Comparators vs. op amps

[[email protected]]

LM339 is as prone to the phase inversion as LM358.

BTW, I have seen a design where a spare section of LM339 was used as
opamp. They put a big RC at the output so it made the LM339 compensated.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultanthttp://www.abvolt.com

I'll second phase reversal as a serious issue when using op amps as
comparators. This is often overlooked.

 

[[email protected]]

mmm I was thinking ua710 comparator, from what I remember of it's diagram
it had few components.

An evil device, now well and truly obsolete. As far as I am concerned,
the first useful comparator was the - relaitvely slow - LM311, which
preceded the LM339 (not to mention the LM3930) which was even slower,
throwing away of a lot of performance for low price and stability.

As the last time I looked at that was < 30 years ago, or perhaps 40? it may well have
been simplified in my brain over time.



Yes you are probably right.. my view was as when you needed less linearity,
then you could use less - , or same number, of stages, with diff amps with higher gain.....

Linearity isn't an issue with a fundamentally non-linear device like a
comparator.

OK let me see, and take this challenge on the ua710:
http://www.datasheets.org.uk/search.php?q=UA710&sType=part
TI has 2 diff stages

And the 709 (from the same era, the 710 seems to be from 1972, so 35 years, I was close )
also seems to have 2 diff stages with some extra stuff added, so my memory checks out OK
that 709 was more complex then 710 (I used both):
http://www.datasheets.org.uk/search.php?q=UA709&sType=part&ExactDS=St...

In fact, I think the uA710 was more nearly contemporary with the uA702
op amp. The uA709 was the first op amp that had enough gain to be
widely useful, but the architecture was evil, and it got replaced by
the uA741/8 and LM301/307 which were much better.

It seems to me that offset in a comparator is very important, but indeed you
are right, there are many opamps with very low offset.
But I would not want a comparator with high offset, or offset drift.

T\If you look at what is made and sold, the market seems to have other
priorities.

 

[[email protected]]

I know of one comparator that has back-to-back clamp diodes on its
inputs! But opamps and comparators use similar processes, so have
similar constraints.

This is a bit misleading. Op amps are constrained by the requirement
that that have to be stable when used with a fair bit of negative
feedback, which limits the number of stages of voltage gain that they
can use. Comparators aren't used with negative feedback, so the
designer can add extra stages of voltage gain - which does add extra
propagation delay, silicon area and so forth, so they don't go for
many more stages.

Many fast comparators have very limited input
differential voltage specs and often have weird common-mode rules.

Comparators tend to have lower gains, because they only need gains in
the ballpark of 1000, just enough to guarantee a solid output when the
inputs differ by more than the offset spec. And they are designed for
speed, another reason to keep the gain and the number of stages low.

But the balance between delay per stage and gain per stage does seem
to optimise around three stages of voltage gain.

 

[[email protected]]

This is a bit misleading. Op amps are constrained by the requirement
that that have to be stable when used with a fair bit of negative
feedback, which limits the number of stages of voltage gain that they
can use. Comparators aren't used with negative feedback, so the
designer can add extra stages of voltage gain - which does add extra
propagation delay, silicon area and so forth, so they don't go for
many more stages.



But the balance between delay per stage and gain per stage does seem
to optimise around three stages of voltage gain.

I'd have to do some digging, but there is a classic equation that
relates the speed of the comparator to the number of stages. Generally
more stages are faster if you are using the comparator in its "linear"
region, i.e. close to zero volts. It is not a situation where you are
propagating logic, but rather you have a number of stages moving
linearly a the same time, each one amplifying the other. If you want
more bandwidth, you use less gain per stage, but more stages. This way
written up generally in strobed comparator design, when you off-set
cancel each comparator stage in one phase, then compare in the next
clock phase.

Of course, now I made a fine mess for myself since I need to dig a
reference on this.

 

[Jan Panteltje]

An evil device, now well and truly obsolete.

I used it in a small oscilloscope design, where it did the trigger level,
and also the delayed timebase slice IIRC.
That was a < 5MHz oscilloscope, my second the first one had tubes,
and the first one I did with dual timebase.
I am not sure if it used DTL logic or RTL..... DTL I think.

There was a related IC to the ua710, the ua711, it had 2 comparators,
so you could make a window comparator.
I liked those chips, they saved many transistors.... reproducable.

As far as I am concerned,
the first useful comparator was the - relaitvely slow - LM311, which
preceded the LM339 (not to mention the LM3930) which was even slower,
throwing away of a lot of performance for low price and stability.

Yes LM??? rings bells of limited to VLF applications ;-)