Comparator and hysteresis

[MRW]

Hello!

I am reading about comparators. I happened to be using this one:
http://www.national.com/pf/LM/LMC6762.html

According to the application hints in the datasheet, "generally, it is
a good goal to reduce the delta voltage on the power supply to a value
equal to or less than the hysteresis of the comparator. For example,
if the comparator has 50 mV of hysteresis, it would be reasonable to
increase the value of
the local bypass capacitor to 0.01 μF to reduce the voltage delta to
10 mV."

I understand this portion clearly. But what I don't understand is: how
do you determine the comparator hysteresis? Does this refer to the
built-in hysteresis in the comparator? Or is this related to any
hysteresis due to external networks?

By the way, I could not find any values associated with the comparator
in the operating tables.

The application hints also has a section on hysteresis, and they show
an example by putting a resistor and capacitor in the feedback network
of the non-inverting input. However, they do not indicate what kind of
hysteresis value you can get with this setup.

Thanks!

 

[Andrew Holme]

Hello!

I am reading about comparators. I happened to be using this one:
http://www.national.com/pf/LM/LMC6762.html
[snip]

what I don't understand is: how do you determine the comparator hysteresis?
Does this refer to the built-in hysteresis in the comparator?
Or is this related to any hysteresis due to external networks?

There is no built-in hysteresis. Hysteresis is provided by external
positive feedback.

The output state changes when the input level (-ve terminal) crosses the
threshold (+ve terminal).

The threshold is set by a resistive divider network, connected to the
output.

There are two different thresholds: one for output high, and one for output
low.

Hysteresis is the difference between the two thresholds.

 

[MRW]

There is no built-in hysteresis. Hysteresis is provided by external
positive feedback.

The output state changes when the input level (-ve terminal) crosses the
threshold (+ve terminal).

The threshold is set by a resistive divider network, connected to the
output.

There are two different thresholds: one for output high, and one for output
low.

Hysteresis is the difference between the two thresholds.

Thanks, Andrew! My next question would be. How do I start designing a
external hysteresis circuit and how would I analyze it?

I'm skimming thru the Arts of Electronics, and it says comparators
always use positive feedback and that the voltage levels at the non-
inverting and inverting terminals are not the same (so opamp golden
rule 1 does not apply page 230).

 

[MRW]

Hello!

I am reading about comparators. I happened to be using this one:
http://www.national.com/pf/LM/LMC6762.html
[snip]

what I don't understand is: how do you determine the comparator hysteresis?
Does this refer to the built-in hysteresis in the comparator?
Or is this related to any hysteresis due to external networks?

There is no built-in hysteresis. Hysteresis is provided by external
positive feedback.

By the way, you are referring to the hysteresis of the LMC6762 right
and not comparators in general?

I found a Maxim application note about adding extra hysteresis to a
comparator. In it, they say, "Hysteresis is designed into most
comparators, usually with a value of 5mV to 10mV. Internal hysteresis
helps the comparator avoid oscillation due to small amounts of
parasitic feedback. Though sufficient to prevent the comparator from
self-oscillating, such internal hysteresis can easily be swamped by
any external noise of greater amplitude. In that case, you can improve
performance by adding hysteresis externally."

 

[Eeyore]

I'm skimming thru the Arts of Electronics, and it says comparators
always use positive feedback

Actually a 'pure' comparator uses no feedback at all.

You get the hysteresis by addding some positive feedback.

Graham

 

[Andrew Holme]

Thanks, Andrew! My next question would be. How do I start designing a
external hysteresis circuit and how would I analyze it?

I'm skimming thru the Arts of Electronics, and it says comparators
always use positive feedback and that the voltage levels at the non-
inverting and inverting terminals are not the same (so opamp golden
rule 1 does not apply page 230).

1. Connect the top of a potential divider to the comparator output
2. Connect the middle of the potential divider to the non-inverting input

The comparator output is either high or low. You can get the high and low
output voltages from the datasheet. Once you have these, you can calculate
the voltages at the non-inverting input.

 

[Andrew Holme]

Hello!

I am reading about comparators. I happened to be using this one:
http://www.national.com/pf/LM/LMC6762.html
[snip]

what I don't understand is: how do you determine the comparator
hysteresis?
Does this refer to the built-in hysteresis in the comparator?
Or is this related to any hysteresis due to external networks?

There is no built-in hysteresis. Hysteresis is provided by external
positive feedback.

By the way, you are referring to the hysteresis of the LMC6762 right
and not comparators in general?

I found a Maxim application note about adding extra hysteresis to a
comparator. In it, they say, "Hysteresis is designed into most
comparators, usually with a value of 5mV to 10mV. Internal hysteresis
helps the comparator avoid oscillation due to small amounts of
parasitic feedback. Though sufficient to prevent the comparator from
self-oscillating, such internal hysteresis can easily be swamped by
any external noise of greater amplitude. In that case, you can improve
performance by adding hysteresis externally."

Yes, some comparators have internal hysteresis.

 

[JeffM]

My next question would be. How do I start designing a
external hysteresis circuit and how would I analyze it?

I'm skimming thru the Arts of Electronics, and it says comparators
always use positive feedback and that the voltage levels
at the non-inverting and inverting terminals are not the same
(so opamp golden rule 1 does not apply page 230).

You're so close, it's painful to watch.
Just a couple of more pages...
http://books.google.com/?q=noting-that-the-hysteresis-equals

A sawtooth on the input is a good test.

 

[neon]

hysterisis is the amount of feedback from the output to the input it can be calculated as voltage F/B lets say out is 10v the input has a divider of 2 10k resistor a 100k F/B resitor will give you the ratio of F/B