How do "You" Match Tubes & Tolerance?

Okay, without getting too much into the amps themselves which can tolerate different ranges for "matched" sets, or seperate bias, etc., when you are matching tubes with your Hickok or other Mutual Conductance tester, what is an acceptable perchantage/tolerance for you to consider being "matched"?

Vintage amps were generally designed with a +/-20% factor but I believe that may be a little wide for tube matching. There's the drift factor to figure in too along with a good, useful life.

Not all my tubes in my possession are mis-matched, but it's easy to end up with singles, odd numbers, etc. and sure, the closer the match the better but what is acceptable?

And lastly, I believe current draw would be more critical than mu but at what range?

Matching isn't as important as many believe. However, if you are determined, just test all the tubes you have and put them in pairs as closely as possible.

Most tubes are expected to be within around 10% of nominal, so if you want them closer than that you can shoot for 5%. I find it hard to imagine that you will be able to hear much difference.

marty59 said:

Okay, without getting too much into the amps themselves which can tolerate different ranges for "matched" sets, or seperate bias, etc., when you are matching tubes with your Hickok or other Mutual Conductance tester, what is an acceptable perchantage/tolerance for you to consider being "matched"?

Click to expand...

I do not match output tubes on a Mutual Conductance tester. I recommend tube bias matching in circuit, where it counts. Placing small value resistors (one to ten ohms) in series with each cathode will allow one to measure the cathode current. One can then try different pair combinations to find the best match with the tubes on hand.

marty59 said:

Okay, without getting too much into the amps themselves which can tolerate different ranges for "matched" sets, or seperate bias, etc., when you are matching tubes with your Hickok or other Mutual Conductance tester, what is an acceptable perchantage/tolerance for you to consider being "matched"?

Vintage amps were generally designed with a +/-20% factor but I believe that may be a little wide for tube matching. There's the drift factor to figure in too along with a good, useful life.

Click to expand...

First off, it is not possible to accurately match power tubes for plate or cathode current with any Hickok (or all but a small handful of other specialized units) tube tester. The voltage applied during testing is simply too low for it to be a meaningful exercise. It also lacks the current capability needed for large power tube testing.

Second, there may have been components with 20% tolerances, but that doesn't necessarily apply to the tubes. It was quite common to run tubes at or near their dissipation limits - and a 20% error would have caused a lot of tube fatalities.

And lastly, I believe current draw would be more critical than mu but at what range?

Click to expand...

Your Hickok does not measure mu! It measures transconductance (Gm) - and that's not the same as mu because there is a 3rd number involved, Rp (plate resistance).

Here's the formula for mu:

Mu= Gm x Rp

Tubes not well matched for mu may (or may not, depending on a lot of things) affect sound quality/tone - but mismatched Gm is not a "mission critical" issue! Mismatched Gm will not cause amp damage.

On the other hand tubes not well matched for cathode current can cause a number of problems, including premature wear and/or damage to a tube; if circumstances are right it can also damage the amp itself.

Take the scenario of a quad of 25 watt dissipation rated tubes - 3 are relatively well matched for cathode current, but the 4th is 25% higher than the others (this is not at all uncommon). Assume the plate and screen voltages are 450 volts.

Under the conditions in the amp the three "matched" tubes all pass 50 ma. So these tubes must dissipate 450 volts x .05 amps (50 ma), or 22.5 watts of heat. For a 25 watt rated tube that'll generally be okay. But the 4th tube of the set passes 25% more, 62.5 ma. 62.5 ma x 450 volts is 28.1 watts, and this tube is in trouble.

As that tube gets hotter it liberates more "gas" molecules from the tube internals. As more gas molecules are liberated the tube current will tend to increase. As the tube current increases the tube temperature increases - which increases the amount of outgassing. The increase in outgassing causes an increase in current which causes an increase in heat which causes more outgassing... well, you can see where this is headed. If and when the tube has its meltdown it may well take out other parts in the circuit.

While this is off topic a bit I hope it'll be helpful. There are a zillion posts here about red plating tubes, and poor tube current matching is often the cause - just the scenario I described above is what happens.

If you have a mechanism to apply realistic voltages to the tubes then you can current match - and I use a standard not in %, but in pure numbers of + or - 1.5 ma. That's tight enough that even if they drift some (despite burn-in and all the other things done to minimize it) the tubes should still work well together. I will occasionally make an exception if circumstances allow, but in general the standard I use has worked VERY well.

My standard for Gm is 10% - although I strive for 5% whenever possible. I also test for Gm when the tube is blazing hot off the match rig. There are some tubes that have a reduced Gm as the tube gets hotter - for one, the SED 6550C tubes tend to have significantly lower Gm (10% - 15%) when hot than when when operating just in the tube tester.

Yes, you can use your amp to match tubes as tcdriver described, but be careful! In addition to the potential for damage from a runaway tube, it's hard on the tube sockets - and since there is some variation in the circuit the results aren't generally as accurate as putting the tubes in carefully controlled environment.

Jim, You've nailed the issue! Your knowledge is respected and appreciated!

We always seem to hear about in the day that NOS tubes were not having to be matched as of now. My understanding is that tubes were batched produced and tested, in racks and under load and their tolerances were tighter. Quality being what it was, when you bought a set of tubes they were fairly matched. But the myths exist too.

Nowadays, not that all new production is bad, but we sure have seen quality vary from brand to brand and at various times.

I still have an NOS pair of Sylvania 6L6GC's that I bought from the local wholesaler back in the day that have been used very little (maybe a few hours) that I keep for reference.

Heck, all these years later I'm still learning/perfecting and always encouraging others when I can. Any differences in my opinion are welcomed, I appreciate the replys!

To Jim's excellent information, I would also add the following:

For power output tubes -- and I stress power output tubes only -- the Gm of the tube is more a factor of it's internal design, rather than it's condition. It will correlate somewhat with condition, but in general, testing, matching, and judging the condition of power tubes by a measurement of it's Gm does not provide very accurate information as to the tube's real condition.

For the best matching, tubes need to be matched fairly closely under two conditions: under static conditions of no signal, and near full power output under high level signal conditions. The two conditions are hardly the same. The cathode's condition determines the ultimate condition of the tube, barring some other mechanical issue, and while it may allow for equal current flow under static conditions, it may hardly allow for full current flow under maximum signal conditions.

Using an amplifier like an original Dynaco MK II, III, or Stereo 70 is excellent for matching tubes, with a slight alteration -- the common cathode resistor would need to be removed and replaced to two separate (say) 10 ohm units. The reason that these are excellent for this application is because they don't provide for individual bias adjustment, and, as long as the phase inverter's resistors are closely matched, they also basically guarantee an equal AC drive to both output tubes.

All that is needed besides one of these amplifiers then is a signal generator (1 kHz is fine), dummy load, and DVM. The matching check is simple enough:

1. The match under static conditions can be easily checked by measuring current draw at the two cathode resistors with no signal. Since both tubes are being supplied with the same bias voltage, if they pull similar current under this condition, then they are matched statically. Use Jim's standard for tolerance.

2. If you are using one channel of an ST-70 with an 8 ohm dummy load, drive the amplifier from the signal generator to 15 vac output across the load, and then measure the cathode current for both tubes again under this condition. If the currents are again nearly matched (within say 5%), then there is a good match under dynamic conditions as well.

3. Once matched, if you want to judge a pair of power tubes as to their actual condition, add a scope to the second test and see if the amplifier will produce full power output before clipping. A power output test is always the best test to judge the condition of any power tube.

Tubes matched under both of these conditions will produce the best results in a P-P amplifier. When a mismatch exists under either or both conditions, distortion of various types all increase significantly.

Dave

To Jim's excellent information, I would also add the following:

For power output tubes -- and I stress power output tubes only -- the Gm of the tube is more a factor of it's internal design, rather than it's condition. It will correlate somewhat with condition, but in general, testing, matching, and judging the condition of power tubes by a measurement of it's Gm does not provide very accurate information as to the tube's real condition.

For the best matching, tubes need to be matched fairly closely under two conditions: under static conditions of no signal, and near full power output under high level signal conditions. The two conditions are hardly the same. The cathode's condition determines the ultimate condition of the tube, barring some other mechanical issue, and while it may allow for equal current flow under static conditions, it may hardly allow for full current flow under maximum signal conditions.

Using an amplifier like an original Dynaco MK II, III, or Stereo 70 is excellent for matching tubes, with a slight alteration -- the common cathode resistor would need to be removed and replaced to two separate (say) 10 ohm units. The reason that these are excellent for this application is because they don't provide for individual bias adjustment, and, as long as the phase inverter's resistors are closely matched, they also basically guarantee an equal AC drive to both output tubes.

All that is needed besides one of these amplifiers then is a signal generator (1 kHz is fine), dummy load, and DVM. The matching check is simple enough:

1. The match under static conditions can be easily checked by measuring current draw at the two cathode resistors with no signal. Since both tubes are being supplied with the same bias voltage, if they pull similar current under this condition, then they are matched statically. Use Jim's standard for tolerance.

2. If you are using one channel of an ST-70 with an 8 ohm dummy load, drive the amplifier from the signal generator to 15 vac output across the load, and then measure the cathode current for both tubes again under this condition. If the currents are again nearly matched (within say 5%), then there is a good match under dynamic conditions as well.

3. Once matched, if you want to judge a pair of power tubes as to their actual condition, add a scope to the second test and see if the amplifier will produce full power output before clipping. A power output test is always the best test to judge the condition of any power tube.

Tubes matched under both of these conditions will produce the best results in a P-P amplifier. When a mismatch exists under either or both conditions, distortion of various types all increase significantly.

Dave

this should be a must read for all tube biasing and other like questions. can someone do that?

marty59 said:

Jim, You've nailed the issue! Your knowledge is respected and appreciated!

Click to expand...

Thank you so much for your kind words. :yes:

We always seem to hear about in the day that NOS tubes were not having to be matched as of now. My understanding is that tubes were batched produced and tested, in racks and under load and their tolerances were tighter. Quality being what it was, when you bought a set of tubes they were fairly matched. But the myths exist too.

Click to expand...

And what you posted is one of them I'm afraid. Old stock tubes were no better matched or tighter tolerance than current production. As an example, I matched over 500 GE 6550As, a tube most all agree was a truly FINE quality, well made tube. Cathode currents varied from as low as 50-ish ma. to as high as about 120 ma.

They followed the same "bell-curve" of tolerance that modern tubes do - the largest proportion of the tubes were centered around a mid-point, and as you got further from the mid-point the number of tubes dropped off - but they were still present in significant numbers. In the case of the GEs, while there weren't many 50 ma or 120 ma tubes there were a LOT of 65 ma and 105 ma - clearly tubes that would not play well together in most amps!

Do NOT count on tubes being well matched just because they are old stock - it simply isn't the case!

I actually built a stereo single ended amp to "test" tubes.

Single dual triode driver cathode biased (non bypassed) cap coupled to triode strapped
7AC pinout octals cathode biased (non bypassed) with sockets that allow for measuring
voltages on every pin. (I carfeully matched every resistor as close as I could get them)

I use the tube rectifier's to run from 5y3 to 5ar4 to vary voltage a decent amount an "test"
from say a 6V6 to KT88, I can get from roughly 300 - 365v, measure an listen to the tubes.

The driver is an octal socket wired for 6SN7 but I have adapters for 9A an 9AJ pin-out so I
can "match" triode sections of many small signal tubes as well, I also have adapters for 807
an 6BG6, plus a few different rectifiers with adapters, it's quite a handy tool, I use a regular
tube tester first to hopefully weed out any truly bad tubes then run through the tester/amp
an keep track of them with an excell spreadsheet so I get an idea of what a certain type of
tube "should" be doing in this particular circuit, it has served me pretty well running them in
there then placing what I've matched up into other amps for pretty good matching.

It's not a perfect tube matching setup but it was built with spare parts an works pretty well.
Anyone who does lot of tube work an has the spare parts laying around could do worse here.

Nice to see a thread on this where people actually know what they are talking about ! a lot of misinformation on it. I hope to not degrade the thread

Rachel has written an interesting article on tube production and matching for modern production Chinese tubes that goes into testing and matching. It's a bit sales oriented to GF, but still worth the read.
http://grantfidelity.com/site/vacuum_tubes_quality_control

We have the Amplitrex AT-1000 tube tester that operates at spec for the tubes under test. It makes short work of matching and testing, after a few months of watching your hair go grey due to it's accuracy and consistency (it recalibrates every time you turn it on).

At some point I will write up my thoughts on testing tubes, maybe in another year after more experience, but for now I suggest 10% plate current, 10% GM for matching is about as good as one can expect from a reseller for a pair or a quad. Test results should be no more than 20% off spec.

A few tips I've learnt the hard way is when installing new or different power tubes, lower the bias 15% from amp spec for the first hour and keep the volume down and watch them for any signs of runaway, a tube can change drastically in the first hour of use and if you don't have any high voltage protection you will lose some internal parts, if you aren't able to shut down in time. After an hour and all is fine, then bring them up to amp spec, again watch them for an hour. if they are biased too high at idle and you crank the volume you can easily watch them make their sacrifice to the tube gods.

When removing the old tubes for a roll, while they are still installed back off the bias a bit, I suggest 20% lower, then shut down and roll. Your old tubes might have the bias cranked up to compensate for tube aging.

As Jim pointed out excess gas is another tube killer (The AT-1000 will test for it) besides being a culprit for runaway tubes it also creates a false current reading, we test for no more that 5% on power tubes and no more than 0.1 ma on other tubes. GM out of match can effect imaging as GM represents the output power in relation to the input current, biasing only controls input.

Hopefully by now most understand that many driver and signal tubes are dual section and both sections need to be matched, so pair of say 6SN7s is really a quad for matching.

The AT-1000 can be programmed to match your amp's circuit design if you have the schematics, this to me is beyond cool and maybe the ultimate tweak for the tube guy that has everything

Rachel won't let me bring it shows anymore since all I do is chat with everyone about it, instead of demonstrating our gear. I don't have any association with Amplitrex, but don't know how I could sell and grade tubes without it.

Cheers,
Ian

"Nice to see a thread on this where people actually know what they are talking about ! a lot of misinformation on it. I hope to not degrade the thread"

Oh, believe me you're not degrading this at all. What a powerhouse of knowledge we have here! You just can't go to school and learn practical tube theory and apply it anymore! Real world applications? This is it!!

iGrant said:

A few tips I've learnt the hard way is when installing new or different power tubes, lower the bias 15% from amp spec for the first hour and keep the volume down and watch them for any signs of runaway, a tube can change drastically in the first hour of use and if you don't have any high voltage protection you will lose some internal parts, if you aren't able to shut down in time. After an hour and all is fine, then bring them up to amp spec, again watch them for an hour. if they are biased too high at idle and you crank the volume you can easily watch them make their sacrifice to the tube gods.

Click to expand...

While some will disagree with the amount of time and the % involved, bringing the current UP to bias spec is always a good idea.

When removing the old tubes for a roll, while they are still installed back off the bias a bit, I suggest 20% lower, then shut down and roll. Your old tubes might have the bias cranked up to compensate for tube aging.

Click to expand...

Might I suggest it be lowered even more... If the old tubes are tired, then it's likely that the amp's owner has been adjusting the bias to compensate for a while. Plugging in fresh tubes with the bias set for the old tired tubes is a recipe for trouble. Drop the bias down a good long way.

As Jim pointed out excess gas is another tube killer (The AT-1000 will test for it)

Click to expand...

Keep in mind that the "gas" in and of itself is not a problem - it's the resultant increase in grid current that is the issue. Good tube testers can detect gas issues by checking grid current.

Hopefully by now most understand that many driver and signal tubes are dual section and both sections need to be matched, so pair of say 6SN7s is really a quad for matching.

Click to expand...

I need to disagree here - for a number of reasons. A few questions/comments...

1. When you say "matched", what parameters are you referring to?

2. Often the two sections of a dual triode are performing completely different tasks in a circuit and matching is completely superfluous.

3. Differences in other circuit part values usually make precision matching much less "precise".

While some circuits have more stringent requirements the vast majority do not need precision matched tubes. Insisting on "matched" tubes when they aren't needed just drives up the tube costs and maybe more importantly it causes tube buyers to disqualify perfectly good tubes from their consideration simply because they aren't "matched". That's a shame, especially when the tubes are in short supply such as with much old stock.

Wildly mismatched tubes rarely are a good idea - but that doesn't mean precision matching is needed.

Just some things to keep in mind.

Good advise Jim,

I'll still go with matched sections, because I can plate current and GM. Most of my gear requires properly matched sections and even if not needed in your current gear, your next piece of gear may require it.

Cheers,
Ian

Many dual triodes are used in phase-splitter applications such as long-tailed pairs, where operating points should be as close as possible to reduce distortion. This means balanced sections.

"While some circuits have more stringent requirements the vast majority do not need precision matched tubes. Insisting on "matched" tubes when they aren't needed just drives up the tube costs and maybe more importantly it causes tube buyers to disqualify perfectly good tubes from their consideration simply because they aren't "matched". That's a shame, especially when the tubes are in short supply such as with much old stock.

Wildly mismatched tubes rarely are a good idea - but that doesn't mean precision matching is needed."

Shhhhhhhhhhh! You'll drive up the prices of non-matched small signal tubes, the ones I buy and keep.

Next thing you know somebody will question using the specs for cascado operation of 6DJ8 for VHF RF amplification for in audio circuits.

Bob

Would like to learn more… rather than start a new thread I’ll ask here. How close to each other is good for power tubes?

I just received 16 Tungsol EL34b tubes. The boxes are marked with plate current and transconductance.
The lowest in 16 tubes for plate current is 24.2 and the highest is 25.3. For transconductance the lowest is 4800 and the highest is 5020.

is that good?

Funky54 said:

The lowest in 16 tubes for plate current is 24.2 and the highest is 25.3

Click to expand...

If those test results are accurate - they seem to be within Jim's 1.5ma tolerance

I use my Amplitrex AT-1000 tube tester, then the final check for balance with the output tube`s AC/DC operation in it`s PP circuit.

Cool tester, very pricy, though.