let´s assume a small triode has coupling 50pF from heater to cathode (i think is much smaller, <10pF)
at 50Hz this Xc is >60Mohm, if you used 1kohm cathode resistor, ~0,1mV of hum "injected" there
(assuming full potential 6,3V across whole filament)
enough to bother about AC?
at 50Hz this Xc is >60Mohm, if you used 1kohm cathode resistor, ~0,1mV of hum "injected" there
(assuming full potential 6,3V across whole filament)
enough to bother about AC?
Interesting discussion. My experience with AC vs DC heaters is with aikido preamp with IDHT tubes. I hear definately cleaner sound, more resolution and tighter bass with voltage regulated DC heaters than with AC. I haven´t tried CCS on heaters yet.
Interesting discussion. My experience with AC vs DC heaters is with aikido preamp with IDHT tubes. I hear definately cleaner sound, more resolution and tighter bass with voltage regulated DC heaters than with AC. I haven´t tried CCS on heaters yet.
Actually, the original post is claiming a sound difference between regulated DC and unregulated DC supplies.
Which as long as both voltages are the same and their are no hidden oscillations, there should be zero difference in signal quality.
If a 70+ year old man not who is not even in the same room can hear the difference then something else was changed besides the heater regulation. There's something else at play here that doesn't warrant another rehashed AC/DC preference debate.
I am ready to believe in such phenomena as soon as I see a controlled comparison with objective data results.
I would explore two possibilities:
1) LTSpice needs a ground reference, otherwise it does not work.
In the real circuit, you also need a reference, should not be left floating.
2) A possible difference between both circuits is the high frequency ripple.
This is more likely with the second circuit, the one with LM317.
1) LTSpice needs a ground reference, otherwise it does not work.
In the real circuit, you also need a reference, should not be left floating.
2) A possible difference between both circuits is the high frequency ripple.
This is more likely with the second circuit, the one with LM317.
I'd love to see a trace of heater power supply , just to know what junk the LM317 is putting out.
Anyway , after inrush , the current needed to supply heater is costant , so i don't see any difference between a Voltage Regulated or Current Regulated and also a well designed unregulated DC supply.
Anyway , after inrush , the current needed to supply heater is costant , so i don't see any difference between a Voltage Regulated or Current Regulated and also a well designed unregulated DC supply.
Try schottky or, better, silicon carbide diodes. I much suspect you could have surprises listening..... also try connecting a dedicated, long, mains supply cable to filament transformer alone. Let us know.....
Why you do not want a constant-current supply.
[1] Because such a supply will force a higher voltage at inrush...
[1a] which is higher wattage...
[1b] which causes higher temperature gradients between cold cathode and hot filaments
[1c] which exacerbates mechanical wear through different temperature coefficients
[2] Because at high inrush wattage or power, filament hot-spots "take the brunt"
[2a] which causes local super-heating
[2b] which rapidly weakens the filament structure
[2c] which with shorter life, results in tube filament becoming ... a fuse
[2d] ... opening, destroying the tube.
[3] Because with CCS supplying many tubes' filaments
[3a] The "death of any tube" becomes a lighter load...
[3b] Which causes CCS to increase voltage output
[3c] Which causes ALL the other tubes to have overdriven filaments
[3d] Which of course shortens their life substantially.
ERGO... "CCS" for filaments is a particularly stupid idea, born (or borne) from the nattering minds of folks who just think that every last circuit topology is adaptable to nearly every situation, from warming filaments to driving C- bias circuits. Please folks ... not everything is a nail, and therefore, not everything needs to be hit with a hammer.
GoatGuy ... PS: CVS is fine. But you might want to make it VARIABLE (which is trivially easy) to facilitate experimentation.
[1] Because such a supply will force a higher voltage at inrush...
[1a] which is higher wattage...
[1b] which causes higher temperature gradients between cold cathode and hot filaments
[1c] which exacerbates mechanical wear through different temperature coefficients
[2] Because at high inrush wattage or power, filament hot-spots "take the brunt"
[2a] which causes local super-heating
[2b] which rapidly weakens the filament structure
[2c] which with shorter life, results in tube filament becoming ... a fuse
[2d] ... opening, destroying the tube.
[3] Because with CCS supplying many tubes' filaments
[3a] The "death of any tube" becomes a lighter load...
[3b] Which causes CCS to increase voltage output
[3c] Which causes ALL the other tubes to have overdriven filaments
[3d] Which of course shortens their life substantially.
ERGO... "CCS" for filaments is a particularly stupid idea, born (or borne) from the nattering minds of folks who just think that every last circuit topology is adaptable to nearly every situation, from warming filaments to driving C- bias circuits. Please folks ... not everything is a nail, and therefore, not everything needs to be hit with a hammer.
GoatGuy ... PS: CVS is fine. But you might want to make it VARIABLE (which is trivially easy) to facilitate experimentation.
Last edited:
Force a higher current at in-rush. What Tosh. It will limit the inrush current.
Usind a CCS for multiple valves may be counter productive and I've never seen a design with individual CCS supplies for each valve but CCS for a single filament will not cause any stress on that filament - exactly the opposite. Even better is to use a slow start CCS.
Usind a CCS for multiple valves may be counter productive and I've never seen a design with individual CCS supplies for each valve but CCS for a single filament will not cause any stress on that filament - exactly the opposite. Even better is to use a slow start CCS.
Last edited:
Tosh indeed... get a tissue and clean your spectacles, Mr. Myopian. I did not say current, but voltage. Giving a wimpy "may be counterproductive" is just saying, "well, Goat, yes in the end you are right, but I wouldn't want to give you any credit here...".
But, not being a ******, I'll give you this straight up: the assertion depends on the resistance temperature coefficient of the filament material. I was under the impression that filament wire was neutral-to-negative (lower resistance with higher temperature), but in fact it is positive, consisting of almost nothing but pure (silicon-free) tungsten. So I've edited the comment to strike out the first and second points. Point 3 remains, and is serious.
GoatGuy
But, not being a ******, I'll give you this straight up: the assertion depends on the resistance temperature coefficient of the filament material. I was under the impression that filament wire was neutral-to-negative (lower resistance with higher temperature), but in fact it is positive, consisting of almost nothing but pure (silicon-free) tungsten. So I've edited the comment to strike out the first and second points. Point 3 remains, and is serious.
GoatGuy
Last edited:
Scope capture of the 6.3V filament voltage. I think this is OK. LM317 is not that fussy.
An externally hosted image should be here but it was not working when we last tested it.
As others have said, that is tosh. You get lower voltage at inrush etc.GoatGuy said:Why you do not want a constant-current supply.
[1] Because such a supply will force a higher voltage at inrush...
A constant-current heater supply might be unnecessary, but it won't do any harm. The only minor problem is that a current supply needs to be a bit more accurate than a voltage supply, as the heater temperature coefficient of resistance works the wrong way.
cold filament/heater has low resistance -> with ccs -> low startup voltage , which is increasing when temperature/resistance goes upERGO... "CCS" for filaments is a particularly stupid idea, born (or borne) from the nattering minds of folks who just think that every last circuit topology is adaptable to nearly every situation, from warming filaments to driving C- bias circuits.
ccs with filaments is "bad" because you have to adjust to each tube (differences between tubes);
going too much complicated, if many tubes used. (1 ccs per tube... $$$$)
Rather go with voltage regulation, can use all heaters in paralel, without any worries (+slow voltage startup circuit)
anyway not fan of linear regs, they can go pretty hot if pushed hard
--------------------------------------
indirect heaters were made to make things easy, not to make headaches
anyway it´s yours baby overmind, you decide 😀
Wrong. The current is constant- that's the point. If it is set to the correct operating value for the heater, then the heater cannot suffer *more* than normal current, whether at startup or otherwise!Why you do not want a constant-current supply.
[1] Because such a supply will force a higher voltage at inrush...
Wrong, for the same reason as above.[2] Because at high inrush wattage or power, filament hot-spots "take the brunt"
Very wrong. The current is constant, so the heaters can never receive more current than the circuit is designed for. In any case, heaters fail open, not short, so the heater supply would simply switch off if one of the heaters failed.[3] Because with CCS supplying many tubes' filaments
[3a] The "death of any tube" becomes a lighter load...
[3b] Which causes CCS to increase voltage output
[3c] Which causes ALL the other tubes to have overdriven filaments
[3d] Which of course shortens their life substantially.
MerlinB ... I'm sorry to say, but you've just demonstrated a BREATHTAKING lack of understanding of basic electronics. ... and you're repeating what other posters have posted, without regard to simple, straight physics.
I never never never said current wasn't constant. I did not. I said voltage was the inconstant, the varying aspect. Therefore your objection to #1 is meritless.
Your objection to #2 is as meritless as #1 ... for the same reason. You fail to see that a "constant current" device, by definition, must force substantially more VOLTAGE to its load as the load current draw drops. Indeed ... the "perfect constant current source" is modelled as a very, very large value resistor attached to a very, very high tension voltage source. Think about what that does when the load tries to change current flow by even a small amount. The voltage across load is [ V = Er/(R+r) ], where E and R are the equivalent very high voltages, resistances ... and 'r' is the equivalent resistance of the load (filament).
Lastly, your objection to #3 is dangerously incorrect, not from a oh-no, high voltage! point of view, but because it is exactly correct. Exactly. Tubes in parallel, being fed from a single CCS would experience exactly what I have shown.
Try again, old bean. You gotta get this right to make your objections stick.
GoatGuy
I never never never said current wasn't constant. I did not. I said voltage was the inconstant, the varying aspect. Therefore your objection to #1 is meritless.
Your objection to #2 is as meritless as #1 ... for the same reason. You fail to see that a "constant current" device, by definition, must force substantially more VOLTAGE to its load as the load current draw drops. Indeed ... the "perfect constant current source" is modelled as a very, very large value resistor attached to a very, very high tension voltage source. Think about what that does when the load tries to change current flow by even a small amount. The voltage across load is [ V = Er/(R+r) ], where E and R are the equivalent very high voltages, resistances ... and 'r' is the equivalent resistance of the load (filament).
Lastly, your objection to #3 is dangerously incorrect, not from a oh-no, high voltage! point of view, but because it is exactly correct. Exactly. Tubes in parallel, being fed from a single CCS would experience exactly what I have shown.
Try again, old bean. You gotta get this right to make your objections stick.
GoatGuy
Tubes in parallel, being fed from a single CCS would experience exactly what I have shown.
Hang on, you're talking about heaters in parallel? I thought they were in series... my mistake.
Heaters in parallel need a voltage supply. Heaters in series need a current supply. Therefore if someone is talking about a current supply it is reasonable to assume that they are using series connection.
Agreed, DF96. All but ["reasonable"] ... since it is, these days, MOST unusual to find a whole lot of tube with series connected DC fed filaments. AC-fed, yes. DC, no.
I doubt it is unusual, as you claim, but if it is then that just shows how far audio (DIY?) practice has departed from sound engineering principles. Parallel heaters need a voltage supply, which may have a slow turn-on if that makes people feel more comfortable. Parallel heaters on a current supply is asking for trouble.
- Status
- Not open for further replies.
- Home
- Amplifiers
- Tubes / Valves
- Effect of Regulated DC Filament Supply