dude, you will have way better quality filament supply, than most people anode supply have.
Yeah. Common sense tells us that it is just filament supply. As long as it is 12V or thereabout then it should be fine. So logical that it took me a very long time to realize my mistake.
I understand what your claim now...
As I have another 6.3Vac tapping on my transformer. So I decided to connect the 9AVc and 6.3Vac together and end up with 15.3Vac. I was going to try 12.6Vdc filament connection.
This time I had about 20Vdc after the rectifier, and I used CRCRC to lower the voltage to 12.6Vdc. On simulation, I have very little ripple, with CRCRCR or even CRC.
CRCRCR = 6800uF / 3R6 / 6800uF / 2.7R / 4700uF /2.5R / 4700uF
CRC = 6800uF / 8.2R / 6800uF
I connected all filament of all 5 tubes in parallel. I first started using stranded 0.75mm cable for the connected and started with CRC first.
The sound is indeed "flat, mechanical, unmusical, uninvolving" as Salectric commented. The sound became very PA like.. very clean, fast, with a hint of metallic. No extension, no depth. My 125watts per channel tube power amp had no more juices suddenly.
CRCRCR - very lightly better nevertheless still not good enough.
Disappointing indeed.
Upon consulting a friend, I changed all wire to solid core 16awg. Well the results is better, but still not good enough. Less PA like, slightly more depth. Basically still having the same characteristics, but slightly better.
I was thinking maybe I can tried the DC with high ripple. As I am looking to improve the sound with 6.3Vdc unregulated for the filament supply.
I tried "RC" 3.3R / 470uF.
On simulation, I got a ripple of +/-3V.. Yes this is crazy..
Surprisingly, I liked the out come. More involving, better extension, Less PA like. But is it good enough? No. Most likely I will switch to back to 6.3Vdc
I forgot to mentioned with 12Vdc I sensed that there is always a emphasis of hi-mid (frequency) in my system, thus the hint of metallic.
I am not saying that 12Vdc is not good. This is system subjective. If you seek to improve the speed of your systems or seek a cleaner bass, this could be good.
That will take a while.. I will revert the circuit to 6.3Vdc.. I will post the scope capture when I am ready..
You should try to elevate it to +30..+50v. This increases the tube's life and prevents noise coupling to cathode.
Use a voltage divider from B+, the lower R must be decoupled with a capacitor.
With this mod, all the heating methods should sound the same...
My theory is that you like unregulated DC best because of the high order harmonic noise it introduces, which is perceived as "detail" and "live sound".
Use a voltage divider from B+, the lower R must be decoupled with a capacitor.
With this mod, all the heating methods should sound the same...
My theory is that you like unregulated DC best because of the high order harmonic noise it introduces, which is perceived as "detail" and "live sound".
Interesting statement, could you explain the theoretical details supporting:
1) Life extension due to increased potential difference
2) Reduce noise coupling by increased potential difference
Looking forward to your reply!
P.S. In my experience, details and life sound is negatively impacted by higher order harmonics, but that is another (subjective) discussion
Sure.
1) Research and experiments conducted in the 50's led to this conclusion. Please read Robert B. Tomer's "Getting The Most Out Of Vacuum Tubes", a very interesting book.
2) See "Valve Amplifiers" by Morgan Jones, third edition, page 362.
Attachments
In my experience, some kinds of intermodulation distortion can lead to "extra high end" and sparkle (which is, of course, entirely artificial). Or to harsh, unpleasant sound, if the amp had enough treble in the first place...
Last edited:
Old Tomer knew his tubes...
A certain A.G.F. Dingwall also published its research on heaters in "Electron Tube Design" by RCA. It can be found online here: Full text of "Electron Tube Design"
Let's think about what happens between the filament and the cathode, when the latter is at a higher potential.
This is especially obvious for topologies where a cathode sits high above zero: cathode follower, SRPP, totem pole, cascode, Aikido.
Hot bodies emit electrons. The filament is red hot, as is its aluminium oxide insulation. They are enclosed by the cathode sleeve. The cathode being electrically positive, electrons are attracted to the cathode: this is the LEAKAGE CURRENT.
This leakage current is modulated by the filament PSU noise: 50Hz for AC, or 100Hz + a lot of higher harmonics for unregulated DC (just put a scope on it, you'll see). This is why you can actually hear differences between heater PSUs, I think.
After some time, a chemical migration takes place between the cathode and the filament. The insulation becomes polluted by metallic ions, and the leakage current increases: the tube becomes noisy!
When the filament is elevated at a higher potentian than the cathode, all this doesn't happen.
A certain A.G.F. Dingwall also published its research on heaters in "Electron Tube Design" by RCA. It can be found online here: Full text of "Electron Tube Design"
Let's think about what happens between the filament and the cathode, when the latter is at a higher potential.
This is especially obvious for topologies where a cathode sits high above zero: cathode follower, SRPP, totem pole, cascode, Aikido.
Hot bodies emit electrons. The filament is red hot, as is its aluminium oxide insulation. They are enclosed by the cathode sleeve. The cathode being electrically positive, electrons are attracted to the cathode: this is the LEAKAGE CURRENT.
This leakage current is modulated by the filament PSU noise: 50Hz for AC, or 100Hz + a lot of higher harmonics for unregulated DC (just put a scope on it, you'll see). This is why you can actually hear differences between heater PSUs, I think.
After some time, a chemical migration takes place between the cathode and the filament. The insulation becomes polluted by metallic ions, and the leakage current increases: the tube becomes noisy!
When the filament is elevated at a higher potentian than the cathode, all this doesn't happen.
Measurements show some 12A.7 valves with same leakage current when cathode is of either polarity with respect to filament, some show more leakage with one polarity direction, and others show vice-versa. So per se, your proposed leakage mechanism doesn't appear to be dominant, and your final statement to be in error.
I've seen at least one technical document that identified whisker metalisation growth and fusing within the insulation - sometimes exhibiting itself as ad hoc 'pops' in the amplified signal.
I've seen at least one technical document that identified whisker metalisation growth and fusing within the insulation - sometimes exhibiting itself as ad hoc 'pops' in the amplified signal.
I was referring to measurements that show that the resistance between filament and cathode for one particular DC bias compared to the opposite polarity DC bias can be lower - the same - or higher.
Whereas your contention is that the resistance would be much higher for one polarity bias than if it were the other polarity. Which is I guess a follow-on from the inaccurate perception that the cathode-heater interface acts as a 'diode'.
Measurement results show that resistance typically increases as bias voltage increases away from zero (of either polarity), which is not a diode characteristic.
I never said that. Please read again my posts above.
I said that we can control the direction of the flow - the only thing that matters is to keep the leakage current out of the cathode. If it flows into the filament, it doesn't matter.
Of course, AC noise flows both ways... UNLESS we elevate the filament above the cathode. If the difference is more than AC noise peak level, it simply cannot flow into the cathode!
DC leakage doesn't hum, but there is the phenomenon of chemical migration I've mentioned before, which degrades the cathode-filament insulation (after a long time). Again, the direction matters, we can choose which way this happens.
edit: By "DC", I mean regulated, clean DC. Unregulated heater PSUs are usually very noisy, with a sawtooth ripple waveform! Sawtooth = 100Hz noise + all the higher harmonics. Switching mode regulators look pretty bad too on the scope.
Last edited:
Perhaps if you read through the only reference paper on the topic that I have found, and perhaps provide some measurements to support the gross generic asymmetry in polarity that you propose exists, then that may progress the discussion.
http://dalmura.com.au/projects/Heater cathode insulation performance.pdf
The only other measurements that I have seen were made by me, but I'm biased
http://dalmura.com.au/projects/Heater-cathode conduction plots.pdf
http://dalmura.com.au/projects/Heater cathode insulation performance.pdf
The only other measurements that I have seen were made by me, but I'm biased
http://dalmura.com.au/projects/Heater-cathode conduction plots.pdf
I finished reading the paper. It analyzes insulation material performance and the factors that influence leakage current.
I repeat: I never said there is no leakage current. Of course there is!
I said the direction of the flow is important in audio, and with an elevated heater, we can keep the current flowing in only one direction. We can keep the noise out of the cathode. Is this clear or not?
The paper talks about a phenomenon I mentioned before: metal ion migration, which occurs with LARGE DC bias, as in cathode followers, SRPP, Aikido, etc.
Metal ions are positive, they go to the negative electrode; in time, they form paths which increase leakage.
Again, the direction is important: if the cathode is negative, the metal ions from the heater will go into the nickel cathode sleeve - that is not a problem!
If the ion migration is into the aluminim oxide insulation, well, it won't insulate very good any more, will it!
This paper was written in 1936. Tome's book was published in 1960, at the end of the tube golden era, after many more years of research.
My conclusion is: a small positive bias on the heater (25-50v, no more) is beneficial for both noise and logevity.
I repeat: I never said there is no leakage current. Of course there is!
I said the direction of the flow is important in audio, and with an elevated heater, we can keep the current flowing in only one direction. We can keep the noise out of the cathode. Is this clear or not?
The paper talks about a phenomenon I mentioned before: metal ion migration, which occurs with LARGE DC bias, as in cathode followers, SRPP, Aikido, etc.
Metal ions are positive, they go to the negative electrode; in time, they form paths which increase leakage.
Again, the direction is important: if the cathode is negative, the metal ions from the heater will go into the nickel cathode sleeve - that is not a problem!
If the ion migration is into the aluminim oxide insulation, well, it won't insulate very good any more, will it!
This paper was written in 1936. Tome's book was published in 1960, at the end of the tube golden era, after many more years of research.
My conclusion is: a small positive bias on the heater (25-50v, no more) is beneficial for both noise and logevity.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.