I remember when I built Melquiades I detected that quietly of wire that feeds the grid of output stage with bias, quality of soldering or any other minute detail on this  wire was very auditable. I also detected that the proximity of this wire to cassias was auditable. Understanding that anything that feeds grid is crucially important (try to play with shunt cap after the gas tube and you will kill the Milq sound) I ask myself: how about the “quality” of the gas tubes?

In Melquiades the gas tubes live after 9000uF capacitor (I do not use C12 as the PS sits very closed from output stage) and they are basically driven by DC. The gas tubes are wonderful “dumper” themselves and whatever own MF noise then generate blocked but the following stunt cap. Still, how do they “sound”, if then have any sound? The negative supply gas tube feeds the both grids and I presumed that should have some effect. The positive supply is kind of irrelevant and I did run the connected Milq without the positive supply, letting the amp to be grounded on the preamp output. I did not detect any sonic differences with or without the positive supply.

So, took some gas tubes that I had in my home now and decided to listening with all of them, trying to detect if it were any difference. As I usually do I kept a small log of my experiments and since, there WAS a sonic different in Milq with different gas tubes I decided to publish the results. A few things should be taken for consideration:

1) In Milq the gas tubes do not used like regulators but rather as “garbage absorbers” and I have no idea if in a different application the tubes might behave similarly

2)  All tubes were new and “cold”. There were identically not pre-burned and before auditioning were running for 5min in hot amplifiers (the got warmed VERY fast when you drive 25mA thought them)

3) There is an army of the 150V 7-pins gas tubes: 0A2, SG1P, STV150/30, StR150/30, 150C1, G150m 150C2, 150C4, CV10664, CV1832, CV2903, CV8161, HD51, M8223, QS1207, GD150/4K, QS1210, SR56. I did not try the larger octal 150V tubes (I very much would like to and I will)

4)  The differences that I detect were VERY minor, much less than a typical flipping of a signal tube from a “good brand” to “bad brand”.

5) Some related threads on the site:

http://www.goodsoundclub.com/TreeItem.aspx?postID=1818#1818

http://www.goodsoundclub.com/TreeItem.aspx?PostID=1339

So, I took my box of Cohiba and placed Verdi’s Rigoletto of 1956 into the game (Callas, Gobbi, Di Stefano by La Scala under Tullio Serafin. The result was following: (from left to right on the picture)

1) Holland Made Amperex with flat shield - All together correct sound, slightly more round then I would like to. The similar sound was by US’s General Electric (not on the picture). However, General Electric subjectively sounded louder but not more dynamic, juts louder.

2) German made National Servisemaster - in a way similar to above by has slightly more “yellow” tone.

3) USA’s Sylvania – very-very nice tube. Perfectly balanced and with no surprised. Dynamic vs. tone, sharpness vs. elegance… everything is perfect.

4) USA’ Tung-Sol –very interesting tube. It has some softness but some interesting attitude in it. It going down very elegantly but then suddenly jump on you but also in a very elegantly way. I do not get and I do not know if I like it but the effect is very interesting.

5) Russian SG1P – horrendous tube, absolutely the worst sound. It has tremendous slam but at the same time it completely washed out any tone. It almost sound like overly contrasty b/w picture of the a colorful landscape

6) USA’s Raytheon – a nice well balanced tube. For whatever reasons sound got “further” with this tube. Very strange affect indeed.

7) USA marked Nations – this tube is a contemporary bitch. The tube might be identified by the glossy sharp-ended pins. (on the second image 3 tubes in the middle) This is a contemporary production of Russia and the mark them as anything you can imagine (including the “original” boxes) I have seen them as Ampere, Raytheon, CVS, FEN, and God knows what. Pretty much everyone push this tube today as anything. Ironically the tube is not really bad. Sonically it is “almost” OK but it has some sharpness and aggressive that I do not like. This tube is heads and shoulders better then the Russian-made #5. Still, I feel that very-very slightly it loosing tonally to the Western tubes.

8) German made Telefunken – superbly tonally but overly soft. Also it eats some dynamic. The good parts about this tube (and completely irrelevant from my point of view) that it locks voltage closest to 150V then any other tubes – 149.98V

9) USA’s RCA – good tube but a bit too dull and not contrasty. At the same time it is not soft and has some “punch”, with a contradiction, isn’t it?

10) Another USA’s Raytheon Registered. Very similar to the #6. The effect of “further” does not manifest itself. Perhaps it is because it is a different batch?

11) Holland Made Amperex with screen shield – it is on the second image on the left – the best 0A2 that I have. Sonically it is identical to Sylvania but it has more dynamic. Tonally it is also superbly reach. A truly phenomenal performer!

The value/type of the capacitor after the gas tube hugely affects sound. I made quite a number of experiments in past with those caps and I have no idea why it is so. Well, I hardly understand (and I did not see anyone who does) why the “gassed resistor” biasing generality took the sound 6E5P way out the… So why should I understand WHY the value/type of small capacitor after the gas tube affects sound. However, it does affect it.

 If you look at the Melq’s circuit:

http://www.goodsoundclub.com/PDF/Melquiades_SET.pdf

then the cap C4 and C5 are very tricky sons of the bitches. Theoretically those caps should be irrelevant as they decupled form the tube grid by a large 390K resistor. Well, not so simple…

The circuit is very simple: a current regulating resistor, a gas tube with the following R10 and the cap across them. One of the cap’s legs sits with at the gas tube’s leg (it is critical). A gas tube creates own MF noise and the cap kills the noses. This is simple and very ordinary…

Now, I ran the amps with many different gas tube including the few versions of low current 5651 and I ran them… without C5 cap. No capacitor no problem other say – well, not so simple. Ironically when I did not use the C5 cap I did not have intolerable noise. I have some very minor MF noise that did not bother me at all. However, I did not have good sound either. The sound was vicious, erratic and reminded me a drunk driver driving. Applying, a small cap right across the tube itself (as it suggested in classic books) queried some noise but did not affect sound in necessary direction. However, placing the shunt cap after the R10 resistor did not necessary magic…

The value and type are other mysteries. 1uF, 2.2uF, 5uF did not sound properly. The 10uF for whatever reasons did. It was not that something got better or worse with raise of capacitance. The nose was killed at 2.2uF and further increase of capacitance did not affect anything… unit I reached 10uF. At 10uF is all come together making the 6E5P to sing in this application. I did true a good quality”expensive” plastic caps but it looks as “it” wants only electrolytic….

As I told before when I discovered this way to operate the 6E5P in Melquiades then was afraid to “loose sound” and did not go for anything further. Still, if a higher voltage gas regulator exists then it will be possible to use higher nominal of resistors to grid... It would always help, and particularly in the super Melq…. Sometimes in future, I plan to try two gas tubes at 300V. I do not know if pairing of the gas tubes will affects sound.  I also wonder if the increase the voltage will lead to the increase of the “best for sound” capacitor values after the gas tube…

Rgs,
Romy the Cat

Who know how they die, at least I was not able to figure it out. The literature suggest that when gas regulator at the end of their life they stop regulate or maintain regulation but increase noise. I was not able to detect it.

I have many gas regulators and measured their mutual conductance. I take the measured 100% gas tubes and the very weak tubes (the same brand) and put them in Milq – they sound identical. I went further and was trying to intentionally kill one 0A2. I drove it for 3 hours with 150mA (instead of 35MA max). The tube went hot like hell but do you think it killed it? Very much not. It still regulated 150V and it after it cooled down it sounded perfectly fine.

Since I use gas regulation in my Melquiades to damp bias and bias compensator supplies I really would like to learn how the gas tubes die and what would be a signifying fact that a gas tune at the end or it life…

Hi Romy,

Still playing with Gas zeners, isn't it ?

I once seen, tested and scrapped "dead" ones....
Clue#1: they fire and (try to) regulate at higher than nominal voltages at normal current and are erratic.
Clue#2: they have sections (at top and bottom, usually, where is the gap between the outer shell and the mica spacer) of the internal glass internally "plated" by the anode metal ions that have drifted reverse to the electron flow inside. BTW, the center rod is the Anode (positive) the external "plate" is really the cathode (negative): the reverse of a "normal" heated-cathode tube. That "ionic" plating looks like the top getter, but is more aluminum like in colour.

Such condition, I guess, cannot be acquired otherwise than by operating few THOUSANDS hours, maybe in excess of 10,000 ? I don't know... I never had enough patience to test that.

The ones I tested a long time ago (military surplus, again) that behaved like that are now lying in some burial site, I guess.

May they Rest In Pieces.....

Best regards,

N-set wrote:

Looks like one more argument for the thesis that caps ssuck! Sorry for a naphtalene-smelling idea, but how about VR tubes *here*? It's an light-weight MF channel. The current variation VR's will see should be very small so perhaps charge redistribution time-lags become unimportant? Would/could you share your friend's views on the last cap operation (separate thread perhaps)? Thanks!

The caT

Thanks, jk

I have given a lot of thoughts to this topic lately and have a number of conversations with various people about the subject. A friend of mine from Europe, who is currently at fitness protection program, suggested that:

“.. in my "real" systems (that is not commercial) I have been running "electrolytic free" for the last nearly a decade and if I had to use them [commercial gear] I have always minimized their influence on the signal circuits as much as possible...”

He offers quite a few techniques how small film caps still might be used and does not screw sound. There were some other proposals and suggestions from different sources, which made me to think: why I ever need capacious? The single-stage Melquiades is as close to cap-free solution as it could be so why do not take it more racial – making PS capacitance free?

I spoke with Dima and asked him to suggest me solutions how it might be accomplished.  The PS caps do two duties: short signal to ground and kill ripples.  After a long conversation he came up with numerous options including those the Steve Bench proposes. The solution one is to use choke, then 600R series resistor (or a choke with 660R DCR) then a shunt of two back-to-back VR 105 tubes.  The 0B2 has very low impedance and will do a perfect path to ground for signal and will kill quite a lot of ripples. My idea was to run the Fundamental and HF channels from my current B+ and to use “no-caps supply” only for MF channel what 30mA the gas tubes will be able marginally to handle. The second that Dima proposed was to run 6C19P as default shunt regulator.

http://www.goodsoundclub.com/pdf/6C19p.pdf

.. having it’s cathode on ground, anode driving the output transformer and having approximately -33V on it’s grid. Of course we got carried away considering application a feedback from amp to drive the regulator’s grid. There were some further solutions that would keep private for now.

I most likely will try it all on my MF but I have an “ethical” problem with it. My usual rule is to make any changes ONLY if I specifically identified a specific sonic problem with the sound that I already have:

http://www.goodsoundclub.com/Forums/ShowPost.aspx?PostID=432

To be honest, I was playing the amp today with new dedicated power supplies on the single-stage channels and it was remarkably good, with me complaining about nothing. So, I might have some intellectual objectives to get rid of caps but I have no natural sonic temptation to pursue this direction. I know myself and I know that it is not the environment where my mind operates productively. So, I do not know where I will be staying on it.  The gas tube might be good, the shunt regulator might be good… but if the current cap sounds “properly” then what would be a definition of “better” for me? I am not a Moron from DiyAudio web site. Perhaps instead of experimenting with no-cap shunt I need to listen more my current MF channel and wait until I will be able to identify it’s weakness… and THEN to see if the no-cap solution would address those specific sonic faults.

Romy the Cat wrote:

3) There is an army of the 150V 7-pins gas tubes: 0A2, SG1P, STV150/30, StR150/30, 150C1, G150m 150C2, 150C4, CV10664, CV1832, CV2903, CV8161, HD51, M8223, QS1207, GD150/4K, QS1210, SR56. I did not try the larger octal 150V tubes (I very much would like to and I will)

Romy the Cat wrote:

Did anyone ever see the ready to do tube sockets adapaters that would let to pit the large gas tubes ingot the small sockets? Melquiades use 7-pin small 0A2 tubes but I would like to try the 0D3. They are the same 150V tubes but I got some VERY old 0D3 and I do not mind to try them on negative supply. The negative supply is bias line on driver tube and it is VERY critical as it sits right in grid. I would like to see what it would do with sound. The different 0A2 do affect sound as I described above, what not to try the larger or older gas tube?