Replace volume pot

Sorry, I mean the fault lies inside the XLR output stage, of course.

The preamp is single end only. So the cables I had were single ended for preamp side and XLR in the other end.
Went back to using a cable that single ended on both ends. Issue resolved. My guess is that if the preamp had XLR outputs and I went XLR to XLR this issue would not occur.

Marakanetz, 50k and 100k pot set at exactly same angle provide almost exactly same loudness since its operation is ratiometric - resistance have nothing to do with it, I used word "almost" because there is not two exactly same pots, even with the same value. Value is roughly selected by the reasons I mentioned before. One designer might select 50k while another might select 100k since it is not that important. It does not change gain.

Kijanki,
I've just finished restoring Accuphase E202 with faulty stereo carbon volume control. When I removed the volume control, I made a series of measurements of resistance per particular rotor position and ch1 vs. ch2 was showing differences from several upto 8KOhms. Also measured the point where rotor position presumed to have same and balanced resistance and the angle between the rotors was almost 90deg. The volume control had an interesting structure and was connected in-series with balance control. Both potentiometers(volume and balance) are 250K. Wires from phonostage are connected to 50KOhm portion of carbon surface so that volume angle will be in sync with line level components.
It was a hard struggle to find matching remote control, but instead I purchased the Alps that had sufficient enough shaft length. The size of leads were unbelievably small and it was quite hassle to put all wires back together in place of substantially better built quality vintage one. I wanted to strip similar built volume from Fisher receiver, but owner changed his mind and now paying me to restore this one too. I didn't want to do any 'surgery' to the new volume control in order to divide it onto 50Kohm portions for phonostage so that the phonostage actually picks up volume from 10...11 O'clock and onwards.

Specialized high-end repair shop would quote $450 restoration job + another $200 for fixing the balance problem. They were insisting with '100% sure' attitude that it's not fault of volume control. When techies speaking to you, they assume that you don't even know the Ohm's law. When I started to argue with technical knowlege and experience, I didn't feel that they even belong to this business. I may only assume that I spoke to the techie that only knows how to intake the job, replace things, but has no idea how to diagnose.

Visualizing the components and circuit of Accuphase that I've restored, I can easily explain how the volume is being adjusted with balance. Balance adds a resistance to one channel or another making the voltage drop higher on one channel vs. another and the volume control makes resistance higher or lower to both channels at a time. The bottom line is that the higher the voltage drop accross potentiometer the lower signal you get so I would highly highly doubt that you can mount 50K volume pot instead of 250K.

I'm leaving electronic repair tech of stage sound and studio equipment including electronic musical instruments, amps, processors and pedals. Occasionally providing restoration of vintage audio and repairs for small audio dealers if I have time or need more money:-)

Marakanetz, It looks like it is not straightforward but rather some convoluted combination of Volume and Balance with addition of Phono. I wouldn't touch it with a stick.

As for technicians - it is different (from designer) set of skills. Technician should look for faulty or misplaced component or open/short circuits but should not modify circuit under any circumstances (worst kind of technicians) since it is designer's job. Designers on the other hand don't have enough troubleshooting experience to find problems efficiently and to apply proper fix. So, when you talk to techs about circuit modification remember that they don't do it (or at least shouldn't).