Re: Need an op-amp expert !

In article <8hfr24d0uk1bojue1r4rkl8at4mqnmsdem@xxxxxxx>,
Jim <troubleatwork@xxxxxxxxxxx> wrote:

All,

I think I've messed up an op-amp (actually a comparator) design at
work.

I'm using an LMV321 (dual low voltage op-amp) operating at 4V.

Using one half as an amp, it seems to amplify a 50KHz square wave to
about 1V peak to peak quite happily.

Then comes the trouble !

The 1V square wave is connected to a clamp (DC restorer) with the
output square wave bottom side approx 1.6VDC (so top side should be
2.6V). That connects directly to the - input of the other half of the
LMV321 used as a comparator.

The reference voltage (2.1V) connects to the + input of the comparator
via 22K. There is a 330K feedback resistor from output to + input for
Schmitt trigger action.

The output of the comparator seems to change the mark/space duty cycle
(weird). The problem gets much worse at cold. Adjusting the voltages
slightly on the comparator inputs seems to vary the output duty cycle
(weird).

My best theory is that an LMV321 is not fast enough as a comparator
(even though it amplifies the 50KHz to 1V peak to peak quite happily).

Everything I've read through the years, strongly suggests that using
op amps as comparators is a poor idea. The symptoms are usually as
you have experienced - the op amp's speed as a comparator is often
much poorer than its gain-bandwidth product would suggest, and the lag
time can vary unpredictably.

I gather that the reason is that when you use an op amp as a
comparator (i.e. open-loop) you're pushing its internal circuitry into
overdrive/saturation. When the signal turns around, it takes the
differential input stage quite a while to pull enough charge back
through the various internal capacitances to cause the output to
change state.

Op amps are designed to be used in a closed-loop environment, where
the voltage difference between the inverting and noninverting input is
very close to zero (maintained by negative feedback). When you use
'em open-loop (as a comparator) the differential input voltage is much
higher, the internal circuitry saturates, and (in some cases) you can
damage the differential input stage.

"Real" comparators are designed for fast open-loop operation, with
very high differential gain, high tolerance to large differential
input voltages, and fast (and predictable) output switching behavior.
They don't have to be designed to be stable in a low-gain closed loop
environment (they make lousy op amps!).

Overdrive recovery time for an op amp is often _not_ a characterized
number - you won't find it in the data ***, and can't depend on
whatever value you happen to measure for a given part. The
next batch of the "same" parts you get from the same manufacturer,
might behave differently.

I have cunning plan. An LMH6658 is an RF op-amp that is pin compatible
and shouuld drop in for the LMV321.

Has anybody any experience with using the LMH6658 as a comparator ?
Will it make an OK fast(ish) comparator ? Are there any circuit tweaks
to speed up the LMV321 ? Do op-amps get slower at cold (-30C) ?

The only really reliable solution I can suggest, is that you use a
real comparator.

--
Dave Platt <dplatt@xxxxxxxxxxxx> AE6EO
Friends of Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!
.