Unreliable transistor issue

[ACharnley]

Hi all,

Just wanting to run this by you.

I've a simple voltage based switch which turns off a mosfet. The fet is rated to 60V with 20V gate. The transistor is the common MMBT3904 rated to 40v.

I'm currently testing with a voltage up to 20v.

It's working correctly, turning off the fet around the 8v mark, but not for long. After some time the transistor creates a short between c+e. Sometimes I can clear it by shorting base to ground but eventually it makes no difference.

I can't see a problem. Even at 10v over the zener the resistor is dissipating 0.01W. The C+E current is much less.

Any ideas?

Andrew

 

[Nigel Goodwin]

There's far too much of the circuit missing to offer any suggestions.

 

[ACharnley]

OK, I can't post the rest of it - sorry.

But this is the only bit that matters. I've checked for over and reverse voltages with a data logger.

 

[rjenkinsgb]

A bipolar transistor used for switching (or almost any purpose) should always have a base-emitter (or base to emitter side power rail) resistor to ensure proper turn-off despite any stray leakage currents in the transistor itself or elsewhere.

Try something in the 1k - 10K range base-emitter on the transistor.

 

[ACharnley]

I thought that may be an issue but wasn't sure if it only applied to fet's.

However, I've tried it earlier today. I had to go down to 500 ohm and it seemed to work for a bit, then the transistor degraded further so that not even a short would turn the collector-emitter off. It's happened to two transistors now and I'm about to fit a third.

Only things I can think off;

1. I'm using 260c on the hot air-gun to reflow. Don't hold it for long, just long enough.
2. the transistors are crap fakes. I'm not 100% I bought these from the UK. I've ordered some from RS today to eliminate that.

 

[dknguyen]

The MMBT3904 is not a FET.

Are you saying that the circuit runs fine as long as the voltage supply stays high and the BJT stays conducting and only starts to fail when you start lowering the supply voltage? Or does it just always eventually fail no matter what operating condition it is in?

Are you aware that zener diodes require a minimum current in order to do their job? A minimum current is required so they are operating in the vertical portion of the curve where their reverse voltage is approximately independent of current.
https://upload.wikimedia.org/wikipe.../450px-V-a_characteristic_Zener_diode.svg.png

What is this circuit supposed to do anyways? Short across a load when the something when the voltage is too high?

 

[ACharnley]

I know it's not a FET, but rjenkinsgb is suggesting a resistor is still required?

The circuit works fine but not for long, it's as if the transistor is degrading. It's difficult to know exactly due to the supercaps on the PCB but I believe if there's a voltage at the collector it's latching the transistor on (regardless of base current). This is erratic but occurs more often over time. It does sound like leakage but I would have thought a 500 ohm would resolve it.

Operation wise it's shorting the fet, which is charging a couple of super-caps. It prevents the caps over-charging and protects the fet gate from over-voltage.

I'm aware of zener voltage, the curves aren't as much of a problem at this reverse voltage, plus the transistor gain + cap voltage tolerance is enough.

 

[dknguyen]

Hi all,

Just wanting to run this by you.

I've a simple voltage based switch which turns off a mosfet. The fet is rated to 60V with 20V gate. The transistor is the common MMBT3904 rated to 40v.

I'm currently testing with a voltage up to 20v.

It's working correctly, turning off the fet around the 8v mark, but not for long. After some time the transistor creates a short between c+e. Sometimes I can clear it by shorting base to ground but eventually it makes no difference.

I can't see a problem. Even at 10v over the zener the resistor is dissipating 0.01W. The C+E current is much less.

Any ideas?

Andrew

I know it's not a FET, but rjenkinsgb is suggesting a resistor is still required?

The circuit works fine but not for long, it's as if the transistor is degrading. It's difficult to know exactly due to the supercaps on the PCB but I believe if there's a voltage at the collector it's latching the transistor on (regardless of base current). This is erratic but occurs more often over time. It does sound like leakage but I would have thought a 500 ohm would resolve it.

Operation wise it's shorting the fet, which is charging a couple of super-caps. It prevents the caps over-charging and protects the fet gate from over-voltage.

I'm aware of zener voltage, the curves aren't as much of a problem at this reverse voltage, plus the transistor gain + cap voltage tolerance is enough.

Show us the part including the FET.

 

[Ylli]

Just to clarify, you are saying that the 3904 fails with a C-E short, right? Or are you saying that the transistor appears to turn on when it isn't supposed to, but the transistor is not damaged?

 

[ACharnley]

That's correct Ylli, it's failing C-E short.

 

[dknguyen]

That's correct Ylli, it's failing C-E short.

Does it fail more and more the longer it is held off or does it fail more and more everytime it enters the MOSFET off-state? Maybe something is making it through the drain-gate capacitance and damaging the BJT everytime it switches. Can you scope it? Maybe add a gate series resistor. 100 ohms?

 

[ronsimpson]

Maybe add a gate series resistor. 100 ohms?

Probably a good idea. Some times a MOSFET with no gate resistor will oscillate at 100+ mhz. But I don't see how that kills the transistor.

What kills a transistor?
..Too much voltage:
....C-E voltage. What is the voltage that charges the super caps? Is there a large capacitor on this supply? Is there a large inductor on this supply that will fly up. (like a flyback or boost power supply) The only way I can see the C-E voltage getting to 50 volts is that there is a incuctor and no capacitor on this supply.
....B-E voltage. If the base is driven negative beyond -5V then there will be base current and maybe damage.
..Too much current:
....Collector current. The only DC current is through a 47k resistor., so very small If the MOSFET way a 200A monster then maybe the Gate capacitance could kill the little transistor. I don't see how this current will harm the transistor. (maybe oscillation) Adding a 100 ohm gate resistor should help.
....Base current. Again I don 't see much Base current unless the supply voltage flys up when it is unloaded.

Pretty clearly the supply jumps up when the capacitors are cut open. That is how the transistor latches. Again, with out seeing more of the circuit.......

 

[Ylli]

I'd agree with the suggestions to add a 100 ohm series gate resistor. Also, add a resistor from the 3904 base to ground to redirect any leakage - perhaps 47k. If speed is not an issue, add a small cap across the 3904 CE to bypass any transients.