MOSFETs IN SERIES

[arg733]

To the OP, what switching frequency do you plan to use?

Well... most of the time it won't exceed 200hz.

 

[KrisBlueNZ]

OK, and what about the times when it DOES exceed 200 Hz? What's the maximum?

 

[arg733]

Absolute maximum should be 300hz

 

[arg733]

Also what is the maximum value of the resistors i can use so to have the minimum leakage?

 

[KrisBlueNZ]

OK, 300 Hz is reasonable.
The value of the voltage balancing resistors will affect how well the MOSFETs share the total voltage. Lower values will improve the voltage sharing. I can't give you a maximum value.
Zener diodes specified for slightly less than the Vds(max) voltage of the MOSFETs, connected across those resistors, might be a better way to voltage-share. You could increase the resistances or remove them altogether if you use a zener across each MOSFET. Anode to souce, cathode to drain, i.e. "pointing upwards". Again, 1W zeners should be fine.

 

[(*steve*)]

Keep in mind, the gate voltage has to be relative to the source voltage.

I did a project like this once where we ended up switching 200kV at over 10A.

We had to use a seperate, ISOLATED, gate drives for each device.

This is the only way it will work flawlessly!

I can only reveal this, because that company is now out of business.

Good to hear from you.

I would recommend people consider this advice.

My issue with the design we're talking about is that it relies on (or results in) avalanche conduction of the mosfet. The addition of source/drain resistors eliminates this in the off condition, but I'm not so sure about during switching.

Mosfets are typically rated for the energy of the avalanche, and I would be very keen to ensure that I kept it well below that.

 

[arg733]

Hi.
I was well enough today to try and it worked i also tried it with 6 and it worked! So here is the new circuit.


I though that if i put zener or avalanche diodes (zd11-zd16) across each MOSFET's drain source and rated slightly bellow each MOSFET's breakdown voltage that would save the MOSFETs from over-voltage. Is this a fair assumption?

Thank you all very much.

 

[KrisBlueNZ]

I was well enough today to try and it worked i also tried it with 6 and it worked! So here is the new circuit.

Good.

I though that if i put zener or avalanche diodes (zd11-zd16) across each MOSFET's drain source and rated slightly bellow each MOSFET's breakdown voltage that would save the MOSFETs from over-voltage. Is this a fair assumption?

Yes, that's probably a good idea.
Your circuit is right apart from the gate drive. For cleanest switching and best safety, you should have a 10k current-limiting resistor in series with EACH diode, instead of a single resistor in series with the 12V source. If possible, use FUSIBLE resistors for these positions.

 

[arg733]

I just tried it with 6 2n7000 (60v) = 360v. I turned on and off an 15w incandescent lamp at 350V (as much as my simple power supply could give which just an 220/220v insulting transformer and an bridge rectifier followed by an 300uf electrolytic capasitor) and it worked nice (no earth shattering kaboom and no overheating) and i actually tried in on the breadboard... the leakage was 6v because i used 10k resistors across each mosfet's drain-source.

 

[KrisBlueNZ]

Good.

 

[arg733]

Ok it works so far with resistive loads but when i try to use it with inductive loads it can't exceed the voltage rating of 1 mosfet.
I don't know what else to do i tried to protect my mosfets with zeners (rated at the same voltage as the mosfets) across each mosfet's drain, source and flywheel diodes anti-parallel to the coil (which is the primary of a 500W step down transformer). This also happens with any other inductor. Is there a way to protect the mosfets from bemf or the "inductive spike"?
Thank you.

 

[(*steve*)]

When you say that with an inductive load you can't exceed the voltage rating of 1 mosfet, are you talking about the voltage applied to the load, or the magnitude of the spike generated when current is suddenly turned off?

 

[arg733]

Sorry about that...
I mean that the voltage i give to the load can't exceed the voltage rating of 1 mosfet.
Edit : This effect also happens with air core inductors.

 

[(*steve*)]

Is there a snubber on the load?

Otherwise the voltage across the mosfets could rise to a voltage many times the voltage applied to the load when current to it is suddenly stopped

 

[arg733]

Ooops , i forgot to try the snubber.
I will definetly try it out. After some digging i found that when the current flow in an inductor suddenly stops it tries to maintain it by inducing a voltage sometimes 10x the supply voltage so that's what's killing my mosfets (i think) but the strange thing is, that it appears that there is no difference if i put 3 mosfets in series or 10 in series, the result is the same : i can't exceed the voltage rating of 1 single mosfet , so if i use 1 or 100 mosfets it is the same
Is there a way of making them work (I'm not so sure that the snubber will solve that, i suspect that this is far more complicated.)

Thank you very much.

 

[Electrobrains]

If you use a Free Wheeling Diode directly in parallel to the load, you wouldn't need any snubbers or other surge suppression circuitry!

That simple diode makes a perfect work, if it's rightly dimensioned.

It needs high enough reverse blocking voltage - the full supply voltage (and a bit of security on top of that).
It should also be fast enough to catch the spikes that might appear (although I doubt that it would be a problem in this sequentially switched, rather slow circuit).

The only reason not to use a free wheeling diode would be for speed reasons. If the inductance of the load is big, it will take time for it to switch off.

Of course, if the supply rail voltage is not stable, the MOSFETs might need over voltage protection.
I would then put some transorb diodes or varistors in series to reach a suitable protecting voltage and place them directly over the MOSFET chain. In case of an over voltage, the load might then have to jump in and take a part of the blow.
A snubber over the transistors would only help at short surges and a snubber over the load would worsen the situation for supply carried over voltage surges.

 

[arg733]

I used 1n4007 and the voltage is fairly steady :
( 220vac mains -> variac -> insulating transformer -> recifier bridge -> 220uf 500v capasitor across the + and - )

 

[Electrobrains]

I used 1n4007 and the voltage is fairly steady :
( 220vac mains -> variac -> insulating transformer -> recifier bridge -> 220uf 500v capasitor across the + and - )

I looked back and saw that you write that the load is:

the coil (which is the primary of a 500W step down transformer)

That shows everything!

1. Have you measured the DC resistance of that 500W transformer? My guess: you don't kill your transistors because of over voltage, but because of over current! Also, be aware of that "a test", without full frequency switching, will load the transistors through the dc resistance of the transformer winding.

2. For 200-300 Hz you will probably reach an enormous current in your coil. You need a very high inductance for that low frequency! Normally, such circuits are switched at "many" kHz. You need to calculate the max current, resulting from your switching frequency, duty cycle and inductance!

3. A normal free wheeling diode can not be used! That will "lock" the current in the coil blocking the transformer function. The energy stored in the primary inductance need to be transferred to the secondary side and not stuck in the primary winding.

4. Instead, a snubber might indeed be necessary in this circuit or a combination of Resistor, Capacitor and Diode.

5. To use the circuit with a "step down transformer", I suggest you look at the timing issue. Probably you need to a) raise the inductance (much), using another transformer, b) raise the switching frequency (much), c) limit the duty cycle d) add a current feedback to a circuit that will automatically switch off the transistors at a set current. There are good, cheap circuits available for such controlling (for instance the UC3842 series).

 

[arg733]

I looked back and saw that you write that the load is:
That shows everything!

1. Have you measured the DC resistance of that 500W transformer? My guess: you don't kill your transistors because of over voltage, but because of over current! Also, be aware of that "a test", without full frequency switching, will load the transistors through the dc resistance of the transformer winding.

2. For 200-300 Hz you will probably reach an enormous current in your coil. You need a very high inductance for that low frequency! Normally, such circuits are switched at "many" kHz. You need to calculate the max current, resulting from your switching frequency, duty cycle and inductance!

3. A normal free wheeling diode can not be used! That will "lock" the current in the coil blocking the transformer function. The energy stored in the primary inductance need to be transferred to the secondary side and not stuck in the primary winding.

4. Instead, a snubber might indeed be necessary in this circuit or a combination of Resistor, Capacitor and Diode.

5. To use the circuit with a "step down transformer", I suggest you look at the timing issue. Probably you need to a) raise the inductance (much), using another transformer, b) raise the switching frequency (much), c) limit the duty cycle d) add a current feedback to a circuit that will automatically switch off the transistors at a set current. There are good, cheap circuits available for such controlling (for instance the UC3842 series).

Ok my mosfets are irf540 (33A version) and i tried it with an air core coil and it also didn't work (it draws about an amp, so no over-current) why can't it work with a simple air coil?

 

[Electrobrains]

Did you use a coil with R=350 Ohm, Supply=350V and a free wheeling diode installed in parallel to the coil (1N4007 - turned the right way around)?
And your MOSFETs died?