Some valves were designed for series heater strings e.g. European TV valves. They had special heaters, designed to all heat up at the same rate so voltage hogging did not occur too much.
If a heater is a constant resistance, then 5% change in voltage or current will cause approximately 10% change in power. A heater is not a constant resistance. 5% change in voltage causes less than 10% change in power. 5% change in current causes more than 10% change in power. That is why some datasheets will say something like voltage sources should be within 7% and current sources within 5%.
If a heater is a constant resistance, then 5% change in voltage or current will cause approximately 10% change in power. A heater is not a constant resistance. 5% change in voltage causes less than 10% change in power. 5% change in current causes more than 10% change in power. That is why some datasheets will say something like voltage sources should be within 7% and current sources within 5%.
Voltage runaway can occur, hopefully only for a few seconds, when a series string of heaters is fed from a voltage source.Kay Pirinha said:
CCS heater supplies can significantly reduce the amount of overhead you need to build into a heater supply. 12x 6.3V 300mA tubes can easily demand 20+ amps of cold-starting current. You need much larger rectifiers to reliably handle that vs a steady-state 3.6A. The 6BQ7 I happen to have at my desk here at work has a cold resistance of 2.5 ohms making for 2.52 amps at 6.3V.
Similar to tube rectifiers vs diode bridges or DHRs meaning the difference between 200V and 400V supply filter caps.
I do not want to be unkind with you, but I think that you confuse "constant voltage" with "constant current"
I have relatives in Germany and they confirm that there, mains is a voltage source like here, so your example is into the "constant voltage" context.
No, electron flow does not depend on our view, current flow does.
Bingo! Speaking of the OP, his issue is feeding the 6922 with constant current.
No, as I said before the resistance is not constant, it is a function of temperature.
The heaters are not fried because steady state is reached due to heat transfer, mostly by radiation.
With a constant voltage connected across a chain of N series-connected heaters, the impedance driving each heater is the impedance of the other N - 1 heaters. Assuming similar heaters, the driving impedance is larger than the impedance of a single heater when N > 2. Hence, it is more similar to constant current drive than to constant voltage drive.
Yes, you are right, but valve manufacturers are/was clever enough to use materials with tiny and very similar temperature coefficients, so the whole string of N heaters behaves like a single one constant voltage connected, and hence self-protected.
If the previous condition is not met, e.g. the j-th heater increases its temperature coefficient and also does its temperature, self-destruction is more likely.
In case of failure on a string of N heaters is/was expected one heater fried.
More than "constant current" I would say "same current".
A series resistor is there to allow the adequate current, never saw a series capacitor however.
Last edited:
The series resistor may or may not increase the overall resistance "substantially". Obviously it will have greater effect when the heaters are all cold.
At the instant of switching on from cold all is well. Once everything is warmed up all is well. It is the transition which creates problems, if the small valves heat up more quickly than the big valves.
That is the crucial assumption, which be OK for audio using the same valve type, but not OK for a TV with a UHF triode on the same heater string as a fat rectifier. That is why TV valves were designed to have controlled warmup. Otherwise the 'constant current' you mentioned may, for some seconds, be three or four times the correct current.MarcelvdG said:
At the instant of switching on from cold all is well. Once everything is warmed up all is well. It is the transition which creates problems, if the small valves heat up more quickly than the big valves.
I think we give the valve manufacturers more credit than they deserve. "Controlled warmup" was rarely perfect in TVs, there was often one valve that glowed really brightly until the others warmed up.
Also there was a specific order of valves, in the UK it would typically would be:
Top:
Boost/ damper diode - that has a ceramic tube insulation between heater and cathode.
Horizontal output
Vertical output
Video output
Other timebase valves
Sync separator
IF stages
Audio
Tuner valves
CRT.
Chassis ground.
There were some variations but most followed this pattern.
And yes, I do remember encountering the use of a capacitor rather than resistor for power dropping. Power factor anyone?
Some models used a silicon rectifier to reduce the RMS as well as a resistor.
Also there was a specific order of valves, in the UK it would typically would be:
Top:
Boost/ damper diode - that has a ceramic tube insulation between heater and cathode.
Horizontal output
Vertical output
Video output
Other timebase valves
Sync separator
IF stages
Audio
Tuner valves
CRT.
Chassis ground.
There were some variations but most followed this pattern.
And yes, I do remember encountering the use of a capacitor rather than resistor for power dropping. Power factor anyone?
Some models used a silicon rectifier to reduce the RMS as well as a resistor.
You don't even have to go as big as TV rectifiers. My daily driver at work is a 7-tube series string AM/FM set and the 17EW8 incandesces like a light bulb when you first fire up the set for the day. Same for a 2x 50C5, 12AX7, 82 ohm series stringer at home but on the 12AX7. It used to bother me but with almost a decade of use on that 7-tuber the 17EW7 still runs fine, in-fact it's playing christmas tunes right next to me!
Anyway, series connected heaters definitely act more like current-controlled tubes than voltage controlled. Datalog the voltage across a heater string containing a variety of small signal and power tubes and you'll see that it can be rough for the small signal guys. That's even with controlled warm up but it's not for long and it evidently doesn't affect life much.
Anyway, series connected heaters definitely act more like current-controlled tubes than voltage controlled. Datalog the voltage across a heater string containing a variety of small signal and power tubes and you'll see that it can be rough for the small signal guys. That's even with controlled warm up but it's not for long and it evidently doesn't affect life much.
Sure! Especially when tubes' count went downhill, capacitors of about 4.5 µF were seen in some TV's.
Best regards!
I thought that was the issue you worried about, that if due to tolerances or whatever a heater gets slightly hotter than intended, positive feedback will aggravate that to some extent.
By the way, I have an ECC81 that lights up yellow at power-on even though its heater is connected to a normal 6.3 V transformer winding. There is a part of the filament sticking out of the cathode tube and that heats up much faster than the rest, causing a yellow glow for a few seconds.
Yes, I want to avoid that condition at all cost.
Constant voltage does not avoid transient at start-up, that's why I use soft start voltage regulators for heaters.
The "mullard flash" is harmless but widely discussed artifact of some
phillips/mullard filaments.
Didn't realize it was called the Mullard flash! My flashy 12AX7 is, in fact, a Mullard tube.
The resistance change might be somewhat of a positive feedback but I feel it is being a bit exaggerated here. Power = I^2 R. Current being constant means that the heater power is directly proportional to the heater resistance.
Resistance change in a heater is fairly linear from ~2 ohms at 20C to 21 ohms at, lets say, 900C, middle of the road for a small signal tube. That's a change of ~0.0216 ohms per degree C.
There just isn't enough resistance change to cause a runaway.
The resistance change might be somewhat of a positive feedback but I feel it is being a bit exaggerated here. Power = I^2 R. Current being constant means that the heater power is directly proportional to the heater resistance.
Resistance change in a heater is fairly linear from ~2 ohms at 20C to 21 ohms at, lets say, 900C, middle of the road for a small signal tube. That's a change of ~0.0216 ohms per degree C.
There just isn't enough resistance change to cause a runaway.
Assuming that the temperature difference between heater and ambient is proportional to power and that the resistance has a first-order relation with temperature (second- and higher-order temperature coefficients neglected), you can calculate the loop gain from those figures:
Suppose the temperature difference increases by 1 % for some reason. That's 8.8 K more temperature, so resistance increases by 0.19 ohm. The resulting increase in power under constant current drive is (0.19 ohm/21 ohm) * 100 % ~= 0.9047619 %.
Hence, the loop gain estimate is 0.9047619. That's less than 1, so you get no thermal runaway, but it does amplify inaccuracies by a factor of 1/(1 - 0.9047619) ~= 10.5.
Suppose the temperature difference increases by 1 % for some reason. That's 8.8 K more temperature, so resistance increases by 0.19 ohm. The resulting increase in power under constant current drive is (0.19 ohm/21 ohm) * 100 % ~= 0.9047619 %.
Hence, the loop gain estimate is 0.9047619. That's less than 1, so you get no thermal runaway, but it does amplify inaccuracies by a factor of 1/(1 - 0.9047619) ~= 10.5.
See Stefan-Boltzmann law. It is not as simple as you think, because of thermal time constants. The heater can heat up much more quickly than the cathode can radiate the heat away. Fortunately heaters are fairly robust.
I'm sure my assumption of temperature difference being proportional to power is overly conservative because of radiated emission, but then again, my assumption of a first-order relation between temperature and resistance is overly optimistic. I mean, 2 ohm at 293.15 K and 21 ohm at 1173.15 K extrapolates to the heater superconducting at 200.518421 K and getting a negative resistance below that temperature. That clearly doesn't make any sense, so there must be a higher-order relation between temperature and resistance as well. (In fact you can find it on Wikipedia: Electrical resistivity and conductivity - Wikipedia ).
In any case, there are plenty of physicists on this forum, so I hope one of them will come up with a better loop gain estimate.
I don't see why thermal time constants would matter much. If loop gain were greater than 1, thermal time constants would only affect the amount of time it takes for the heater to self-destruct.
In any case, there are plenty of physicists on this forum, so I hope one of them will come up with a better loop gain estimate.
I don't see why thermal time constants would matter much. If loop gain were greater than 1, thermal time constants would only affect the amount of time it takes for the heater to self-destruct.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.