Hi folks.
This is on the tails of a circuit I made to adapt an old fuel gauge to match a new style tank sender - which works in reverse. The old gauge uses two coils of 100 ohms each and a 150 ohm resistor to 10v. The old style sender pulls down the centre point to move the needle between full and empty.
My circuit reverses the action so that full reads full and empty reads empty by operating a N-Fet in the linear region as a current sink. This seems to work fairly well.
My problem comes when turning the ignition off. When the gauge is static at a reading of the tank and the supply is turned off, the voltage supply to the gauge and also the electronics begins to fall. Somewhere along the curve to zero the Fet ceases to conduct - at which point the gauge needle pegs to the Full position as it sees an open circuit which equals full. I am trying to stop this behaviour.
I've tried adding a P-Fet controlled by an op-amp as a schmitt trigger to spot when the supply falls and conduct to bring the needle down, but in the Spice model I still see a phantom peak as the op-amp gives up.

I need to find a way of pegging the control to the gauge down to zero, or removing the 10v supply to the gauge when it starts to fall below about 8v. Basically it doesn't have to work below 8v supply.

Thank you for your time and suggestions.

 

Schematic?

 

Basically it doesn't have to work below 8v supply.

Below 8V on the supply, what happens? The needle will stay in the location that it is in and it does not respond to the voltages anymore?

If you can delay the turn off of the FET longer than the decay of the supply from 10V to 8V, are you suggesting the problem would go away?

 

My crystal ball is in for cleaning.
As Alec noted, we need a schematic.

 

Sorry I'm not so good at uploading files here yet.
Try this from LtSpice, or I'll take a photo tomorrow.

 

Try with C4 connected from the U2 non-inverting input to ground instead of to V+.

 

Try with C4 connected from the U2 non-inverting input to ground instead of to V+.

I tried this in LT spice, and looked good, but in practice the gauge now pegs to Full when first turned on, which is not an improvement. Maybe something else as well as moving C4?

I've tried changing the design to use another Nfet to keep a Pfet in the on state, the idea being that when this Nfet stops conducting (which is the main issue) the Pfet gate is pulled down and conducts. In practice this doesn't work either.

I have the feeling there is something blindingly simple and I can't see the wood for the trees.

 

Found a solution today.
The main difference was to slow down the action of the main output Fet by putting a 10k resistor in front of the gate, and a 100uf cap to ground from that. When turned off it had the desired effect of keeping charge on the gate and keeping the gauge reading low until the supply voltage dropped. However this created a new problem of the gauge pegging full as soon as the power was turned on because the cap was charging slowly through the 10k resistor and I had created a slow turn-on circuit here. The solution to this was to inject a supply straight onto the gate at turn on via another PNP. The base being used to charge another cap, so when full it turned off. a 1Meg resistor was used beside the PNP to drain the cap after use. It was enough to keep the Fet saturated quickly at turn on.
After that swapping C4 to the new position as Alec_t suggested worked even better.

 

Good to know you have a solution and that the C4 position swap was at least a step in the right direction .