground. The DME, via these transistors, allows current to flow through the coil for around 5ms at a time to charge the primary winding.

I can't see how I had any wiring problems on the positive side of the circuit, but it is possible that the ground side wiring was exposed/shorted to 12 Vdc, due to a coolant leak. Those wires go into the lower engine area and I had a lot of water flowing that could have somehow created a short to 12 Vdc.

My question: in this setup, if the negative side of the transistor was shorted to 12 VDC, would that cause the transistor to overheat and explode? There was enough heat to melt the solder and physically destroy the transistor. I am pretty certain that something shorted and not a bad coil, as the problem occured a couple of seconds after the coolant leak and after a minute of running after that could smell the burnt transistor.

Second question: Is it likely that this event could have just taken out the transistor and not damaged anything in the control circuit? I can't find any other physical damage and it seems as though the DME was functioning, except the coil circuit driven by the affected transistor.

Thanks so much for your assistance.

Larry

 

That transistor is handling a large current at low voltage and the resistance of coolant will not be that low so it would probably not directly affect the transistor. I think there are two possibilities. The water could have affected the CPU circuitry which then left the transistor switched on continuously which caused it to overheat. Alternatively, the timing of the water leak and the transistor failure was a coincidence (they do happen) and the transistor simply self-destructed (they do that from time to time).

In either case, drying out and cleaning the CPU area and replacing the transistor should fix it unless the dying transistor fed voltage back from its base connection to the CPU pin and has damaged the CPU.

 

I had experienced even 1500 Amp capable igbt explosion where even so small as 5 Amp mains fuse is not blown out. Transistor shot to the ceiling and hanging down in the wire looks very `sexy`. Sad side is that it cost rather much.

 

The DME has 8 IGBT's, so not sure how it could have only held one of them open for an extended period. The CPU stay'ed dry. There is no condition where the s/w would hold one IGBT open for more than the dwell period. Clearly other wires, got wet, but don't see how any of them would cause this. Also, the IGBT is rated for the current drawn by the coil, even for an extended period.

So are you saying that if I got 12V on the ground/emitter side, that would not cause the heat / destruction?

Thanks again for the help here.

Larry

 

It would short the coil and transistor so no current could flow through them. It would short the battery so it's going to either blow a fuse if there is one or burn out the wire.

From what you say, it sounds like the transistor had had enough of life and just committed suicide.

 

The DME has 8 IGBT's, so not sure how it could have only held one of them open for an extended period. The CPU stay'ed dry. There is no condition where the s/w would hold one IGBT open for more than the dwell period. Clearly other wires, got wet, but don't see how any of them would cause this. Also, the IGBT is rated for the current drawn by the coil, even for an extended period.

So are you saying that if I got 12V on the ground/emitter side, that would not cause the heat / destruction?

Thanks again for the help here.

Larry

You are confusing terms. You say you are using IGBTs but also mention a base pin. IGBTs don't have a base pin, they have a mosfet gate.

The gate is a high impedence voltage control with a small amount of capacitance to charge - no other current flow needed.
Depending on how your microcontroller is programmed and your exact circuit design, that little bit of capacitance can cause an issue and cause enough shoot-through current on an H-Bridge to kill the IgBT over time.

Make sure your gate has a pull-down resistor path to ground (Low side) and a pull-up resistor path to Vcc if you have P-Channel (PNP) high side device.

A 10k to 100k resistor is usually enough but you'll have to calculate for your gate capacitance.

Capacitance can keep the gate open and current flowing. Also, a microcontroler set to a TriState high impedence input can make the gate essentially float and stay open. Just because no current is flowing in does not mean it is off. True for a simple BJT but not true for a MOSFET or IGBT.

 

Most of the ECUs ive pulled apart use Darlington transistors for ign coil switching not IGBT. I would remove a good component & check with a component tester to see what it realy is. Its not common for these components to fail. But a shorted coil primary may do it or if an ignition lead falls off for a period of time, can cause high voltage spikes & rip off an ign switching device.

 

Most of the ECUs ive pulled apart use Darlington transistors for ign coil switching not IGBT. I would remove a good component & check with a component tester to see what it realy is. Its not common for these components to fail. But a shorted coil primary may do it or if an ignition lead falls off for a period of time, can cause high voltage spikes & rip off an ign switching device.

It is a 14C40L. At least that is what I am replacing it with (original was Bosch 30021 and it is a Bosch ECU). I was under the impression that it was an IGBT, but could be wrong. It was also my understanding that it was a triple darlington. The open source ECU's that I have built used BIP 373's. I will look harder for a short on the coil primary. Given that this occurred the second the water started flowing and has run fine otherwise for years leads me to believe that the water caused a short somewhere. The leads from the transistor to the coil are well protected and did not get wet. However, at least a couple of sensor and other wires must have shorted, as I threw a bunch of codes. Reality is, now that it is dried out, it should be fine. I will test the coil and wires associated with the blown transistors to be sure.

 

Most of the ECUs ive pulled apart use Darlington transistors for ign coil switching not IGBT. I would remove a good component & check with a component tester to see what it realy is. Its not common for these components to fail. But a shorted coil primary may do it or if an ignition lead falls off for a period of time, can cause high voltage spikes & rip off an ign switching device.

You appear knowledgeable about ECU's. The Bosch ECU I am repairing has 30021 IGBTs/transistors. It appears these are the same as infineon BTS2140's. Do you know if the 14C40L IGBT is compatible with that? I have the 14C40L's now to replace, but could get the 30021's in a week.

Larry

 

Cant find any specs on BTS2140 that say they are suitable for Ign. But it appears the 14C40L is an IGBT for ign use & is Vce 430V max @ Ic @ 110 deg C 14Amps. It appears on landrover forum that the Bosch 30021 does fail, also a Bosch 30057 will replace it.