Is MOSFET in series is possible?

I think the earlier suggestion of using a contactor is the correct way to go. Trying to switch 80000W with Mosfets is crazy!!
Especially is it is at very low frequency / dc.
connect your contactor to the micro via a transistor/mosfet/ whatever and away you go. Trying to do it with silicone will almose guarantee lots of smoke
until you get it exactly right , if ever.
Ned

There are two problems with MOSFETs in series. One is the difficulty of driving the gates with different floating voltage references, and that problem is what has been discussed here at length. But the other problem that has received very little attention is the problem of equalizing the voltage drop across the MOSFETs when they are off. MOSFETs are very high resistances because they have almost no leakage current when they are off. That means it is very difficult to match the leakage resistances of two MOSFETs in series. The voltage of 1000v will not equally divide with 500v on one MOSFET and 500v on the other. The 1000v will divide according to the leakage resistance. If one MOSFET has an effective leakage resistance of 10,000,000 Ohms and the other MOSFET has a leakage resistance of 2,000,000 Ohms, then the 1000v will split with 833v across the first MOSFET and 167v across the second MOSFET. And if the MOSFET is only rated for 500v then 833v is too high for the first MOSFET and it could be damaged. What you could do is to degrade the leakage resistance of both MOSFETs by placing a resistor in parallel with each one, maybe 1,000,000 Ohms in this case, or whatever resistance it takes to ensure that most of the leakage current is flowing in the parallel resistor and not the MOSFET. That will balance the voltage across the MOSFETs so each one sees about half the applied voltage (500v). Of course this also degrades the overall leakage performance of your circuit. Depending on what you are driving with these MOSFETs, you might not want to increase the leakage current through the circuit when the MOSFETs are supposed to be off. But that is the trade-off you need to make if you want to ensure that the applied voltage is evenly divided between two MOSFETs in series.

Thanks goldsmith.
It was not my intention to advice it for series transistors but for a single N-channel high side mosfet. If I understand correctly a pulse transformer (and associate circuit) is a suitable driver for series transistors ?
I would like to know more about the feedback circuits that you mention here:

Yes you can , but you need a pretty precise driver ! because if both of them don't be on at the same time , both will born ! you'll need some feedback loops to be sure about on time and off time .

Click to expand...

Best Regards

Tunelabguy
Very interesting point.
Do you think that placing RC snubbers would form a capacitor voltage divider to equalize the voltages? As capacitor voltage dividers work in DC too.
Can varistors in parallel do the job?

Best Regards

Thanks goldsmith.
It was not my intention to advice it for series transistors but for a single N-channel high side mosfet. If I understand correctly a pulse transformer (and associate circuit) is a suitable driver for series transistors ?

Click to expand...

Hi Albert
Yes i've already understood that you're referring to the H side driver . but don't forget that , author of this thread wants to use a mosfet or IGBT ( or some of them in series together ) as an on / off switch ! ( his aim isn't a switching system with square wave ! as i've understood his purpose is just on / off with a command signal ! so it means there is a long time between each command ! hence the capacitive driver isn't a good idea . the best way is using a floated ground that has been created with a transformer to create individual PSU ( two kind of ground )

I would like to know more about the feedback circuits that you mention here:

Click to expand...

Certainly . For such a application ( if we use a power element in high speed system ) we've to take feedback from current and voltage and also from driving signals . but in this application ( a power element just as a key , it can be neglected and with some simple tricks he can handle this aim )
But i'm still in this believe that switching this power , via a semiconductor switch , isn't reasonable !
Of course i've used switch mode systems in pretty high powers , until now , but i've never used them as a key for high powers because of high value of loss power ( at conduction mode ) but in a high speed system , the loss equation will be completely changed .

Best Wishes
Goldsmith

The discussion seems to me somehow bloodless without considering application parameters. The problem discussed in the original post would be usually solved with a single 1000 - 1200 V N-MOSFET or IGBT and isolated gate driver. It's no good example for series connected switches.

Other applications have quite different requirements. I've seen series connected MOSFET e.g. for kV/ns pulse switches. Power electronic would refer to HV IGBT (6.5 kV rating is state-of-the-art) and think about series connection above e.g. 2 or 3 kV.

The discussion seems to me somehow bloodless without considering application parameters. The problem discussed in the original post would be usually solved with a single 1000 - 1200 V N-MOSFET or IGBT and isolated gate driver. It's no good example for series connected switches.

Other applications have quite different requirements. I've seen series connected MOSFET e.g. for kV/ns pulse switches. Power electronic would refer to HV IGBT (6.5 kV rating is state-of-the-art) and think about series connection above e.g. 2 or 3 kV.

Click to expand...

Hi dear FvM
I think you've missed the point that his power rating is very high and if he use a mosfet or an IGBT , he will need more price + large heat sinks . ( i think that contactor idea would be better instead of semiconductor switches ) . and about gate driver , i think he can't use gate drivers ! because he wants to use it as a key instead of a switch with square wave as command , hence i think a bootstrap driver won't be a good idea . ( i think creating a floated DC supply with an opto coupler would be ok ) am i right ?
Best Regards
Goldsmith

I don't think that I missed any of the requirements mentioned by the original poster. I understand that he thought a PMOS switch would be a simple solution to switch a load connected ground. The assumption might be true, if the voltage can be handled by a single transistor. But it doesn't work with PMOS due to the smaller voltage range.

As soon as you need to consider series connected switches, that require some kind of floating driver, you can better go for a single NMOS or IGBT. It utilizes a floating switch too, but with less overall effort. That's my simple point related to the original post.

Your comment about bootstrap drivers apparently doesn't refer to my post. I mentioned isolated gate drivers in general, bootstrap drivers aren't isolated.

I wonder, if forum members suggesting contactors for switching 800 V DC have ever seen a high current DC contactor. It's a really bulky component, involving a large arc quenching chamber, or made as a rather expensive vacuum contactor.

P.S.: It's not clear at all, if the nature of application (switching frequency, speed) allows contactors at all. Thus I didn't refer to it in my post.

If contactors are an option, this would be a nice one (not really cheap) but worth the money http://relays.te.com/datasheets/ev200.pdf

albert22 said:

Tunelabguy
Very interesting point.
Do you think that placing RC snubbers would form a capacitor voltage divider to equalize the voltages? As capacitor voltage dividers work in DC too.
Can varistors in parallel do the job?

Best Regards

Click to expand...

I don't think capacitive voltage dividers or RC snubbers will be effective in equalizing the voltage drop if the circuit ever has to maintain the "off" state indefinitely. For DC, capacitors cannot compensate for a mismatch in open resistance. Varistors may do the job if their leakage resistance is matched. But how are you going to ensure that? They don't generally spec two-sided limits for varistor leakage resistance.

I understand that he thought a PMOS switch would be a simple solution to switch a load connected

Click to expand...

Hi dear FvM
I think if you re look at his other posts , you'll see that he accepted to change it's mosfet to the N type . and then to the IGBT .

As soon as you need to consider series connected switches, that require some kind of floating driver, you can better go for a single NMOS or IGBT. It utilizes a floating switch too, but with less overall effort. That's my simple point related to the original post.

Click to expand...

I agree ( completely ) .

I wonder, if forum members suggesting contactors for switching 800 V DC have ever seen a high current DC contactor. It's a really bulky component, involving a large arc quenching chamber, or made as a rather expensive vacuum contactor.

Click to expand...

Arc ? why arc ? speed of switching isn't high , and there are many many contactors available for high voltages too .

P.S.: It's not clear at all, if the nature of application (switching frequency, speed) allows contactors at all. Thus I didn't refer to it in my post.

Click to expand...

He has mentioned in his other posts that just like a key .
Sincerely
Goldsmith

Can a IRG4PH50U driven by a pulse transformer with a snubber network can't do the trick?

http://www.irf.com/product-info/datasheets/data/irg4ph50u.pdf

Can a IRG4PH50U driven by a pulse transformer with a snubber network can't do the trick?

Click to expand...

Hi Alertlinks
I hope you know that the driving signal for this aim isn't pulse ! it is just a DC signal ( a command ) . as he have mentioned before .
Regards
Goldsmith

I think we agree about preference for single switches if applicable.

Arcing will mostly happen when breaking a circuit of sufficient operation voltage or inductive loads. In AC switching, the arc is quenched during the next current zero crossing, for DC, special means are required., e.g. magnetic blowouts and arc quenching chambers.

A simple method for isolated MOSFET gate drive would be a photovoltaic coupler. Unfortunately it's rather slow (ms range), so it causes additional switching losses unless combined with some kind of bistable waveform shaper. Pulse transformers can be supplemented by circuits that convert pulse trains to static gate signals. Both methods (photovoltaic and transformers with signal conversion) are used in DC solid state relays.

I think we agree about preference for single switches if applicable.

Arcing will mostly happen when breaking a circuit of sufficient operation voltage or inductive loads. In AC switching, the arc is quenched during the next current zero crossing, for DC, special means are required., e.g. magnetic blowouts and arc quenching chambers.

A simple method for isolated MOSFET gate drive would be a photovoltaic coupler. Unfortunately it's rather slow (ms range), so it causes additional switching losses unless combined with some kind of bistable waveform shaper. Pulse transformers can be supplemented by circuits that convert pulse trains to static gate signals. Both methods (photovoltaic and transformers with signal conversion) are used in DC solid state relays.

Click to expand...

Yes of course we are both agree about this issue . and about arc effect , thank you , i think i missed the point that high voltage contactors are not cheap .
ABout driving such mosfet or IGBT : i'm using this method in many of my designs
Consider please that we have an N Mosfet and then we design a low power DC supply with a tiny transformer and then connect it's ground to the source and then an opto coupler with this floated supply and then a totem pole current driver .
Best Regards and sincerely
Goldsmith

In post#4 he is talking about non repetitive pulse.

i have study IRF840 and other mosfets. those have PULSED Current parameter. i want to use that.

Click to expand...

In post#4 he is talking about non repetitive pulse.

Click to expand...

You are right. It's difficult to get a clear picture about what the original poster wants. If it's pulsed switching for a limited period, a gate transformer might work.

FvM said:

You are right. It's difficult to get a clear picture about what the original poster wants. If it's pulsed switching for a limited period, a gate transformer might work.

Click to expand...

After a quick skim through after the weekend , the thread seems to have lost the fact that he wants to be able to switch 100A @ 800V.
You can devide the voltages across mosfets how you like , or drive them any number of ways , the current is the killer here.
Normal mosfets(irf840 was mentioned earlier on) are just not going to work. I think he is going to be hard pressed to find a silicone solution!
I still think a contactor is the way to go.

Hi again
A year ago i've seen an IGBT with ability of 1200 A and 1200 v voltage and current handling . but i think it's price will be higher than a high voltage contactor .
Best Wishes
Goldsmith

That's quite impressive figures , but I think you are right , cost may be an issue. A contactor is such a simple solution to this problem , and provides isolation if needed.