Why do old tubes sound better than current tubes?

A question for the assembled experts here with respect to small signal dual triodes.

I'm wondering how significant a variable in all of this gain or gm mismatches between the two triode sections within a tube might be, and if that might account for some of the divergence of experiences with these kinds of tubes.

I recently had occasion to test about a half-dozen absolutely NOS 1950's RCA 5963's (12AU7 equivalents) on my Hickok 800A tester. All of the tubes measured well and matched each other just about perfectly. However, all of them had close to a 20% difference in gm between the two triode sections.

As someone with minimal tube-rolling experience, I'd be interested in comments on how common that kind of difference (or worse) tends to be on both vintage and current production tubes, and on how significant it might be for each of the kinds of circuit functions dual triodes are used for.

Re Grant's question, fwiw I'll mention that in my Paxthon VTA-160 I've just replaced a pair of 1960's vintage used Telefunken 12AX7's, which had measured well until one of them recently developed a problem, and a pair of the NOS 5963's, with matched and balanced pairs of Genalex Gold Lion reissue ECC82's and ECC83's. My early impressions of the new tubes are very favorable, with improved clarity on high volume peaks, and lower noise. (The power tubes I've been using all along are a matched octet of SED "Winged C" EL34's).

Thanks!

-- Al

tung sol 12ax7 gold pins, very musical, sovtec lps 12x7 , very open great bass , but maybe a little less body than the tung sol.

Almarg, if you will open NOS, on Mapmans post dated 08-23-10, and then read my post above his which was posted before his post; you will be able to reach some very interesting conclusions.

Orpheus10, yes Brent Jessee's article is certainly an interesting read. However, the concerns about genuineness and condition are not applicable to the 5963's I referred to.

The 5963's were part of a large collection of tubes I purchased about 20 years ago, from a very nice elderly gentleman in my local area, whose career had been servicing radios and tv's. All of the tubes had been stored for many years in his nicely appointed house, and he obviously had no interest in misrepresenting anything in order to maximize his profit. He sold me the entire collection of about 1400 tubes for a total of $65, and he obviously just wanted to get them out of his house and to have them find a good home. I am an antique radio collector, which is mainly why I purchased the collection (very few of the tubes he had are applicable to audio).

Brent's article also contained the statement that:

"Five percent is doing really good for a match of vintage NOS tubes. Most amplifiers that are designed correctly should be able to operate just fine with even a 25 percent or greater mismatch, provided the bias is set correctly.

Which provides an interesting quantitative perspective, but the reference to 25% is obviously addressing power tubes.

So my questions about section-matching of small-signal dual triodes remain.

Thanks,
-- Al

Hi Al, it of course depends on how the tube is being used in the circuit. If the two sections are cascaded single-ended transconducance amplifiers, or a triode driving a split-load phase inverter, then I'd say the matching matters not at all, so long as each section of the tube is within specs to work for its own part of the circuit.

But for a differential amplifier or cathode-coupled ("long-tailed pair") phase splitter, then the matching of the two sections affects the production of even-ordered distortion products. Both of these types of circuits are sensitive to both quiescent and dynamic balance between the two active elements. The extent to which the circuit can tolerate an imbalance is generally directly linked to its transconductance -- the higher the transconductance, the more critical the balance. It's also wrapped up a bit in the tail current - a true current-source will make the balance less critical for large-signal distortion. One can also place degeneration resistors in the cathodes, and increase the tail current to keep the same transconductance . . . and make the balance less critical.

Kirk, thanks very much for your informative and characteristically knowledgeable post.

To complete the picture, can anyone cite significant measurement experience that would provide a feel for the degree of section imbalance that can be expected for various makes of vintage and current production dual triodes?

Thanks,
-- Al

Almarg, a 5963 is equivilant to a 12au7. Would a pair of 12au7's Gm 1, 2800 Gm 2, 2850, NOS RCA Cleartop be in the ballpark of what you are referring to.

Orpheus, yes, what I am looking to establish is a feel for how widely Gm1 and Gm2 tend to differ from each other. And if any correlation can be established between the degree of that difference and whether the tube is current production vs. vintage, or among brands.

The 5963's I was referring to are RCA's date-coded 57-31 and 57-35, with long black ribbed plates and square getters. My tester indicates a nominal Gm for 12AU7's of 2200. The tubes in question typically measured about 2700 for one section and 2300 for the other, which means that one section was about 17% higher than the other.

In most cases the section having the grid on pin 2 was higher than the section having the grid on pin 7, but in one or two cases it was the opposite.

Regards,
-- Al

Thanks, Viridian. Anyone else?

Best regards,
-- Al

I have great things about the new gold lion 12ax7 also, I am going to try them soon.