Fixed bias and filament bias have the same point that's not using Rk, Ck (Rk-few ohms takes almost no effect on quality sound), I think they are quite similar in terms of electricity and sound quality. So the quality of bias circuit have important role there. Someone did these careful circuit by tube rectifier and CLC filter with very good results.
Last edited:
Yes, exactly. It should be possible to get similar quality sound with fixed bias. Strangely, though, I never managed to do this. I tried battery bias on the grid - didn't like the idea much of about 70v of battery DC! Anyway, I wasn't convinced. I tried CLC though not with tube rectifier, again not convinced. Rod Coleman suggested a good idea which was to use one of his regulators across a fixed resistor to generate the bias voltage, but for 70v that would require modifying the regulator. I don't know if glow tubes can be used for a negative voltage - haven't tried that. I'm sure there's a solution for uber-smooth DC on the grid. I just haven't got to it with my own experiments. I'm open to persuasion.....
Fixed bias can surely be better if properly done. It is not just a piece of power supply, usually poorly made. Has to be integrated in a global way. It is not good if you want to swap tubes and other components to see which one sounds better or other similar things. An amp made like that will always be no meat no fish.....
The advantage of filament bias is that someone has done a proper job and there is nothing left for mucking about except soldering and chosing a resistor....
I'm very happy with filament bias on my 4P1L PSE outputs. The amp is on 24hrs and has been for months and months. Perfectly stable. So one thing about a fixed bias supply is that it should be bomb-proof and fail-safe. This shouldn't be an issue in a careful design but it has to be designed in.
If you look at a lot of fixed bias supplies they are, as you say, very basic. Even just a solid state rectification and a big cap. I've tried that kind of thing and I don't think it is good enough. I have no ability to design a complex and better supply but there are very clever people about who can do these things. But you would really have to put some thought into the whole design.
If you look at a lot of fixed bias supplies they are, as you say, very basic. Even just a solid state rectification and a big cap. I've tried that kind of thing and I don't think it is good enough. I have no ability to design a complex and better supply but there are very clever people about who can do these things. But you would really have to put some thought into the whole design.
I use fixed bias in my SE amps with great results. I also love filament bias, but agree with 45 that if fixed bias is correctly design is really good.
Here is an idea I was playing with for the 6C4C. Can be used for 4P1L but needs voltage tweaking given bias voltage levels:http://www.bartola.co.uk/valves/2013/05/18/fixed-bias-supply/
Cheers,
Ale
Here is an idea I was playing with for the 6C4C. Can be used for 4P1L but needs voltage tweaking given bias voltage levels:http://www.bartola.co.uk/valves/2013/05/18/fixed-bias-supply/
Cheers,
Ale
Some disadvantages of filament bias for my opinion:
- only few tubes can run, e.g. 26 or 4P1L... they are small tube and have low grid voltage.
- hard to get exact resistor and these get plenty dissipation. I don't like hot anyway.
- hard to adjust the current's tube. Not much tube can run like datasheet, including curve response and aslo filament.
All of above can be solved easily by fixed bias
I actually don't like DC heater, sound is hash and a bit cold. AC heater with hum balance pot is my favorite. There is only 3-4mV AC noise in my 300B amp, even 1-2mV AC noise in my 2A3 amp, so you cannot hear hum from speaker with <= 92-93dB sensitivity at your chair.
Noise will be more down if using negative feedback or ultrapath capacitor, out of phase filament for PSE or some other methods...
Let take AC heater, especially 4P1L (2V parallel filament).... you can hear some more interesting and no noise
- only few tubes can run, e.g. 26 or 4P1L... they are small tube and have low grid voltage.
- hard to get exact resistor and these get plenty dissipation. I don't like hot anyway.
- hard to adjust the current's tube. Not much tube can run like datasheet, including curve response and aslo filament.
All of above can be solved easily by fixed bias
I actually don't like DC heater, sound is hash and a bit cold. AC heater with hum balance pot is my favorite. There is only 3-4mV AC noise in my 300B amp, even 1-2mV AC noise in my 2A3 amp, so you cannot hear hum from speaker with <= 92-93dB sensitivity at your chair.
Noise will be more down if using negative feedback or ultrapath capacitor, out of phase filament for PSE or some other methods...
Let take AC heater, especially 4P1L (2V parallel filament).... you can hear some more interesting and no noise
Last edited:
Some disadvantages of filament bias for my opinion:
- only few tubes can run, e.g. 26 or 4P1L... they are small tube and have low grid voltage.
- hard to get exact resistor and these get plenty dissipation. I don't like hot anyway.
- hard to adjust the current's tube. Not much tube can run like datasheet, including curve response and aslo filament.
All of above can be solved easily by fixed bias
I actually don't like DC heater, sound is hash and a bit cold. AC heater with hum balance pot is my favorite. There is only 3-4mV AC noise in my 300B amp, even 1-2mV AC noise in my 2A3 amp, so you cannot hear hum from speaker with <= 92-93dB sensitivity at your chair.
Noise will be more down if using negative feedback or ultrapath capacitor, out of phase filament for PSE or some other methods...
Let take AC heater, especially 4P1L (2V parallel filament).... you can hear some more interesting and no noise
I agree that 26 and 4P1L are the best tubes for filament bias, but also include 10Y and 01A if you wish. But they are superb tubes, so that's not a handicap and PSE 4P1L can be used in the output. But 300b and 2a3 are not practical. As for the resistor - yes it has to be a quality wirewound, but there are plenty of those. And bias is exact and stable. So I really don't see these as issues, apart from the obvious limitations of tube choice. And don't forget the advantages - no bypass or ultrapath caps or input caps on the first stage.
The sound of DC with Rod Coleman's filaments isn't hard or cold - but you need a REALLY good DC supply. If you try it with even a LM1084 it won't sound as good. And you can add chokes in the filament supply if you want to go uber-clean. Rod uses the 300b and he designed his regs to be superior to AC.
So my actual experiences don't agree with your opinion of filament bias. Have you tried PSE 4P1L outputs in filament bias with Rod's regs and a top wirewound resistor (like vitreous enamel)? I don't think you can know how good this sounds until you actually make it and listen to it. If you do make it you will be surprised. It's unlike any other output stage I've tried for transparency. Maybe this can also be done with a top quality fixed bias circuit - Ale's looks the kind of thing we need. But until then I've actually found a solution and I'm very happy with it. If you look at the "26 pre amp" thread you will see that more and more builders are now using filament bias - it's becoming more accepted as a solution of choice for tubes like the 26. And in terms of "driver options for the 300b" 4P1L in filament bias is a strong contender if a mu of around 10 is enough for your purposes. And as an output, a higher mu tube like the 4P1L gives very welcome gain, and that can give you better input tube options.
Last edited:
Any filament heating method (dc, RF, ultrsonic, solar panel, battery, CCS, off-line converter, etc) should be compared against ac to see whether you have a good solution or bad solution.
It only takes a minute to compare.
Any solution that does not sound better in every way, compared to ac, can be thrown away. ac sounds muddy and confused, with poor articulation. Stereo presentation is only modest.
Ordinary plain-rectified dc fails this test, which is no surprise - the filament is sharing the rectifier recharge and recovery pulses with the raw dc's capacitors.
DC with regulation using LM317 or other chips with noisy bandgap reference voltages may be slightly better or slightly worse than ac, depending on implementation.
But every amplifier deserves better than ac heating, with its hum and mediocre sound.
It only takes a minute to compare.
Any solution that does not sound better in every way, compared to ac, can be thrown away. ac sounds muddy and confused, with poor articulation. Stereo presentation is only modest.
Ordinary plain-rectified dc fails this test, which is no surprise - the filament is sharing the rectifier recharge and recovery pulses with the raw dc's capacitors.
DC with regulation using LM317 or other chips with noisy bandgap reference voltages may be slightly better or slightly worse than ac, depending on implementation.
But every amplifier deserves better than ac heating, with its hum and mediocre sound.
Hi Andy
I'd be wary of using a regulated fixed bias supply on a 300B if the B+ supply is unregulated. The latter will vary with mains voltage, but the bias won't, leading, probably, to unacceptable variations in plate current. When both bias and B+ supplies are unregulated, both track changes in mains voltage, and plate current is (surprisingly) stable. I say "surprisingly" as bias and supply voltage won't vary in exact proportion in practice
Though I've never tried regulated fixed bias- hence the "probably". I have some suitable Paul Hynes negative voltage regulators; I was going to try using them in conjunction with shunt regulators on B+, but I've put this on hold. I managed to destroy the shunt regs ; and setting the latter to e.g. twice the plate current would make an already very inefficient amplifier twice as bad!
I’ve attached a schematic of one channel of my power amp, for reference
Paul N
I'd be wary of using a regulated fixed bias supply on a 300B if the B+ supply is unregulated. The latter will vary with mains voltage, but the bias won't, leading, probably, to unacceptable variations in plate current. When both bias and B+ supplies are unregulated, both track changes in mains voltage, and plate current is (surprisingly) stable. I say "surprisingly" as bias and supply voltage won't vary in exact proportion in practice
Though I've never tried regulated fixed bias- hence the "probably". I have some suitable Paul Hynes negative voltage regulators; I was going to try using them in conjunction with shunt regulators on B+, but I've put this on hold. I managed to destroy the shunt regs
; and setting the latter to e.g. twice the plate current would make an already very inefficient amplifier twice as bad!I’ve attached a schematic of one channel of my power amp, for reference
Paul N
Attachments
I know that hard to compare sound quality of 2 amps/ tubes/ filament burns if they are not the same building and reportage by two people with distinct taste and different system. I really like LCL filament filter if I use DC heater
Let your ears be your choice ;-)
2A3, 6A3, 6B4G have the same construction and electrical data, so why 2A3 is more expensive than 2 others in same manufacturer 2-3 times ? It is easy to say that it can run in AC heater with least ac hum.
Legendary 300B Western amps built Ac heater for power tubes.
Let your ears be your choice ;-)
2A3, 6A3, 6B4G have the same construction and electrical data, so why 2A3 is more expensive than 2 others in same manufacturer 2-3 times ? It is easy to say that it can run in AC heater with least ac hum.
Legendary 300B Western amps built Ac heater for power tubes.
Last edited:
hi Andy,
as Paul notes: any grid bias voltage should aim to have the same dependency on the Mains supply Voltage as the Anode supply (B+), or else the Anode current will vary too much as the public supply fluctuates.
for instance, a 300B set up for the WE operating point of 350V 60mA (-74V) burns 21W idle. If the mains supply increases by 10% (which is allowed in the UK, and probably elsewhere, too) the anode current will increase to about 90mA, and increase anode dissipation to 31.5W. If the 350V/60mA point were less conservative, we risk overheating the 300B.
But there are ways around this problem:
- connect a negative supply, RC or LC filtered, and add a cap-multiplier at the end. This way, you can avoid electrolytics.
- or, you can make a regulator for the grid bias. But instead of using chips with bandgap references, or even gas tubes, which are needlessly noisy, you can use the B+ supply, tapped through a RC filter and cap-multiplier to generate a negative reference voltage. this reference follows the variations in B+, but has the noise cleaned out of it. Apply the reference to a PNP follower connected to bias supply V- ...... and you have a grid bias voltage, compensated for B+., and of low dynamic impedance (so you don't need an electrolytic). I'll draw it when I get a minute.
Last edited:
I know that many DIYers believe that removing the hum is the only design-aim for a DHT filament supply. But even if you mask out the hum by nulling with a hum-pot, you are still left with huge quantities of intermodulation (50/60 and 100/120Hz sidebands on every note in your music).
Removing this intermodulation (without introducing even worse problems, like many dc heating solutions) will make a major difference to the clarity and natural sound of DHT amps - as well as removing the hum.
But since I supply DIY kits of the DHT Regulators, you won't take my opinion too seriously. Maybe the view of many of the 2A3-based TRAM-2 preamp builders will help:
http://www.diyaudio.com/forums/diy-hifi-supply/156854-dht-otl-linestage-tram-2-a-17.html#post3576846
Then yes, let your ears choose!
Hi Rod,
Great idea, didn't think about referencing to +B. My current circuit doesn't regulate bias voltage at all. It has a capacitor multiplier and a follower. I thought this way it should follow the mains variation as well. Haven't simulated the response to raw supply variation as I'm not at home now, but will do it later. Hopefully should be good, alternatively should add a reference to +B as you suggested?
Cheers
Ale
hi Ale,
The advantage of referencing from B+ is that the operating point is preserved, whatever the cause of the change in B+. So maybe you have a valve rectifier that varies its drop with age, or because you plugged in a different rectifier: the grid bias will be compensated.
And if you have more than one power tube running on the same B+, an open-circuit failure of one tube might cause B+ to rise; again the compensating action of the B+ reference prevents overstress of the good tube.
The advantage of referencing from B+ is that the operating point is preserved, whatever the cause of the change in B+. So maybe you have a valve rectifier that varies its drop with age, or because you plugged in a different rectifier: the grid bias will be compensated.
And if you have more than one power tube running on the same B+, an open-circuit failure of one tube might cause B+ to rise; again the compensating action of the B+ reference prevents overstress of the good tube.
Naturally, it is not much harder to actually sense the anode current of the power tube, and servo the grid bias accordingly. Such a circuit can also start with a very negative voltage, to prevent applied B+ damage to the DHT when the filament is cold.
in this case, a lot of care is needed to keep the music signal completely out of the grid bias.
in this case, a lot of care is needed to keep the music signal completely out of the grid bias.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.