Question about high-side MOSFET circuit w/o ground connected

[dwolsten]

Hi,
I'm working on a circuit which involves a MOSFET being used in a simple relay. The FET is N-channel and connected in a high-side configuration, and being driven with a driver chip that uses a charge pump (LM9061). Overall, it works just fine, except when the ground connection is disconnected. When this happens, the FET conducts power to the load.

I'd like to modify this design so that this doesn't happen; when the ground is disconnected, the FET should remain off. Is there a way to do this? Strangely enough, when I tried shorting the Gate to the Source, it stayed on, on two different types of FETs.

Any help would be greatly appreciated!

 

[Tahmid]

Hi,
Connect a resistor in between gate and ground, maybe. Usually a resistor is placed between gate and source, but there the ground isn't disconnected.
Tahmid.

 

[keith1200rs]

Re: Question about high-side MOSFET circuit w/o ground conne

Hi,
Connect a resistor in between gate and ground, maybe. Usually a resistor is placed between gate and source, but there the ground isn't disconnected.
Tahmid.

It would be useful to see the circuit. It the circuit is ON when the gate is shorted to source it sounds like you have diode conduction.

Keith

 

[dwolsten]

Re: Question about high-side MOSFET circuit w/o ground conne

Hi all,

Thanks for the tips. I tried various resistors between the gate and source, to no avail. I also tried this on other identical units with the same result. I even tried shorting the two and it still conducts, which seems very strange to me.

I've drawn up a simplified schematic showing the relevant parts:


The circuit is powered by +12V, and has a 1N4007 diode protecting everything but the MOSFET. There's a 5V regulator powering a microcontroller, which switches the LM9061 high-side driver on and off. There's various components associated with the LM9061 which provide features like overcurrent protection (by sensing the voltage drop across the MOSFET) and a delay for that overcurrent protection to kick in.

Thanks again!!

Added after 42 minutes:

I've done a little more experimenting and have found a few data points:

1) If I lift pin 5 of the LM9061 high-side driver, nothing changes.

2) If I remove the resistor connected to the source of the MOSFET, nothing changes.

3) If I disconnect the MOSFET's gate, it stops conducting. (Obviously, this isn't a solution!)

I've also tried connecting various resistors and capacitors between the source and gate, with no success.

Added after 7 minutes:

Ok, I've found something pretty significant. If I remove the MOSFET altogether, but leave all the other components in place, it still conducts. It looks like it's going through the LM9061 chip somehow. It's really odd though: if I look at the internal diagram on the LM9061 datasheet, it shows the inputs from the source and drain (connected through ~1K resistors) going into the inputs of a comparator. Comparators and op-amps usually have very high input impedances, so they shouldn't be able to conduct across their inputs, I would think.

It looks like this might be more of a problem with National's chip than with any MOSFET.

 

[keith1200rs]

I was going to suggest a resistor from gate to source, but you have already tried that. Also, the LM9061 can't supply much current so the value would have to be quite high.

If it conducts without the MOSFET then your load current must be going through the 1k sense resistor? If so, you load must be quite high resistance. Maybe you could increase the sense resistor?

Keith.

 

[stefannm]

Re: Question about high-side MOSFET circuit w/o ground conne

So, I guess I don't really understand the use-case here, but can you answer a few questions.

1. When you say ground is disconnected, do you mean to your whole board?
2. If the above is true, you can only get conduction because the load is still connected to ground. Does disconnecting ground from your board not result in ground being disconnected from the load? Can you "connect" the load and board grounds?

It seems to me that with ground disconnected, the gate of your NMOS will float (pulled toward 12V i would imagine) which could lead to conduction. Of course I cannot explain conduction to the load with no mosfet on the board. Are you sure the current is going to the load?

stefannm