High side Mosfet swich design

[AdrianM]

I need to switch a 30Amp resistive DC load from a 5V microcontoller pin. This isn't PWM just occasional on/off.

I'm familiar with the design considerations for high speed switching - rapidly getting charge in and out of the gate to minimise time spent in the linear region but in this application I'm pretty sure I can safely simplify the design a great deal because it's only used as a discrete on/off control.

What I've come up with does seem uncomfortably slow so I thought I'd ask around in case I'm missing something!



Q1 would actually be IPD50P04P413ATMA1 a 12mOhm P-Channel MOSFET, 50 A, 40 V. The total gate charge is less than 50nC so charging/discharging at around 5mA should take no more than 50nS x 1/0.005 = 10uS

This keeps it in the top right corner of the safe operating area:



The DC input could be as high as 30V so to limit the gate voltage low swing, the 470R source resistor makes M1 act like a constant current sink (its threshold being around 2.5V). R1 maintains around 10V across it until the supply drops below 12V

LTspice sim of ~30uS switch-off for part with even higher gate charge (120nC) seems OK:


Green is gate voltage, red ouptut voltage and blue power dissipation in Mosfet. Switch-on should be a bit sharper and it is at <10uS:



Too good to be true? I'm so used to the pitfalls of Power PWM design that this simple application has me jittery

[eccerr0r]

Damn. I was looking for some ideas of high side switching and found this thread... grr PFET on V+ rail is low side switching. Not what I want. I want an IRF640 on the upper rail switching 150+ volts, controlled by a microcontroller. Need tens to hundreds of microsecond switching speeds - not insanely fast but not leisurely. Came up with a solution with no special components, but need to simulate it, and perhaps see some other ideas that don't involve custom driver chips...

Anyway I don't understand R2 other than perhaps setting up a voltage divider. Having that there will kill performance of both transistors as it would limit how fast you can get the charge off of their gates.

But do worry that the voltage divider is somewhat necessary due to exceeding the maximum gate voltage of Q1. Perhaps just moving R2 to the drain side of M1 will speed up performance and still maintain gate voltage limits.

[budm]

1) Since M1 control the ON/OFF switch for Q1. How fast is M1 switching?
2) "R1 maintains around 10V across it until the supply drops below 12V" If M1 switch is OFF, then I cannot see how you can maintain 10V across it.
3) When M1 is OFF then the Q1 Gate will be discharged through R1so if you want to discharge it faster the put diode in parallel with R1, Anode on top end of R1.
4) BTW, are you loading the SIM library with the actual spec of the component?
https://www.vishay.com/docs/70611/70611.pdf

[redwire]

OP's circuit, R2 is silly as it is limiting to 5mA and R1 is too low and a P-ch mosfet? not sure how the circuit even starts, it's prob. the substrate diode always on
I always put in a gate zener, because you need to overdrive the gate to get good speed.
Something else that works for driving high-side mosfets is photovoltaic opto-coupler, these have a tiny solar cell to make 10V for gate drive, at very low current though.

[eccerr0r]

Original circuit is fine with the P-channel with source to V+, that's how it's normally implemented, problem is that M1 won't saturate due to R2 - at first when the gate voltage is turned on, it works fine, but once it's on it will shut off because source voltage will increase... This is actually horrible and no wonder why it's working so poorly. So: short out R2, and if you don't want to dissipate as much power, put R2 on the drain lead so it will also reduce gate voltage of Q1 (though I've not looked at the data sheet for its maximum Vgs.)

Alas I'm sorry, I necroed an old unanswered thread, but still wondering about my issue - finding other solutions to driving a high side N-channel mosfet switching 150V from a 5V signal with no custom ICs : just parts out of a 2-transistor switching power supply or at least that's the type of constraint I'm thinking about. So no photovoltaic isolators, though a photocoupler is acceptable.

I suppose winding a transformer for the high side drive is also a possibility though I'd like to avoid transformers if possible, and if something comes up good with a transformer, I guess I can go with that - just no photovoltaics and no 'high side driver' chips...

[budm]

Yep, R2 should be on the Drain of M1.

[eccerr0r]

Okay, finally looked at data sheet but I shouldn't have bothered: Not only that R2 needs to be on the Drain of M1, it needs to be increased to 1KΩ if V1 is really 30V.

and #@*%(#! can't get away from using p-channel MOSFETs or PNP transistors in my high side design (I was tempted to figure out how to incorporate a CD4069 but of course those have p-channel MOSFETs in them). I guess that's the way things go, but at least I don't need high power devices for the driver, just enough to pass maybe 1A for a very short amount of time...

[stj]

if you use a single "logic level" fet with a 2v gate voltage, the microcontroller can hard shift the gatepin from 5v to gnd - that will switch the fet much faster than using a pullup resistor or a current limiting resistor to ground.

[eccerr0r]

But will that work well with a p-channel switch on the power rail that's well above 5V say 12V?

The second mosfet is there specifically for the level shifting needed to drive the p-channel. At least this is much simpler than a high side n-channel driver...

[stj]

good question, not sure.

[budm]

Originally Posted by eccerr0r

But will that work well with a p-channel switch on the power rail that's well above 5V say 12V?

The second mosfet is there specifically for the level shifting needed to drive the p-channel. At least this is much simpler than a high side n-channel driver...

The Gate Voltage of the P-CH has to be < the Source Voltage = to the Vth for full turn on of the MOSFET but it cannot be higher than the G-S breakdown Voltage. It is P-CH so we are pulling the Gate down not pulling up.
See the attached PDF in previous post on how it is done.

[eccerr0r]

The answer is that the lower Vth devices do not help if the source voltage of the MOSFET is not referenced to ground, like the original post.

Any MOSFET on the high side has special problems to deal with. A P-channel with source at a fixed value (rail) is a bit easier to work with, at least the voltage is fixed.

Now my problem, a N-channel device on high side, the source voltage changes and likely above the voltage of microcontrollers, making it a much tougher problem. I've yet to find any good solutions other than requiring photovoltaic isolators, transformers, or special high side driver ICs. It would be interesting to figure out how to make a circuit that works like the high side driver IC.