and this just in
KORG INC and Noritake Co., Limited Release Innovative Vacuum Tube: the Nutube | News | KORG
Makes sense that it would be a manufacturer of vacuum fluorescent displays.
This "rich harmonic" stuff is a bit much. I like linearity, which is a natural outcome of constant mu.
KORG INC and Noritake Co., Limited Release Innovative Vacuum Tube: the Nutube | News | KORG
Makes sense that it would be a manufacturer of vacuum fluorescent displays.
This "rich harmonic" stuff is a bit much. I like linearity, which is a natural outcome of constant mu.
"...it will not need to be replaced regularly like a 12AX7 ."
🙄
I strongly suspect they are after the guitar effects box market.
🙄
I strongly suspect they are after the guitar effects box market.
Is there an easily defined geometry that encourages linearity in vacuum triodes? Or does it come as a secondary consequence of something else? Or something still else?
Much thanks for any comments, as always,
Chris
Last edited:
I have a really nice reference in storage, Gewartowski and Watson, Principles of Electron Tubes, from 1965 (late!) which probably talks about this. I know it discusses the state-of-the-art in gain-bandwidth and the effects and limitations of close grid-cathode spacing.
Probably the initial assumptions for models are an infinitely-extended grid and cathode and a uniformly spaced grid pitch. Nowadays I'm sure some Comsol etc. modeling software would be a powerful tool for design.
I have H. J. Reich of that title (from the old MIT post-war series) but not that one. I'll try to find a copy; thanks for the recommendation.
Most books give a model of edgeless parallel planes, and this is the derivation of the 3/2 power law. Other, anecdotal accounts wax poetic over cylindrical structures, large cathode structures, all kinds of often contradictory things. Can't be only accidental that some types, like 12AX7, or 6SN7, or 211, are truly linear if lightly loaded, but they're all different constructions.
Much thanks, as always,
Chris
According to Van der Ziel, one important source of 1/f noise is a borosilicate interface layer that may form at the cathode. Cathodes without silicon should not have this.
I haven't got any evidence, but I also expect that large cathodes have less 1/f noise than small ones, just like large MOSFETs have less (input-referred) 1/f noise than small ones.
Yes, it's intuitively the effect of paralleling individual elements, a square root of N dependence, unless something else goes wrong.
Yes, for linearity what counts is for the transconductance to vary inversely as the plate resistance varies, so the product mu is constant. The 3/2 exponent is ideal, and I see that it can get to 2 with closely-spaced grid-cathode, but I haven't yet found the associated plate resistance variation function.
The accounts seem to focus on improvement in transconductance (understandable) but not so much on distortion. It may be that in the early days, the cumbersome means of realizing high impedance current sources, other than brute-force high resistances and large voltages, and the assumption of application of negative feedback, made designers tend to reject active current sources out of hand. In fact I have another book, author's name starts with an A, who gives examples of big voltages and big R before saying Of course this is impractical. In Valley and Wallman (1948) the easier job of hollow-state sinking of current for a cathode follower is given as an example, using a ballasted-cathode pentode as the sink. Of course we're using such stages to drive lower impedances, which then proceeds to spoil things!
For the friend's desire of using 6C45s as MC stepup (pretty crazy but he hates solid state) I was getting quite low distortion using a Boxall compound current source as the plate load, for reasonable and even rather large input voltages (in the context of MC anyway!). Of course the only other load on the plate was the 100k and parallel C of the Ap.
You and jcx have both mentioned this recently, and of course it's good sense, but I can't figure out what to do with it. Maybe there's a clue tied in with the slope of the load line, or even the slope of the "cathode resistor" line. Or maybe not - I'm sure smarter folks than me have worried this into the ground. But I'm keen to try to learn something.
Very much thanks, as always,
Chris
The answers are out there. What would be nice, but I believe it to be unlikely, is if the effect of the approach to the square law of gm from the Langmuir-Child 3/2 power law, as the grid gets closer, were to be complementary with heavier loading of the plate! But I don't believe that occurs. I don't have time right now to look at this but I'm sure someone has, or maybe jcx simply knows the answer.
For the load line, in the limit of zero incremental conductance it's easy to see what happens. And no triode is perfect, but over a limited range some are awfully good. And some are not.
I'm not feedback-averse like a lot of folks in diyaudio. But to do it with d.c. coupling and tubes, or even tunes with some sand-state assistance, borders on the extravagant. It is kinda cool to get high performance out of the little bottles. I do wonder how those Noritake/Korg things look on the curve tracer.
Afterthought: a stretch, but maybe some active load in the plate could render the voltage gain constant for the supertriodes.
That’s my ambition too, and one of several reasons I started the noise measurements. But to get low enough noise for a MC cartridge, I figure I need at least 4 6C45s in parallel. A MC preamp needs to get down to the noise level of the 2SK209 JFET in my list, or preferably lower (the 2SK209 is a bit noisier than a Jensen JT-44K-DX step-up transformer).
But if we’re wiring a bunch of tubes in parallel, I wonder if the 6C45 is technically the right choice. Wiring the same number of 12AX7 (or 6H9C) envelopes in parallel would be about as good, and avoid some of the technical problems of the 6C45 (matching, microphonics). I like the 6C45 better (trendy and more exotic), but….
Did you mention that one of your friends described the 6C45 as sounding metallic? Did he already try it in this application?
Scott
That metallic remark was from the guys at Audio Note UK, who are about as excited about sand state as my friend. BTW one of the tubes they like is the 6463, which was designed for computers! But I have no idea about its noise levels or microphonics. I have a few but no time to play right now, due to a remarkable and welcome return of paying work 😀
As far as I know the AN folk did not try to do an MC stepup with anything other than transformers.
I did use the 45s as cathode followers with sand state current sink loading. In that very much less demanding application there were no reports of problems with microphonics, and they were astonishingly quiet. This was for the friend's active crossover, which uses a first-order highpass and a lot of bass boost. I wish I could say that the low-frequency channel was quiet, but the 6SN7s with that much LF gain were terrible. One lives and learns.
Hi,
Yeah..........Given up on using 6C45s for that kind of use myself. Too much hassle.
If only there were to be twin triodes with a bit les transconductance (for good behavior) and a mu a tad higher than say a 6DJ8....
Say a twin EC86 or thereabouts.
Anyway, keep up the good work. It's really interesting stuff. 🙂
Cheers, 😉
Yeah..........Given up on using 6C45s for that kind of use myself. Too much hassle.
If only there were to be twin triodes with a bit les transconductance (for good behavior) and a mu a tad higher than say a 6DJ8....
Say a twin EC86 or thereabouts.
Anyway, keep up the good work. It's really interesting stuff. 🙂
Cheers, 😉
Just read through the thread, and have a few comments to make:
To get constant Mu, the tube construction likely has to place the grid geometry on an equipotential surface between the cathode and plate. (giving all sub-portions of the tube equal Mu factors) Modern E&M software should be able to perform wonders for such tube design. One confounding issue however would be space charge variation changing the equipotential surfaces, if it is more concentrated in some areas. So a uniformly spaced (flat cathode, flat grid, flat plate) likely has the best chances for linearity over a wide current range.
Then there is the issue of grid wire spacing versus grid to cathode spacing. We have graphene for grids now-a-days besides frame grids. (and we have room temperature, uniform, thin film thermionic emitters too)
And finally, there are end effects at the extremes of the grid and also from support posts. End effects can be overcome by extending the grid well beyond the active cathode area, or by placing end sheet metal shields around the grid.
The computer oriented tubes (like 6463) were designed to avoid an interface layer on the cathode, so they
-might- avoid the 1/f flicker noise. Lots of computer oriented tube types around.
If larger cathode areas are quieter, there are some other high gm frame grid tubes to consider like: E55L/8233, 12HL7, 12GN7, and 5842 maybe among others (6HB6 a non frame grid, high gm tube). The 12HL7 has a frame grid, but seems to space it further from the cathode than the others. The 12HL7 (a high gm pentode) and 1E7G (a low gm dual DHP) have phenomenal linearity in triode mode (don't know about their noise factors yet though):
To get constant Mu, the tube construction likely has to place the grid geometry on an equipotential surface between the cathode and plate. (giving all sub-portions of the tube equal Mu factors) Modern E&M software should be able to perform wonders for such tube design. One confounding issue however would be space charge variation changing the equipotential surfaces, if it is more concentrated in some areas. So a uniformly spaced (flat cathode, flat grid, flat plate) likely has the best chances for linearity over a wide current range.
Then there is the issue of grid wire spacing versus grid to cathode spacing. We have graphene for grids now-a-days besides frame grids. (and we have room temperature, uniform, thin film thermionic emitters too)
And finally, there are end effects at the extremes of the grid and also from support posts. End effects can be overcome by extending the grid well beyond the active cathode area, or by placing end sheet metal shields around the grid.
The computer oriented tubes (like 6463) were designed to avoid an interface layer on the cathode, so they
-might- avoid the 1/f flicker noise. Lots of computer oriented tube types around.
If larger cathode areas are quieter, there are some other high gm frame grid tubes to consider like: E55L/8233, 12HL7, 12GN7, and 5842 maybe among others (6HB6 a non frame grid, high gm tube). The 12HL7 has a frame grid, but seems to space it further from the cathode than the others. The 12HL7 (a high gm pentode) and 1E7G (a low gm dual DHP) have phenomenal linearity in triode mode (don't know about their noise factors yet though):
Attachments
Last edited:
Good contribution, thanks. I should muster to look at the spectrum of 6463 noise, but I am very preoccupied right now. Qvortrup said the 6463s that my friend obtained were the worst, but were still pretty good 😀
They will have to do for the moment.
The next question: do frame grids entail high-Q microphonics?
I would guess that frame grids exhibit significant microphonics at near mono-chromatic frequencies, from all those tiny elongated grid wires. There probably is some variation in wire tension along the support rods though, which might make for a band of frequencies, as mentioned earlier. The E55L has rather thick support rods, if that makes any difference.
OH, and 6JC6 is another (small, 3 Watt) frame grid tube (pentode) available cheaply.
OH, and 6JC6 is another (small, 3 Watt) frame grid tube (pentode) available cheaply.
Last edited:
I'd suggest the grid wire sections are the only structure that could generate the 12-17kHz observed microphonics by Scott. There are many many short sections of wire - all subtly different than each other so as to be observed as a continuum over a range of frequencies - and of small enough mass for the frequency band observed - and of a metal and attachment structure that would exhibit such low damping.