So the idea is to take an existing tube amplifier operating in class AB and make it run more class A.
In order to sell, we'll assume the amplifier was designed with high gain and as much power output as can be squeezed from whatever output tubes were used. in other words, designed cover a wide range of input levels and speaker efficiencies. The trick is pentodes for gain and high B+ values for high power in class AB.
One could simply turn up the bias current until the plate dissipation is maxed out and conclude that's the most "lean towards A" as can be done with a given design; the B+ value is what it is for the design and tube bias current is thus limited - before the plates start to glow.
What would happen if you took an isolation transformer and used it to buck-down the HV secondary winding of the existing amplifier's power transformer? That would drop the B+ - and not the filament nor bias winding. Drop B+ by 100V or so, turn up the bias current to again approach the max power dissipation of the output tubes. Does doing so lean things more toward class A operation?
Of course, the idea wont work with any amp design, such as one with a regulated screen supply at an independently fixed voltage value. That would have to be modified. Also, the B+ values to the input and driver tubes would go lower in value as well. The presumably higher current would have to be supported by the amps power transformer HV secondary winding and the rest of its power supply.
I did this and the principle worked in practice. Whether it made the amp "sound better" I cannot say. Whether it's sustainable indefinitely or was just for show/demo I cant say either. (The bucking transformer I used was capable of 10X the current being drawn from the output tubes)
Is it a viable idea? At least there's no power being burned that you'd have to design a heat-sink for, like there would be if you built a linear regulator to reduce the B+. Being able to otherwise run the amplifier B+ at various values (with or without a corresponding bias adjustment), will make the amplifier sound differently, corresponding to a setting - of, say, taps on a bucking transformer secondary.
An engineer friend insists "just replace the power transformer with one that has the proper secondary winding voltage". That's not DIY, or as much DIY as this idea. Thanks for considering!
In order to sell, we'll assume the amplifier was designed with high gain and as much power output as can be squeezed from whatever output tubes were used. in other words, designed cover a wide range of input levels and speaker efficiencies. The trick is pentodes for gain and high B+ values for high power in class AB.
One could simply turn up the bias current until the plate dissipation is maxed out and conclude that's the most "lean towards A" as can be done with a given design; the B+ value is what it is for the design and tube bias current is thus limited - before the plates start to glow.
What would happen if you took an isolation transformer and used it to buck-down the HV secondary winding of the existing amplifier's power transformer? That would drop the B+ - and not the filament nor bias winding. Drop B+ by 100V or so, turn up the bias current to again approach the max power dissipation of the output tubes. Does doing so lean things more toward class A operation?
Of course, the idea wont work with any amp design, such as one with a regulated screen supply at an independently fixed voltage value. That would have to be modified. Also, the B+ values to the input and driver tubes would go lower in value as well. The presumably higher current would have to be supported by the amps power transformer HV secondary winding and the rest of its power supply.
I did this and the principle worked in practice. Whether it made the amp "sound better" I cannot say. Whether it's sustainable indefinitely or was just for show/demo I cant say either. (The bucking transformer I used was capable of 10X the current being drawn from the output tubes)
Is it a viable idea? At least there's no power being burned that you'd have to design a heat-sink for, like there would be if you built a linear regulator to reduce the B+. Being able to otherwise run the amplifier B+ at various values (with or without a corresponding bias adjustment), will make the amplifier sound differently, corresponding to a setting - of, say, taps on a bucking transformer secondary.
An engineer friend insists "just replace the power transformer with one that has the proper secondary winding voltage". That's not DIY, or as much DIY as this idea. Thanks for considering!
Last edited:
Correctly biased Class AB works well. Class A works well. Halfway between them is worse, so why choose this?
Perhaps a discussion of proper AB bias point selection is in order. Obviously there is a lot of territory between B and A to choose from but choosing the "best" operating point can seem like magic.
Also with my limited understanding it seems that it would be very hard to move your bias effectively without changing B+.
Also with my limited understanding it seems that it would be very hard to move your bias effectively without changing B+.
I guess the answer to "why" is the same why does anyone make modifications to anything - car, boat, motorcycle, airplane - that changes it from the stock form in some beneficial way for the owner and the owner's specific use case.
I'm making an assumption that compromises have been made in the stock form that allows the product to cover as wide a range of application as possible. And, these compromises can be exploited to improve some aspect for a specific use case.
Maybe in the last 20 years it's all been concisely worked out - if you're going to use tubes to amplify an audio signal, you do this, this, this - or - that, that, that; anything else is just a joke and/or garbage. If you want to improve on the amplifier you already have, there's no other realistic option than to use it as-is - or sell it and then buy/build one implementing one of the two circuit topologies known to work very well...
OK, glad I asked. So what's my best bet to buy 'n build for a < 10W / ch all tube implementation? First place to start is DIYAudio Store? Meanwhile, I'll get the paperwork going on the hopeless crap I mistakenly purchased off ebay, thinking I could make it sound good via simply "tinkering around with it".
I'm making an assumption that compromises have been made in the stock form that allows the product to cover as wide a range of application as possible. And, these compromises can be exploited to improve some aspect for a specific use case.
Maybe in the last 20 years it's all been concisely worked out - if you're going to use tubes to amplify an audio signal, you do this, this, this - or - that, that, that; anything else is just a joke and/or garbage. If you want to improve on the amplifier you already have, there's no other realistic option than to use it as-is - or sell it and then buy/build one implementing one of the two circuit topologies known to work very well...
OK, glad I asked. So what's my best bet to buy 'n build for a < 10W / ch all tube implementation? First place to start is DIYAudio Store? Meanwhile, I'll get the paperwork going on the hopeless crap I mistakenly purchased off ebay, thinking I could make it sound good via simply "tinkering around with it".
I don't think it is an unreasonable question at all. Lots of folks modify console amps for improved performance in one respect or another. Often times an original designer's primary goals may be different from the new owner's goals.
So it seems reasonable to investigate what happens if we move the crossover point to a higher or lower power output point and how to affect such changes.
So it seems reasonable to investigate what happens if we move the crossover point to a higher or lower power output point and how to affect such changes.
The optimum OT primary impedance for class AB and class A is usually somewhat different. (two tubes driving at same time for class A, only one tube driving at a time in the class B portion)
Idle current for class AB is usually (I think) set so the sum of gm's at the zero crossing is equal to that of the single tube gm when one tube cuts off. Roughly trying to get constant gm (sum) through-out the class A portion of the class AB spread. However, Mosfets typically have less total gm variation as the idle currents are increased, so the tube model probably is not set in stone.
Some modeling could show how the gm sum (versus signal) behaves for the tubes versus bias currents.
Idle current for class AB is usually (I think) set so the sum of gm's at the zero crossing is equal to that of the single tube gm when one tube cuts off. Roughly trying to get constant gm (sum) through-out the class A portion of the class AB spread. However, Mosfets typically have less total gm variation as the idle currents are increased, so the tube model probably is not set in stone.
Some modeling could show how the gm sum (versus signal) behaves for the tubes versus bias currents.
Last edited:
just get rid of the crossover distortion in class AB and you should be fine, go for the lowest bias that you can get away with...too much more and you are just hastening tube death...
Could leave the B+ and Vg2 the same as before for class AB, but put a differential driver in front of each tube with "local" N Fdbks for each tube. Then the tube V gains would be held constant, and crossover would be smoothed out. Could run with lower idle currents too.
😀 right handed triode or screen grid drive?
David Berning's EA230 comes to mind...
Sorry guys slightly OT question, for tube amps do you really hear a great difference between a well design class AB over class A amps. Aside bass slam, to my ears a proper design class AB tube amp sounds better then many class A ss amps that I've heard before.
I know its augmentative but some how they're more organic/natural sounding.
Thanks
I know its augmentative but some how they're more organic/natural sounding.
Thanks
no, remember it is not only the amp but the speaker you connect it with....
you do not listen to the amp, you listen to the whole setup...so how you did it is what really mattered...
we discuss amps here to make it engineeringly sound....not let out the magic smoke and have it last a bit longer free from distortions, noise and hum...
you do not listen to the amp, you listen to the whole setup...so how you did it is what really mattered...
we discuss amps here to make it engineeringly sound....not let out the magic smoke and have it last a bit longer free from distortions, noise and hum...
In that case, you need a design brief. Unless you know what it is you want to acheive (more power? less distortion? wider frequency range? all of these?) you have no target to aim at.
Probably correct - compromises will have been made. $ compromises generally, but possibly others including some that wouldn't be viewed as a compromise then, but may be now with better or different technologies available.
No, the existence of this forum attests to the fact that there are several thousand ways to skin the tube cat. There are a couple of handfuls of best practise approaches though.
And we are back at my original point. Define what you mean by "sound good" in the context of any capablities and limitations presented by your skills, your budget, and the equipment you have. With a defined goal, your task becomes easier, as does your chance of success.
Most of the time a class AB amp is running in the class A portion with constant gm sum if biased correctly.
The class AB amp then picks up on it's gm as it goes up further into the class B section, maybe giving a bit more "slam" effect. Then reaching tube saturation (with consequent compression) so avoiding hard clipping. Not a bad combo.
The class AB amp then picks up on it's gm as it goes up further into the class B section, maybe giving a bit more "slam" effect. Then reaching tube saturation (with consequent compression) so avoiding hard clipping. Not a bad combo.
OK, thank-you for the opportunity. I bought an amplifier, which is said to be a Dynaco Stereo 70 clone. It's close, except no ultralinear taps on the output transformers - they instead ran the stock 6CA7 screens from 2 independent regulated supplies, one for each tube pair...
Goal 1. - I dont need all of 35W/Ch and would gladly sacrifice power output capability for improvement in other qualities. I've taken a shot at this goal by running the stock output tubes in "triode mode". There goes at least 1/2 of the original 35W/Ch for whatever sonic benefit triode mode provides. I'm willing to go even further with the reduction of output power, if I can exchange that for a sonic quality I might like better.
Goal 2 - Ensure the basics, like a flat 20 - 20k frequency response is maintained, despite the effect of such modifications that would reduce the output power as a consequence.
Goal 3 - Reliable operation, all tubes and passive components still operating within safe area limits, no blatant engineering taboos violated (like try to run the original P-P OTs SE)
Goal 4 - Take advantage of any opportunity the original circuit lends to the above. An example might be, since the G2/screens arent connected to the 2 provided regulated supplies anymore, can those be cleverly used to improve operation elsewhere in the circuit? (I'd think so)
Goal 5 - finally, use any simple trick from the "imagineering" book that would allow me to do something beneficial, however generally not possible given the original circuit. Short - far short - of rebuilding the whole thing toward the likes of the Berning EA-230 mentioned earlier.
When I see a table from elsewhere on the web, like
EL34 - 25W
Hot (70%) Avg (60%) Cool (50%)
300V 58mA 50mA 42mA
325V 54mA 46mA 38mA
350V 50mA 43mA 36mA
375V 47mA 40mA 33mA
400V 44mA 38mA 31mA
425V 41mA 35mA 29mA
450V 39mA 33mA 28mA
475V 37mA 32mA 26mA
500V 35mA 30mA 25mA
It makes me wonder if trying these different operating points may have some sonic benefit? I see the only way to do that is if you can change the B+ value, which cannot be done in the amp's stock form - it is what it is.
So if someone were to tell me the output tube is like a voltage controlled resistor and its effective resistance is set by both the quiescent bias current and B+ voltage... Furthermore, this resistance value must match the output transformer primary impedance to be optimum - and that's why you dont want to go messing with the B+, because it all was designed together - maybe I'd begin to understand why my idea wont provide any benefit.
In other words, all those B+ voltages in the table, along with their corresponding bias current ranges, each correspond to a specific output transformer primary winding impedance. As a general engineering rule, one cant take an output transformer designed to run at 500V, 30mA (~17K) and make it work - with any benefit - at 300V, 50 mA (6K). On the other hand, maybe the tube setup as a 10K "source" would enjoy the easy life of driving into a 15K "load" and - if the power output loss can be tolerated due to the mis-match - would provide an audible benefit. I dunno -
Perhaps I should go read the "Designing an amplifier around the output transformer" thread.
Goal 1. - I dont need all of 35W/Ch and would gladly sacrifice power output capability for improvement in other qualities. I've taken a shot at this goal by running the stock output tubes in "triode mode". There goes at least 1/2 of the original 35W/Ch for whatever sonic benefit triode mode provides. I'm willing to go even further with the reduction of output power, if I can exchange that for a sonic quality I might like better.
Goal 2 - Ensure the basics, like a flat 20 - 20k frequency response is maintained, despite the effect of such modifications that would reduce the output power as a consequence.
Goal 3 - Reliable operation, all tubes and passive components still operating within safe area limits, no blatant engineering taboos violated (like try to run the original P-P OTs SE)
Goal 4 - Take advantage of any opportunity the original circuit lends to the above. An example might be, since the G2/screens arent connected to the 2 provided regulated supplies anymore, can those be cleverly used to improve operation elsewhere in the circuit? (I'd think so)
Goal 5 - finally, use any simple trick from the "imagineering" book that would allow me to do something beneficial, however generally not possible given the original circuit. Short - far short - of rebuilding the whole thing toward the likes of the Berning EA-230 mentioned earlier.
When I see a table from elsewhere on the web, like
EL34 - 25W
Hot (70%) Avg (60%) Cool (50%)
300V 58mA 50mA 42mA
325V 54mA 46mA 38mA
350V 50mA 43mA 36mA
375V 47mA 40mA 33mA
400V 44mA 38mA 31mA
425V 41mA 35mA 29mA
450V 39mA 33mA 28mA
475V 37mA 32mA 26mA
500V 35mA 30mA 25mA
It makes me wonder if trying these different operating points may have some sonic benefit? I see the only way to do that is if you can change the B+ value, which cannot be done in the amp's stock form - it is what it is.
So if someone were to tell me the output tube is like a voltage controlled resistor and its effective resistance is set by both the quiescent bias current and B+ voltage... Furthermore, this resistance value must match the output transformer primary impedance to be optimum - and that's why you dont want to go messing with the B+, because it all was designed together - maybe I'd begin to understand why my idea wont provide any benefit.
In other words, all those B+ voltages in the table, along with their corresponding bias current ranges, each correspond to a specific output transformer primary winding impedance. As a general engineering rule, one cant take an output transformer designed to run at 500V, 30mA (~17K) and make it work - with any benefit - at 300V, 50 mA (6K). On the other hand, maybe the tube setup as a 10K "source" would enjoy the easy life of driving into a 15K "load" and - if the power output loss can be tolerated due to the mis-match - would provide an audible benefit. I dunno -
Perhaps I should go read the "Designing an amplifier around the output transformer" thread.
Last edited:
B+ is easily adjusted downward by reducing the value of the rectifier's input capacitor. At the limit of zero you have a choke input. Of course filtering downstream must be increased to suit. Variacs at way, way overvoltage aren't really a good plan.
Triodes are more linear at higher idling currents, always. Maybe not enough to matter to you, or worth the penalties, but still always true.
Triodes are more linear into lighter loads, again always. Ignoring possible output transformer vagaries, the output taps that give you the peak output that you believe you'll need with the lightest output valve loading are the best choice from a distortion perspective. (Also loudspeaker damping.)
Just because amplifiers of the 1950's always used B+ degraded from the output valves' supply for the driving stages, doesn't mean we must slavishly do the same in DIY. Driver stages' linearity benefits from higher supply voltages, especially for "modern" triode outputs.
So: define your goal. Forget about anything anybody tells you that talks about amplifier "watts". That's been wrong since God made mud. Think in dBW. Design to a goal: desired acoustic peak output, speaker conversion efficiency, room loss from near-field to listening position. Add 'em up; that's your amplifier output required. How do we best achieve *that* ?
All good fortune,
Chris
Triodes are more linear at higher idling currents, always. Maybe not enough to matter to you, or worth the penalties, but still always true.
Triodes are more linear into lighter loads, again always. Ignoring possible output transformer vagaries, the output taps that give you the peak output that you believe you'll need with the lightest output valve loading are the best choice from a distortion perspective. (Also loudspeaker damping.)
Just because amplifiers of the 1950's always used B+ degraded from the output valves' supply for the driving stages, doesn't mean we must slavishly do the same in DIY. Driver stages' linearity benefits from higher supply voltages, especially for "modern" triode outputs.
So: define your goal. Forget about anything anybody tells you that talks about amplifier "watts". That's been wrong since God made mud. Think in dBW. Design to a goal: desired acoustic peak output, speaker conversion efficiency, room loss from near-field to listening position. Add 'em up; that's your amplifier output required. How do we best achieve *that* ?
All good fortune,
Chris
Since the OT does not have UL taps, you have the perfect opportunity to try out "shunt Schade" local N Fdbk. (resistor from output tube plate to driver plate)
This may well give better linearity than "triode" mode. I would keep it in class AB mode, avoiding extensive mods. However, the driver/splitter stage will need re-working to provide for some current output for driving the low Z input formed by the "shunt Schade" N Fdbk.
This may well give better linearity than "triode" mode. I would keep it in class AB mode, avoiding extensive mods. However, the driver/splitter stage will need re-working to provide for some current output for driving the low Z input formed by the "shunt Schade" N Fdbk.
In order to do that he would need a swing choke.
"Watts" sound like "Muscle", and "Powerful" sounds good, macho. The more power speakers turn into heat the better they seems to appeal, but it is plainly wrong. Magic of words, like "Global Warming" sounds similar to "Global Warning", plain old hypnosis.
Last edited:
yes at the 1 watt levels, at full power the saturation limitations of the OPT gives you just about half the power....
so that at 35 watt full power at 1khz, 17 watts is available at 40 hz...
Reading that other thread suggests that a lower impedance driving a higher one may be better in some ways, than the case where the two are matched. Perhaps the matched case allows the greatest power transfer through the output transformer, so a mis-matched case has a consequence of reduced power. Like going to triode operation does.
The bias current and the plate voltage set the quiescent resistance of the tube, I hope. So if you had 350V on the plate, with 50 ma through the tube, that's a 7K resistor, dissipating 17.5 watts. So the 6CA7 tube in the above table can go from being a 20K resistor at one end of its bias conditions, to being a 5K at the other.
Some comments have suggested a reduced impedance driving the output transformer would be beneficial, and others have pointed out local feedback as one way to achieve that effectively. But it's clear that the tube naturally can run at different quiescent resistances just by changing the bias, moreso by changing the bias and B+ voltage.
Say the OEM says to bias the amp at 40 ma, each tube. You take a DMM and measure the voltage across the tube and find its 425V. You assume the 10K or so tube resistance matches up well with the output transformer's primary winding and that this bias setting is optimal for transferring power through the transformer into an 8 ohm load.
But other properties are appealing; like "they say there's less distortion with a lower impedance driving a higher" and "Triodes distort less with a larger current passing through them". So if you can turn down the voltage, turn up the current, the tube will appear to be a lower impedance source looking from the output transformer primary.
The consequence is, I assume, reduction of output power due to now having tube-transformer impedance mismatch. I'd guess the negative feedback way is better because the impedance lowering is dynamically effective at the plate - but you keep the original bias conditions and voltage swing.
Simply reducing the B+ and re-biasing to set a particular source resistance is a sticks 'n stones approach versus working out negative feedback along the likes of Mr Berning and that EA 230. I can see how his circuit would reduce the effective plate impedance of the output tubes. Other than copying his circuit exactly, whatever else I'd build would probably just turn into an oscillator. Hence the sticks 'n stones approach I initially suggested.
The bias current and the plate voltage set the quiescent resistance of the tube, I hope. So if you had 350V on the plate, with 50 ma through the tube, that's a 7K resistor, dissipating 17.5 watts. So the 6CA7 tube in the above table can go from being a 20K resistor at one end of its bias conditions, to being a 5K at the other.
Some comments have suggested a reduced impedance driving the output transformer would be beneficial, and others have pointed out local feedback as one way to achieve that effectively. But it's clear that the tube naturally can run at different quiescent resistances just by changing the bias, moreso by changing the bias and B+ voltage.
Say the OEM says to bias the amp at 40 ma, each tube. You take a DMM and measure the voltage across the tube and find its 425V. You assume the 10K or so tube resistance matches up well with the output transformer's primary winding and that this bias setting is optimal for transferring power through the transformer into an 8 ohm load.
But other properties are appealing; like "they say there's less distortion with a lower impedance driving a higher" and "Triodes distort less with a larger current passing through them". So if you can turn down the voltage, turn up the current, the tube will appear to be a lower impedance source looking from the output transformer primary.
The consequence is, I assume, reduction of output power due to now having tube-transformer impedance mismatch. I'd guess the negative feedback way is better because the impedance lowering is dynamically effective at the plate - but you keep the original bias conditions and voltage swing.
Simply reducing the B+ and re-biasing to set a particular source resistance is a sticks 'n stones approach versus working out negative feedback along the likes of Mr Berning and that EA 230. I can see how his circuit would reduce the effective plate impedance of the output tubes. Other than copying his circuit exactly, whatever else I'd build would probably just turn into an oscillator. Hence the sticks 'n stones approach I initially suggested.
- Status
- Not open for further replies.