[sdiy] Newbie questions about summing buffers

[David G Dixon]

> Scott J mentioned the input impedance issue, but otherwise 
> the equations for the op-amp all work out the same. I 
> remember doing an experiment with a 741 when I was first 
> learning about op-amps. I built an inverting amp using 100R, 
> 1K, 10K, 100K, 1M and 10M resistors. What I learned was that 
> apart from the two extremes, I couldn't hear any difference. 
> The extremes were noisy. I also learned that the 10K, 47K, 
> and 100K values that I commonly saw in circuits were right in 
> the middle of the range, and so I've used similar values ever since.

At 100R, you'll get excessive current noise.  At 1M or 10M, you'll get
excessive Johnson noise (I think that's what it's called).

I tend to use mostly 100k in my home-built through-hole circuits, but have
found that 10k to 30k is better in SMD circuits.  I still don't know why,
but figure it may have to do with layout issues.

> In general, it's better to add the capacitor. After all, in 
> general you're not trying to amplify radio waves, so you 
> might as well tame the top end. I usually leave space in a 
> circuit to add the cap, even if sometimes I don't fit it. 
> Stick in something that rolls everything off above whatever 
> you regard as the top end of audio (15KHz, 20KHz, 25KHz?) and 
> you're done. I tend to use a fairly low value for this since 
> I'm not a fan of crisp bright treble anyway, and a lot of 
> 'warm', 'analogue' gear has a frequency response that 
> basically nosedives above 10KHz!

I too almost always make room for caps next to the inverting and output pins
of opamps, which is very easy to do in my layout style.  However, bear in
mind that there are a few situations where you definitely do not want to use
feedback capacitors -- on unity-gain inverters in filter cores, for example
-- as they can sometimes introduce spurious behaviour.  Experiment!