Why is balance in my pre inaccurate at low levels?

Your preamp probably functions as designed. My Counterpoint preamp behaves similarly at low volumes. Here's some info from the FAQ section at the Alta Vista Audio web site regarding this "feature" of volume pots (and high gain tube preamps).

Q: When I turn the volume control down on my preamplifier, one channel goes off before the other. What is wrong and can it be fixed?
A: Continuous-track (non-stepped) potentiometers cannot track perfectly. At some low level, one channel will always go to zero (fully off) before the other. The manufacturers of the parts will tell you this. No manufacturer specifies low level tracking to any degree of accuracy below -50 or -60dB. It's just too hard to manufacture parts that track well at those low levels. Some, however, are better than others. The TKD parts that I use for upgrading Counterpoint gear sound much better than the Noble 25mm parts used in most Counterpoint gear, but track even less well. The tracking of stepped resistor attenuators is essentially perfectly, but the limited number of steps and the large physical size of the parts make them impractical for most Counterpoint preamps. If this mis-tracking at low levels is a real problem for you, you might want to locate another potentiometer and have a technician install it, but it might take a lot of work to find one that tracks nearly perfectly. If you do find a source of good-sounding volume pots that track well, I'd appreciate hearing about it.

VTL pre-s do not require matched tubes. The small signal tubes run at very low bias current thus have the longest life possible.
The volume pot is realy 90% probable cause.

Amps commonly have channel imbalance at low levels. Check out this product which apparently is the solution for this problem.

http://www.rothwellelectronics.co.uk/Rothwell/rothwell_attenuator.asp

Marakanetz: Bias helps to minimize the "knee" in the gain curve of a device. The less bias that you have, the more apparent the differences in gain are between devices. I work with Class C amps all the time and the gain curve on these is more like a "gain slope" than a curve due to the lack of bias. They are either off or on and the transition period is NOT pretty.

As such, gain matching would be more appropriate for a low bias circuit than it would be for a high bias circuit. Otherwise, one device may be at full output whereas another device in the same circuit with the same amount of drive applied might just be going into the point of conduction. Obviously, the levels of output / gain / linearity / distortion at that point would be very different things between conductors with different levels of matching. Bias keeps them turned on so the threshold of conduction has already been passed and the "knee" in the gain curve is reduced or minimized. This improves linearity, reduces distortion, etc... The higher the bias, the deeper we are into the gain curve of the device and the less differences one would have in terms of linearity at various drive levels.

The reason that line level tubes last forever is that they aren't called upon to pass any current. The more current that a tube has to pass, the faster it will deteriorate or flatten out. This becomes even more true as frequency is raised as the bandwidth of the tube falls quickly as the tube deteriorates. This is why tubes that are "flat" sound dull, lack air and top end sparkle, etc... The bandwidth / high frequency linearity of the tube has been reduced. Sean
>