Level matching?

[Skorpio]

When I measure level between channels I found generally differences up to and sometimes more than 1dB. I you NFB tube circuits, at they might generally have bigger differences than solid state with NFB...

Question 1: How close in general must levelmatching be until it is good enough? 0.5dB?

Question 2: Where in the chain is it optimal to do this match (as system)? I am thinking power amp input...?

[DF96]

1dB is 12% in voltage terms, so small but higher than resistor tolerance. Valve circuits don't have the huge open-loop gain of SS, so the closed-loop gain is not entirely set by the feedback resistors but also by the characteristics of the valves (which may vary by something approaching 50% between two samples - one high the other low of the nominal gain). Some valve circuits don't use global NFB, so their channel matching can be much worse unless carefully matched valves are used.

Channel matching depends on context. Unless your listening room and speakers are 'perfect' and symmetrical you might require some electrical imbalance to get acoustic balance.

The best place to put a balance control is around the same place as a volume control, and largely for the same reasons. Not too early, as that can be noisy. Not too late, as that can cause overload. However, these effects are much less of an issue for balance. It just happens that the circuit conditionsneeded for a balance pot are about the same as for a volume pot so convenient to put them together. Yo can combine them, in the sense that the balance pot forms the upper end of the volume potential divider.

[Vinylsavor]

Hi Skorpio,

channel imbalances become audible at around 2dB. So I would see 1dB or less as uncritical.

if you want to correct it, I would find the root cause and correct it there. E.g. mismatch of mu betwenn tube sections of double triodes or individual tubes. Some tubes have a quite high variation in mu with op point, so it could be an indication of sub optimal op point of that particular tube. See if you can move it into a region where the variation of mu is smaller. Some datasheet show curves for mu against op point.

Best regards

Thomas

[Miniwatt]

Isn't the -3 dB standard because 3 dB is the smallest difference we can hear?

[DF96]

-3dB is the standard for roll-off for two reasons: it is half power, and it really does estimate the (rounded off) corner for a first order low or high pass filter. It applies, slightly modified, for other filter orders.

-3dB channel imbalance would certainly be audible!

[Vinylsavor]

Hi!

3dB difference is quite audible. 2dB is the threshold of audibility. For golden ears it might be even lower

Thomas