window comparator, with hysteresis

[AndrewT]

Hi,
I know you can wire up two comparators to give a window comparator.
The lower one will operate a relay and upper one to operate a second relay.
When the voltage is in the window both relays are off.

But I don't want the relays (or fet switches) to chatter.

I would like to know how to put in a bit of hysteresis to hold the relay on/off for a significant voltage change before the comparator makes the decision to change back again.

Any links to a suitable schematic?

[PigletsDad]

Hysteresis can be added by using a small amount of positive feedback.

Simplest way is to add a high value resistor from the output to the non-inverting input, and a small one between the raw value and the non-inverting input.

[AndrewT]

Originally posted by PigletsDad
Hysteresis can be added by using a small amount of positive feedback.

Simplest way is to add a high value resistor from the output to the non-inverting input, and a small one between the raw value and the non-inverting input.

Hi,
does that work for a window comparator?
A three resistor string monitoring the variable voltage with the junctions returned to the input pins of the comparator chips. and fixed voltages, defining the window, to the other two pins.

[PigletsDad]

Sure; provided the source impedance is reasonably low compared with the low value resistor, so the currents injected through the FB path don't upset things.

[Eva]

Apply positive feedback (hysteresis) to each comparator individually. That's all. Feed the + inputs from a medium to low impedance like 1K and add a big resistor (100K to 1Meg) from the outputs to the + inputs (the output is required to swing between known voltages in order to get predictable results).

[AndrewT]

Hi,
I have tried 2M2 from output to upper non-inverting.
The resistor string is 47k, 1k0 (variable 2k5 to 0r0), 47K (~=95k).
That ratio of 2M2 to 95k seems to put in too much hysteresis. But it does seem to work as you suggested.
The lower comparator will get 2M2 to the inverting input to pull it LOWER after detecting going below the window of acceptability.

You're suggestion for low impedance seems to indicate that my first guess for ladder resistors of 47k is a bit too high.

It looks like a standard window comparator + two extra feedback resistors is all that is required. Unless someone sees I am going astray.