"Some valves don't do what we expect. The 6J6 is basically two planar triodes back to back, so it ought to be linear but it isn't. "
Looks similar to the 12AT7 curves. (see the Mu curves at end of data)
I don't have one handy to look at, but I would guess it has the grid 1 very close to the cathode, and maybe a somewhat coarser pitch to the grid wire windup. Both are rated for mixer use. I would guess they are both using grid wire "island effect" to get the non-linearity.
http://tubedata.milbert.com/sheets/093/6/6J6.pdf
http://tubedata.milbert.com/sheets/093/1/12AT7.pdf
Looks similar to the 12AT7 curves. (see the Mu curves at end of data)
I don't have one handy to look at, but I would guess it has the grid 1 very close to the cathode, and maybe a somewhat coarser pitch to the grid wire windup. Both are rated for mixer use. I would guess they are both using grid wire "island effect" to get the non-linearity.
http://tubedata.milbert.com/sheets/093/6/6J6.pdf
http://tubedata.milbert.com/sheets/093/1/12AT7.pdf
Very possible: in computers they were used in logical elements. Multi-vibrators and triggers work with positive voltages on grids.
There appears to be a strong correlation between plate tallness and linearity (for a given tube) - taller plates have better linearity. I am not familiar with the constructional details of the 6J6. How tall are is plates compared to the 12AT7?
Cheers
Ian
It is true when grids completely surround cathodes: edge effects affect linearity. Planar tubes have 4 edges, so taller electrodes do not mean higher area covered by uniform field. Planar tubes usually have denser grids, but grids of 6J6 and 12AT7 seems to be rare.
Member
Joined 2009
Paid Member
Nicely said - and you could write <Solid State device> where you have written 'pentode'
what about Rk? isn't it that there is supposedly a cathode resisistor equal to 1/gm at operating current of the stage?
Typical 6J6 anodes appear to be about 10mm tall, but the active part of the cathode is a bit shorter at about 8mm. There could be some end-effect, but as all electrodes continue past the active cathode surface I would expect this to be small. The grid pitch is not much smaller than the grid-cathode spacing (maybe 2:1 ratio?), so island effect will be fairly strong.
That is similar to the 12AT7. The 6SN7 and its derivatives (6CG7/6FQ7) have an anode height of about 24mm and much better linearity. I tested about 50 6CG7/6FQ7 types some with 24mm plates and some with 18mm plates and the shorter plate types were always less linear.
This also appears to apply to the 12AX7. The EH12AX7LPS (a long plate version of the 12AX7) is a lot more linear than the standard height plate types. Despite its detractors, the 12AX7 is actually intrinsically very linear, even more so than the 6SN7 in the test I did.
Cheers
Ian
I'm not certain how universal is the trend you have seen, but if it generally applies the next question is why? I think it is unlikely to be end effects, although these may play a role. Maybe something else is going on.
If two valves are nominally the same (e.g. both 12AT7) then a longer anode has to be compensated for by something else (wider cathode-anode spacing?) otherwise it will draw too much current to be a 12AT7. Wider cathode-anode spacing then means wider cathode-grid spacing, in order to maintain the same mu. Wider cathode-grid spacing, with the same grid pitch, means less island effect and less distortion.
So, do long and short anode versions of the same valve have the same grid pitch or the same total number of grid wires or something in between?
If two valves are nominally the same (e.g. both 12AT7) then a longer anode has to be compensated for by something else (wider cathode-anode spacing?) otherwise it will draw too much current to be a 12AT7. Wider cathode-anode spacing then means wider cathode-grid spacing, in order to maintain the same mu. Wider cathode-grid spacing, with the same grid pitch, means less island effect and less distortion.
So, do long and short anode versions of the same valve have the same grid pitch or the same total number of grid wires or something in between?
I don't know. I have tried it with only two tube types so far but it is clearly there in both.
Also, you have mentioned the 'island effect' several times but I can find no reference to it. Can you point me to the appropriate literature?
Cheers
Ian
For island effect see here. A long page, so search within it. The phenomenon is sometimes known by its German name: inselbildung.
Areas of the valve near/under a grid wire get cut off before areas between grid wires, so the effective size of the valve reduces near cutoff. Coarse grid spacing or small cathode-grid distance makes it worse.
Areas of the valve near/under a grid wire get cut off before areas between grid wires, so the effective size of the valve reduces near cutoff. Coarse grid spacing or small cathode-grid distance makes it worse.
Anyone know why the D3a is so revered for linearity in triode mode? Looking at the Philips data sheet triode curves it does not look all that great to me (next to last pages).
http://tubedata.milbert.com/sheets/009/d/D3a.pdf
The Mu curve droops significantly off at low current. Its active element is short, its got a small cathode. True it has a frame grid, and maybe some are unduly impressed by the extremely high gm, but an extreme linear tube would use a frame grid with only modest gm. Or are they all cheating and putting a big degeneration resistor in the cathode circuit?
There are some 20 K gm 5 W and 10 W frame grid TV video tubes ($0.50 to 1$ each in quantity) that simply, seriously..., blow away the published D3a curves. If I had a D3a I would put it up against one on the curve tracer and post a comparison, but I don't generally buy expensive tubes.
http://tubedata.milbert.com/sheets/009/d/D3a.pdf
The Mu curve droops significantly off at low current. Its active element is short, its got a small cathode. True it has a frame grid, and maybe some are unduly impressed by the extremely high gm, but an extreme linear tube would use a frame grid with only modest gm. Or are they all cheating and putting a big degeneration resistor in the cathode circuit?
There are some 20 K gm 5 W and 10 W frame grid TV video tubes ($0.50 to 1$ each in quantity) that simply, seriously..., blow away the published D3a curves. If I had a D3a I would put it up against one on the curve tracer and post a comparison, but I don't generally buy expensive tubes.
Because when you set them up with a CCS plate load and LED bias, the experimentally determined linearity is astonishing. The 65 ohm ENR is not unattractive, either.
"when you set them up with a CCS plate load and LED bias, the experimentally determined linearity is astonishing"
Hmm, with a CCS load most tubes are too linear to tell any difference from the datasheet or curve tracer curves. But I generally like to see the triode curves not drooping over at high plate V and the Mu curve not drooping at low current, which does show up on the typical curves (well not so often published for pentodes) or on the tracer. Seems like I have seen -77 dB (2nd harmonic, sound card FFT) fairly routinely, at an optimized operating point, with frame grid video tubes. Haven't really tried to find the best one, probably as much variation, at that kind of level, within a tube type anyway.
Hmm, with a CCS load most tubes are too linear to tell any difference from the datasheet or curve tracer curves. But I generally like to see the triode curves not drooping over at high plate V and the Mu curve not drooping at low current, which does show up on the typical curves (well not so often published for pentodes) or on the tracer. Seems like I have seen -77 dB (2nd harmonic, sound card FFT) fairly routinely, at an optimized operating point, with frame grid video tubes. Haven't really tried to find the best one, probably as much variation, at that kind of level, within a tube type anyway.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.