Switch with a MOSFET

[OlivierNum]

Hi guys and gals!

My name is Olivier and I just discovered this blog a couple of days ago. It's amazing! I have a background in electronic but I have been more involved in the programation side. Now, I have a new assigment and I need some advice to choose the right path.

To make it simple, I need an electronic switch that is fast (ns or us) that can withstand 0-600V at 0-3A. Should be drived by a signal generator at maybe 20kHZ max (most of the time will just be on or off). The load is resistive, from less than an ohm to multiple kOhms. The test is destructive and should not last more than a couple of ms.

I know that with a N Channel MOSFET it is easier to put the load on the low-side but for security reason I would prefer to be on the high-side.

Do you think it's feasible?
Is using a gate drive a viable solution?
Do you have a better solution?

[richcj10]

It is easier to use a high side driver:
http://www.digikey.com/product-detail/en/FAN73611MX/FAN73611MXCT-ND/3025296

[Kremmen]

A gate driver is not only viable/easier but necessary as well. Using a N channel FET as a high side switch means the gate voltage is the highest potential in the system, exceeding the source voltage at least by some ~10V to ensure full turn-on. When the FET is turned-on, Uds is close to 0 implying that the gate voltage must be above Ud. That extra voltage is not directly available usually, necessitating a high side driver with a charge pump or similar arrangement to generate the gate voltage.
Also it means the switch gate potential will be high, i.e. dangerous. So an isolating gate driver circuit is a real good idea.
Many of the commonly available drivers have a maximum isolation level of 600V which is just enough in this case. Personally i would be more comfortable with a significant margin in max permitted voltage. However it seems to be the case that while 600V parts are a dime a dozen, higher than that is not so easy to find. I know of a couple of Ixys chips but neither is available from e.g. Digikey.

A classical way would be to use a pulse transformer to drive the gate. With proper construction the transformer can take the voltage difference and provides a reliable galvanic isolation. Of course that solution has its own limitations an is perhaps a bit cumbersome these days, but it would definitely work when implemented right.

[Corporate666]

I would just use a P-Fet on the high side, and use an NPN transistor with the collector connected to the FET gate, and tied to Vin with a resistor.  Then a resistor on the base of the transistor connected to the output of your function generator.  Am I missing something about the design?  Does it need to be an N-Fet for some reason?

P-FET's are available through Digikey and others cheaply with a working voltage of 600V and >3 amps.

Just need to check switching time/Rds and make sure you have adequate heat dissipation.

[tron9000]

I would just use a P-Fet on the high side, and use an NPN transistor with the collector connected to the FET gate, and tied to Vin with a resistor.  Then a resistor on the base of the transistor connected to the output of your function generator.  Am I missing something about the design?  Does it need to be an N-Fet for some reason?

P-FET's are available through Digikey and others cheaply with a working voltage of 600V and >3 amps.

Just need to check switching time/Rds and make sure you have adequate heat dissipation.

I agree with corporate and rich here, you could use an ITS4142 , purpose built high side switch, or use a p-channel FET on the high side of your load, which will be pulled down by an n-channel FET (2n7000 or 2n7002 are small and cheap enough).

The problem with the latter solution is that you have to connect a resistor network between the high side of the p-fet to the source of the n-channel fet,  so that when the n-fet pulls down, the voltage drop across the resistor between the gate and source (R7) of the p-fet is greater than or equal to its Vgs at the current your pulling through it.

[OlivierNum]

Thanks guys I'll look into that!

[OlivierNum]

I'm back!

I want to try a couple of thing before commiting to a solution.

I made a circuit with a gate drive an I would like to know if it looks right? Can't seem to find any particular information on a basic configuration in the datasheet ... I've also look in this Application Note. I can't seem to find any anwser in it too.

I would like to know if this configuration would work. I choose R1 and R2 from what I saw arround. I saw some schematic where no Rg was used... The datasheet seem to use one. As for Cboot I have no idea if I should use one and if so what value should be used for a simple low frequency switch.

If you could point me in the right direction it would be greatly appreciated!

PS I choosed the part randomly on Digikey so if you think another one would be better just tell me.

[muvideo]

Hello, that should work.

Some random considerations: keep in mind that while on high state, the gate
of the mosfet relies on the charge on Cboot capacitor, so you have
to switch the mosfet off periodically to replenish Cboot. The penalty
is an huge dissipation in the mosfet while Cboot discharges below gate
threshold voltage, tuning off slowly... unless the driver has some
undervoltage safety cut-off.
The time depends on Cboot value and the leackage of the gate/driver itself.
R1 doesnt help in this, you could use an higher value, it's not really needed,
but it keeps the gate to source just in case something goes wrong...
You should use a resistor in series to Dboot, to limit charge current.
If you plan to go fast, better use a fast recovery part for Dboot.
Put another high value resistor in parallel to D1, to make sure that
Vs goes to ground while the part is quiescent and the output is disconnected,
high voltages and high temperatures could increase the leackage trough
the mosfet preventing Vs from going to ground, and keeping Cboot
correctly charged.

Hope this helps, good luck! 

Edit: about the values, the appnote you linked contains all
the details about the values you need. I didnt made any
calculations, but ballpark values could be:
Dboot UF4007,
Cboot 1uF electro in parallel to 100nF ceramic (but you could
try higher valued electrolitics, I didnt check if it's a problem
for the driver),
resistor in series to Cboot 10ohm,
R1 470k,
R2 10ohm,
Resistor in parallel to D1 100k 2W.