An amp staying purely in Class A will still have shifts in average current caused by second-order distortion in the output stage. With fixed bias this does not matter. With resistor bias you get a shift in quiescent current in the opposite direction, but this is not too serious and you can compensate by biasing the valves a little 'hotter' than is ideal. With CCS bias (however achieved) you get a bigger shift in quiescent current so you need either to bias even hotter or accept some crossover distortion on sustained loud passages. Hence resistor bias is the best option if you want to avoid user adjustments, and fixed bias is the best option if the user is happy to fiddle; CCS bias is never best.
cathode bias lose a lot of power and resolution.
to me it is unacceptable for a hifi amplifier.
DF96 is right that bias is higher for cathode bias, but max anode dissipation is lower.
I use a separate transformer and passive supply with individual capacitors to each grid.
The danger with fix bias is on shut down, if the (-) drops too fast you end up pushing for 1 second at maximum dissipation your tubes, especially with big capacitor banks.
so the trick is to have a minimum set up bias if the trimpot fails with a resistor in // then an individual charging capacitor after the trimpot.
Also another mistake with fix bias, especially with big tubes like kt150 is to have too much resistance to ground.
I use grid bias with battery in the phono stage, and grid bias in all stages of my amplifiers, or led bias too is super good.
to me it is unacceptable for a hifi amplifier.
DF96 is right that bias is higher for cathode bias, but max anode dissipation is lower.
I use a separate transformer and passive supply with individual capacitors to each grid.
The danger with fix bias is on shut down, if the (-) drops too fast you end up pushing for 1 second at maximum dissipation your tubes, especially with big capacitor banks.
so the trick is to have a minimum set up bias if the trimpot fails with a resistor in // then an individual charging capacitor after the trimpot.
Also another mistake with fix bias, especially with big tubes like kt150 is to have too much resistance to ground.
I use grid bias with battery in the phono stage, and grid bias in all stages of my amplifiers, or led bias too is super good.
I never built amps with Triode connected output stages for the same reason, i.e. deliberately wasting power. Different strokes for ditto folks I suppose
I never built amps with Triode connected output stages for the same reason, i.e. deliberately wasting power. Different strokes for ditto folks I suppose
Fair, but at least you get something for the power loss in triode mode.
10% power loss, of what decibels, voltage output?
i worry a lot about losing precious power that the output stage can develop, it sound a lot better with grid bias.
resolution I mean that the local feedback or the capacitor color the sound and it is absolutely not needed.
You are worried about Efficiency, Right? Do not take the following too seriously, but do think about it: Tubes are not efficient in most of the configurations and topologies of audio amps. Output transformers have a power loss, so do not use an OPT. OTL has a large power loss due to impedance mismatch, so do not use OTL. LED power dissipation is the same as resistor power dissipation (V x A). Battery Bias is great for safe power up (cold tubes are already biased); and for safe power down (hot tubes are biased during power down in case B+ is still there).
A 2A3 consumes 6.25W filament, 15W plate, 21.25W total. 6.25/21.25 = 29.4% filament, the rest 70.6% is the plate.
Don't use DHT unless it is a Thoriated filament (some are a little higher efficiency). Push Pull AB1 amp: do not use Individual LED for cathode bias, the LED will turn on and off, and will be very non-linear over the large changes in current. Use a common LED array.
Then use 2 elevated voltage 1 Ohm cathode resistors to monitor the cathode currents (and you may find that your "Matched" output tubes are not so well matched. These amps are big and heavy, they waste power being transported across the country or the ocean. The massive parts also take lots of energy to produce, so use solid-state amps. Linear solid state amps are too heavy. Use solid state switcher amps. All designs have tradeoffs, live with it.
A 2A3 consumes 6.25W filament, 15W plate, 21.25W total. 6.25/21.25 = 29.4% filament, the rest 70.6% is the plate.
Don't use DHT unless it is a Thoriated filament (some are a little higher efficiency). Push Pull AB1 amp: do not use Individual LED for cathode bias, the LED will turn on and off, and will be very non-linear over the large changes in current. Use a common LED array.
Then use 2 elevated voltage 1 Ohm cathode resistors to monitor the cathode currents (and you may find that your "Matched" output tubes are not so well matched. These amps are big and heavy, they waste power being transported across the country or the ocean. The massive parts also take lots of energy to produce, so use solid-state amps. Linear solid state amps are too heavy. Use solid state switcher amps. All designs have tradeoffs, live with it.
Last edited:
well, i know already about all this, nothing new here, I already use many of your 'suggestions' because power is important to me and has audible benefits. (edit: my conscience require that I build something which uses the resources to their full potential, I feel compelled to get that 10% or whatever it is power out of those output tubes.)
It is still fun to build amps with triodes and tube rectifiers.
maybe you didn't like what I said about cathode bias.
I was truthful to the question about his amplifier autobiasing, which many of you said that it was a cathode resistor, which he could modify to fix bias and enjoy it even more.
It is still fun to build amps with triodes and tube rectifiers.
maybe you didn't like what I said about cathode bias.
I was truthful to the question about his amplifier autobiasing, which many of you said that it was a cathode resistor, which he could modify to fix bias and enjoy it even more.
Last edited:
Please do not get me wrong, Fixed bias is a good thing. Like all other topologies there are tradeoffs.
Output stages for example: You can use fixed bias and interstage transformers. You can use fixed bias and RC coupling. You can use fixed bias and DC coupling.
If you use RC coupling, keep the signal below the point of drawing grid current, or you have the same effect as cathode R plus bypass cap self bias.
If you use interstage transformers, for a push pull amp, be sure to use matched tubes, or separate grid windings for individually adjusted biasing. Note that drawing grid current will change the load to both the interstage and the driver, account for that.
If you use DC coupling, be sure to deal with mis-matches, and drift of all causes. Again take account of any grid current.
If you use fixed bias, do notice what was said (before my posts): take care of power-up, and of power-down. I am sure lots of people have heard very excellent systems that used fixed bias, self bias, battery bias, and whatever.
Some topologies get a bad reputation, because they are not done properly. That is why I and 2 others did over 600 extensive measurements, multiple listening venues with different audiences and different loudspeakers, math proofs, and physics proofs. Then we wrote the results about parallel tubes for the last issue of Glass Audio, "Paralleling Tubes Effects" Volume 12, Number 5, 2000. All started as a lark, when we asked the question about the sound of parallel tubes: Why? The Michelson-Morely Experiment (they tried to prove the existence of Ether; but proved it did Not exist). We started with the premise that parallel tubes sound Bad, but proved with correct design and attention to details, that parallel tubes sound Good. Hmm.
"For every engineering problem there are 100 solutions, of which at least 3 will work . . . but only if they are implemented properly" - Me
More power sounds better, because of the Fletcher Munson effect. Fletcher's Munson effect causes extreme frequencies not to be heard by our ears when they are played at low levels.
Over 40 years ago, I took a friend to a Stereo warehouse. He was interested in those original Bose 901 speakers. We listened to music, including some low frequency organ music. Sounded pretty good.
Then I asked the salesman to play that organ music again, but through a pair of AR3a Acoustic Research speakers. Suddenly we FELT the Bass notes. My friend brought the AR3a speakers home with him.
In reality both the Bose 901s are not very efficient, because the low frequencies are boosted in order to level the apparent frequency response (and a majority of the power in music is quite often in the bass frequencies. Of course, the very high frequencies are also boosted, again to make up for the driver rolloff (but music highs require lower power than bass frequencies).
AR3a speakers are not noted for efficiency either.
Perhaps we should be using the 'last design' of the EV patrician with horn loading, 30 inch woofers, 12 inch bass, 8 inch lower mid, horn higher mid, and T350 horn tweeter. Then we can all use 45 amplifiers, which take a lot less power than many solid state non-switching amps.
Output stages for example: You can use fixed bias and interstage transformers. You can use fixed bias and RC coupling. You can use fixed bias and DC coupling.
If you use RC coupling, keep the signal below the point of drawing grid current, or you have the same effect as cathode R plus bypass cap self bias.
If you use interstage transformers, for a push pull amp, be sure to use matched tubes, or separate grid windings for individually adjusted biasing. Note that drawing grid current will change the load to both the interstage and the driver, account for that.
If you use DC coupling, be sure to deal with mis-matches, and drift of all causes. Again take account of any grid current.
If you use fixed bias, do notice what was said (before my posts): take care of power-up, and of power-down. I am sure lots of people have heard very excellent systems that used fixed bias, self bias, battery bias, and whatever.
Some topologies get a bad reputation, because they are not done properly. That is why I and 2 others did over 600 extensive measurements, multiple listening venues with different audiences and different loudspeakers, math proofs, and physics proofs. Then we wrote the results about parallel tubes for the last issue of Glass Audio, "Paralleling Tubes Effects" Volume 12, Number 5, 2000. All started as a lark, when we asked the question about the sound of parallel tubes: Why? The Michelson-Morely Experiment (they tried to prove the existence of Ether; but proved it did Not exist). We started with the premise that parallel tubes sound Bad, but proved with correct design and attention to details, that parallel tubes sound Good. Hmm.
"For every engineering problem there are 100 solutions, of which at least 3 will work . . . but only if they are implemented properly" - Me
More power sounds better, because of the Fletcher Munson effect. Fletcher's Munson effect causes extreme frequencies not to be heard by our ears when they are played at low levels.
Over 40 years ago, I took a friend to a Stereo warehouse. He was interested in those original Bose 901 speakers. We listened to music, including some low frequency organ music. Sounded pretty good.
Then I asked the salesman to play that organ music again, but through a pair of AR3a Acoustic Research speakers. Suddenly we FELT the Bass notes. My friend brought the AR3a speakers home with him.
In reality both the Bose 901s are not very efficient, because the low frequencies are boosted in order to level the apparent frequency response (and a majority of the power in music is quite often in the bass frequencies. Of course, the very high frequencies are also boosted, again to make up for the driver rolloff (but music highs require lower power than bass frequencies).
AR3a speakers are not noted for efficiency either.
Perhaps we should be using the 'last design' of the EV patrician with horn loading, 30 inch woofers, 12 inch bass, 8 inch lower mid, horn higher mid, and T350 horn tweeter. Then we can all use 45 amplifiers, which take a lot less power than many solid state non-switching amps.
Fletcher Munson is very interesting, they pose a serious dilemma to hifi when playing any other level than at the mastering level.
It is exactly as described, play lower volume, sound becomes mid-rangy, play higher sounds become bass and extreme heavy.
Some use equalizers and constantly play with them, never satisfied. The other solution is to get the sound to sound right at lower volume with a compensating curve and for less than ideal room acoustics. I think it is a fine solution instead of playing the sound engineer.
I had a good solid state amplifier with tone controls, and Fletcher Munson effect compensation, sounded hifi but very irritating no matter how I would adjust it.
It is extraordinary that you write in the Glassware, I am an avid reader and built many of the innovative circuits there.
In a world of deception and less than ideal, non inspiring, colored sound, Glassware audio is a real gem, a diamond shinning and showing the truth.
I never heard the AR3a but they are considered some of the best vintage speakers ever built.
You are right that in the ideal world we would all be using 30 inch woofers etc and use 45 amps. it is probably better sounding as efficiency, no matter what they say, and small voice coils, some of the most natural systems.
I keep sinning with my all scan-speak new system (almost ready).
I can only add one thing smart to what you said, I still have no explanation as to why my new OT which are quad cores kinda sound so much better than anything else...
As soon as I listened to an ensemble and heard the oboe and bassoons clear and woody, with my other OT I need to pay attention and they are buried under other bass sounds, voices and everything else.
The improvement makes no scientific sense to me as the IMD, THD, bandwidth is similar, however the sound is just incredibly natural with perfect texture, transparency, even double bass has more detail. The sound was so good, it draws attention and sound just like wow it is there.
It just defeats anything I thought, a 2700scan speak tweeter has no right to play that good. The amplifier is just supposed to be another part with trade/offs. just an annoyance. It isn't right.
It is exactly as described, play lower volume, sound becomes mid-rangy, play higher sounds become bass and extreme heavy.
Some use equalizers and constantly play with them, never satisfied. The other solution is to get the sound to sound right at lower volume with a compensating curve and for less than ideal room acoustics. I think it is a fine solution instead of playing the sound engineer.
I had a good solid state amplifier with tone controls, and Fletcher Munson effect compensation, sounded hifi but very irritating no matter how I would adjust it.
It is extraordinary that you write in the Glassware, I am an avid reader and built many of the innovative circuits there.
In a world of deception and less than ideal, non inspiring, colored sound, Glassware audio is a real gem, a diamond shinning and showing the truth.
I never heard the AR3a but they are considered some of the best vintage speakers ever built.
You are right that in the ideal world we would all be using 30 inch woofers etc and use 45 amps. it is probably better sounding as efficiency, no matter what they say, and small voice coils, some of the most natural systems.
I keep sinning with my all scan-speak new system (almost ready).
I can only add one thing smart to what you said, I still have no explanation as to why my new OT which are quad cores kinda sound so much better than anything else...
As soon as I listened to an ensemble and heard the oboe and bassoons clear and woody, with my other OT I need to pay attention and they are buried under other bass sounds, voices and everything else.
The improvement makes no scientific sense to me as the IMD, THD, bandwidth is similar, however the sound is just incredibly natural with perfect texture, transparency, even double bass has more detail. The sound was so good, it draws attention and sound just like wow it is there.
It just defeats anything I thought, a 2700scan speak tweeter has no right to play that good. The amplifier is just supposed to be another part with trade/offs. just an annoyance. It isn't right.
Last edited:
10% power loss, which is what I said.gabdx said:
Don't lose sleep over less than 10% loss in the cathode resistor when you are losing maybe 20-30% in the heaters.
Maybe you are using the wrong capacitor value? Lots of people miscalculate a cathode bypass cap value, thinking that they are merely bypassing the cathode resistor when in fact they are operating with the typically much smaller cathode impedance.
But, here, others may be interested, in any case, do not read all this and will continue with your doubts "in eternum"....
AutoBias Adjustment — PrimaLuna USA
AutoBias Adjustment — PrimaLuna USA
6L6 - Wikipedia
It is wonderful, the PL autobias works perfectly, I have changed several types of valves, and also enjoyed the triode switching to pentade Ul "hot", for many years. On one occasion a channel began to distort, after a simple exchange of valves, I detected that it was a defective KT88. (It should have lasted longer, they are Chinese (Shuguang) Superseded it, everything went back to normal.
Last edited:
Most worry about a parallel RC network: cathode resistor and capacitor in parallel; while actually capacitor plus internal series resistance/impedance is way more important.
Rk or RC is trivial, do you really think a 100uf or 150uf , f3 at -5hz or 1hz would make a difference? I don't.
The worst that can happen is that the capacitor value is higher than required by like 50%, it is not the end of the world.
Academia, this is interesting, auto bias ok, how do you know that the OT has the maximum inductance at this 'adjustment'?
Answer: you have no clue
we do balance the currents in the OT to obtain 2n order reduction of harmonics and most important to reduce the resultant currant which eats your transformer power.
Try an Edcor (the best ones) and an audionote, take a lundhal, take a toroidal.
Measure the THD at 20hz while you adjust the bias, you will see that the inductance has the biggest effect on the sound quality.
If the magnetic fields are not equal at equal bias, you need to change it a bit.
Then the 2n harmonics are not cancel perfectly right? So you unbalance slightly the drivers to compensate and get the best of both worlds. This is very advanced.
Rk or RC is trivial, do you really think a 100uf or 150uf , f3 at -5hz or 1hz would make a difference? I don't.
The worst that can happen is that the capacitor value is higher than required by like 50%, it is not the end of the world.
Academia, this is interesting, auto bias ok, how do you know that the OT has the maximum inductance at this 'adjustment'?
Answer: you have no clue
we do balance the currents in the OT to obtain 2n order reduction of harmonics and most important to reduce the resultant currant which eats your transformer power.
Try an Edcor (the best ones) and an audionote, take a lundhal, take a toroidal.
Measure the THD at 20hz while you adjust the bias, you will see that the inductance has the biggest effect on the sound quality.
If the magnetic fields are not equal at equal bias, you need to change it a bit.
Then the 2n harmonics are not cancel perfectly right? So you unbalance slightly the drivers to compensate and get the best of both worlds. This is very advanced.
gabdx,
I wrote for Glass Audio, not Glassware.
A large enough un-balenced current in a push pull transformer will not only increase 2nd order distortion;
it will also increase Intermodulation Distortion. All the signals at the extreme of one direction of a Bass alternation will be squashed by the lamination saturation.
At the other alternation direction, the other signals will not be squashed.
Agreed, many people use the self bias resistor resistance to calculate the needed capacitance of the bypass cap.
But the resistor is in parallel with the cathode impedance.
The cathode impedance of a triode is larger than 1/Gm, because the plate load is not zero, except for a cathode follower. The plate load/u is in series with 1/Gm.
Or is it 1/Gm + plate load/(u+1)? Anyway, they get the idea.
2nd order HD is not only dependent on the quiescent current balance of the output tubes, it is also dependent on the 2nd HD of the earlier and driver stages, and the output tubes transconductance balance too (it is a function of the sums and differences of those distortions, more excursion in the positive or in the negative direction of each stage, relative to the others). This too is a bit advanced, and requires a better write up.
The Heathkit W5M instruction manual has a graph of distortion of 20Hz through 100 Hz tones, versus un-balanced current in the OPT primary.
It does not seem to be real bad, but just remember that the negative feedback is there compensating, until the core saturates - then making the situation even worse.
I wrote for Glass Audio, not Glassware.
A large enough un-balenced current in a push pull transformer will not only increase 2nd order distortion;
it will also increase Intermodulation Distortion. All the signals at the extreme of one direction of a Bass alternation will be squashed by the lamination saturation.
At the other alternation direction, the other signals will not be squashed.
Agreed, many people use the self bias resistor resistance to calculate the needed capacitance of the bypass cap.
But the resistor is in parallel with the cathode impedance.
The cathode impedance of a triode is larger than 1/Gm, because the plate load is not zero, except for a cathode follower. The plate load/u is in series with 1/Gm.
Or is it 1/Gm + plate load/(u+1)? Anyway, they get the idea.
2nd order HD is not only dependent on the quiescent current balance of the output tubes, it is also dependent on the 2nd HD of the earlier and driver stages, and the output tubes transconductance balance too (it is a function of the sums and differences of those distortions, more excursion in the positive or in the negative direction of each stage, relative to the others). This too is a bit advanced, and requires a better write up.
The Heathkit W5M instruction manual has a graph of distortion of 20Hz through 100 Hz tones, versus un-balanced current in the OPT primary.
It does not seem to be real bad, but just remember that the negative feedback is there compensating, until the core saturates - then making the situation even worse.
Last edited:
gabdx,
I use speakers that are mid efficiency.
I have many different models.
Much of my listening is very near field, and so I do not need very much power from the amp. I mostly use only one channel, as only one ear is working.
I never was very well satisfied by tone controls and loudness compensation circuits either.
For over 2 decades I have not used either.
My major "tone controls" are what loudspeakers I use (I have many types), and which recordings sound good with my systems. The other "tone control" is how far I turn up the volume control . . . getting further up the set of Fletcher Munson curves.
I can get serious about the sound of a particular system; but then I can relax and just listen to enjoyable music too.
I find that some of the best sounding instruments to me are in a lower to mid frequency range. I theorize that is because the harmonics of those tones fall at frequencies where the ear is most sensitive.
A good Bassoon, Cello, Bass Clarinet, Bass Trombone, etc. are some that seem to qualify for this effect. I like the bite of a Cello, the Brrrapp of a Bass Trombone, etc.
I have been at listening sessions where some people get too serious, and talking about the accuracy of the system.
At that point, I may say that the Clarinetist was using a # 2 reed, and should have been using a # 1 reed.
That brings it back to reality for those who understand that.
I use speakers that are mid efficiency.
I have many different models.
Much of my listening is very near field, and so I do not need very much power from the amp. I mostly use only one channel, as only one ear is working.
I never was very well satisfied by tone controls and loudness compensation circuits either.
For over 2 decades I have not used either.
My major "tone controls" are what loudspeakers I use (I have many types), and which recordings sound good with my systems. The other "tone control" is how far I turn up the volume control . . . getting further up the set of Fletcher Munson curves.
I can get serious about the sound of a particular system; but then I can relax and just listen to enjoyable music too.
I find that some of the best sounding instruments to me are in a lower to mid frequency range. I theorize that is because the harmonics of those tones fall at frequencies where the ear is most sensitive.
A good Bassoon, Cello, Bass Clarinet, Bass Trombone, etc. are some that seem to qualify for this effect. I like the bite of a Cello, the Brrrapp of a Bass Trombone, etc.
I have been at listening sessions where some people get too serious, and talking about the accuracy of the system.
At that point, I may say that the Clarinetist was using a # 2 reed, and should have been using a # 1 reed.
That brings it back to reality for those who understand that.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.