Preamp tube tolerance

[ckpop]

Does anyone know what the standards of 12AX7 tolerance use to be when manufactured in the early days ? Example RCA, GE, Sylvania, Ect. Was it 5%, 10% ? I'm sure all did not measure out at to be 100% gain factor The same goes for the balance of each side of the triode.

Need some tube experts to chime in on this

[Andy Le Blanc]

tubes are a 20% device, thats why you make them in large batches and then
put the tubes through a grading process to insure matching and consistency.

That's how you end up with, for example, a 12AX7 WA, WB, and WXT.
And how you end up with a color system to choose the response of 6L6's.

[FYL]

tubes are a 20% device

That would be wonderful. Boxed small signal tubes can be off by ±40% when it comes to transconductance (current isn't really important), large bottles are nearly as messy - here we have to deal with current and transconductance. Bulk tubes are even worse.

In the good old days, tubes were qualified using a very rigorous system. For instance European standards for safety tubes were ±12.5% for Ip and ±18.5% for slope (transconductance)

Notation is different depending on which side of the pond you're sitting, but the relation is straightforward: 1 mho = 1 siemens = 1 A/V (ampere per volt), for tubes we use micromhos and mA/V, with 1 000 micromhos = 1 mA/V - nobody really uses the siemens, even if it's the standard SI unit.

Let's have a look at a 6V6GT, aka VT107-A. The specs are: Ip = 49.5 mA, s = 4.1 mA/V, end of life at 33 mA and 2.7 mA/V.

Picking randomly a test report from my archives, a batch of 48 late '60s RCA's gives the following spread (min - max):
Ip 43.8 - 55.7 mA
s 3.6 - 4.8 mA/V

This batch resulted in 6 matched quads (±5%), 11 matched pairs (±5%) and 2 rejected tubes. Excellent results, RCA knew their turf...

If you're interested in US MIL specs and testing, check http://www.dscc.dla.mil/Downloads/MilSp ... td1311.pdf

And here's a link to New Sensor's tubes specifications: http://www.newsensor.com/TubeSpecification.aspx

Wot? An empty page?

That's how you end up with, for example, a 12AX7 WA, WB, and WXT.

New Sensor bull^h^h^h communications. W was a JAN suffix for US MIL tubes, A, B and C designated revisions, not gain levels.

And how you end up with a color system to choose the response of 6L6's.

Mesa, Groove Tubes et al. use the color code as a rough current indicator. I've seen differences of more than ±20% between supposedly identical tubes...

[iknowjohnny]

being electronically retarded, can anyone explain what the difference i would hear from a 12ax7 thats low and one thats high in transconductance?

[FYL]

The higher the transconductance, the higher the gain.

We have E = I * R per Ohm's law. Keeping a constant plate load R, more current across it means more voltage is developped across it, ie. more signal.

A triode with s = 1 mA/V will draw 1 mA more current for a 1 V grid change; 1 mA across a typical 100K plate R gives Eout = 100 V; with s = 1.5 mA/V we have Eout = 150 V; etc.

Higher s tubes will sound louder and show more touch sensitivity - a very simplified view, but OK for here.

[iknowjohnny]

So are you saying that theres no other sonic difference besides the higher gain? If so i would think the word wouldn't even come up because if higher gain was the only difference i would think it would only be mentioned as "gain".

[FYL]

So are you saying that theres no other sonic difference besides the higher gain?

I'm not, far from it. Transconductance is *one* factor conditioning the transfer function of a tube, there are many others.

[ckpop]

FYL What I am trying to make sense of is If I am checking a pile of NOS 12AX7's on lets say B&K 747 tube tester for example and I get a reading of 85......... Does that mean the tube has been worn down or maybe the because of tube tolerance it never met the 100% spec ? All my sylvania's just slammed 100 or way above, Most GE's around 85 to 90 %. The variations of the balance of the dual triode and gain and you wonder how much tuning you can do with preamp tube alone.

[FYL]

The 747 is quite peculiar: when well maintained and properly calibrated - quite easy BTW -it does an excellent job at measuring various params, including transconductance but displays the results on a silly English dial (Good/Bad) with a 0 - 120 scale.

Min for a usable tube is 65, max is 120, but does a bogey 12AX7 display 100? What is the basis for 100%? And what are the actual tolerances, ±10%, ±15%?

I've measured NOS and new production 12AX7 ranging from 1.1 to 1.3 mA and 1.2 to 2.3 (!) mA/V using a reference-grade tester and standard settings (Ep = 250 V, Eg = - 2.0 V), while specs are Ip = 1.2 mA and s = 1.6 mA/V for a bogey.

So I wouldn't worry too much...

[ckpop]

FYL what tube tester model do you perfer ? I borrowed the 747 and am in the market to make an investment. I opened a can or worms after measuring preamp tubes and tuning several Wrecks. The tubes that were 100% didn't sound as good and Bloom the same as some lower reading tubes.

I guess the first rule is........all other rules don't always apply.

[FYL]

what tube tester model do you perfer ?

It all depends on what kind of tubes you're testing and how you intend to test them.

I use a B&K 650 and an Eico 667 as qualifying testers. They can be configured really fast and allow for fine measurements of grid emission, gas and leakage - very important factors.

Real-world tests are done with custom built gear: a digital analyzer/tracer (Amplitrex-style) and a high voltage current/Gm tester - most commercial models stop at 250V or so, some go up to 500V, barely enough for KT88's and other big bottles.

The tubes that were 100% didn't sound as good and Bloom the same as some lower reading tubes.

The proof of the pudding is in the eating.