Apheared 47 bypass cap

[tiagoncalves]

Hello! I'm rebuilding a A47 amp for my Grado's S80. With my previous layout I get some hiss, or hum, when nothing connected to the amp, and it increased if I connected just the RCA cable to the amp, but not to the source. Can it be picking up RF? Also, If I connected an wall powered source, I get some noise also, that was reduced if I touched the case...
So, now I'm reducing the gain of the amp, and I would like to get the bypassing right. Hope someone could help me understanding, because I read a lot, but can't understand what's the best option for this case. This is the schematic of what I have thought to build:
 

 
I was thinking in 4x 0,10uF ceramic bypass caps from rail to ground near the op-amp pins, and 2x 10uF tantalum also from rail to ground before the op-amps. But after reading many posts about it, and Tangent's documentation, I notice that with this rail splitter option, it's not advisable to use capacitors after the buffer. So, what's the best way to bypass in this situation?
And, about the grounding location, should I use one single point, near the input RCA jacks, or use one point for the output jack ground and another point between the TLE2426 and the op-amp for the other grounds?
This is the board layout I'm starting to build, but I accept sugestions.
 

 

[greyhorse]

Wow, this one brings back memories...
I remember making one without the GND buffer a long time back, and it sounded good.  I basically used just the TLE2426, with 100uF bypasses.
I personally haven't had much luck with the sound of GND buffers, and prefer a solid stack of capacitors instead.
 
Grounding:
I'd ground to the case right near the input connectors, and run the thickest ground bus that I could to everything else.
As for GND bypasses I guess in this case you'd only have rail to rail bypasses, and forego much capacitance to GND?
 
To Note:
-Everyone says tantalum capacitors short out when they fail, so I suppose it's good practice never to use them.  Enough people have said this anyway, that I've never used one.
-That 1N4002 diode is in the exact WRONG position.  I'm assuming it's there as reverse battery protection, but it's doing the exact opposite for your batteries.  While your circuit might survive, it's basically dead short for your batteries.  That should be placed in series with the power switch.
 
Best of luck.

 

[tiagoncalves]

Yes, this is not new stuff, but I've built it some years ago and I'm very pleased with the sound quality. Now I'm building a new case, and I'm reassemble the board layout. I would like to fix the noise that I was having with the previous one.
So, ceramic bypass caps from pins V+ to V- near each op-amp, is it the best way? I would like to know the difference between rail-to-rail and rail-to-ground bypassing, especially in this situation. Any modifications suggested for this schematic?
 
About the diode, thanks for the advice! I used it in the previous one just like this, but I don't know why. For this to work between the rails, I would need also a fuse in series with one side, right? Well, it appears that I had never used the cell in the wrong way. So, I'll remove the diode, it's not really necessary...