building an active balanced ground?

[porthillsbomber]

Power supply is just the one I drew using a tle2426 with a couple of caps on the output. Haven't tried with both inputs connected, I'll hook in up to my computer now so I can play some test tones.
 
My guess is just a mistake in the wiring or a dodgy bread board.

 

[porthillsbomber]

I just ran some test tones and there is a little bit of crosstalk not much though, I suspect this is either due to the bread board or possibly the input currents of the ground channel amp allow some cross talk through R1 and R2, I'll have a think about it.

 

[Armaegis]

Would buffering the three channels make a difference?

 

[porthillsbomber]

Yip I just thought about it and input currents wouldn't make a difference but you could possibly get some voltage bleeding through R1 and R2 of the order of the (source impedance)/R1 so yes buffering the inputs will make a difference. I've got one more opa2132 so I'll use that to buffer the inputs and see what the results are.

 

[Armaegis]

I originally ran my L/R inputs through some opamps (I think in inverting unity gain config), but it didn't seem to do much. How about adding current buffers to the three outputs?

 

[porthillsbomber]

Ok I just added a non inverting input buffer and still had crosstalk but then I removed one channel from where it connects to the headphones on the breadboard and the crosstalk was gone so the bread board is definitely causing the two channels to couple. Having rectified that I tried without the input buffer and the crosstalk appeared again so an input buffer is needed if you're driving it from something with a high output impedance ie a pot or unbuffered dac

 

[Armaegis]

I posted this over on amb and this was his reply...
 
Quote:

As soon as you make the ground channel's output swing voltage which are derived from the left and right channels, your add stereo crosstalk (because that "ground" is shared between the two channels). Not only does this scheme not remove "ground contaminants", it produces them.

 

 

[porthillsbomber]

Well yes I would agree this method is never going to give you less crosstalk but whether it adds any significant amount of crosstalk is a different story, I would doubt it adds anymore than you get with a headphone cable that has a shared ground.

 

[Armaegis]

Well I've got all the parts to build a B22... maybe once I complete a regular one and feel more confident, I'll order another kit and try this active balanced ground thing.

 

[porthillsbomber]

I don't think it would be possible with a beta22, you need opamps to do the summing and subtracting. I guess you could put unity gain beta22s inside the feedback loops as current buffers but that's probably a huge waste of money. If you wanted to make a good one of these I would just use some good opamps and use some BUF634s or similar inside the feedback loops.

 

[Armaegis]

I think you misunderstood how I'd use the b22. I would use the opamps in unity gain (or maybe 2x) to create the OG/OL/OR, then feed those three signals into three separate b22 boards to drive a headphone. Does anything seem wrong with that?
 
Along those same lines, are any opamps better at driving high impedance/line level inputs (ie: the b22 input impedance) rather than headphones which are comparatively low impedance? I think I've read in a couple spots that opamps work fairly well either way, but some perform better in that application than others.
 
If I do a full b22/s22 build, the s22 has a fourth output which I could use to power the "active balanced ground" circuit (maybe run it through a s25 transformer to bring it down if needed).

 

[porthillsbomber]

Ok yes you could do that but you would end up with crosstalk again if the gains were not well matched. To do it properly you would create OG with the opamp then feed it to a beta22 board which would create OG2. You would then use OG2 for the left and right opamps instead of OG in the schematic I drew and feed OL/OR to 2 more beta22 boards which would be referenced to OG2.
 
Yes some opamps are definitely more suited high impedance loads than others

 

[Armaegis]

Wait, what? I'm not sure I understand what you're doing here. This is how I had it pictured...
 
source (L/R) --> volume pot --> active balanced ground circuit (OL/OR/OG) unity gain --> 3 board b22 fixed gain
 
Alternatively, put the volume pot in front of the b22, but then you need to control all three simultaneously (do they make 3 way pots?) and volume matching becomes critical. Stick the pot at the beginning and mismatch won't affect the ABG circuit.

 

[porthillsbomber]

You're right about the volume pot but I mean like this
 
source --> volume pot--> (buffer) --> ABG ground OG gain G1 --> beta22 ground OG2 gain G2 --> ABG left/right OR/OL gain G1 --> beta22 left/right OL2/OR2 gain G2
 
The gains G1 and G2 could be whatever you want (the total gain being the product of the two)

 

[Armaegis]

So you want to feed OG2 back into the ABG circuit... but doesn't that amplify the G portion again? It's not going to balance out on the headphone end unless everything is unity gain.
 
Your OG2 is also simultaneously feeding the headphones and as well as back into the b22. In my case OG2 is untouched except for the headphone out.