GEC KT66, what one to look for?

[StingRay85]

What kind of data you want to see? My Maxi Matcher measures idle current at 400V -36V bias, and also transconductance. I have a data set of more than 250 EL34, mostly NOS.

 

[StingRay85]

For example these are my new in box Blackburn Mullard tubes. You can clearly see a large variation in idle current at realistic plate voltages.

 

[StingRay85]

My new in box RFT. You can already see how different these tubes behave, and also that transconductance of these tubes are considerable lower compared to a Mullard.

 

[StingRay85]

Also, maybe a bit controversial statement, but the matching that is performed at 250V -13.5V doesn't always hold up at 400V. Tubes that are very close together at 250V can drift apart easily 5 mA on 400V. Does it matter much? Perhaps not, but I prefer my method of matching at 400V, and pick a tube which is also very close in transcondunctance value. Basically connect the dots with the smallest distance between them to make pairs and quads. But you need a lot tubes to do this, and buying a few random tubes on ebay or reverb will have a high risk of not being suited. And for some tubes you just have to be patient. For example the 60+ mA RFT, could take a long while before I find a good match for that one.

 

[2L man]

To me it make more sense to balance tubes at lower voltage and higher current and build a balance adjust to bias circuit to set high voltage low current point to match tube pair.

However most users just complain bias difference.

 

[Dblgun]

I prefer vintage tubes over newer production, particularly in pre-amp tubes. I find the consistency of quality in new tubes suspect. I certainly understand it is not always possible/practical to use vintage tubes in every amp. I recently saw a quad of new Gold Lions where 2 had open filament circuits out of the box. The supplier had no problem replacing them no questions asked. To me this simply means more effort on my part to get something built or re-tubed. It's the world we live in I'm afraid.

If you have the time, knowledge, wherewithal and funds to get a quality set of vintage KT66's I think you should. If for no reason other than they will provide good service and you will feel good about having them in your amp. If not, do your homework about current offerings and expect to inspect what you are sent for proper performance in your amplifier.

 

[Matthews Guitars]

Thanks for the correction on my transconductance/gain error, although I note that consistently as I read tubes with higher transconductance, they correlate to higher gain in circuit based on onset of clipping.

I have no specific information on why some ultralinear amps are very hum balance sensitive, and others are not, but Fender amps certainly are. At least the 70 and 135 watt versions are. They are sensitive to tube matching for sure.

The next time I have a reason to open up my Pro Reverb I'm going to measure the balance pot, set it to perfect center balance, and start checking some tubes that are allegedly tightly matched against both the amp and the transconductance reading on my B&K 707.

I would personally prefer that tubes be matched at actual operating voltage. But it's harder to build an economical matcher that works at 470 volts than 250.

 

[Pete Farrington]

back in the "golden days" of tube manufacturer, when Genelec, Mullard, Telefunken, Siemens, GE, RCA, Philips, Sylvania, Brimar, MO Valve, RFT, etc. were all active and making tubes by the truckload, tube quality was very high, with rigid quality and performance standards.

Agreed, but notice how none of their valve type info ever states a tolerance on their electrical characteristics?

The tubes that came out of the shipping dock of those factories were of a uniformly high specification and you could just pick four new KT66s or EL34s out of the bin, put them in your Marshall, set the bias, and go. They'd work.

Maybe, maybe not. My experience of buying such valves from bulk suppliers in the early - mid 80s was that it was pot luck, idle currents could easily have a 50% disparity, such that there’d be idle hum and the upper and lower parts of the waveform would clip at different signal levels to each other. When revalving an output stage, I had to buy 50% more than the amp required, in order to avoid using outliers.

They were well matched no matter which ones you pulled off the shelf

Really, no, as per above, I think that’s a myth.
As evidence, see notes about independently adjustable bias with a range of +/-25% for fixed bias, or independent cathode resistors (rather than shared) for cathode biased push pull arrangements, p2&3 of https://frank.pocnet.net/sheets/084/k/KT66_GEC.pdf

I have no specific information on why some ultralinear amps are very hum balance sensitive, and others are not, but Fender amps certainly are. At least the 70 and 135 watt versions are. They are sensitive to tube matching for sure.

We all know what’s meant, but being my usual nit picky, I don’t think those Fender amp with primary taps for their screen grids should be referred to as UL.
The goal with that range was probably to make them more powerful, rather than more linear. Hence the purpose of the primary taps would be to reduce screen grid dissipation (by reducing the voltage differential between anode and screen grid at high power output). The taps are at about 12.5%, rather than the usual, higher UL tapping points.

Anyway, those Fenders need bias balance to get a reasonable idle hum because the HT supply was not very well smoothed, and UL requires a well smoothed HT because by its nature, the screen grids get fed from the same HT node as the anodes, ie the OT CT.
Screen grids are, er, grids, they have a gm, a lot less than the control grids, but even so, feed VAC to them and it will modulate anode current. The VAC in this case being HT ripple.
That’s why all regular pentode amps have an extra stage or 2 of HT smoothing for the screen grids. Unless the anodes are already fed from a very smooth HT, eg pi filter (CLC).
Fender swapped an expensive pi filter for a bias balance pot (and highish value reservoir caps).

 

[playloud]

What kind of data you want to see? My Maxi Matcher measures idle current at 400V -36V bias, and also transconductance. I have a data set of more than 250 EL34, mostly NOS.

These are fantastic! Exactly what I had in mind.

One interesting metric to consider might be the average ratio between any two tubes from the sample, for each parameter (i.e. min(x1,x2)/max(x1,x2)). You could take this further and do it for four tubes (min(x1,x2,x3,x4)/max(x1,x2,x3,x4)).

This would estimate how closely-matched a random pair/quad of new tubes from a given brand were. Would then be interesting to compare the numbers for these Mullard and RFT tubes with a new-production brand (although I don't blame you if you don't have many/any of these, given you have so many NOS examples!)

In particular, we could test this hypothesis directly:

The tubes that came out of the shipping dock of those factories were of a uniformly high specification and you could just pick four new KT66s or EL34s out of the bin, put them in your Marshall, set the bias, and go. They'd work. They were well matched no matter which ones you pulled off the shelf due to the high quality standards they had to meet to be accepted for sale.

Also @Ivan H, just wanted to say I really appreciated this:

Ok, let's just get a few things straight. Transconductance (aka mutual conductance) is NOT gain. It is actually expressed as the opposite of resistance, hence the unit of transconductance being expressed in mhos, the opposite spelling of ohms.
Transconductance is a ratio of the incremental change in plate current divided by the incremental change in the control grid voltage producing it (taken directly from my copy of The Radiotron Designers Handbook, 4th edition).
This brings us to "gain", usually associated with pre-amp "voltage amplifying" tubes, aka the "common cathode gain stage". The "amplification factor" of these type tubes is often INCORRECTLY referred to a the tube's "gain", but actual "gain" is calculated. Gain = (mu * RL)/(rp + RL + Rk (mu + 1))
Where;
mu = amplification factor of tube.
RL = Plate load resistance.
rp = internal plate resistance.
RK = Cathode resistance
(RK = 0 if cathode is fully bypassed).

 

[StingRay85]

That's how my Maxi Preamp 2 measures it. I have interesting data sets about 200+ NOS ECC83 gain values too

I judge tube life based on transconductance

 

[Sigs]

So here is a pair of matched tubes-

KT 66 (90810) : emission: 93,6 mA ; mutual conductance: 7,0 mA/V (= 7000 µmhos)

KT 66 (7830 Z) : emission: 92,3 mA ; mutual conductance: 6,9 mA/V (= 6900 µmhos)


Another set of matched tubes-


KT 66 : emission: 81 mA ; mutual conductane: 7,0 mA/V (= 7000 µmhos)

CV 1075 : emission 87 mA ; mutual conductance 7,1 mA/V (= 7100 µmhos)

(GEC datasheet values for new tube : emission 88 mA mutual conductance = 7000 µmhos)

What does this tell me about the two sets ? the bottom one shows a KT66 and a CV1075, is there any difference in a KT66 and a CV1075 ?

Is one set better than the other ?

 

[mickeydg5]

Seems to be various types of GEC KT66's, double halo getter, cv1075 and others, what's the best ?
I'm look for a quad set but difficult to get at a reasonable price, so what if I purchased a pair of matched tubes and later purchase another pair with the results of the second pair were almost the same as the first pair, would this be considered as matched set of quads ?
Would these run in an amp without any issues ? if biased correctly.

Nope, there is only one GEC KT66.

Try some and then compare to other equivalents and copies.

 

[neikeel]

So here is a pair of matched tubes-

KT 66 (90810) : emission: 93,6 mA ; mutual conductance: 7,0 mA/V (= 7000 µmhos)

KT 66 (7830 Z) : emission: 92,3 mA ; mutual conductance: 6,9 mA/V (= 6900 µmhos)

Another set of matched tubes-

KT 66 : emission: 81 mA ; mutual conductane: 7,0 mA/V (= 7000 µmhos)

CV 1075 : emission 87 mA ; mutual conductance 7,1 mA/V (= 7100 µmhos)

(GEC datasheet values for new tube : emission 88 mA mutual conductance = 7000 µmhos)

What does this tell me about the two sets ? the bottom one shows a KT66 and a CV1075, is there any difference in a KT66 and a CV1075 ?

Is one set better than the other ?

As I am sure someone will point out the mA (Ia) ia the quiescent current draw at idle. Older tubes may have higher current draw to give the required output and if you are looking to match tubes ideally this number should be very close as a starting point. My tester is an AVO and gives quiescent current and Gm readings (ie the current being required to provide 1mA/V and there are normal and minimum acceptable readings)

So it will have a Gm figure (normally a new KT66 should be around 7 mA/V) and a current figure (Ia, should be < 87mA) . The data sheets suggest discarding a KT66 if its Gm is < 5.5mA/V.

So Ia is an idle match and Gm is whether the valve is capable of doing its amplifying bit.

Looking at your pairs above, assuming both tested on same machine (all vary a little) they are unlikely to match up as a quad, you could of course go dual bias circuits to use not quite matched pairs.

Not trying to confuse you but near in mind those figures are what most people in the UK with AVO valve testers will use and will have used 'back in the day'.

 

[playloud]

Nope, there is only one GEC KT66.

Yep. From what I can tell, CV1075 is just the UK military designation for "KT66". It isn't some special variant like the CV4004 is to ECC83 (although this was also available commercially as "M8137").

 

[mickeydg5]

Yep. From what I can tell, CV1075 is just the UK military designation for "KT66". It isn't some special variant like the CV4004 is to ECC83 (although this was also available commercially as "M8137").

Yes because from what I remember noticing they all looked to be constructed the same and same data characteristics. The labeling was different.
However, maybe I am wrong.

 

[neikeel]

There are variations (types of getters etc) but I think performance is the same.

 

[mickeydg5]

There are variations (types of getters etc) but I think performance is the same.

Is that variations between KT66 and CV1075 manufactured in the same time frame?
I would suppose that available materials, machinery and construction may differ over time for all.

 

[Amadeus91]

Is that variations between KT66 and CV1075 manufactured in the same time frame?
I would suppose that available materials, machinery and construction may differ over time for all.

Here are a pair of dark glass CV1075 from 1957 and a pair of GEC clear glass 1963 and early 1964 both with the same internal structure. Single cup getter.
I believe they went to clear glass with x2 halo getters late in 1964.
If, I remember correctly my CV1075 came from NATO surplus.

 

[Lance Chambers]

I recently pulled out the oem KT tubes and installed a matched quad of Premium GT-6L6-R tubes in my VM2466. It is whisper quiet now (no hum) and brings the noise when required.

 

[playloud]

What do people make of this quad? https://reverb.com/item/40938855-no...-kt66-cv1075-matched-quad-from-1970-s-uk-made

Obviously priced optimistically as they've been sitting for a year (on Ebay too). Test values look reasonably well-matched though and they do look "new in box" to my untrained eye. Would still be concerned that they haven't been tested at high-voltage.

I've seen matched quads of used but decent-testing KT66s go for £4-600 on Ebay in the last few months.