My 339 datasheet says comparators in general have a tendency to oscillate due to stray capacitance. I'm assuming that would be at or near the threshold of the switching point and since I'll be watching the discharge rate of a battery bank with a 339 I would expect to be in that area for considerable lengths of time. 2mv sensitivity sounds a little touchy.

The fix suggested in the Applications Hints section is a positive feedback resistor that develops 1-10mv.

My V+ input to gnd is 4 to 5K depending on how much I need to get above the zener on my Vref. In order to develop an additional 10 mv I'll need roughly 2uA from the 12v on the output of comparator. 12v/2uA is 6 megs and I've never seen a resistor that large!

No one will be more surprised than me if I've got this right. Something tells me I don't ....

SP

 

don't you think 1mv hysteresis will do fine? afterall it does affect the accuracy, so you'll want to suppress the hysteresis as low as possible, right?
600k is fair, although 6megs is ok if you decide to use it.

 

don't you think 1mv hysteresis will do fine?
600k is fair, although 6megs is ok if you decide to use it.

With 600K being 10 times less than 6 megs, won't I get 10 times more current? That'd put an extra 20uA thru my Vin and develop 100mv of hysteresis? I'm mentally double checking my zero's as I type .... but I think that's right.

SP

 

that's right.
i thought 1 mv is better for the accuracy, we don't have to settle with 10mv, to get rid of the oscillation in the slow transition period.
the value of 600k is easy to find and fair with respect to noise immunity and all, but can you size down the input resistors 100 times?
nevertheless, 6megs is ok to go. the combined impedance will still be low.