comparator with hysteresis, upper and lower trigger points

[PG1995]

Hi

Please help me with **broken link removed** query. NOTE: In the reference, given at the bottom in the attachment, there is no mentioning of upper hysteresis limit. So, perhaps there is a typographical error.

Regards
PG

 

[MrAl]

Hi,

I think you need to show that other figure in the lower reference so we can view that circuit too.

The major mechanism of the comparator is that it changes the state of the output when the input difference changes from positive to negative or from negative to positive (the input difference is the non inverting input voltage minus the inverting input voltage).
The major mechanism of the hysteresis is that is makes that input difference greater after a change in state so that it becomes harder for the input to force yet another change in state.

Also, when the input difference is positive, the output swings (or stays) positive, and when the input difference is negative, the output swings (or stays) negative (or to the lowest possible voltage level).

The whole idea with hysteresis in a comparator is to increase the voltage DIFFERENCE between the input and feedback signals to the comparator inputs.
For example, if the non inverting input of the comparator is at 2.000v and the inverting input of the comparator is at 1.000v and ramping up, once the input reaches 2.001v the two inputs are nearly at the same voltage level (inverting input slightly higher) and that causes the output to change to a new state and now the hysteresis should immediately increase the difference between the non inverting input and inverting input in a manner that KEEPS the output at that new state, which in this case has to be a low. So the output goes low and that drives the non inverting terminal lower and that makes the non inverting terminal lets say go down to 1.000v and now the difference in inputs is 1.000v and that ensures that the comparator does not react to a tiny decrease in the input at the inverting terminal. With both inputs at close to 2.000v, a tiny change in the inverting terminal could cause the output to swing back and forth but once the non inverting terminal is brought down to 1.000v (by the feedback) a tiny change in the inverting terminal now has no effect, so the output state remains stable.

Now should the inverting terminal start to ramp down (due to the external process driving the input) it will have to go all the way down to 1.000v before the comparator will again change states. Once it reaches down as low as 1.000v (or just a tiny bit lower than that) the comparator inputs are again close to equal (with the inverting terminal just slightly less than the non inverting terminal) and so the output increases (non inverting higher than inverting terminal means output goes high). The output causes the non inverting terminal voltage to rise to 2.000v through the feedback, so again we have 1.000v on the inverting terminal and 2.000v on the non inverting terminal so the output stays high, and tiny changes on the inverting terminal dont change this.

So the idea then is to make the hysteresis (which is measured as the difference between the non inverting terminal and the inverting terminal voltages after the change from one state to the other) greater than any expected noise level that could cause undesirable fluctuations of the output.
Also, the hysteresis is not always symmetrical, in that it may be different for each output state.

So here is a quick run though of the above hypothetical circuit...+ and - used to show non inverting and inverting inputs...
1. Input + is at 2.000v, output is High.
2. Input - is at 1.000v, difference is 2.000-1.000=1.000v output is still High.
3. Input ramps up to 2.001v, difference is now 2.000-2.001=-0.001v, output shoots low.
4. Output is now low, so non inverting terminal voltage falls to 1.000v, difference is now 1.000-2.001=-1.001v (much bigger difference thanks to the hysteresis).
5. Output stays stable until the input falls to 0.999v and then output goes high again.
6. Etc., etc.

It's interesting to look at step 3, 4, and 5 and more without hysteresis (no feedback just a fixed 2.000v reference for the non inverting terminal):
3. Input - ramps up to 2.001v, difference is now 2.000-2.001=-0.001v, output shoots low.
4. Input - is ramping up slowly, so it is still at 2.001v, difference is still -0.001v, output is low.
5. A little noise riding on the inverting input causes that input to go down to 1.999v (2mv of noise).
6. Input - is at 1.999v, input + still at 2.000v, difference is 2.000-1.999=0.001 so output shoots high.
7. A tiny bit of noise from the output shooting high couples back to the inverting input, cause it to go a little higher to 2.002v.
8. Now the input difference is 2.000-2.002=-0.002 and so the output shoots low again.
9. Output shooting low partly couples to the inverting terminal and pulls it very slightly low, to 1.998v.
10. Output changes state again and again!
11. Finally after the inverting input ramps high enough to reach beyond the noise the output stays stable.

So the difference was that in the first scenario we reached stability within a very short time, while in the second case we had to wait for the input to rise high enough to see stability. Obviously the first case where we use hysteresis is much better.

 

[steveB]

I could be misinterpreting the question. But, as I read it, the explanation is quite simple. One text is talking about a non-inverting input configuration and the other is talking about an inverting input configuration.

The first text shows the input going into the negative terminal of the OPAMP/comparitor. Hence, it's clear that the output will go low when the input goes above the upper hysteresis limit, and it will go high when the input goes below the lower hysteresis limit. For a non-inverting configuration, the output goes high when the input goes above the upper limit and goes low when the input goes below the lower limit.

EDIT: Also, the second text does seem to have a typo. It still seems to be talking about a non-inverting Schmidtt trigger which goes high when the input crosses the UPPER limit and goes low with the input crosses the LOWER limit. However, it would help if the Figure for the circuit is shown to be sure.

 

[PG1995]

Thank you, MrAl, Steve.

I think you need to show that other figure in the lower reference so we can view that circuit too.

The **broken link removed** given in that reference does not show any circuit.

One text is talking about a non-inverting input configuration and the other is talking about an inverting input configuration.

Yes, that seems to be the case. I didn't think about it. I had thought that Schmitt trigger always used inverting configuration.

Thank you.

Regards
PG