It's like a lot of things: it's not the use, but the mis-use, that causes detractors. Like any triode, the 12AX7 operates with more linearity the larger the plate load becomes relative to the r(p). Since r(p)= 90K (nominal) it takes quite a large plate load to do that, and there are a helluvalot of 12AX7 designs out there with plate loads that are too small.
You either need a large margin of DC rail or an active plate load to really realize the linearity of the type. Similar considerations apply to the 6SL7, though to a lesser extent (r(p)= 70K for the 6SL7).
So far, I haven't had to use 12AX7s, but I wouldn't disregard the possibility if I needed that much gain. Same for the 6SF5 or 6AV6 (High gain singletons).
yes, i got the 6AG9 i got for a dollar each...thanks for the tip, don......
"12AX7"
I kinda wonder if high Mu automatically gives some advantage toward linearity, at least with a CCS load. For a given voltage swing on the output, the plate's field effects at the cathode are attenuated by 1/Mu. So it effectively operates over a smaller part of the characteristic for the same signal output.
The D3a also has a high Mu, around 80 I think. For the extreme gm of it's g1 grid, it must have a very fine mesh that shields the cathode well from "plate" (or screen) voltage variation. Provided the high gm does not come from excessive proximity effect (to the cathode), it probably gets this linearity advantage versus output V when operating with a CCS load.
For linearity with a real R load though, with varying current, I would expect a lower gm frame grid (but still fine mesh), using reduced proximity effect, would give better results. This would show up as a flatter Mu versus Ip curve.
-------------
re: Tony
I think the other video compactrons are also on the $1 sale, as well as a bunch of the ones on the 9DX and 9PM bases.
I kinda wonder if high Mu automatically gives some advantage toward linearity, at least with a CCS load. For a given voltage swing on the output, the plate's field effects at the cathode are attenuated by 1/Mu. So it effectively operates over a smaller part of the characteristic for the same signal output.
The D3a also has a high Mu, around 80 I think. For the extreme gm of it's g1 grid, it must have a very fine mesh that shields the cathode well from "plate" (or screen) voltage variation. Provided the high gm does not come from excessive proximity effect (to the cathode), it probably gets this linearity advantage versus output V when operating with a CCS load.
For linearity with a real R load though, with varying current, I would expect a lower gm frame grid (but still fine mesh), using reduced proximity effect, would give better results. This would show up as a flatter Mu versus Ip curve.
-------------
re: Tony
I think the other video compactrons are also on the $1 sale, as well as a bunch of the ones on the 9DX and 9PM bases.
Last edited:
Interestingly, the 12BZ7, which specs like paralleled 12AX7 (u = 100, rp = 31k) and has much longer plates, is much less linear. Lots of high order harmonics. Nasty tube.
The long plate- linearity correlation hasn't worked well for the data I've taken, though admittedly I run the tubes as linearly as I can (CCS load, LED bias).
The long plate- linearity correlation hasn't worked well for the data I've taken, though admittedly I run the tubes as linearly as I can (CCS load, LED bias).
I used 8AL9 as a driver to single ended MOSFET amp. Nice lamp!
Also, guess the tube: gm=14 mA/V Mu=135
An externally hosted image should be here but it was not working when we last tested it.
Also, guess the tube: gm=14 mA/V Mu=135
Last edited:
I have been testing them in mu followers which provides a reasonably high plate load (in the region of 10*rp) without KV PSUs!
Here are some comparative figures from tests I have made. In all cases the distortion was measured at 20V rms output into a 100K load and at 2KHz. In all cases, distortion is mainly 2nd harmonic with the third and lower harmonics at least 20dB lower.
Code:
6CG7 (tall plate) 0.3%
6CG7(short plate) 0.4%
12AX7 (tall plate) 0.1%
12AX7 (standard plate) 0.5%
12AU7 0.8%The regular 12AX7 is not outstanding but the tall plate version is in a class of its own.
Cheers
Ian
A mu follower might not be a totally fair test, as the negative feedback will given an advantage to high gain valves.
When running open-loop, high mu is more likely to be a disadvantage unless the comparison is done at constant output voltage rather than constant input voltage. This shows that you have to try to compare like with like.
Remember that the 12AU7 is basically a pair of 6C4 - a VHF power triode. Never intended for audio or TV, but low power transmitters. I guess when there was a demand for a twin triode for TV timebase, it was cheaper to just use an existing design and put two of them in the same envelope.
When running open-loop, high mu is more likely to be a disadvantage unless the comparison is done at constant output voltage rather than constant input voltage. This shows that you have to try to compare like with like.
Remember that the 12AU7 is basically a pair of 6C4 - a VHF power triode. Never intended for audio or TV, but low power transmitters. I guess when there was a demand for a twin triode for TV timebase, it was cheaper to just use an existing design and put two of them in the same envelope.
When you don't make a tube with an rp of 60-70k drive a 100k load, the linearity (not surprisingly) is 10-20 times better than that. Interestingly, in a test where the tubes weren't asked to drive a heavy- for them- load, the no-name ultra short plate 12AX7 (innards look like 12AT7, Eastern European source) tested as well or better than the much-vaunted smooth plate Telefunkens.
I'm surprised that the 6CG7 doesn't do better than that.
I'm surprised that the 6CG7 doesn't do better than that.
There is no NFB in a mu follower except for that around the top CF.
A mu follower gives a stage gain close to mu - hence their name - it is open loop.
Note: I quite clearly stated that all the result were at a 20V rms output voltage so I was comparing like with like.
Cheers
Ian
That's the point of using a mu follower.
Can you provide figures and describe your test circuit?
Morgan Jones in his tests with a CCS load got figures only 3dB better than this for the intrinsic distortion in a 6SN7. Most tubes do worse.
The good thing about a 6SN7 is that it does as well driving a 10K load but the 12AX7 obviously does not.
Cheers
Ian
http://www.diyaudio.com/forums/tubes-valves/164920-guess-tube.html#post2150434
Also, see post 52 in that thread.
OK, that ties up well. Since distortion is directly proportional to output level your 8.5V rms is -7.5dB relative to my 20V rms. I got 2H at -60 which means at your level I would get -67.5dB cf your -70db. And as I said before, the mu follower seems to be no more than 3dB worse than a CCS.
So, the key question is exactly which 12AX7 was it you tested?
Cheers
Ian
Yes, you got -65dB 2H cf my -60dB so that's a bigger 5dB difference - more than I would have expected.
Which does not tally with my tests where the small (less tall) plates gave distinctly worse results. That said I only tested two tubes of each type all current production. I also tested a couple of EH 12AX7 WA and one gave the amazing result of -68dB 2H at 20V rms - but this was at the expense of higher 3H. I am not quite sure what I was seeing there but I have noticed some other tube examples that have a better/worse 2H component accompanied by a similarly worse/better 3H component.
You mentioned the effect of mu follower load on distortion. This is quite true but also seems to exhibit this trade off between 2H and 3H.
For example, I tested some 6CG7 at 20V rms into loads varying from 100K down to 10K. At the extremes the 2H was identical but 3H increased by about 10dB at 10K load. There was a pronounced 2H minima of several dB at 15K load but again at the expense of 3H (although THD was also at a minimum). Generally from 100K load downwards to 15K, 2H decreased at the expensive of 3H (though THD decreased too) thereafter both increased. I also measured higher order harmonics which were not present at the 100K load but began to appear with heavier loads. I repeated the tests at 10V rms and there were no higher harmonics than the 3rd detectable at any load.
Cheers
Ian
What i find common on hi linearity tubes like 6e5pi, D3a,C3m,(6h1p) is the obvious distance p-g and having flat plates pieces.
See, no plate at the sides, where grid bars are. Rather a openings. Maybe electron beams around gridbars are "dirty", distortion rich
Yes. Uniform field. Consider a tube as a multiple tubes with different parameters in parallel and take an integral. Planar tubes are winners. As well as tubes with strictly concentric electrodes.
Also, tubes with longer electrodes and denser grids like 6P15P are more linear than tubes with shorter electrodes and rarer grids like 6L6, it was designed for video amplifiers in TV, based on already existing EL84 design.
I expect if to make a version of GU-50 with at least half longer electrodes it would be a killer 60W tube for audio. But it was designed to work in mobile equipment, that's why it was so relatively short. RCA for mobile RF equipment designed, for example, 832 that resembled 2 of 6L6 with very short electrodes in a single envelope.
Also, tubes with longer electrodes and denser grids like 6P15P are more linear than tubes with shorter electrodes and rarer grids like 6L6, it was designed for video amplifiers in TV, based on already existing EL84 design.
I expect if to make a version of GU-50 with at least half longer electrodes it would be a killer 60W tube for audio. But it was designed to work in mobile equipment, that's why it was so relatively short. RCA for mobile RF equipment designed, for example, 832 that resembled 2 of 6L6 with very short electrodes in a single envelope.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.