Mr Al,
The LM358 output will swing within a few mV of its Vss pin (ground), which is not even a requirement for my circuit. It has an input common-mode range which includes ground (actually up to 0.5V below ground), so the circuit I posted will work from 0 degC on up... It will work with a tiny (~2mV) of hysteresis, which is only about (o.2degC). I specifically tried to make it as bulletproof with respect to the relay switching transients (note the source of the positive feedback) as possible by referencing everything in the circuit to Gnd (not Vdd like in the OP's circuit). I'm relying on the LM358's PSRR so that its supply voltage does not have to be regulated. I notice that I forgot to show a bypass cap across its Vdd/Vss pins.
Hi Mike,
Oh really? Well i havent been able to get one to go perfectly to ground yet and the spice
model doesnt either. Another trick that sometimes worked was to put a diode in series
with the output to drop some voltage, when it is possible of course in the app.
Also, for some values of feedback (hysteresis) resistor the output of that op amp will have
to swing negative, not just to 0.000v, so it's not a bad idea i think to use the two diodes.
I've used one diode for this same purpose too in the past because even 0.6v gets you
there, allowing close to -0.65v output without the bother of having to add another power
supply rail that is negative.