Part of a design of mine has an NPN transistor that during a -100V negative transient (with the schematic as per a datasheet application example), experiences -100V on it's emitter, and 0V on it's base. There is a diode with the cathode connected to the base, which is connected to GND via a 10R resistor.
However, I've modified the schematic to achieve something slightly different, and during the same negative transient - the emitter sees -100V, whilst the base see's ~-30V, with a diode connected cathode to base, in series with a 10K resistor, before again grounding the base with 10R.
My question is, is either of these considered reverse biasing of the B-E junction, as I have been told reverse biasing a B-E junction ultimately destroys the transistor?
These values are obtained from LTSpice by the way, using models supplied by the manufacturers.
jtw11:
Part of a design of mine has an NPN transistor that during a -100V negative transient (with the schematic as per a datasheet application example), experiences -100V on it's emitter, and 0V on it's base. There is a diode with the cathode connected to the base, which is connected to GND via a 10R resistor.
The implication of the above is that you have a peak current of 10A though the 10 ohm resistor and the diode. Is that correct? Or is the 100V a theoretical value that would occur if you didn't provide the diode and resistor?
jtw11:
However, I've modified the schematic to achieve something slightly different, and during the same negative transient - the emitter sees -100V, whilst the base see's ~-30V, with a diode connected cathode to base, in series with a 10K resistor, before again grounding the base with 10R.
My question is, is either of these considered reverse biasing of the B-E junction, as I have been told reverse biasing a B-E junction ultimately destroys the transistor?
The first one reverse biases it by around 1V (the forward voltage of the diode at the peak current), which is not enough to worry about. The second one reverse biases it by 70V. It will probably break down at between 5V and 10V, and unless the current is limited to a few millamps at most, this will destroy the transistor.
the resistor is between D1 and base (or emitter and D1) ?
I'm afraid the transistor won't like it very long .
Well, between D1 and the base - but between the emitter and D1 is the same electrically is it not?
The implication of the above is that you have a peak current of 10A though the 10 ohm resistor and the diode. Is that correct? Or is the 100V a theoretical value that would occur if you didn't provide the diode and resistor?
No, typo - the resistor during this test is 10K, so the peak current through the diode is only a few mA. No worries there.
The first one reverse biases it by around 1V (the forward voltage of the diode at the peak current), which is not enough to worry about. The second one reverse biases it by 70V. It will probably break down at between 5V and 10V, and unless the current is limited to a few millamps at most, this will destroy the transistor.
How are you calculating these reverse bias values?
What i'm trying to achieve, is an improved response to fast negative input transients of the schematic presented in the datasheet. Certain proportions of a negative transient end up appearing on the output with the application example schematic, as Q2 cannot switch off fast enough.
Q2 cannot turn off fast enough with a 10K resistor to ground, so I switched it for a 10R resistor which allows the FET to switch off quickly - and none of the negative input transient appears on the output - however now of course there is a large pulse current through the diode, so I put a 10K series R in with the diode to take the peak current back down to a few mA, only now I've got this forward biasing problem...
jtw11:
Q2 cannot turn off fast enough with a 10K resistor to ground, so I switched it for a 10R resistor which allows the FET to switch off quickly - and none of the negative input transient appears on the output - however now of course there is a large pulse current through the diode, so I put a 10K series R in with the diode to take the peak current back down to a few mA, only now I've got this forward biasing problem...
Try going back to the original circuit, and change the resistor from 10K to 1K or a little lower, but not as low as 10 ohms. BTW the 2N3904 is not rated for 100V collector-to-emitter.
Roughly 50R is the limit before negative transients start to come through still - at which point the current in the diode is far too high.
I wonder perhaps if the solution would be simply to have a low resistance, and a power Schottky - these transients arn't all that common. It's not like space/dissipation is a primary concern considering I've got 2 pairs of double paralled FETs back to back to handle current in excess of 100A. (that's another thing, I've got double the gate capacitance to discharge at turn-off, as I've got parallel FETs)
The transient is -100V with a rise time of 1uS, that decays back to -10V in 2mS, and back to 0V within 200ms, all times absolute from the start of the pulse. This is pulse 1 of ISO7637-2 in case one is interested.
The design still needs protection from low energy, high voltage (>100V) transients by means of a TVS, but that's not a problem.
I was aware of the 2N3904 situation, it's simply in there as a starting point from the application for simulation purposes - it will be replaced by a suitable device, but thanks for the heads up.
How are you going about calculating that forward bias voltage? Or was that a rough answer based on general behaviour of diodes, is this a parameter to be found from a specific datasheet - forward voltage vs current? EDIT - It seems to be, looking at some sheets for the 4148 that's in there now.
Maybe you should try a different transistor, instead of the ancient 2N3904? I don't know how good the SPICE model is, however one of the problems I can see is that the transistor needs to pull charge out of the mosfet gate very rapidly, which implies a high peak collector current. The 2N3904 isn't good above 100mA. Try a higher current transistor, such as BC337 or ZTX851, or even 2N2222A.
You could also consider using two transistor/resistor/diode combinations, one for each mosfet gate.
Part of a design of mine has an NPN transistor that during a -100V negative transient (with the schematic as per a datasheet application example), experiences -100V on it's emitter, and 0V on it's base. There is a diode with the cathode connected to the base, which is connected to GND via a 10R resistor
If this is the circuit and -100V on the emiter with base at 0V and NPN, then you are trying to spike a FORWARD biased junction, the diode will protect any neg bias but not forward.
Base more Positive than Emitter on NPN, then its forward biased, not good for EB junction that only conducts 0.7V forward bias.
Even with emitter -100V and base -30V, the junction is still forward biased.
Where is the spike coming from, another circuit, because I can't see anything in that circuit that would produce such a spike.
Tom
PS Come on guys where did you learn about transistor bias to let that one through?
Tom is right. The potential reverse bias of the base-emitter junction occurs during normal operation of the circuit (positive input voltage), not during the transient. However, with a 100V negative transient and a 10 ohm base resistor, the forward base current would theoretically be about 10A, which is way too much for the poor old 2N3904.
Base more Positive than Emitter on NPN, then its forward biased, not good for EB junction that only conducts 0.7V forward bias.
When you say more positive however, do you mean a greater potential - or greater positive potential? What do you mean by "not good for EB junction that only conducts 0.7V forward bias" - I thought it was reverse bias that damages the BE juntion?
Where is the spike coming from, another circuit, because I can't see anything in that circuit that would produce such a spike.
External, on the supply line. The rest of the circuit deals with positive transients, with the higher voltage transients to be dealt with by circuitry yet to be added.
Right, I believe I'm with you - just rereading lots of material on p-n junctions etc.
So both conditions are forward biasing of the base-emitter junction, emitter at -100V and base at 0V, and emitter at -100V and base at -30V, as in both scenarios the the base is at a more positive potential than the emitter.
So, forward bias isn't the damaging scenario - but rather reverse bias when the p-n junction breaks down and a large current flows.
So, how is the diode protecting the transistor from reverse bias at the b-e junction?
I think I've almost moved the point of this conversation, but going back to a solution - to keep a fast turnoff, I need the low resistance to ground of the NPN base to ground - so perhaps I just need a power diode that can handle the large peak currents. Also, I need a diode whose forward voltage stays low at large peak currents, unlike the 4148s, which seems to be around ~70V.
jtw11:
I think I've almost moved the point of this conversation, but going back to a solution - to keep a fast turnoff, I need the low resistance to ground of the NPN base to ground - so perhaps I just need a power diode that can handle the large peak currents. Also, I need a diode whose forward voltage stays low at large peak currents, unlike the 4148s, which seems to be around ~70V.
Like I said before, to turn the mosfet off quickly, you need a transistor that can suck a lot more current out of the mosfet gate than a 2N3904 can.
Also bear in mind that if you reduce the base resistance to 10 ohms, then it will pass more than 1A under normal operation, and 10A when the -100V transient occurs. This means that:
(a) the resistor will dissipate more than 10W in normal operation;
(b) the transistor will need to handle 10A peak base current. I very much doubt that you will find a transistor that can handle this base current and has a fast enough turn-on time.
The LTSpice model apparently doesn't allow for the fact that the circuit is is simulating will blow the transistor very quickly. I wonder how accurate its prediction of the fed-through transient is.
Like I said before, to turn the mosfet off quickly, you need a transistor that can suck a lot more current out of the mosfet gate than a 2N3904 can.
Ah - I think something has just clicked. BJTs arn't at all like MOSFETs are they, in that they cannot just have as much current through them as the rest of the circuit resistance allows, but the BJT itself is the main limiting factor.
Ah, and of course - I had not thought of the steady state issues in normal operation with the low resistance.
As for the actual current route during discharge of the MOSFET gates, is it through the collector --> base to ground, or through the collecter --> emitter --> diode to ground? Must be through the base given your comments.
It's beginning to look like I need to device some other circuitry to allow me quick turn off. Perhaps something whereby a BJT turns on a smaller MOSFET to discharge the main MOSFETs.
EDIT - Well, LTSpice dosen't directly flag up an error screen telling you you've destroyed a component no, but you can easily view traces that show you current through devices, energy absorbed, power dissipation etc - so in that respect, once you look you can tell.
jtw11:
As for the actual current route during discharge of the MOSFET gates, is it through the collector --> base to ground, or through the collecter --> emitter --> diode to ground? Must be through the base given your comments.
The path for discharging the mosfet gate is collector -> emitter. Current between the base and emitter is necessary to make the collector -> emitter current flow. The diode doesn't come into it, other than that its capacitance will slow down the turn-on of the transistor.
jtw11:
It's beginning to look like I need to device some other circuitry to allow me quick turn off. Perhaps something whereby a BJT turns on a smaller MOSFET to discharge the main MOSFETs.
If you are not careful, then the more devices you add, the slower the circuit will be. It's possible that two NPN transistors connected as a darlington pair might turn the mosfet off faster. However, I would choose a higher-current transistor first, especially as the 2N3904 has nowhere near the current rating you probably need (let alone a good hfe at that current). Try re-running the simulation with a BC337, a ZTX851, and a 2N2222A.
Hi, you say that the -100V transient is coming from another source, what is that source and have you tried suppressing it at the source.
If its a coil or solenoid have you got reverse bias diode across it to suppress the back emf?
Tom.
The path for discharging the mosfet gate is collector -> emitter. Current between the base and emitter is necessary to make the collector -> emitter current flow. The diode doesn't come into it, other than that its capacitance will slow down the turn-on of the transistor.
Ah, I'm with you - I was thinking the current route would go through the diode to ground - but given the positive rail is negative during the transient - the current will actually dicharge to the positive rail, correct?
Hi, you say that the -100V transient is coming from another source, what is that source and have you tried suppressing it at the source.
If its a coil or solenoid have you got reverse bias diode across it to suppress the back emf?
Tom.
The supply is an automotive 12V supply... - I hope that should answer it This transient is a standard pulse from ISO7637-2 that deals with protection against common transients.
I've dealt with auto transients before - just looking for a solution that's somewhat more robust and intelligent, hence use of the LT4356 chip.
As for using a Darlington, I shouldn't need two diodes (D1), but rather one connected in forward bias between the emitter and base of the device as if it were an integrated Darlington?
I can see that you are trying to use the reverse connection protection circuit as transient protection.
But the chip will only with stand -60v.
The chip will be on the transient side of the circuit and will still get the -100V spike even if the transistor circuit protects what it is supplying.
Have you looked up transient suppressor diodes, varistors, etc. Try google for (transient suppression techinques).
This transient is a standard pulse from ISO7637-2 that deals with protection against common transients.