Switch for 200kHz, 600W, 2% duty

mosfet for 200khz switching

Colleagues,

Newbie to the power electronics here. I need to design a switch that will connect a 200kHz signal with 600W 2% duty power to a load resistor. The switch will connect the load resistor for a 100us at most. The initial signal comes from the unbalanced source to a hand-wound balun, so the voltages are flexible.

Could you point me to topologies for such switch?

Thanks,
- Nick

The specification isn't completely clear to me. You have a 200 kHz AC voltage that has to be connected to a load by the switch? Or do you want to design the 600 W power stage? In any case, has the load impedance a defined value?

FvM said:

The specification isn't completely clear to me. You have a 200 kHz AC voltage that has to be connected to a load by the switch?

Click to expand...

Yes, that's the case.

FvM said:

In any case, has the load impedance a defined value?

Click to expand...

The load will be purely resistive. Nominally R=230ohm, but that can cnahge depending on the transfromer design.

Maximum permitted switch on-state voltage, isolation in off-state and switching speed are additional parameters to know.

A single FET with s bridge rectifier or two antiparallel FETs are a common solution. For short on-time, as specified, a transformer coupling of gate voltage could be most simple.

FvM said:

For short on-time, as specified, a transformer coupling of gate voltage could be most simple.

Click to expand...

Frank, could you explain this a little further. I've never heard of this technique.

I don't know yet, if a FET switch (respectively an antiparallel series circuit of two FETs) is suitable for the application.

Using a pulse transformer to drive the FET gate has some limitations:
- the gate voltage must have zero mean value
- the on-time is limited by transformer maximum ∫udt
- the transformer leakage inductance in connection with gate capacitance forms a resonant circuit, a series R is necessary for damping and the rise/fall time is limited
- stray capacitance and isolation voltage may be problem when switching high voltages

But it's a robust and simple method, if the limitations are tolerable by the application.

I assume, that a transformer driver at the primary side isn't an issue, but the transformer specification must be well thought.

FvM give two good ways. Given that you have a 370 V, 1.6 A load it isn't so difficult to find MOSFET for this application.

If you prefer to drive only one MOS you can use a bridge of high speed diodes. Anyway to drive two MOS in this application is not so difficult because they works in the same way so you can simply make a transformer with two secondary windings.

About the parallel: MOSFETs have a built-in diode, normally it's reverse biased, but when you connect two in antiparallel you have two diodes in antiparallel too. And they're ready to conduct. So you have to add a diode in series with each MOS.
Or you could use two MOS in antiseries. You could connect the sources togheter and drive both with a center tap transformer. Never heard but I don't find anything wrong.

Switching speed: a MOS for this application could have less than 1 nF of total stray capacitance, with 10 ohms in series to avoid ringing it seems not impossible to have short switching time.

s bridge rectifier or two antiparallel FETs

Click to expand...

FvM could you explain these two circuits a little further? I googled both, and did not turn up anything useful for an s bridge rectifier.

And, to the best of my knowledge, an antiparallel connection for FETs would be the Drain of Q1 connected to the Source of Q2 and vise versa. Is this correct? What is the advantage of this circuit?

Thanks

kboltz said:

s bridge rectifier or two antiparallel FETs

Click to expand...

Click to expand...

I think, "s bridge rectifier" was a typo for "a bridge rectifier".

a bridge rectifier

Click to expand...

yes, obviously. The circuit is simply a bridge rectifier with N-FET drain at + and source at -. Rectifier AC pins are the switch terminals.

antiparallel is probably the wrong term, I fear. I don't remember the origin, but some guys used the term for back-to-back diode circuits since ever. Anti-serial would be more exact. I apalogize for possible confusion.

However, trapoe got it right. I mean a series connection of two FETs with common source. The drains are the switch terminals then.

Regarding transformer coupling of gate drive, an active electronic pulse former could be used between transformer and gates, but this would be beyond the simple circuit I wanted to suggest.

I'm probably side-tracking this conversation, but theoretically antiparallel MOSFETs could work too. The topology would be similar to a transmission gate. However, there could be some practical problems:

- MOSFETs with exposed substrate (4-terminal MOSFETs) are needed. They are relatively rare.
- Negative supply rail
- Level shifters (U3A and U3B)
- Switched signal has to always stay between V+ and V-

- MOSFETs with exposed substrate (4-terminal MOSFETs) are needed. They are relatively rare.

Click to expand...

I think 4 terminal MOS are only signal MOS, I never saw power MOS with 4 terminal. This could be the main problem. And when the MOS are off you need a "strong" driver or you risk that stray capacitance (Cgd) switch on the MOS during fast edge of 200 kHz signal.

Switched signal has to always stay between V+ and V-

Click to expand...

Switched voltage is around 370 V !

If you use "normal" 3 terminal power MOS (with source and substrate connected) you can use antiparallel adding diodes or antiseries, see my previous message.

In a transfer gate, the complementary FET is used for performance purposes (lower and constant Ron, cancellation of gate charge). A single FET is baiscally able of switching bipolar signals. However, as all power FET are three-terminal type, you need the discussed anti-serial circuit or a rectifier.

Just found an article on antiseries FET switch. Here it is **broken link removed**