View Full Version : output tube matching - real pita.


gogofast
09-11-2009, 04:22 PM
just got a gadget that matches output tubes in working voltage, and spent last 6 hours straight trying to match my tubes - actually just one type so far, and i'm beginning to feel that this whole tube matching thing is a really pain in the a**.

for a few amps, i've matched the tubes in circuit, but for my other amps, it's just a lot of work, for me at least, and don't like poking through deadly voltages, too.

anyways, i've gone through about 3 dozen EL84 tubes and was barely able to find ONE quad that'll be within 5%, actually more like 3%, both in plate current and transconductance. 3 freaking dozens! :no:

and i was really surprised to see that my small batch of nos tubes have plate current range with difference of 30%!!!

http://img25.imageshack.us/img25/5581/dsc0041on.jpg

sloober
09-11-2009, 04:33 PM
I feel your pain. I don't have a tester like you just got, but have had several tubes all score veryy good on my Hickok 6000 but in circuit were all over the place. If you can get them within 5% I'd say that was pretty good. You can send me all the wonky ones you have. John

gogofast
09-11-2009, 04:46 PM
I feel your pain. I don't have a tester like you just got, but have had several tubes all score veryy good on my Hickok 6000 but in circuit were all over the place. If you can get them within 5% I'd say that was pretty good. You can send me all the wonky ones you have. John

i'll send you all my tubes only if you can guarantee that this one quad will last forever.:D

JonL
09-11-2009, 05:01 PM
Since you have so many tubes at your disposal that have been measured and characterized, it would be a very interesting experiment to see the audible differences between a closely matched set and sets that are mismatched in various ways... matched on current but far apart on transconductance, vice-versa, etc. I know that in the guitar amp world there are some people who feel that the need for closely matched tubes is overstated. Those people are in the minority, and tend to keep quiet for fear of being royally flamed. Certainly guitar and hi-fi amps have different desired sonic qualities but I think it would be a great experiment.

mhardy6647
09-11-2009, 05:06 PM
When matching NOS tubes, statistics is pretty much against you, unless you bought out a distributor (i.e., bought a large quantity of tubes of similar age and/or from the same lot). Much better shot with new production (but then you can just buy matched pairs or quads as needed).

Best bet: for a fixed bias amp, modify it so that each tube can be individually biased. Much easier in the long run; cheaper too. You can also think about bias strategies like the Blumlein Garter autobias circuit (some of the New England tubies have been kicking this around for the "hot" biased Radio Shack SAF-40... AKer JoZmo has one which is, sadly, sitting in my basement in a state of disrepair).

http://www.tubecad.com/2005/May/blog0046.htm

gogofast
09-11-2009, 05:12 PM
Since you have so many tubes at your disposal that have been measured and characterized, it would be a very interesting experiment to see the audible differences between a closely matched set and sets that are mismatched in various ways... matched on current but far apart on transconductance, vice-versa, etc. I know that in the guitar amp world there are some people who feel that the need for closely matched tubes is overstated. Those people are in the minority, and tend to keep quiet for fear of being royally flamed. Certainly guitar and hi-fi amps have different desired sonic qualities but I think it would be a great experiment.

i haven't done the extensive testing yet, but so far, i can definitely tell you this: i had a quad of EL84 tubes in one of my main amp that had exactly identical plate current (maybe .02v different - that's about 0.05%) and heard slight distortions time to time even at low output. when i measured those tubes for transconductance, they were off by 15-20%, and that probably was the cause for distortion.

one of the effects of difference in plate current i guess is obviously red plating - which i had plenty of problems with before.

When matching NOS tubes, statistics is pretty much against you, unless you bought out a distributor (i.e., bought a large quantity of tubes of similar age and/or from the same lot). Much better shot with new production (but then you can just buy matched pairs or quads as needed).

Best bet: for a fixed bias amp, modify it so that each tube can be individually biased. Much easier in the long run; cheaper too. You can also think about bias strategies like the Blumlein Garter autobias circuit (some of the New England tubies have been kicking this around for the "hot" biased Radio Shack SAF-40... AKer JoZmo has one which is, sadly, sitting in my basement in a state of disrepair).

http://www.tubecad.com/2005/May/blog0046.htm

i think you're right. ...and about the individual biasing, don't you still have to have all tubes within the range of each biasing pot? i'm not sure what the range of bias pots can be, but from what i've seen so far, some used, nos, new production tubes have plate current difference of almost 100% sometimes.

Jim McShane
09-11-2009, 05:48 PM
just got a gadget that matches output tubes in working voltage, and spent last 6 hours straight trying to match my tubes - actually just one type so far, and i'm beginning to feel that this whole tube matching thing is a really pain in the a**.

Aren't you glad you decided to do it yourself? :D

I spend about 20 hours/week matching tubes - virtually every week.

anyways, i've gone through about 3 dozen EL84 tubes and was barely able to find ONE quad that'll be within 5%, actually more like 3%, both in plate current and transconductance. 3 freaking dozens! :no:

Yep, that'll happen. More frequently than not.

and i was really surprised to see that my small batch of nos tubes have plate current range with difference of 30%!!!

That's tight! Variance of 100% is not uncommon even on NOS tubes. Now with your post maybe people will believe me when I tell them that. A lot of people think old stock tubes were perfect, right on the money every time! You want to tell them how wrong they are or should I?

The good tube vendors out there do a lot of really difficult and tedious work to get guys a properly matched tube set. Hopefully your post helps people understand how it really goes down.

Thanks a LOT for posting!! :thmbsp:

becketma
09-11-2009, 08:09 PM
My favorite monoblocks, balanced, PP 6KD6s with 6FQ7 cathode follower has adjustable bias. I bet I have 30-40 6KD6s so that I can make matched pairs in the amp for that 6FQ7. Getting a quad requires a LOT of tubes to play with!

The cathode follower tube also effects the balance.

Best
Bob

gogofast
09-11-2009, 11:04 PM
Variance of 100% is not uncommon even on NOS tubes.


yes, i tested some other tube type NOS and some do have close to 100% difference in plate current, and the readings actually continued to change on NOS tubes. i had a pair stayed on the matcher for two hours and their plate current changed almost 15% after that brief burn-in, and i'm sure it will continue to change as i use them in the circuit. on the other hand, almost all used tubes maintain pretty stable plate current just after 5 or 10 min warm up.

i used to just pop in NOS pairs of EL34's in eico amps and get puzzled about red-plating. i wish all tube sellers are as sincere as you and few others out there.

audiodon
09-12-2009, 07:30 AM
One of the wonders of tube designs is their tolerance of both component variations and their forgiveness when a tech messes up.

That's both good and bad. For the good, the designs tolerate considerable variation among tubes. There was no computer matching back in the day.

Jim McShane has stated the tube vendor's dilemma well. I know for myself that I tolerate a 20% variation in a push-pull pair. Going tighter is better, but it's expensive to acquire enough tubes to pull it off. To those of you who do or have, I salute you.
Don

DENNYDOG
09-12-2009, 07:55 AM
You have to kind of wonder then how many sets of so called matched NOS or good testing originals people are selling are really closely matched. Maybe it isn't that big of a deal.

Dave C
09-12-2009, 08:02 AM
S.E.T. amps, fewer tubes, fewer worries.

Fast_Eddie
09-12-2009, 11:13 AM
Best bet: for a fixed bias amp, modify it so that each tube can be individually biased. Much easier in the long run; cheaper too.

Is this done simply by replacing the resistor to ground with a pot? Anyone have a source for an appropriate pot?

Thanks,

Ed

Tom Bavis
09-12-2009, 01:24 PM
Is this done simply by replacing the resistor to ground with a pot? Anyone have a source for an appropriate pot?

Thanks,

Ed

No, that existing grid resistor goes to a pot with variable voltage. One pot per tube.

Ty_Bower
09-12-2009, 02:47 PM
anyways, i've gone through about 3 dozen EL84 tubes and was barely able to find ONE quad that'll be within 5%, actually more like 3%, both in plate current and transconductance. 3 freaking dozens! :no:

I could believe that. What's even more disheartening is that you are attempting to match two electrical characteristics under one specific set of conditions. What happens if you take your "matched" quad and run them at a plate or screen voltage +/- 30% of where you did your testing? Are they still matched? You almost need to trace a full set of tube curves to characterize each one. Good luck getting those to match under every condition.

Fast_Eddie
09-12-2009, 03:25 PM
No, that existing grid resistor goes to a pot with variable voltage. One pot per tube.

Thanks - there I go threadjacking again.

Steve O
09-12-2009, 04:00 PM
just got a gadget that matches output tubes in working voltage, and spent last 6 hours straight trying to match my tubes - actually just one type so far, and i'm beginning to feel that this whole tube matching thing is a really pain in the a**.

for a few amps, i've matched the tubes in circuit, but for my other amps, it's just a lot of work, for me at least, and don't like poking through deadly voltages, too.

anyways, i've gone through about 3 dozen EL84 tubes and was barely able to find ONE quad that'll be within 5%, actually more like 3%, both in plate current and transconductance. 3 freaking dozens! :no:

and i was really surprised to see that my small batch of nos tubes have plate current range with difference of 30%!!!

http://img25.imageshack.us/img25/5581/dsc0041on.jpg

IMHO, 3% matches are breaching the realm of meaningless precision. IME, tubes matched to that precision don't stay that close for long. I generally go for 5% on bias if the amp doesn't have individual bias adj and up to 10% if it does. 5% on gm is about all that's nec if the amp uses feedback in the output stage.

FWIW, I've not found many "factory" matched pairs of NOS output tubes that were matched better than 5%, some are worse. The factory matching provided by a famous mfgr/distributor of new Russian tubes is routinely off by way more than this, again IME.

Bottom line: A robust amp design should be able to operate per spec with commonly avail tubes not matched to an extraordinary degree.