my external-input amp (revised)

[MACHINE MEDIA]

I stayed out of this the first time, but I am glad to see Matt has given 
up on inventing 150db single stage mic pre :)

> >that; any lack of connection as you turn the pot will open
> >the feedback loop causing the output to slam to one of the power
> >supplies.  Ie., it'll be *really* scratchy as you turn the pot.

Yes, it would.  Is there any reason why this thing must have variable 
gain by placing a pot in the negative feedback loop??  Couldn't you place 
an attenuator (a voltage divider based on a pot) at its output??  If you 
place it at it's input it will amplify the noise of that pot (and if I'm 
thinking about this correctly vary its input impedance).  Output 
impedance can be kept low by placing a second op-amp at unity gain after 
the pot -- or have I lost my mind??

> Good point. OK, I've changed the value of the resistor going from (-)input
> to gnd, and changed the pot value, and added a resistor in parallel. With
> the pot at max resistance, the combined resistance will be 833k. With the
> 82k resistor to ground taken into consideration, we get a gain of 11 or so.
> With the pot acting as a short, we get a gain of 0. If the pot goes dead
> momentarily (i.e. it's scritchy), we have an effective resistance of 1M,
> which would give us a gain of 13. So scratchy pots aren't going to be all
> that noticeable.

Wouldn't that throw off the response curve of your pot??  What about 
using a fixed resistor in series instead of in parallel??

> >Also, the point of the clipping zeners is to keep the opamp from
> >raw clipping, but that won't happen when the zeners are after the
> >opamp.
> 
> So where would they go? Before? In the feedback path? You had mentioned
> diodes (apparently of the non-zener variety) in some of your comments,
> which I added at the input stage and quoted you. Should these be zeners
> instead? As for the ones on the output...won't these protect my other
> modules in case something dire goes wrong?

This thing seems to have an awful lot of overload protection.  You 
already have the input AC coupled and limited with a pair of diodes -- 
what were you planning on sticking into this thing?? :)  Nothing is 100% 
foolproof.  Just don't build any edison-to-1/4" adapters!! :)

If your input minus any offset is limited to 1.2V p-p just keep it so 
that your maximum gain times 1.2V is less than the maximum output voltage 
of the op-amp (which in all cases is less than the supply voltage).  
Placing zener diodes at the output won't keep the op-amp from clipping 
internally, and will distort the signal under a varying load (zener 
regulators bend under current -- if you must use a pair of zeners 
(something I've never seen in a pre-amp) you can follow it with another 
op-amp as they have low and stable input impedance).

Gene Stopp and Don Tillman both brought up the term "audiophile" -- a 
word I don't like that can't apply to your situation.  Why?? Because your 
application is too general.  You are designing an amp to be used with an 
unkown input to drive an unknown load.

However, if you are interested in producing a stable low noise design, 
there are a few suggestions that I would like to offer.  First off, low 
noise often has more to do with what is built around the design, than the 
gain stage itself.  Books like "Noise Reduction Techniques In Electronic 
Systems" by Henry Ott and "Grounding and Shielding" have whole chapters 
on things like PCB layout and the geometry of placing bypass capacitors 
(the idea being that increasing the distance and increasing the 
interaction of components both increase noise).

Since you aren't building some sort of pico-amp instrument gizmo for the 
Space Shuttle you need not be so fanatical.  Even in audio terms, your 
requirements aren't that critical -- your input signals aren't that low, 
you aren't using tranformers (imho, still the best way to build a mic 
pre), and your system isn't balanced (although using high impedances and 
short cable runs seems fine for what you are building).

These are my suggestions:

1.  Use low loise, low drift, op-amps.  Choose op-amps screened for low 
noise rather than high-precision.

2.  Use an uncompensated op-amp with the extra cap -- this may reduce 
unwanted oscillations.  You also do not need megahertz bandwith.

3. Use an op-amp with high PSRR -- this is the amount noise rejection at 
the power supply pins.  I've audibly reduced line hum in modifying gear 
just by swapping the op-amps to those with better PSRR.

4.  Use op-amps with high CMRR -- especially if they DC coupled.   

5.  Even with high PSRR, use bypass caps anyway, especially if you . . .

6.  Keep circuitry carrying the signal away from the power supply.

7.  Use the highest supply voltages you can -- usually +/- 15V.

8.  Despite #3 & #5 -- build a quiet power supply.

9.  Use only shielded cable (grounded at one end) to carry the signal 
even for short internal runs.

10. Use a grounded metal chassis -- preferably made from a ferrous metal.

Well, that enough for now :)

I hope this has been helpful.

PEACE OUT :)
MARK