Amplifiers vs Comparators

hi guys, i designed a comparator and would like to test it for features like : aol,gain,overdrive,offset,slew rate and more

my question is are the test benches for amplifiers is good for comparators? and if so , why ??

tnx

They are not the same test bench as some of the measurements are different.
For example, overdrive, and response time are not normally measured for amplifiers.

Comparator overdrive is usually a spec condition for
prop delay tests. I've never seen it tested as a primary
datasheet param in its own right. Sometimes have seen
multiple prop delays at multiple overdrives (basically
small signal and large signal input prop delays).

Testing Vio, Iib/Iio, AVOL, PSRR, CMRR tends to be
done in the same way as op amps, at least the topology
of the ATE load-board is the same; an integrator and a
divider feedback. The op amp may be expected to be
stable (but this is not necessary), the comparator is
expected to chatter and the integrator just takes the
duty cycle about the logic threshold imposed at the
integrator ref input and smooths it out.

A good summary of bench measurement methods, which
I find also work well in simulation (esp. if your goal is in
part, to set / assess spec compliance to limits for product
test - small signal analysis is often too idealistic):

Analog Dialogue 45-04, April (2011)
https://www.google.com/url?sa=t&rct...BuBrNoH5BoqALXIOlfyx-Q&bvm=bv.112064104,d.eWE

tnx for the replies , i understand the tests are different and even the spec are different
but do anyone can direct me to a formal test benches especially for comparators ? or post a link for some literature of , it would be nice.
P.s - i tryed to use test benches from Phillip E Allen book : Cmos analog circuits design

tnx for the replies , i understand the tests are different and even the spec are different
but do anyone can direct me to a formal test benches especially for comparators ? or post a link for some literature of , it would be nice.
P.s - i tryed to use test benches from Phillip E Allen book : Cmos analog circuits design

Click to expand...

Hi Toli_knigin

I suggest you to have a look into the datasheet of LM311 and LM741 and compare them together , and try to simulated the suggested circuits in ther datasheets . that would help you so much .

Best Wishes

Goldsmith

Hi,
From my point of view OpAmp and CMP have similar tests in terms of time, but when the desired tests are for small-signal (AC) it is not so clear.
I think CMP literature is very underrated comparing to OpAmp.

goldsmith : i'll try to look there

ENMA : i agree about the underrated luterature

Hi,

OPAMPS are meant to operate with the voltage between both inputs to be (near) zero. Comparators are made to handle a lot of differential input voltage.
Some OPAMPs suffer from phase reversal with overdriven inputs. Comparators don´t have phase reversal.

OPAMPS are meant to operate in linear (output) region. Comparators usually work with saturated outputs.

OPAMPs usually have feedback to the inverting input. Comparators usually work without or with positive feedback (hysteresis).

OPAMPs usually have a push pull output stage. Comparators often have a open collector output stage.

*****
Therefore I say: Use a comparator if you want digital (switching) outputs, use an OPAMP whan you want linear outputs.

*****
So although they share the same schematic symbol, they are designed for different kind of operations.

*****
Now your question is how to test comparators. My recommendation is to look into a (high performance) comparator´s datasheet. There you find how the manufacturer tests them. Usually with test circuits and test conditions. Also you can see the different test parameters in the table.

I don´t think it´s usefull to test a comparator with the same parameters a an OPAMP.

Klaus

Keep in mind that many modern comparators will have small amounts of hysteresis built into them, making small signal analysis like open loop gain meaningless.

mtwieg said:

Keep in mind that many modern comparators will have small amounts of hysteresis built into them, making small signal analysis like open loop gain meaningless.

Click to expand...

I can agree with you, but the question is how would you suggest to perform a gain test?

Hi,

as there is hysteresis (i think) it is impossible to perform a gain test.
(At least the result is meaningless)

Klaus

KlausST said:

Hi,

as there is hysteresis (i think) it is impossible to perform a gain test.
(At least the result is meaningless)

Klaus

Click to expand...

Could you please explain?
Thanks.

Gain is a small signal measurement, and therefore it only has meaning when the system is operating in a linear. But hysteresis is a large signal behavior which will always manifest when the output is changing. Therefore a comparator with hysteresis has no definable gain. It also can't be characterized by CMRR or PSRR either, since those are also small signal metrics.

mtwieg said:

Gain is a small signal measurement, and therefore it only has meaning when the system is operating in a linear. But hysteresis is a large signal behavior which will always manifest when the output is changing. Therefore a comparator with hysteresis has no definable gain. It also can't be characterized by CMRR or PSRR either, since those are also small signal metrics.

Click to expand...

Thank you for your quick reply!
So how can you explain datasheets of comparators with hysteresis that still have PSRR/CMRR info?
for example - http://cds.linear.com/docs/en/datasheet/1540fas.pdf

A comparator has a switching input threshold voltage that is partly produced by its power supply voltage because the output becomes saturated to it and the output provides the hysteresis. Then the power supply voltage changes cause PSRR and CMRR effects to the switching threshold voltage.

If you do not insist that your measurements be DC small
signal, but allow the comparator to chatter away inside a
test loop (as the link in my previous post) you can get
true-enough Vio, AVOL, PSRR, CMRR results. The only fly
in the proverbial ointment is rising/falling asymmetry at the
input (e.g. one-sided hysteresis) or output (unequal drive
strengths or slew rates). These are likely to be second
order, not a big deal, assuming you test it like you'll use
it (output logic threshold, etc.).