Agree, and seems to me that you did not read my statement correctly
If current is exceeded, then we are not talking about a CCS.
I always design (constant voltage heater) PSUs with a proper soft-start, warm-up time is chosen from design.
For other reasons, I am designing a CCS for heaters, but, as I said before
If current is exceeded, then we are not talking about a CCS.
Yes, and that's exactly the reason why thermal runaway never will occur.
Best regards!
Are you kidding, right? Just in case you don't, let me explain
Heater temperature is related to power
W = I V = I² R = V² / R
For V=constant, an increase on R yields a decrease on W. Good thing.
For I=constant, an increase on R yields an increase on W. Bad thing.
I hope that now you get it.
Heater temperature is related to power
W = I V = I² R = V² / R
For V=constant, an increase on R yields a decrease on W. Good thing.
For I=constant, an increase on R yields an increase on W. Bad thing.
I hope that now you get it.
Last edited:
Increase of W leads to increase of temperature, and much higher heat radiation losses. As the 4th power of temperature! It is not easy to raise temperature a lot.
If the regulated current is pretty-near the rated value it will not run-away.
It helps that we "usually" have a voltage compliance. Say I get my 12AX7 heat from a 18V supply and 150mA current limiter. It could go over 12.6V, yes; but it can not get to 18V. 12AX7 will not die quick at 18V heat.
A reasonable worst-case *was* the series-string TV set. Here we threw 117V down a row of 38V and 16V tubes with the little 4V RF tuner at the end. If the little tube warmed-up faster than the big tubes, it could take a mighty large fraction of the 117V for some seconds. Indeed it got to be a problem so everybody had a meeting and defined "Controlled Warm-Up"; for series-use tubes, but the designs and materials became common for all small tubes.
I am thinking maybe put the LM317 have a slow start eventhough I have a CCS. The slow start should not be too difficult to implement.
That's wrong. Stefan–Boltzmann law says that the total energy radiated per unit surface of a black body across all wavelengths per unit time is
E = σ T⁴
T is the absolute temperature of the black body [T]=ºK
In real world the heater is surrounded by the cathode, and it is not a black body, so you must modify Stefan-Boltzmann law to something like
E = A {[T(heater)]⁴ - [T(cathode)]⁴}
As both absolute temperatures are close, radiation is not *that* high.
Another little detail: The absolute temperature of the heater is not uniform, Stefan-Boltzmann law does not apply.
To add injury to insult, radiation is not the only mechanism for heat transfer, because heater is in mechanical contact with cathode.
Voltage hogging would occurs regardless of current, unless it belongs to some authors imagination, Morgan Jones included.
From "Valve Amplifiers-Fourth Edition"
Just for the record, I never said that thermal runaway is a common failure, but saying that it is non existent is wrong.
This was a classic, in my childhood I saw it a couple of times on my hybrid Philips B&W TV.
I am not *that* old, color TV came here in 1978, even so I did see the football world cup in B&W.
Last edited:
Well, neither I ever heard of thermal runaway in series heater chains, fed by a CCS, nor did I ever experience that by myself. In 2001 I've built a set of monophonic Circlotrons, featuring a pair of PL519's each. Each tube's heater is fed from a 82VAC winding through a 12.7 µF series capacitor. This cap has an impdance of 250 ohms @ 50 Hz, which corrsponds to a maximum current of 330 mA @ 82 VAC. So it roughly can be defined as an AC CCS w.r.t.the tube's heater. After a one minute's warm up the voltage of each heater settles at roughly 40 V, with a voltage drop of 72 V across the capacitor.
At the end of tubes' heydays, tubes count in German b/w TV sets dramatically decreased to eventually four tubes. Several manufacturers decided to replace the resistors in the series heater chain by a capacitor of about 4 µF to maintain a heater current of 300 mA. This wouldn't have worked well if thermal runaway were to be suspected.
Today, and especially for you, I did the following: I set one half of my bench PSU to it's maximum output voltage of 32.5 VDC and it's current limiter to 0.3 A, addidtionally measured by a DMM. Then I grabbed a hand full of tubes from my large box of old pulls, connected a DVM in parallel with the heater pins, fed them, one after the other, from my supply, and watched the DVM's readings. The results were as following:
1st tube was a Siemens EF80. Reading started at 0.8 V and settled at 6.9v after about 50 seconds. A little bit more than the expected 6.3 V, but, remember, this tube primarily has been designed for parallel feeding with only the option of series heating. The main observation was the settling of the heater voltage, though.
2nd tube, a Valvo PCC88, also started at 0.8 V and, after 50 seconds, settled at 7.1 V, well within it's specs.
3rd tube, a Valvo PF86, started at 0.5 V and settled at 4.1 V within 45 seconds.
A Telefunken PCL82, my 4th tube, started at 2.4 V and, within 70 seconds, settled at 16.1 V.
Q.e.d., I'd say. Your claim of voltage hogging, or thermal runaway, has well been rebutted. There are two possible reasons that may explain it, the NTC characteristics of the heaters, and, as yet been said by PRR, the thermal radiation which grows with the 4th power of temperature.
Best regards!
At the end of tubes' heydays, tubes count in German b/w TV sets dramatically decreased to eventually four tubes. Several manufacturers decided to replace the resistors in the series heater chain by a capacitor of about 4 µF to maintain a heater current of 300 mA. This wouldn't have worked well if thermal runaway were to be suspected.
Today, and especially for you, I did the following: I set one half of my bench PSU to it's maximum output voltage of 32.5 VDC and it's current limiter to 0.3 A, addidtionally measured by a DMM. Then I grabbed a hand full of tubes from my large box of old pulls, connected a DVM in parallel with the heater pins, fed them, one after the other, from my supply, and watched the DVM's readings. The results were as following:
1st tube was a Siemens EF80. Reading started at 0.8 V and settled at 6.9v after about 50 seconds. A little bit more than the expected 6.3 V, but, remember, this tube primarily has been designed for parallel feeding with only the option of series heating. The main observation was the settling of the heater voltage, though.
2nd tube, a Valvo PCC88, also started at 0.8 V and, after 50 seconds, settled at 7.1 V, well within it's specs.
3rd tube, a Valvo PF86, started at 0.5 V and settled at 4.1 V within 45 seconds.
A Telefunken PCL82, my 4th tube, started at 2.4 V and, within 70 seconds, settled at 16.1 V.
Q.e.d., I'd say. Your claim of voltage hogging, or thermal runaway, has well been rebutted. There are two possible reasons that may explain it, the NTC characteristics of the heaters, and, as yet been said by PRR, the thermal radiation which grows with the 4th power of temperature.
Best regards!
I don't assume anything about the constant current source, except that it sources a constant current.
I meant the positive feedback mechanism you identified, whether that will lead to disasters depends on the loop gain. An initial temperature difference DELTA T_initial between the filament (or a hot spot at the filament) and the intended temperature will lead to a resistance increase DELTA R compared to the intended resistance, which will lead to a power increase DELTA P which will cause a new temperature increase DELTA T2. As long as DELTA T2 < DELTA T_initial, the temperature increase will remain bounded - as is the case with all valves Kay measured.
The interesting thing is that the temperature should still be less well-defined than under voltage drive, because even though it stays bounded, the temperature difference does increase to some extent due to the positive feedback effect. So all in all, I have to agree with Kay that current driving filaments can work, and I have to agree with you that voltage drive with current limiting should in principle be better than pure current drive. No idea whether the difference is big enough to be of practical importance, though.
Worst case scenario can be avoided, by using a comparator to trigger a decrease in CCS current once a certain voltage threshold is reached. This solution would be reserved for only the most die-hard, belts-and-suspenders, anal-overkill types of builds though.
I have to agree that the series chain issue is not common enough to worry. I've never lost a series heated filament, but I've had parallel filaments go open circuit during normal operation on two occasions, and it wasn't even during start up.
I prefer parallel AC heating, all things considered.
I have to agree that the series chain issue is not common enough to worry. I've never lost a series heated filament, but I've had parallel filaments go open circuit during normal operation on two occasions, and it wasn't even during start up.
I prefer parallel AC heating, all things considered.
Doing an analogy, last night I did hear a strange noise on my backyard, when I went there nothing happened, but the dog.
I did stay watching for an hour! Nothing did happens!
Conclusion: Aliens does not exist. Q.E.D. I'd say.
That's wrong, tungsten has a POSITIVE temperature coefficient, take your multimeter and measure cold vs hot heaters and convince for yourself.
Thermal radiation issue was already covered on post#48.
Well observed, but nothing is affected. Due to it's POSITIVE temperature characteristis, the tube's heater controlls it's power consupmtion automatically.
@ all the others: There's no need to provide additional zeners or other belts and braces. Believe me. I've succwith their series heater stringsessfully done CCS fed heaters multiple times, not to count millions of old b/w TV sets with their series heater chain from the 1950ies to the beginning of the 1970ies .
Best regards!
@ all the others: There's no need to provide additional zeners or other belts and braces. Believe me. I've succwith their series heater stringsessfully done CCS fed heaters multiple times, not to count millions of old b/w TV sets with their series heater chain from the 1950ies to the beginning of the 1970ies .
Best regards!
OK, then we agree on that.
The formal treatment of this issue would be much more complicated, even if treated as a classical effect.
Simplifications always help, and you did put things better than me.
The positive feedback, I think, is due to the positive temperature coefficient of tungsten, and its deviation from unity loop gain is, on the long term, due to hot spots.
I agree with you, again, both mechanisms should work quite well, but on the long term, paralleled heaters at constant voltage would be safer, especially if you are using a shunt regulator for +B.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.