Although there is a spread of anode-to-cathode voltage across the length of the filament, and there may be some current skew..... it does not have much influence on lifetime, if any.
Why not? Because the migration of barium from the subsurface coating to the surface is controlled by filament temperature, not by current or voltage. And it is the speed of this migration that determines the wearout rate of DHTs. Of cource, current x voltage raises temperature - but by how much? The heating current of a 300B is 1.2-1.4A, which burns ~6.5W in the whole length of the filament; meanwhile the offset overcurrent due to biasing skew might be 10 or 20mA at most, and runs through only a fraction of the filament's resistance. If we are generous, this might add 50mW of power.
Non-negigible temperature differences would make one end of the filament (or sections) brighter, and having observed EML 300Bs in a darkened room, and seen slight variations in brightness with AC or DC.... but these variations are not at one end of the filament, but rather in various small spots, unrelated to voltage spread.
Without measurements showing substantial temperature gradient, matching the voltage spread, we can forget about the idea of AC heating giving longer lifetime. Ordinary Line AC heat gives shorter lifetime, because the voltage, and so the temperature varies with the mains line voltage (as much as 10% in either direction - both of which will degrade lifetime).
Kevin Kennedy has reported here on diyAudio to getting 10000 hours from JJ 300Bs (on DC, using my regulators), which speaks for itself.
As for the difference in performance between AC and DC heating, it is not so well illustrated in the tube curves, as with the output spectrum of a real amplifier.
To keep the testing fair, these plots were taken on the same day, with the same amplifier, same EML 300Bs. Only difference is the the one is with AC heating and the usual 2x 33Ω resistors from each end of the filament, and the other is DC heated with my RC-V9 regulator. The scale is in dBV, so you can see that the noise floor is quite low enough without the resistors.
In the case of AC heating, the intermodulation distortion is shown by the spurs on each side of the 1kHz signal, and its harmonics at 2, 3, 4 ...(etc) kHz.
Intermodulation with 50/60 and 100/120Hz is of course completely alien to the original music signal, and does not fit in with my idea of 'high fidelity'.
The noise floor of the amplfier here is very low: the anode supply uses large capacitors followed by a capacitor multiplier.
With the AC heat sample, the 50Hz fundamental can be reduced with careful nulling; but the IMD remains in all cases.
To keep the testing fair, these plots were taken on the same day, with the same amplifier, same EML 300Bs. Only difference is the the one is with AC heating and the usual 2x 33Ω resistors from each end of the filament, and the other is DC heated with my RC-V9 regulator. The scale is in dBV, so you can see that the noise floor is quite low enough without the resistors.
In the case of AC heating, the intermodulation distortion is shown by the spurs on each side of the 1kHz signal, and its harmonics at 2, 3, 4 ...(etc) kHz.
Intermodulation with 50/60 and 100/120Hz is of course completely alien to the original music signal, and does not fit in with my idea of 'high fidelity'.
The noise floor of the amplfier here is very low: the anode supply uses large capacitors followed by a capacitor multiplier.
With the AC heat sample, the 50Hz fundamental can be reduced with careful nulling; but the IMD remains in all cases.
Last edited:
Probably the problem is that in a DC heated, filamentary cathode the whole cathode current flows to the negative terminal, regardless of which terminal is grounded. This means that if not the negative terminal is grounded, the cathode current will flow through the heater supply.
This can be avoided by AC, ultrasonic or RF heating and grounding the cathode via a pair of chokes.
@Windcrest77: Please don't forget the PTC characteristics of heater filaments. The inrush current will be about seven times as high as the nominal current. I bet that there aren't too many AF power amplifiers that can cope with this.
Best regards!
I have used a crude, 30kHz converter in a test supply:
https://www.diyaudio.com/community/...for-testing-tube-projects.327831/post-5555313
I only used iit for indirectly heated tubes though; with directly heated tubes, the result would probably be extremely messy
https://www.diyaudio.com/community/...for-testing-tube-projects.327831/post-5555313
I only used iit for indirectly heated tubes though; with directly heated tubes, the result would probably be extremely messy
PP may cancel some of the 50(60)Hz IMD, but the + and - 100 (120)Hz spurs remain.
This happens because the AC heating waveform is imposed on g-k as well as a-k; so second harmonic of the mains waveform are generated in the DHT, and do not cancel.
Connecting PP sides & L and R channels all together corrupts the signal (anode current swing causes a differential voltage across the filament).
If you want to take this shortcut, best to use a KT88 (indirectly heated), since the result will not be 300B sound.
This happens because the AC heating waveform is imposed on g-k as well as a-k; so second harmonic of the mains waveform are generated in the DHT, and do not cancel.
Connecting PP sides & L and R channels all together corrupts the signal (anode current swing causes a differential voltage across the filament).
If you want to take this shortcut, best to use a KT88 (indirectly heated), since the result will not be 300B sound.
If you put a transformer between the sine wave leaving the Class D amplifier and filament you have the possibility of blocking common-mode interference from the mains. More to the point, if your transformer operates at 50kHz (say) it doesn't need much primary inductance and can be quite small, and that means it can be made to have low primary to secondary capacitance, reducing common-mode interference. There are some really tricky things that can be done by linking a pair of ferrite toroids with a common winding but with intervening electrostatic screen, and this was done by the UK's National Physical Laboratory to power very sensitive measurement kit.
A good DHT stage will have the cathode return point directly at ground, or at AC ground via the cathode bypass capacitor.
In this case, the common mode mains noise becomes single ended, and easily eliminated. See the spectrum I posted above, for the DC case, where the 50Hz spur is below the (low) noise floor.
A cheap EI PT has a leakage current of about 6 - 10uA at 50Hz 240V; lower for 120V mains. It's not a real problem unless you want to use high values of Rk without a bypass capacitor, but there is no good reason to do that.
No, that would not work. The filament heats the cathode and causes the electrons to become capable of migrating to the anode as current flows. Second, there are a lot of technical articles about the effects of the cathode not being up to the correct temperature to activate the emmissive material that is deposited on the cathode during manufacture. If you attempted to pass current through a conventional Vacuum tube, it will actually strip the emmissive material off the cathode and you kill the tube.
Actually, you are better off going into the MC( MHz) region. At 100 KC, you are still too close to frequencies of interest. even though the filament is "Just the heater", you start putting frequencies in the bottle that are close enough, you will get a mixing effect in the tube and it could get ugly.it would be actually easier to make a network at 3 or 7 MC that at 100 KC. The example I used above shows the problem. Most RF generators ( and 100KC is RF) have a nominal of typically around 50 Ohms. The load you are going to have to match to to transfer 4.2 Amps is 1.7 ohms. To make that kind of impedance transformation you filament supply is going to be the size of a small refrigerator!! Just a thought.
Jamie
You would be better to get an older Radio Amateur's Handbook and build a transmitter from it that is in the MHz region. Just in the example I used above, the load mismatch for a small SE El34 amplifier is going to be 29 to 1. I have seen it and you are not going to be able to do it at 100 KC.
The idea of an Ampere-level 10MHz generator, connected by long wires to an unshielded bottle is crazy. Radiated emissions. We would fail EMC regulations like EN55022(CISPR22) by a huge margin.
If any of it gets in the DAC, or anything with older opamps, the sound will take another step down.
Sure it's fun to imagine solutions for problems, but here we don't even have a problem to solve.
If any of it gets in the DAC, or anything with older opamps, the sound will take another step down.
Sure it's fun to imagine solutions for problems, but here we don't even have a problem to solve.
Last edited:
The coupling capacitances in tubes aren't exactly small so you'll get significant leakage of that MHz heater voltage. So now the heater supply will need to drive those capacitances as well as the heater. At some point I hope someone will ask, "why are we doing this exactly?"
As Rod points out above, DC on the heater does a wonderful job of eliminating mains hum and its associated intermodulation products.
I bet an SMPS or Class D amp will have higher mains leakage due to the EMI filter(s).
Heating up the glass isn't likely to cause a lot of heating inside the tube. Vacuum is a pretty good insulator. Inductive heating could possibly work but you'll be heating up much more than the cathode. That's probably not what you want.
Tom
Well, lets try this again. It is not an imagining of a solution. This was an actual published work by John Atwood and it worked really good. As far as emission, you obviously don't understand RF very well. When you match the generator to the load , it terminates in the load. The load being the filaments.Now as far as the 10 MC signal getting into your DAC, You would think that the DAC in my rather high end player should have been dead 20 years ago. After all, I have HAM Station not more than 10 feet from my listening area. One of my station linear's runs at just around 850 Watts RMS out. the other runs at 3KW at full tilt. I don't ever run it that high but, I have run it at just around 1.5KW. Still no issues. This is not a "Flight of Fantasy" but rather a completed and published work by John Atwood. A rather brilliant man and the founder of One Electron. Just thought I would let you know. Funny though, snide opinions just seem to propagate. Lastly, if you have older opamps in you equipment, why are you concerned about the sound. Just the slew rates of the older opamps degrades the sound so much I would hardly think you would notice!
Tom,
Again in John Atwood's amplifier, there was no significant leakage. As a matter of fact there was virtually no leakage at all. If you think about it,terminating the 10 MC signal to the filaments correctly, all the signal is terminated at the load( the filaments) and if you think that a tiny portion of filament much less than a 1/4 of an inch when a wavelength at 10 MC is right about 93 feet. That is less of a possibility of it getting into the DAC or opamps. The radiation resistance alone would have to be something on the order of 1 time 10 to the -12 or close. RF propagation 101. The lower the radiation resistance the less likely of signal propagation. The answer to your question. "Why are we doing this exactly"? It is like an artist looking for a new shade of green. To have the possibility to enhance the art another step. it's easy to be complacent and very hard to step out and do something new and different. I guess if I had the why attitude, I wouldn't have 8 home brew amplifiers here within my reach.
Jamie
Again in John Atwood's amplifier, there was no significant leakage. As a matter of fact there was virtually no leakage at all. If you think about it,terminating the 10 MC signal to the filaments correctly, all the signal is terminated at the load( the filaments) and if you think that a tiny portion of filament much less than a 1/4 of an inch when a wavelength at 10 MC is right about 93 feet. That is less of a possibility of it getting into the DAC or opamps. The radiation resistance alone would have to be something on the order of 1 time 10 to the -12 or close. RF propagation 101. The lower the radiation resistance the less likely of signal propagation. The answer to your question. "Why are we doing this exactly"? It is like an artist looking for a new shade of green. To have the possibility to enhance the art another step. it's easy to be complacent and very hard to step out and do something new and different. I guess if I had the why attitude, I wouldn't have 8 home brew amplifiers here within my reach.
Jamie
Nobody has demonstrated any real problems that can't be addressed with a properly implemented DC solution.
I have debunked the one about wear out - which makes no sense, and is not supported by any measurement.
Saying that someone published something does not prove the need for such a solution, or that is has merit.
Rod,
I would really like to know how you debunked the DC wear out statement. Especially since RCA and Sylvania made exactly
those statements about Metal migration in filaments running on DC. That app note from RCA I do have in my library. What is the issue about need of such a solution or merit or anything else for that matter. I am really glad that I have not become jaded about looking at, discussing it and even considering the possibilities of different concepts. More power to ya but, that is why I have all the amplifiers I have built, around 20,000 Vacuum tubes, and close to a half ton of old iron here. Not to mention all the other components I have so I can grab the things I need to build and keep building until I don't. You see, I was under a different impression of what these forums were all about. Not to just keep doing it the old same way but, rather, to look at other ideas. Especially if they were successful. Tell me what does it harm you for these types of discussions to go on here? If you think it is for you to say because you don't think it's right or there is a need or even that it holds merit. SO WHAT??? There are obviously some that do want to look into it and be able to discuss it. There ole buddy, is your MERIT.
Good Day,
Jamie