Here's my latest listening results on the CNC.
I compared it directly to my battery powered P06, using it as a yard stick.
It was very good but I noticed it did not have the gain of the P06 and also some uncomfortable graininess in the middle to upper regions, which was only noticeable in direct comparison. But as my wife says now that I noticed it, it has to go.
I tried a few different values for the gain resistors from 75ohms upwards and found 100 ohms a perfect match in the first gain stage. It has the gain now of the P06, which has given it a bit more beef.
On to the graininess, working on the assumption that it's probably something simple as the design is sound, I started experimenting.
I found the input resistors made no difference so they were left as is, I then tried the output caps and it was better with them in and this value seems ideal. So I then removed the output loading resistors with the idea of swapping values. I tried listening without them in first and viola sweet sweet sound, no graininess at all now.
I then did another comparison between the "CNC" and the P06 expecting them to be nearly identical.
I was wrong it is the P06 that is now slightly on the grainy side compared with the "CNC". The sound stage is extremely wide and 3d with as much detail as I have heard mixed with a ever so slightly mellow sound that is very pleasing to the ear. Exactly what the AD823's should ultimately sound like or even what the venerable OPA627/637's are described as.
Who's the idiot that put those loading resistors in the first place?
It seems they don't agree with the output of the op amps.
Back for more listening. The sad thing is I am going out of town tomorrow until next Wednesday/Thursday so I won't be able to do any more testing until then.
Oh and I tried a LED connected across the batteries with a 8.2k resistor and it works fine, not sure how much juice it takes so I will have to measure it with my amp meter, it should be very low, but like others have said it's pretty necessary.