You should assume that the variability is just as big between sections of the same tube, as between tubes. Also some types or even manufacturers will be better than others. Best is to try the tubes you have and see which shares the current best.
Conceivably, there could even be a systematic error between the two sections, for example if an assembly machine was misaligned.
/Kenneth
Conceivably, there could even be a systematic error between the two sections, for example if an assembly machine was misaligned.
/Kenneth
Agreed, matching is a bit hit-and-miss with double tubes, except for specially selected ones (e.g. 6SU7 is a well-matched 6SL7). In Morgan Jones's Valve Amplifiers, he built his "Crystal Palace" amp using 6J5s in carefully selected pairs, instead of 6SN7s, so as to facilitate matching.
Expect up to 20% difference, with some worse than this. My findings, on a random set of ECC83 (mainly NOS but some modern), are that there is some correlation between sides. So if one side is low, then the other side is more likely to be low than a random half picked from any other valve. This is what you would expect, if some of the variation is due to differences in manufacturing processes rather than just random error.
One of the worst I have seen was a new Tesla E83CC (I think it was genuine) - emission 43%, difference between halves 28.9%. Unfortunately I didn't discover this until some time after I bought it so too late to ask for a refund from a well-known UK dealer. Some I bought on eBay for much less money were much better. Also bad was a Pinnacle ECC83 (not sure of the make) - emission 107%, difference 42.5%.
One of the worst I have seen was a new Tesla E83CC (I think it was genuine) - emission 43%, difference between halves 28.9%. Unfortunately I didn't discover this until some time after I bought it so too late to ask for a refund from a well-known UK dealer. Some I bought on eBay for much less money were much better. Also bad was a Pinnacle ECC83 (not sure of the make) - emission 107%, difference 42.5%.
Matching, schmatching... good circuits will accommodate a ******* transistor and still work as well. If you need matching, you either need particularly high accuracy (or low distortion, much less than audio needs), or you need a circuit that actually works.
As I recall, tubes in spec are around 10% tolerance (Gm / mu / Rp, Ip, Ig2, Vg1, etc.), and the two halves of a dual are better than that (I guess 5%?). Tubes are remarkably accurate; most transistors have hFE's varying by +/-50%, or Vgs(th) by +/-1V.
Tim
As I recall, tubes in spec are around 10% tolerance (Gm / mu / Rp, Ip, Ig2, Vg1, etc.), and the two halves of a dual are better than that (I guess 5%?). Tubes are remarkably accurate; most transistors have hFE's varying by +/-50%, or Vgs(th) by +/-1V.
Tim
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.