iam sure this question has come up before or is even unanswerable but i will ask regardless.
Iam planning to build a kt88 amp, i found the circuit for the modified williamson circuit on the Plitron website. A friend mentioned to me that fixed bias was for guitar amps and i should convert to auto bias. Should i keep the fixed bias or convert it back to the williamson? Your thoughts, opinions and past experiance would be greatly appreicated.
Iam planning to build a kt88 amp, i found the circuit for the modified williamson circuit on the Plitron website. A friend mentioned to me that fixed bias was for guitar amps and i should convert to auto bias. Should i keep the fixed bias or convert it back to the williamson? Your thoughts, opinions and past experiance would be greatly appreicated.
Fixed is best for high-power, high voltage amplifiers such as KT88 PP, as you get a lot more control and more power. Also you don't bleed off a lot of power through large cathode resistors.
If you are going for triode-connected with a lower B+ , then it might be ok as autobiased.
I recommend fixed bias with individual adjustments for each tube, that way you can use unmatched tubes and adjust the idle current to be equal. This is very important with PP amps as the output transformers do not like unbalanced DC in the primary- even more crucial for toroidals like the plitrons I believe.
Its a bit more work but worth it. Also you should try regulated screens (g2) using a tube regulator if you don't like ultralinear...
If you are going for triode-connected with a lower B+ , then it might be ok as autobiased.
I recommend fixed bias with individual adjustments for each tube, that way you can use unmatched tubes and adjust the idle current to be equal. This is very important with PP amps as the output transformers do not like unbalanced DC in the primary- even more crucial for toroidals like the plitrons I believe.
Its a bit more work but worth it. Also you should try regulated screens (g2) using a tube regulator if you don't like ultralinear...
I generally prefer fixed bias in my designs both SE and PP as it allows me to easily fine tune and find the "sweet spot" operating point for the output tubes I use, and perhaps more importantly it eliminates the need for high quality cathode bypass capacitors. This also makes more efficient use of the available B+ and wastes less power than autobias. (The power dissipated in the cathode resistor is wasted energy and can be a very significant % of the overall power consumed in an output stage. 
)
Good fixed bias designs recover much more gracefully from sustained clipping where the output tubes in a pp autobias design may be driven so far into class b that when signal drive is reduced to normal (not clipping) levels the output tubes actually cut off for some period of time - this is called "blocking" and sounds truly terrible. (Larger cathode caps needed for better bass response exacerbate this due to the longer time constant of the bias network) Note this is a simplification and this phenomena can be made even worse by bad driver stage design, but in general fixed bias designs can and do recover more gracefully from overload.
I have designed power amplifiers around everything from a 6BQ5 to K90 as well as dhts such as the 45/2A3/300B using fixed bias.
IMHO Autobias amplifiers to me usually (not always) sound more euphonic and less detailed, but for this reason may be more forgiving on bad sounding material.
edit: typo, and content
)Good fixed bias designs recover much more gracefully from sustained clipping where the output tubes in a pp autobias design may be driven so far into class b that when signal drive is reduced to normal (not clipping) levels the output tubes actually cut off for some period of time - this is called "blocking" and sounds truly terrible. (Larger cathode caps needed for better bass response exacerbate this due to the longer time constant of the bias network) Note this is a simplification and this phenomena can be made even worse by bad driver stage design, but in general fixed bias designs can and do recover more gracefully from overload.
I have designed power amplifiers around everything from a 6BQ5 to K90 as well as dhts such as the 45/2A3/300B using fixed bias.
IMHO Autobias amplifiers to me usually (not always) sound more euphonic and less detailed, but for this reason may be more forgiving on bad sounding material.
edit: typo, and content
Hi Kevin, it sounds like the blocking you're talking about is the cathode bypass capacitor charging up?
The blocking that I've always believed to be a problem is the coupling cap charging when the grid is driven positive causing grid current to flow. It gets charged such that the grid becomes more negative (the same result you spoke of.) Obviously, it takes some time for the cap to discharge through the grid resistor (with some time constant RC.)
Anyway, the way that I've understood (and believe) it is that a self biased amp will tend to increase the bias under clipping, thus reducing the amount of blocking caused by the coupling cap. This is the popular explanation for why self-biased amps often seem as though they can play louder than a similar fixed-bias amp even though with fixed-bias the maximum output can be measured as being higher.
A grid choke or tranformer coupling will provide a low DC resistance for the coupling cap to discharge with a resulting reduction in the time it takes to recover from blocking.
-- Dave
The blocking that I've always believed to be a problem is the coupling cap charging when the grid is driven positive causing grid current to flow. It gets charged such that the grid becomes more negative (the same result you spoke of.) Obviously, it takes some time for the cap to discharge through the grid resistor (with some time constant RC.)
Anyway, the way that I've understood (and believe) it is that a self biased amp will tend to increase the bias under clipping, thus reducing the amount of blocking caused by the coupling cap. This is the popular explanation for why self-biased amps often seem as though they can play louder than a similar fixed-bias amp even though with fixed-bias the maximum output can be measured as being higher.
A grid choke or tranformer coupling will provide a low DC resistance for the coupling cap to discharge with a resulting reduction in the time it takes to recover from blocking.
-- Dave
Hi Dave,
Yes, that's what I was talking about, I have run into this scenario in amplifiers I have troubleshot for friends that used really large (>470uF) cathode bypass capacitors.
The time constants I use in my grid coupling circuits usually don't result in audible blocking under any circumstances I have encountered, however the mechanism you mention does exist, and can be a major problem if very large values of coupling capacitance are used and/or the grid bias resistors are large values. (I have seen just this problem in Williamson based amplifiers with inappropriately chosen first and second stage phase splitter time constants when oscillating at subsonic frequencies.)
Interesting your experience with autobias has been more or less the exact opposite of my impression for a given steady state power rating, I have always felt that the fixed bias amplifier played louder, and dynamic headroom on burst waveforms has generally has measured better on amplifiers running fixed bias in my experience. (only) I suspect that other factors are also playing a role here, my regulated power supply designs are very stiff, and my amplifiers employ fixed bias, whereas the comparison amplifiers had beefy, but unregulated supplies and autobias.
What you say is true in terms of grid cap blocking, however I think the actual mechanism with autobias is that the effective increase in bias may actually prevent positive grid current from flowing.
I am definitely with you on the grid choke thing, I have designed such amplifiers (SE) for others and they have a lot of other advantages as well.. Using a 50 DHT this is the only practical way to go other than a IT..
Kevin
edit: fix typo, add comment
Yes, that's what I was talking about, I have run into this scenario in amplifiers I have troubleshot for friends that used really large (>470uF) cathode bypass capacitors.
The time constants I use in my grid coupling circuits usually don't result in audible blocking under any circumstances I have encountered, however the mechanism you mention does exist, and can be a major problem if very large values of coupling capacitance are used and/or the grid bias resistors are large values. (I have seen just this problem in Williamson based amplifiers with inappropriately chosen first and second stage phase splitter time constants when oscillating at subsonic frequencies.)
Interesting your experience with autobias has been more or less the exact opposite of my impression for a given steady state power rating, I have always felt that the fixed bias amplifier played louder, and dynamic headroom on burst waveforms has generally has measured better on amplifiers running fixed bias in my experience. (only) I suspect that other factors are also playing a role here, my regulated power supply designs are very stiff, and my amplifiers employ fixed bias, whereas the comparison amplifiers had beefy, but unregulated supplies and autobias.
What you say is true in terms of grid cap blocking, however I think the actual mechanism with autobias is that the effective increase in bias may actually prevent positive grid current from flowing.
I am definitely with you on the grid choke thing, I have designed such amplifiers (SE) for others and they have a lot of other advantages as well.. Using a 50 DHT this is the only practical way to go other than a IT..
Kevin
edit: fix typo, add comment
kevinkr said:
Kevin, I agree. I made a autobias based on 30 mA CC source with 200 muF on a EL84 (in fact ECL86 on a Revox Modell 40) and observed that after clipping the capacitor could load up to 30 volt - and stay there for the next cycle!
. So Dave, the -Vbias is obtained by filling the cap fast and slowly, throttled, release. In my case, very slowly.So what I did is bypass the whole CCS with a resistor taking another 5 mA (at the normal op point of 10V bias/300 V anode) and this really was able to bleed the capacitor in a gracefull manner
.I use the CC source just to force the tube into balanced mode because transformers are less forgiving for inbalance; should the tubes, which are now paired, be replaced in some futrure, then the owner will not be harmed by imbalance.
I also observe that small caps have just the same punch as large ones: even interstage capacitors can be smaller than one would like on the calculator.
triode_al, your observations are interesting, mainly because you're using a CCS instead of a simple resistor.
My first thought is that the time constant for discharging after blocking is going to be MUCH higher. Normally the cap would see the cathode resistor in parallel with the impedance looking up through the tube. The CCS is by design a very high impedance path, so the only place for the extra charge to go is up. (If the tube is cutoff that becomes a high impedance path too, so the charge can't go that way very quickly either.) With a 330 ohm resistor instead of the CCS the time constant would be much smaller. Obviously you recognized that already - and I'm not trying to argue anything, I'm just trying to understand what was going on and thinking out loud.
My other thought is more important (I think.) What caused the bypass cap to charge up in the first place? With an ordinary resistor we might guess that a larger than normal cathode current caused a larger than normal voltage drop, but with a CCS that can't happen. So something else must have caused the tube itself to bias up so high. More specifically, something caused the CCS to raise the cathode voltage so that the tube would be at some operating point that allowed 30mA of cathode current to flow. Off the top of my head, the only thing I can think of that would cause that is if the grid were made more positive for some strange reason.
Again, I'm not really trying to argue anything. I'm just trying to figure out what was going on.
-- Dave
My first thought is that the time constant for discharging after blocking is going to be MUCH higher. Normally the cap would see the cathode resistor in parallel with the impedance looking up through the tube. The CCS is by design a very high impedance path, so the only place for the extra charge to go is up. (If the tube is cutoff that becomes a high impedance path too, so the charge can't go that way very quickly either.) With a 330 ohm resistor instead of the CCS the time constant would be much smaller. Obviously you recognized that already - and I'm not trying to argue anything, I'm just trying to understand what was going on and thinking out loud.
My other thought is more important (I think.) What caused the bypass cap to charge up in the first place? With an ordinary resistor we might guess that a larger than normal cathode current caused a larger than normal voltage drop, but with a CCS that can't happen. So something else must have caused the tube itself to bias up so high. More specifically, something caused the CCS to raise the cathode voltage so that the tube would be at some operating point that allowed 30mA of cathode current to flow. Off the top of my head, the only thing I can think of that would cause that is if the grid were made more positive for some strange reason.
Again, I'm not really trying to argue anything. I'm just trying to figure out what was going on.
-- Dave
Really a good point !
I was looking at a CCS as a "elegant" substitute to a "complicated" servo in order to fix bias in power tubes.
Obviously it's a monstruous mistake.
Dave, with positive peaks on the grid, the tube WANT to pass more current but the CCS cannot accept that !
Thus it pushes the cathode hi to the climbs to "shut down" the tube.
This is my interpretation of the huge increase of voltage accross the bypass cap.
And of course, no way to discharge quickly
Yves.
I was looking at a CCS as a "elegant" substitute to a "complicated" servo in order to fix bias in power tubes.
Obviously it's a monstruous mistake.
Dave, with positive peaks on the grid, the tube WANT to pass more current but the CCS cannot accept that !
Thus it pushes the cathode hi to the climbs to "shut down" the tube.
This is my interpretation of the huge increase of voltage accross the bypass cap.
And of course, no way to discharge quickly
Yves.
Yes, CCS biasing works great on power tubes provided that you always operate in pure class A - this of course is not really a problem in an SE amplifier, but it is a different case in a PP amplifier where you may migrate into class AB2 when overdriven, at that point the grids are driven positive and the output tube delivers additional current which then charges the cathode bypass capacitor, the only thing that constrains the charge on the capacitor is the time that the tube is conducting and its internal impedance..
Limited experience........
I am not sure which bias method of bias sounds better. I know with fixed bias on triode connected 6V6 sounds good. With 325 volts on the plates, -25 volts bias sounded better than -24 volts.
For cathode bias, I use just enough cathode capacitance to get good low frequency response. I only use Solen polypropylene type.
I am not sure which bias method of bias sounds better. I know with fixed bias on triode connected 6V6 sounds good. With 325 volts on the plates, -25 volts bias sounded better than -24 volts.
For cathode bias, I use just enough cathode capacitance to get good low frequency response. I only use Solen polypropylene type.
In the late 70's there was alot of UK HiFi wisdom which claimed better sound quality from amps using cathode bias instead of fixed, whatever the design. I came across several custom cathode bias amps using KT88s and 66's run at 500V & 400V and at 80mA quies and they sounded good. Under such conditions, a quick reckoner tells one to turn the room radiator off and expect to replace tubes every so often.
It's interesting to note most of us for the high power end i.e over 30W are using fixed bias and are content with the sound quality.
The beauty about fixed bias for easy listening is that one can reduce the quiescent current down to conserve without a serious detriment in quality.
I would add that the KT90 sounds better in hifi using cathode bias but with such a tasty sized anode I slam it in fixed bias for MI.
richj
It's interesting to note most of us for the high power end i.e over 30W are using fixed bias and are content with the sound quality.
The beauty about fixed bias for easy listening is that one can reduce the quiescent current down to conserve without a serious detriment in quality.
I would add that the KT90 sounds better in hifi using cathode bias but with such a tasty sized anode I slam it in fixed bias for MI.
richj
jmartins said:
With AB1 the max sig current with music # cathode bias is probably around the 65mA ballpark. Not quite class A. It's tempting with such conditions to reduce the cathode resistor value even further until one gets near class A, then listen to the sound quality. The thd would also decrease.
At 50 mA quies (bit underrun) expect at least min 15000 hours (10+ years) running under light conditions with such a generous sized tube. Assuming 4hrs /day running with only one on/off cycle.
In contrast, on trumpet (MI) I'm running KT90's at 650V # 65mA quies fixed bias; under such grilling conditions in AB2, I replace 'em every 6 months.
richj
Yvesm said:Really a good point !
- Dave, with positive peaks on the grid, the tube WANT to pass more current but the CCS cannot accept that !
Thus it pushes the cathode hi to the climbs to "shut down" the tube.
Yves.
Yes it must be that the grid charges the coupling capacitor (like Dave mentions) , and I forgot to look at the voltage on the bias reistor: it is some 68k.
A resistor would not bee too bad, .. But in the Revox the grid goes directly to the balance potmeter, and this might give (not audible) open circuit once in a while. I have seen that the volume control which goes into the preamp (cascade ECC81) in this old Revox cracked; and I attached the scope and a 1.000 Hz signal and observed that the signal went positive and to zero while turning. Then I attached a 390k to the grid of the entrance, and the signal only went to zero while turning, with NO audible impact. No cracking, even though visible on the scope.
My lesson then, that I should add such a reistor in place for the entrance grid section (ECC83) of the output section as well!
Tnx, albert
I generally prefer fixed bias in my designs both SE and PP as it allows me to easily fine tune and find the "sweet spot" operating point for the output tubes I use, and perhaps more importantly it eliminates the need for high quality cathode bypass capacitors. This also makes more efficient use of the available B+ and wastes less power than autobias. (The power dissipated in the cathode resistor is wasted energy and can be a very significant % of the overall power consumed in an output stage.
Good fixed bias designs recover much more gracefully from sustained clipping where the output tubes in a pp autobias design may be driven so far into class b that when signal drive is reduced to normal (not clipping) levels the output tubes actually cut off for some period of time - this is called "blocking" and sounds truly terrible. (Larger cathode caps needed for better bass response exacerbate this due to the longer time constant of the bias network) Note this is a simplification and this phenomena can be made even worse by bad driver stage design, but in general fixed bias designs can and do recover more gracefully from overload.
I have designed power amplifiers around everything from a 6BQ5 to K90 as well as dhts such as the 45/2A3/300B using fixed bias.
IMHO Autobias amplifiers to me usually (not always) sound more euphonic and less detailed, but for this reason may be more forgiving on bad sounding material.
edit: typo, and content
Interesting observations.
Actually this thread must have the topic
"output stage with bias adjust vs. output stages without bias adjust", because "Autobias" means for me the use of a bias servo unit, where I can exact determine the value of the wanted idle current, independend of the individual tolerances of the tubes.
I note also, that the quality standart of the cathode bypass capacitor for the associated resistor makes large influence of the sound character. I can good play with smaller values, if there are loudspeakers with too much level in the low frequency aera - in certainly cases an advantage.
Can I assume, that for not matched tube pairs of push pull output stages (both "ultra linear" and "PPP") the autobias topology is to prefer?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.