buffer-splitter / opamp protection diodes

[Brett Sinclair]

Howdy,

I want to build a good buffer/splitter that will be the first in my chain and that may have to drive some amount of cable / go though quite a few TB effects.
Heck i want to build something that i know that is certainly over-specified so i do not need to be paranoia 

So i came up with this, which is basically the AMZ Super Buffer:


Can someone confirm i got it right concerning the +9/-9 supply (i'll use a MAX1044 with 2x 10uF caps) and GND? The input impedance is about 1.8 Meg, right?

Also, i thought it would be a good idea to use them diodes as described in RG's article "When good opamps go bad"... does it make sense or not to use them in this circuit? Maybe it's overkill, but does it hurt?

Anything else i can add to drive up the part count? 

cheers

[R.G.]

Oh, there's always a way to use more parts! 

If you want to make sure you can drive cable, use the LM833 opamp. It has enough current capability for driving 600 ohm lines. So does the NE5532, but the 5532 doesn't have as high an input impedance.

Better, along the lines of using more parts, the ne plus ultra of opamps is still discrete. Look up "JE-990 opamp" on google. And have a look at http://www.10000cows.com/JE990.pdf. The JE-990 was designed specifically for low noise high performance audio in difficult situations. It's still arguably the best opamp for the applications if made well.

[Brett Sinclair]

 

Thank you for the reply RG!

[head_spaz]

I would do it this way.
This provides input protection to +0.7V and -0.7V of their respective rails...
which should protect your opamps... without limiting your input signal.
You had diodes across the inv/non-inverting inputs... but in normal operation
these inputs will always have the same differential. If not, then you have problems
that those diodes won't fix, or protect.
Lowering the input impedence will reduce the johnson noise, which will be substantial
even at 1 Megohm.
The output pot is optional. Without gain you probably won't need it... but with
+/-18 volt rails, and the headroom those rails will provide, you might do well
to add some gain to the circuit and then attenuate with a pot when needed.
Just my 2 cents.
Hope this helps.

[head_spaz]

BTW... I seriously doubt that a single MAX1044 will drive 4 opamps, especially if you plan on driving 600 ohm impedance loads.
You might want to consider using two 9V batteries instead, or build an external powersupply using a wall wart.
If you decide to use two batteries, I'd recommend hooking them in series and then split them with a "Virtual_Ground" circuit, this way if the batteries fatigue at different rates, you will still have a balanced supply. You can use just two resistors, or two resistors with an opamp to provide more current upon demand (minimizing the drain on your batteries), or even better yet check out Texas Instruments' nifty little TLE2426 rail splitter in a simple T0-92 package. Super efficient!http://focus.ti.com/docs/prod/folders/print/tle2426.html

Hope this helps.

[Brett Sinclair]

Thanks for that!

A question though... isn't the 47n input cap going to roll off a lot of bass?

I also saw the TLE2426 used in the MINT headphone amp (http://www.tangentsoft.net/audio/mint/) which gave me an idea for the BurrBrown BUF634 in the feedback loop instead of stacking opams?