Mosfet problem (possibly)

Thread Starter

rjjenkins

Joined Apr 16, 2011
162
Hi I'm having trouble with the circuit below. The idea is that an 5V input is regulated down to 2.5 V to charge a supercapacitor, C3.
The supercapacitor then powers the load (an MCU). However, the load is disconnected when the supercapacitor is charging. This is to allow the load to be hard reset in case of a brownout (which is likely as the supercap voltage drops).

The disconnection happens because the P channel mosfet gate is taken high when the input voltage is applied, and stops conducting from S to D. The rest of the time, the gate is pulled low through R2. This all works fine, and indeed I've used the same principle when charging a LIPO cell (using a LIPO charging regulator rather than a voltage regulator). While the voltage regulator is absent from the circuit, the mosfet does what I would expect, turning on most of the time unless the gate is brought high. However, when the regulator is in the circuit, I find that current seems to flow from C3 back through the MOSFET to the gate, incidentally illuminating the LED so long as there is charge in C3, keeping the gate high and shutting off current to the load.

I'm not sure whether the problem is with the regulator or the mosfet or what.


1599848932728.png
 

jpanhalt

Joined Jan 18, 2008
10,221
The problem is the voltage regulator:

1599851480263.png

Vout which is connected to the C3 can back-feed to the gate via the body diode of the control mosfet in the MCP1700 and Vin. You might try a Schottky diode in the Vout line, of curse any diode will give another voltage drop.
 

Chris65536

Joined Nov 11, 2019
224
You might try a Schottky diode in the Vout line, of curse any diode will give another voltage drop.
He could choose a slightly higher regulator voltage to compensate for the added diode.

Edit: Or would that trickle charge the supercap to a higher voltage, potentially damaging it?
 

jpanhalt

Joined Jan 18, 2008
10,221
That would depend on the diode and the supercap's self-discharge rate. If that rate is a mA or so, then an ordinary low-voltage diode should be OK.
 

Thread Starter

rjjenkins

Joined Apr 16, 2011
162
Thanks. I don’t know the exact self discharge rate of the supercap, but it’s low (which is what I want). If I went up to a 3.3v regulator and used an ordinary silicon diode, would that work?
 

jpanhalt

Joined Jan 18, 2008
10,221
I think so. You need to consider the self-discharge rate of the cap, the diode's Vf/current curve, and the P-mosfet's Vth. I have done none of those adjustments, because we don't know the super cap's self-discharge and whether that is even a reliable metric (it's easily manipulated). However the forward-biased diode is an easy and common solution to back-feeding. Worth a real world try, IMHO.
 

Thread Starter

rjjenkins

Joined Apr 16, 2011
162
Thanks. What puzzles me about this is that I've used the below circuit, which seems identical in all relevant respects, without difficulty. Of course in itself this circuit makes no sense - I already had the boards made up for charging a LIPO but changed it to charge a supercap by shorting out the LIPO programming resistor, and adding the voltage regulator. But it works.

1599859901328.png
 

jpanhalt

Joined Jan 18, 2008
10,221
MCP73831:
1599861767388.png

Do you know what "direction control" consists of? Another diode?

EDIT: This is the TI bq21040 -- a similar genre. Notice the blocking diode.

1599862113939.png
 
Last edited:

Thread Starter

rjjenkins

Joined Apr 16, 2011
162
I think I get it - the diode in the MCP73831 prevents current flowing back to the mosfet gate? In which case can't I put a diode in my original circuit on the VIN of the regulator, where voltage drop doesn't matter?
 
Top