I have wondered if a filament would draw a non constant amount of current. It would seem to me that it shouldn't , but if you wanted to make sure a CCS would be the way to go...
But isn't one of the concerns here that the heater (DC driven) is unevenly heated - hot on one end and cold on the other? This doesn't make too much logical sense to me since the current going in is the current going out, but I've been surprised more than once on this forum
But isn't one of the concerns here that the heater (DC driven) is unevenly heated - hot on one end and cold on the other? This doesn't make too much logical sense to me since the current going in is the current going out, but I've been surprised more than once on this forum
I use LT1083/84/85 depending on the filament currents required, and these will deliver 7.5A, 5A and 3A respectively and come in TO-220 and TO3P packages. They make a good CCS which is the best compromise I have come to WRT to dc heating. They operate well with as little as 2.5V - 3V across the regulator and 1.25V across the current setting resistor. The higher current devices perform a little better wrt drop out voltage at a given load current.
WRT to heating DHT filaments with AC, the only ones I consider acceptable for AC heating are the 2A3 and 45, I use 2.5VCT transformers generally with very low winding resistance and good balance, neither hum or IIRC hum induced IMD have been an issue with these amps. Using either one of these types as a driver I would use DC via CCS generally or use choke isolation.
I remember reading an article years ago in VTV about RF heating DHT tube filaments and the conclusion was that the audible results did not justify the trouble - and they had a huge amount of trouble with filament inductance and uneven heating of the filament along its length. These problems can start at relatively low frequencies. I attempted it at ~500kHz (SQ wave) and found it did not work that well, and the harmonics radiated right up into the lower end of the shortwave band.
Yeah, conventional H bridge drivers are digital only. Seems like some relatively low frequency resonant types might be a candidate, but even at 2MHz even filament heating can be an issue.
The answer to an earlier question is that they heat evenly with DC on the filament but the voltage gradient across the filament results in uneven emission across its length - this problem becomes progressively worse at higher filament voltages and/or lower grid bias voltage levels.
WRT to heating DHT filaments with AC, the only ones I consider acceptable for AC heating are the 2A3 and 45, I use 2.5VCT transformers generally with very low winding resistance and good balance, neither hum or IIRC hum induced IMD have been an issue with these amps. Using either one of these types as a driver I would use DC via CCS generally or use choke isolation.
I remember reading an article years ago in VTV about RF heating DHT tube filaments and the conclusion was that the audible results did not justify the trouble - and they had a huge amount of trouble with filament inductance and uneven heating of the filament along its length. These problems can start at relatively low frequencies. I attempted it at ~500kHz (SQ wave) and found it did not work that well, and the harmonics radiated right up into the lower end of the shortwave band.
Yeah, conventional H bridge drivers are digital only. Seems like some relatively low frequency resonant types might be a candidate, but even at 2MHz even filament heating can be an issue.
The answer to an earlier question is that they heat evenly with DC on the filament but the voltage gradient across the filament results in uneven emission across its length - this problem becomes progressively worse at higher filament voltages and/or lower grid bias voltage levels.
I'm not convinced the gradient problem is a big deal except in a few special cases like the 811 - I've been running 300B on DC for over 10yrs now and have not encountered any life issues or measurable performance issues either for that matter. (I run everything 5V and above on DC.)
I think that 5Hz is low enough that you may encounter new and unanticipated problems due to the thermal time constant of a typical dht filament.
Not to be curmudgeonly about this, but I think this is a solution looking for a problem.
No one has conclusively demonstrated to my satisfaction that any of this is audible in a properly designed implementation, some effects are measurable and should probably be designed out as George has commented, but beyond that it becomes a possibly baseless obsession.. Some people will tell you the only way to heat is with AC, others will say only DC with a CCS, and that voltage source based heating even with chokes sounds like *!%#@!!, switchers can never be cleaned up enough to be acceptable, and still others are completely agnostic. I have used all of these techniques as appropriate, plus low frequency RF and I have to say that I couldn't hear any demonstrable difference between any properly implemented scheme if the amplifier's noise spectra and other measured performance parameters were generally similar. (That's my caveat. And as a point of reference I am an EE with generally subjectivist tendencies.)
I actually prefer DC CCS based heating for the simple fact that it limits the heater current to nominal at all times and avoids inrush current related filament stress - which in theory should result in longer filament life, but who knows?
I think that 5Hz is low enough that you may encounter new and unanticipated problems due to the thermal time constant of a typical dht filament.
Not to be curmudgeonly about this, but I think this is a solution looking for a problem.
No one has conclusively demonstrated to my satisfaction that any of this is audible in a properly designed implementation, some effects are measurable and should probably be designed out as George has commented, but beyond that it becomes a possibly baseless obsession.. Some people will tell you the only way to heat is with AC, others will say only DC with a CCS, and that voltage source based heating even with chokes sounds like *!%#@!!, switchers can never be cleaned up enough to be acceptable, and still others are completely agnostic. I have used all of these techniques as appropriate, plus low frequency RF and I have to say that I couldn't hear any demonstrable difference between any properly implemented scheme if the amplifier's noise spectra and other measured performance parameters were generally similar. (That's my caveat. And as a point of reference I am an EE with generally subjectivist tendencies.)
I actually prefer DC CCS based heating for the simple fact that it limits the heater current to nominal at all times and avoids inrush current related filament stress - which in theory should result in longer filament life, but who knows?
Actually I do play around a lot with the 71A, and find currents in the range of 15 - 20mA appropriate depending on load and plate voltage. (And horrors I use a voltage source supply to heat the 71A filaments in my headphone amplifier.)
I have had lousy luck finding nice matched pairs, but most of them seem to sound very good and very similar if not completely dead.. The filament supports seem to be quite fragile and I have quite a number that I am afraid to use because the filament is so clearly off center that touching the control grid as the tube ages (even slightly) is a very real possibility. I guess these tubes are rather sensitive to shock or just had abysmal QC during the depression years which is when most of the ones I own were made. Globes seem a bit better in this regard, but they are older, and good ones even more expensive. I have bought several new pairs from different sources and in each case one of the two tubes in the pair had this problem. (NIB/NOS! in unopened boxes)
I have had lousy luck finding nice matched pairs, but most of them seem to sound very good and very similar if not completely dead..
The filament supports seem to be quite fragile and I have quite a number that I am afraid to use because the filament is so clearly off center that touching the control grid as the tube ages (even slightly) is a very real possibility. I guess these tubes are rather sensitive to shock or just had abysmal QC during the depression years which is when most of the ones I own were made. Globes seem a bit better in this regard, but they are older, and good ones even more expensive. I have bought several new pairs from different sources and in each case one of the two tubes in the pair had this problem. (NIB/NOS! in unopened boxes)- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.