N-channel MOSFET problems

[microbug]

I am building a boombox which runs off a 3s Li-Poly battery. These are damaged if they are discharged too low, so I have designed a circuit which has two indicators (low and critical). This uses an LM358 configured as a comparator, driving an amber and a red LED. The anode of each LED is hooked up to a 5V source and the cathode goes through a 150 ohm resistor to the output of the LM358.

Since the most power-hungry device in this project is the amplifier, I want to disconnect it when the critical LED comes on. I have some N-channel MOSFETs (datasheet attached), and have been trying to get them to work, using an LED as the load. I attach the gate to the output of the LM358, the drain to the cathode of the LED and the source to 0V. When I do this the LED turns on and the MOSFET blows (the drain and source short). I have gone through 4 of my 10 MOSFETs so far so I think there's something wrong!

The fact that the LED turns on when the gate is "low" could be because the output of the LM358 goes from 0V to about 1.5V under the load of an LED, so the gate voltage is around the threshold voltage of the MOSFET. However, the datasheet says this can vary between 1.3V (min), 1.7V (typical) and 2.15V (max). It seems unlikely that the threshold would be below 1.5V for all 4 MOSFETs.

Complicating matters is the fact that my bench is next to our house's main switchboard so there is lots of 50Hz noise. My power supply is a LASCAR PSU 130, which is a fairly cheap switch-mode supply (datasheet attached). I am aware that I need to be careful with static around FETs, so I have been regularly touching my scope's ground banana terminal. When I touch this at the same time as the cathode of the LED which the MOSFET is driving, the LED turns on! I think that this could be related to the problem but I don't know that much about ground loops or whatever... I imagine the power supply is not isolated.

Any ideas on what to do?

Thanks!

[madires]

Please take a DMM and measure the voltage between the PSU's negative terminal and the scope's ground.

[HackedFridgeMagnet]

For developing circuits you probably want to use a current limited supply.
It is easy to make a bad circuit and the current limit will save you. If not now then soon.

Surely your devices are failing on over power or over current.
You seem to want to be using your FET as a switch.
If you are running near the gate threshold then you aren't using the FET as a switch. Use a higher gate voltage.
With the FET properly turned on, if it is still too much current or power then add some more series resistance.

As always a schematic image will get you more help.

[idpromnut]

A couple of things:
- The LED should probably have a current limiting resistor on it (at least while testing to avoid this particular problem).
- Could you attach the exact circuit as you have it now?
- The gate of the MOSFET may also be building up a charge, so attaching a 100K resistor to ground to it may help keep it low when it needs to be (but that is probably not the "right" solution, just a noob's suggestion).
- I would also probe the gate pin on an oscilloscope and see if the op-amp is oscillating.
 

[Simon]

Yes you never mentioned any resistors on the LEDs and yes a floating gate will charge up you need a grounding resistor

[microbug]

The two resistors running from the LM358 each have a 150 ohm series resistor. The LED running from the MOSFET has a 560 ohm resistor.

The gate pin of the MOSFET was not oscillating last time I checked, but I will again.

I am in the process of putting my schematic into Upverter, and will post it here when I'm done.

@madires: the voltage between the negative terminal of my PSU and earth (the scope ground) oscillates at 50Hz and is about 100mV.

[microbug]

I attached the schematic. The text on the bottom-right should read, "Amp will be connected in place of this".

You can find it on Upverter here.

EDIT: I will try adding a grounding resistor.

[idpromnut]

Disconnect D2 and try it again (with the grounding resistor on the MOSFET's gate).. that resistor+LED (R2+D2) pathway to V+ is charging up and slamming the MOSFET on regardless of what the op-amp is trying to do. This is assuming that REF != GND, so the best that the op-amp can do is slam the output/MOSFET gate as far down as the value of REF (ish) which is probably not enough to shut off the MOSFET.

[Zero999]

This probably isn't the problem here but there's no hysteresis and the LM358 isn't the best IC to use as a comparator. Ideally you should use a proper comparator IC such as the LM393.

[microbug]

It works! Thank you very much @idpromnut!

REF is 1.222V but without a load the op-amp can go very close to 0V... I am aware that an LM393 would be a better choice but this is just a rough indication that the battery is low.

I'll figure out how to get LED back in.

Thanks!