This design sprung as the output of a discussion from the shortcommings of another dumpload controller. It used a 7812 regulator (which immediately limited the application to 12/24V loads), had way too many bits, used some chips that I hate, and had little (if any) hysteresis.

This design was a case of "put your money where your mouth is" - and was a 5-minute throw-together, followed up by a 10 minute prototype on pegboard to get some values for someone who was actually making one.

Yes, a 15V zener would be better.... if we were only making 48V units. But one of the design criteria was a controller that would work quite happily on 12, 24, 36 or 48V systems.

It has shorcommings, I admit. But it was meant to be a design that was sufficiently "non-critical" in components and layout that the average mug could scrounge bits from his/her junkbox and make it work. The Opamp is nothing special, and just about any will work. Two single opamps, or a dual. My prototype was a quad that I only used half of.

Zener regulator isn't very efficient, but at least it still works at 48V (or 60V) whereas a 7812 would blow up. I could have used a transistor as "power zener" but that was more complexity than was probably warranted in the application.

If you can only get 3.55 to 4.48 volts, you are not getting high enough to trigger the unit into turning on. What values do you have for R2, the pot and R4?

If it's turning on at all, it's probably from noise.

I'm suspecting either the 68K is too small, or the 820K too large....

Errr..... or the hysteresis resistor is too small! I tested it at 12V where the reference divider was much smaller. It is possible there is now too much hysteresis, and when its "off" it's pulling the input down too far.