Curious what people's thoughts are on this for a fixed bias PP amp. I can see the argument that a pair of tubes used in one channel of a push pull amp it might be best if they are a matched pair, but is there really a need for all 4 tubes to be matched? Obviously on this type of amp the bias is adjustable on each tube.
My initial thought has always been "Have the pairs matched but left/right don't need to be".
My initial thought has always been "Have the pairs matched but left/right don't need to be".
I have seen more amplifiers with Fixed Adjustable Bias that did not have separate pots for the push and pull output tube pairs.
But . . . Those amplifiers came with very well matched output tubes, that were matched at the same quiescent operating conditions as the amplifiers that they went into (I hope most manufacturers actually did that).
And, if replacing is required, the manufacturer suggests using very well matched tubes.
Dyna Stereo 70 is one example.
I always used to design with individual self bias resistors and bypass cap for each tube.
Now, with my balanced amplifier, I use a single self bias resistor, and Extremely well matched JJ output tubes (From Eurotubes.com), I drive right over to pick them up (lucky me).
If you do use a common self bias resistor, or use a common fixed bias to two tubes . . .
Then measure the two plate to center tap winding DCRs, remember which one is which, and then CAREFULLY measure the DC voltages across each of the plate to center taps. Then calculate the current in each winding, Voltage / DCR.
Trust but Verify.
Your output transformers will love you.
Most amplifiers, push pull, or single ended that have moderate global negative feedback, will have closely matched gain of the Left and Right channels.
I do not use global negative feedback.
With very well matched tubes, my results are very good.
Your Mileage May Vary.
But . . . Those amplifiers came with very well matched output tubes, that were matched at the same quiescent operating conditions as the amplifiers that they went into (I hope most manufacturers actually did that).
And, if replacing is required, the manufacturer suggests using very well matched tubes.
Dyna Stereo 70 is one example.
I always used to design with individual self bias resistors and bypass cap for each tube.
Now, with my balanced amplifier, I use a single self bias resistor, and Extremely well matched JJ output tubes (From Eurotubes.com), I drive right over to pick them up (lucky me).
If you do use a common self bias resistor, or use a common fixed bias to two tubes . . .
Then measure the two plate to center tap winding DCRs, remember which one is which, and then CAREFULLY measure the DC voltages across each of the plate to center taps. Then calculate the current in each winding, Voltage / DCR.
Trust but Verify.
Your output transformers will love you.
Most amplifiers, push pull, or single ended that have moderate global negative feedback, will have closely matched gain of the Left and Right channels.
I do not use global negative feedback.
With very well matched tubes, my results are very good.
Your Mileage May Vary.
Last edited:
Single Bias pot ? The manufacturing adage in my early days selecting tubes for parallel P-P; select out of three tubes the nearest pair by measuring near sim anode currents (for given bias setting and screen ignoring UL), and do same for other P-P half with bias setting unaltered. That selection follows standard manufacturing deviations. Going to parallel pairs or triple sets, one has far more tolerance leverage and distortion is further reduced.
Many decades ago I made a homemade rig to create easy selection without the need to interfere with amp circuit. With such a tester every parameter is alterable, one soon discovers that many vendors who specify matched tubes, are not so good with their matching services.
Important in the tester fitment is a IO socket or another just with heater connections, as one tube can be soak tested while the other preheated. With heat resistant gloves, this saves considerable time. For those without such, one has to rely on good reputation sources. I found EH, JJ, Chelmer & Billington particularly consistent and there are others but with the car boot market or ebay, it´s caveat emptor.
I get good results selecting power tubes by measuring the anode current at two selected neg bias points, i.e for KT series for a given set anode voltage, measure anode current with grid bias at, -40 and other-60V giving near 2:1 anode current change. This will weed out the wildest.
Bench Baron
Many decades ago I made a homemade rig to create easy selection without the need to interfere with amp circuit. With such a tester every parameter is alterable, one soon discovers that many vendors who specify matched tubes, are not so good with their matching services.
Important in the tester fitment is a IO socket or another just with heater connections, as one tube can be soak tested while the other preheated. With heat resistant gloves, this saves considerable time. For those without such, one has to rely on good reputation sources. I found EH, JJ, Chelmer & Billington particularly consistent and there are others but with the car boot market or ebay, it´s caveat emptor.
I get good results selecting power tubes by measuring the anode current at two selected neg bias points, i.e for KT series for a given set anode voltage, measure anode current with grid bias at, -40 and other-60V giving near 2:1 anode current change. This will weed out the wildest.
Bench Baron
Attachments
Matching tubes seems over rated to me. Wouldn't imbalanced output tubes be the source for 2nd order harmonics? Isn't that what SET's have gobs of and people rave about? Serious question. That said, I incorporate a balance pot on my PP amps so I can set quiescent current between the two OT halves to zero. I've convinced myself that's a feature when using Toroidal output iron.
Ignoring the premise that any distortion is generally desirable (with which I'd disagree, but to each their own) we must first define "matching".
Pairs of output valves matched for idling current at some hopefully useful and representative set of applied DC voltages will minimize the dominantly 3rd order (but all odd) distortion of the OPT's core at high levels and low frequencies.
If instead, the valves are matched for gain (Gm, but usually at very unrealistic voltages, often not even DC, for example TV-7, other Hickoks, really most tube testers) they may idle at different currents for the same applied DC voltages, but may well balance out even order harmonics, at some expense to 3rd order issues in the OPT core. Also gives peak output power, but that's usually of very minor effect.
Matching for both is a gold standard and the bigger starting population the better the possible matches.
All good fortune,
Chris
Pairs of output valves matched for idling current at some hopefully useful and representative set of applied DC voltages will minimize the dominantly 3rd order (but all odd) distortion of the OPT's core at high levels and low frequencies.
If instead, the valves are matched for gain (Gm, but usually at very unrealistic voltages, often not even DC, for example TV-7, other Hickoks, really most tube testers) they may idle at different currents for the same applied DC voltages, but may well balance out even order harmonics, at some expense to 3rd order issues in the OPT core. Also gives peak output power, but that's usually of very minor effect.
Matching for both is a gold standard and the bigger starting population the better the possible matches.
All good fortune,
Chris
Last edited:
We don't ever want imbalanced idle currents, especially not with toroidal transformers, but with any of them if we don't want them saturating prematurely...so individual adjustments, or a balance pot (per pair). Either way gets you matched current. It is good to start with 4 close to each other, in bias and in Gm as well. But if the bias is balanced per channel, and both channels are biased the same, could you really hear the difference even if the Gm's were slightly off? Probably not.
Would I sport for all 4 matched each time? I definitely wouldn't pay more for it. Even if I mix up the tubes (once I bias, I'll try to take my time and avoid that) and re-insert them in a different order....Once I take any bias imbalance out of the equation I'm going to be good. I also like to keep my bias adjustments visible and marked somehow. Eventually the differences do become enough to matter as the tubes age, and I start to notice that someone's grid is way further pos/neg than the other 3...
Individual bias pots just make it a little easier. With a master bias control and balance pots, there is a lot more going back and forth to get them all settled in, equally biased. And I am all for matching tubes with a more rigorous test than simply one plate voltage and Ip. Though 2 sets each should suffice. When I do that, now I'm getting to see how that matched (or not) Gm comes into play....
Loren
Would I sport for all 4 matched each time? I definitely wouldn't pay more for it. Even if I mix up the tubes (once I bias, I'll try to take my time and avoid that) and re-insert them in a different order....Once I take any bias imbalance out of the equation I'm going to be good. I also like to keep my bias adjustments visible and marked somehow. Eventually the differences do become enough to matter as the tubes age, and I start to notice that someone's grid is way further pos/neg than the other 3...
Individual bias pots just make it a little easier. With a master bias control and balance pots, there is a lot more going back and forth to get them all settled in, equally biased. And I am all for matching tubes with a more rigorous test than simply one plate voltage and Ip. Though 2 sets each should suffice. When I do that, now I'm getting to see how that matched (or not) Gm comes into play....
Loren
For cases where the bias of each valve/tube was not possible, I always just used matched pairs. Nobody could 'hear' the difference.
For those kind of amplifiers where individual bias was not possible, it helps to de-gauss the OPT's (especially if you suspect the amp had been run for some time with mis-matched valves/tubes). There was nearly always an audible improvement after de-gaussing the OPT's.
1. I use individual self bias, and very well matched tubes on my push pull amplifiers.
Disadvantages:
Self Bias tends robs a little output power capability from the amplifier (less plate to cathode voltage).
Self Bias tends to have a bias shift, like when a 32Hz organ pipe plays for 30 seconds, and the amplifier is fully into Class AB1, with both tubes alternately cut off during parts of the complete cycles of the music.
Advantages:
Measuring the voltage across the self bias resistors allows you to instantly see how good the current balance is between the tubes . . .
When you first install the tubes,
And then when you check them 6 moths later, 1 year later . . . etc.
2. I use a single self bias resistor on my output tubes, no bypass capacitor, and extremely well matched output tubes.
Advantages:
I do not have to use a bypass capacitor.
There is differential negative feedback, to help correct any error in the balance of push and pull.
Yes, the single self bias resistor does reduce the plate to cathode voltage, so the power output capability is lower.
I make those tradeoffs, knowing what they are.
Your Mileage May Vary.
Disadvantages:
Self Bias tends robs a little output power capability from the amplifier (less plate to cathode voltage).
Self Bias tends to have a bias shift, like when a 32Hz organ pipe plays for 30 seconds, and the amplifier is fully into Class AB1, with both tubes alternately cut off during parts of the complete cycles of the music.
Advantages:
Measuring the voltage across the self bias resistors allows you to instantly see how good the current balance is between the tubes . . .
When you first install the tubes,
And then when you check them 6 moths later, 1 year later . . . etc.
2. I use a single self bias resistor on my output tubes, no bypass capacitor, and extremely well matched output tubes.
Advantages:
I do not have to use a bypass capacitor.
There is differential negative feedback, to help correct any error in the balance of push and pull.
Yes, the single self bias resistor does reduce the plate to cathode voltage, so the power output capability is lower.
I make those tradeoffs, knowing what they are.
Your Mileage May Vary.
Thanks for all your comments. A tube manufacturer reached out to me asking my thoughts on providing matched pairs vs matched quads. It's clearly more work to match 4 tubes than a pair and would add to the cost to consumers to go through that process, even if it appears to "not cost more". Someone has to pay for that sorting/matching time and it's us 
I think often when someone is buying a quad it's for a stereo PP amp, or perhaps stereo parallel SE right? It would be nice to be able to request 'two pairs pretty close together, but not necessarily a perfect quad.' That's usually what I would be looking for. I'd rather not have L & R channels having tubes with wildly different performance, but if I'm not doing some kind of simultaneously parallel push-pull setup, I don't care so much about a quad.
Similarly with preamp tubes, often I want two tubes where each tube has really good balanced sections, and again the two tubes would hopefully be close-ish, but I don't need 4 perfectly identical sections.
Similarly with preamp tubes, often I want two tubes where each tube has really good balanced sections, and again the two tubes would hopefully be close-ish, but I don't need 4 perfectly identical sections.
This is the sting in the tail when using non-standard tubes. I hope to build something with EL38s, but to get two pairs that are reasonably matched I have to buy at least 12. Out of the 12 I will probably have two outliers, and 5 pairs fairly close, and from those 2 quads (so one quad spare).
I have a uTracer for matching and look for similar curves between pairs.
I don't think it is imperative to have a matched quad, but it is not good to have two pairs with very different characteristics.
I have a uTracer for matching and look for similar curves between pairs.
I don't think it is imperative to have a matched quad, but it is not good to have two pairs with very different characteristics.
Match the gain of 2 matched pairs of tubes?
Yes, but do not loose sleep at night.
I hope there is not a problem with the 1dB match of L and R channels on a phono cartridge, or if you are rich, lucky, or both . . .
you have a 0.5dB match.
And that is at 1kHz, but not that good at low frequencies, and not that good at high frequencies.
Or is the problem the record cutter . . . it does not matter which, the performance you get is the cutter plus the phono cartridge errors.
if you are feeling confident, RSS anybody? (root sum of the squares of the errors).
By the way, were your ears tested within the last year? are they matched within 1 dB across the audio band?
Things work much better than the error factors would appear to spoil things . . . our brains have enormous correction analysis power.
Happy Listening!
Yes, but do not loose sleep at night.
I hope there is not a problem with the 1dB match of L and R channels on a phono cartridge, or if you are rich, lucky, or both . . .
you have a 0.5dB match.
And that is at 1kHz, but not that good at low frequencies, and not that good at high frequencies.
Or is the problem the record cutter . . . it does not matter which, the performance you get is the cutter plus the phono cartridge errors.
if you are feeling confident, RSS anybody? (root sum of the squares of the errors).
By the way, were your ears tested within the last year? are they matched within 1 dB across the audio band?
Things work much better than the error factors would appear to spoil things . . . our brains have enormous correction analysis power.
Happy Listening!
