[digger945]

First, you won't change the gain on the ground channel, only the two amp boards.
For the Zobel, you need one resistor and one cap per channel(two of each for you).
Have you checked to make sure everything is grounded properly inside the amp? Diagrams for wiring are at the link amb posted above. One missed ground connection will cause the problems you are having.
Could you post a pic of the insides, or do you have a pic from the sale?

 

[oneplustwo]

For full disclosure, this is the B22 that I had posted about previously that had oscillation problems with my IEMs (they were handy so I had just used them for testing purposes). I never had any hum or oscillation issues with my Grados or Senns however. Neither did any of my friends with HD650's or HD800's. I've been talking with boozcool to figure out the best way to proceed and have pretty much come to the same conclusions as folks here. Gain is too high and a zobel network is likely required. I had also asked him to confirm the ground loop breaker was still installed correctly.

 

[boozcool]

Here are some pics of the inside

I was unaware about the oscillation issues with IEMs prior, but the Audio Technica AD2000s exhibit the most distortion and have the issue at the higher or lower volume settings. Oddly enough, the impedance of the Audio Technica AD2000s is higher, at 60, while the impedance of the Denon D2000s is 25, but the distortion is significantly less with the Denon D2000s than the Audio Technica AD2000s
My experience with soldering is minimal at best, so hopefully lowering the gain and adding a zobel network is not difficult?

 

[linuxworks]

I've always (well, on both my builds) added zobels.

they can only help and cannot (just cannot!) hurt. they should be in every build, imho.

also, on that black chassis b22, you may be better off running the audio cables along the side of the chassis, AWAY from the trafo. those are the sensitive inputs and you want them away from the ac hum area.

I would remove the molex connectors on the left (center chassis) area on all 3 b22 boards and move them to the right side closest to the metal chassis side. then run wire along there.

for the pot, I would use shielded wiring since its a 'long-ish' run.

 

[AtomikPi]

Even though the Denons are lower impedance, the AT's are higher sensitivity, so they exhibit hiss more easily. The AD2000's are particularly revealing of hiss, because they're some of the most sensitive full-size dynamics.

 

[digger945]

Why is there a wire going to the input of all three channels?
The ground channel(the one closest to the PS board I presume, since the other 2 have what looks like ground wires going to the attenuator) should have nothing connected to it's input. It should be grounded accordingly, and to the other boards(daisy chained or star like amb explains), but have nothing connected to board signal input.

 

[digger945]

Maybe the plug is on backwards, and you can just turn it around(if that's the ground board of course). I would confer with the builder first.
It may only fit one way, so you might also try to shift it over towards the other board and just let the wire contact the ground pin.

I would still make sure everything is daisychained on the ground side, even if this fixes the hum problem for now. The gain changes won't fix things that are not grounded.

 

[digger945]

Do you have an DMM?

 

[boozcool]

I will confer with the builder
My friend has a DMM I could borrow

 

[digger945]

I'm referring to the black wire. It looks to me like it is just run from the ground of board 1 to the input of board 3.

I gotta hit the sack. I'll check with you guys sometime in the morning.

You get a dmm and we can have this thing fixed by tomorrow evening.

 

[boozcool]

Perhaps this picture could show the run better

It seems the 'GND' from Board 1 splits to 'GND' on Board 2 and 'Input' on Board 3
The 'Input' from Board 1 and 'Input' from Board 2 runs to the pot

 

[oneplustwo]

The board closest to the PSU is indeed the ground board. (EDIT: boozcool's picture shows the gnd wire connected to the input, not the GND... should be an easy fix.)

It's possible the attenuator is not grounded to the enclosure. I honestly can't remember for sure.

As for the wiring, the reason I had the input wires down the middle in the first place is that I had originally had output wires to speaker posts (which are no longer there) going around the right side of the enclosure. I wanted to keep those separate from the input wires as well, thus the routing. (This was also the shorter route.)

Now that those output wires are no longer there, it would be make sense to have the wires run on the outside vs. down the middle.

There should be enough slack in the wires to at least move them to the other side most of the way to see if it makes any difference at all directionally at least.

boozcool and I will figure out next steps. Thanks everyone for the input.

 

[boozcool]

Switched the GND wire from input to gnd. No change in hum, and hiss with the AD2000s at highest/lowest volume setting is still there, but has improved slightly

 

[digger945]

The boards are isolated from the attenuator. I see signal in(white) and signal out(purple) of the attenuator, but no ground. Check that next, and make sure the attenuator ground is connected to board ground.

 

[hiker101]

From the photos, the ground channel seems to have R1 populated. AMB's specs for a unity gain ground channel involve leaving out R1.

Is the gain setting of that channel intentionally non-standard?