questions from a newbie (Op Amps again)

[Harry Bissell]

Tim Daugard wrote:

> 2) Opamps differ in many ways. Important specs include
>
> >Gain Bandwith Product... This means that an op-amp can be used for a low
> >frequency signal with very high gain, or a high frequency at low gain
> >(1 * 1000  =  1000 * 1) again, bigger is sometimes better
>
> A recent piece of paper from TI argues against this point. For audio
> applications they said that smaller is better. Wide gain bandwidth means
> more noise gets amplified while not actual having any benefit for the
> signals being used. Noise can add to distortion and modulate signals or at
> least continue to be fed down the path.
>
> To reduce noise they recommended an input filter to restrict inputs to the
> frequency range desired and tailor the feedback network to keep the gain in
> the band of frequency desired.
>
> Reasonable or just an engineer who had to write some thing for the latest
> advertising piece?

NO! its reasonable indeed... that's why I said "sometimes better".  The trade
offs
might involve phase shift in the band of interest... can you stand that ???
(probably)
and will you get slew-limited (not strictly a function of GBW, but related...)

Just arbitrarily using a faster op-amp is a recipe for oscillation (and noise)
A phaser I designed using a (yuck) LM324 and upgraded to a TL084, oscillated
ultrasonically because of the higher gain at high frequencies... Drove the dog
nuts !!!  I almost left it that way TO drive the dog nuts... I hated that dog
!!!!

Many vintage synth circuits will not prosper with a faster op-amp than
originally used, because the circuits are often designed to boost the high
frequency gain of the opamps involved (see LM301 with feedforward comp... used
in a LOT of filter circuits). Remove
the feedforward and just use a faster op-amp and you might be fine...

My point was that you NEED to decide on which specs are appropriate, and use
that op-amp.  As I said... the 741 is still useful for me.  Not too much
gain... LOTS of output
current, short circuit limited, and clips without "phase reversal" which is
from nasty (not good nasty) to fatal !!!

>
>
> >Single vs Dual supply... some opamps need a plus/minus supply, some are
> >happy with one...
>
> I don't see the difference in this. The best I can figure, A dual supply op
> amp will put out a signal balanced around the arbitrary center  (+V-V)/2
> while the single supply op amp isn't optimized for balance. ??

NO! not the same !!!  The LM324 (yuck) has a crossover notch in the output
response
that can be severe, especially on a bipolar supply. Where it occurs (right near
ground) it adds a constant distortion that is level independant... so with a
small signal it is a HUGE distortion....

Some list members (to the 324's credit) use it for voltage processing apps...
where the crossover is virtually unnoticible. It is very low priced....

>
>
> >Output Current... Can it drive the load you will be using ???
>
> I have never looked at this spec. Every thing I've done so far has had a
> reasonable input impediance. I hooked the output of an LM324 to a 10Hz low
> pass filter today and it seems to be working okay - I'm not sure what the
> impediance is - I forgot to check - should be on the order of 100 to 200
> ohms.

The LM324 (yuck) has a nasty trait... which is why I look at output drive
current. It has
different source and sink currents... could be a problem in symmetrical
circuits (like limiters) or if you are driving a capacitive load.  Again... you
got to watch your @ss here.

>
>
> >Output Short Circuit current protect... will it blow up if you shory it
> >out... This can be real important in a modular synth, where you cannot
> >predict what will be connected...
>
> So far I have hung a 1K resistor off all outputs to avoid problems with
> this. What is everyone else doing?

Yep... thats what I do... but sometimes I want to drive a capacitive load as
fast as I can
(like peak detector...) and I can use an op-amp with output current limit...
and allow the
output to go into current limit on the fast transient.  So I would get a linear
response instead of an RC response.  With careful design you can get around the
output current problem (scale up impedances, etc...) but sometimes cheap and
dirty wins...

The 1K resistor shows up in series with the input of the next stage... with
100K summing
inputs that would be 1% error... not too good for a CV. Even worse with 50K
inputs...
I've seen some synth inputs as low as 10K... Maybe you care... maybe you don't
...

BTW the 741 has plenty of drive and bandwidth... as long as you don't need gain
too...

I think its time for an OPAMP primer here... whaddya think???

H^)  harry

>
>
> Tim Daugard