Bilateral Switch AC/DC 1000V/10A

[BitSplitter]

Hello all,

I need to make one bilateral switch that is controlled by micro-controller.
When the switch is ON, it must have very low resistance (<100mOhm) and to be able to withstand current of 10A.
When the switch is OFF it must be able to withstand 1000V.
Switch is bilateral so it needs to function both for AC and DC.
Also, switching speed should be high.

Does somebody have some idea which components to use (some MOSFETs...) or is there such component
already available on the market.

Thanks in advance!

 

[iop95]

What is switching speed?

 

[schmitt trigger]

Back to back n-chan mosfets driven by a photovoltaic gate driver like the FDA217.
They are not fast, though.

 

[mtwieg]

A pair of antiseries mosfets is the simple way, but you're looking at some very pricey components to get 1kV and 0.1ohms, since even high quality 1kV FETs typically have resistances of several hundred mOhms.

 

[BitSplitter]

Hello all,

thanks for your replies.

Yes, I had the same idea to use N-channel MOSFETs. This solution is fast enough, robust and small enough but very expensive.
If fact to achieve the required specification I had to use combination of 8 MOSFETs (some in parallel to reduce the RdsON, and some in series to achieve the 1000V spec.).

Do you think there is some cheaper solution?
Also, the size matters, using some big relay or similar is not possible for my application.

Do you think that combination of some "smaller" relay, some MOSFETs, PTCs, varistors and similar could do the trick?

Thanks!

 

[FvM]

Yes, I had the same idea to use N-channel MOSFETs. This solution is fast enough, robust and small enough but very expensive.

It's probably the cheapest. What are you dreaming of?

The specification is still vague. Besides switching speed, which you apparently don't want to specify exactly, how about switched power? 10A * 1000V can imply upto 10 kW, do you intend to operate the switch under load? In this case, a simple photovoltaic gate drive won't work (immediately kill the MOSFET by exceding SOA), you need to go for some kind of active gate drivers, possibly DC/DC converters for it's power supply etc.
Also specification of off-state behaviour may be important, leakage current, acceptable capacitance.

 

[BitSplitter]

Hello FvM,

thanks for your interest in my problem.
I'll provide more details:
In normal operation, I need to measure the current (1-10A) in some circuit.
Therefore, my switch is normally closed and the shunt resistor (10mOhm) is right next to it.
It is necessary that voltage drop on the switch and the shunt (when I=10A) is <= 1V.
That means that switch resistance must be <= 90mOhm.

Now the problematic part:
The rest of the circuit is problematic and that's just the way it is, and no changes can be done on this matter.
It can happen that sometimes (by an accident) 1000V can appear between switch contacts (actually, switch + shunt).
This condition must be quickly detected and the switch must be turned off immediately.
I'll have one fuse next to the switch. Actually, accident situation, mentioned above, can happen frequently.
Fuse is too slow, and even if it would be fast enough, changing fuses frequently is not an option.
By project specification, measurement circuit should recover fast from this accident and to be able when needed to start
measuring current again.

This is not an easy problem, possibly it is not feasible at all.
I need to analyze it, and to have good arguments if I decide to say NO to the customer.

P.S.
Driving a switch is another subject. First, I need to find out what kind of switch to use.

Thanks!

 

[mtwieg]

Now the problematic part:
The rest of the circuit is problematic and that's just the way it is, and no changes can be done on this matter.
It can happen that sometimes (by an accident) 1000V can appear between switch contacts (actually, switch + shunt).
This condition must be quickly detected and the switch must be turned off immediately.

You need to clarify this more. You're saying you sometimes expect to see 1KV appear across the switch+shunt alone, while the switch is turned on? What kind of extraordinary event could force 1KV across 0.1ohms? A direct lightning strike? Or do you actually mean 1KV across the load which is being switched?

 

[FvM]

Presuming the information in post #7 is correct so far (otherwise we should simply stop the discussion), this additional information is needed, I think.

- maximum current that the "1000V" source could drive into the circuit, in other words source inductance
- dV/dt of the switch connecting the 1000V source
- maximum dI/dt that can be generated in the circuit, in other words circuit impedance
- peak current and fusing integral ∫I2dt tolerated by the circuit

 

[KerimF]

Is it hard monitoring the voltage of the switch in your project?
I use this method to protect instantly the MOSFET of my inverters (from current overload).

 

[BitSplitter]

You need to clarify this more. You're saying you sometimes expect to see 1KV appear across the switch+shunt alone, while the switch is turned on

That's exactly what I'm saying. It can happen when the load gets short circuited. Then in the branch you measure current you only have switch + shunt alone. Then, switch must be turned off quickly.

- - - Updated - - -

Is it hard monitoring the voltage of the switch in your project?
I use this method to protect instantly the MOSFET of my inverters (from current overload).

Voltage from the shunt will be monitored by some hardware. If this voltage is out of range, the switch will be turned off.
Please, tell me what method exactly are you using?
Thanks!

 

[KerimF]

Please, tell me what method exactly are you using?

I meant simply by interrupting the MCU.

But in my case I do it via a buffer circuit since I should disable this detection when the inverter transformer and the MOSFETs work as charger.