What is the theoretical advantage of direct heated triodes?

[DF96]

I have in my possession a number of ex-British Post Office STC 3A/109s. These DHTs were especially designed for telephone repeaters, and the samples I have were made in 1968.

This appears to be another 1930s low mu design? The 4021A data sheet is dated 1937. This is equivalent to 3A/109A. Presumably the 3A/109 is slightly earlier.

 

[SY]

Also, the 300B needs about twice the drive voltage of the KT88, so really we have only shifted the linearity problem from the power valve to the preamp.

Yes. The reductio ad absurdum is a cathode follower output stage.

 

[Chris Hornbeck]

This discussion of low mu triodes being more linear than otherwise equivalent higher mu triodes is new to me. We can all easily name several exceptions (12AU7, 12AX7, etc.) so I wonder what's the theoretical underpinning.

Thanks for any help,
Chris

 

[Wavebourn]

This discussion of low mu triodes being more linear than otherwise equivalent higher mu triodes is new to me. We can all easily name several exceptions (12AU7, 12AX7, etc.) so I wonder what's the theoretical underpinning.

Geometry. Fields and distances.

 

[SY]

Well, remember that low mu triodes have higher feedback intrinsically. High mu triodes can have the same feedback applied externally to reduce the mu to the same value, and generally the result is better linearity.

 

[Chris Hornbeck]

Geometry. Fields and distances.

You're not giving me much to work with here....

Thanks,
Chris

 

[Wavebourn]

You're not giving me much to work with here....

It's the starting point. And I believe, the major one in the question.

 

[Chris Hornbeck]

Well, remember that low mu triodes have higher feedback intrinsically. High mu triodes can have the same feedback applied externally to reduce the mu to the same value, and generally the result is better linearity.

I could more easily accept a correlation with 1/gm than with 1/mu. Here's a pair that's observably the same except for grid pitch: the 211 and 845. What can we conclude from that pairing?

Thanks,
Chris

 

[SY]

You're not giving me much to work with here....

That's the origin of the intrinsic negative feedback.

 

[cotdt]

The low mu theory doesn't explain the poor performance of the 12AU7. And the EL84 has relatively high mu for an output tube yet is still quite linear.

 

[Wavebourn]

EL84 is tall. We already discussed. Uniform field, many "mini-tubes" with the similar characteristics in parallel

 

[Chris Hornbeck]

That's the origin of the intrinsic negative feedback.

If that's the basis of the argument I'll just have to agree to disagree. Y'all seem fairly set, so I'll not try to dissuade you. The obvious exceptions, otherwise identical constructions differing only in grid pitch, like 12AX7/12AU7 and 211/845, are too big a stumbling block for me.

Thanks, as always,
Chris

 

[Wavebourn]

I learned that Bonch-Buievich at the beginning of 20'th century developed and taught the theory of vacuum triodes, with all math apparatus needed for optimization.

 

[SY]

The obvious exceptions, otherwise identical constructions differing only in grid pitch, like 12AX7/12AU7 and 211/845, are too big a stumbling block for me.

Don't know about 845/211, but 12AU7 and 12AX7 differ by a lot more than just grid pitch. Disassemble a few to see.

 

[SY]

EL84 is tall. We already discussed. Uniform field, many "mini-tubes" with the similar characteristics in parallel

Interestingly, I have some very short plate 12AX7 which seem to be every bit as linear as the "standard" constructions. The only 12AX7 I tried that had noticeably better linearity was the smooth-plate Telefunken. The ridged plate Telefunken acted about the same as the others, which is pretty darn good, one of the most linear triodes out there.

 

[astouffer]

I have in my possession a number of ex-British Post Office STC 3A/109s. These DHTs were especially designed for telephone repeaters, and the samples I have were made in 1968.

The Western Electric 300B was released in 1937, later than the release of the 6L6.

So, this claim is also quite unfounded.



I answered this point in an earlier post, above.

Repeatedly claiming that things are Not So because you haven't seen them highlights the weakness of your argument.

Western Electric shut down their last tube plant in 1988. Just because your tubes were made in 1968 does not mean they were meant for new designs.

 

[Miles Prower]

The low mu theory doesn't explain the poor performance of the 12AU7. And the EL84 has relatively high mu for an output tube yet is still quite linear.

Design criteria determine this. The 12AU7 was not designed to be an audio tube, and audio applications were an after thought. Look at the schemos where 12AU7s appear. These are most frequently RF circuits: oscillator/buffers, frequency multipliers, PP, Class C intermediate drivers, low power RF finals, and quasi-digital and digital applications: plate or cathode coupled multivibrators, one-shots, RS latches and clocked flip-flops. For this, you don't require linearity, and for some RF apps (mixers) good linearity is a detriment.

It's the same, but more so, for the 12AV7, whose audio loadlines are positively hideous. Another RF type not designed to be a good audio performer.

As for the negative reputation of the 12AX7, this seems to be related to poor design. Since the type has an Rp= 90K, it's all too common to see passive plate loads that are too small. Back in "the day" this may have been a problem, but these days, active plate loads made from transistors and ICs solves that problem. Loaded appropriately, it's a good audio performer.

 

[Wavebourn]

Interestingly, I have some very short plate 12AX7 which seem to be every bit as linear as the "standard" constructions.

Is it wider? See DF's post about rods, on the previous page. The field is affected by rods that hold the grid.

 

[Pano]

Wrap 6dB of feedback around that KT88 triode and its curves would probably beat the 300B,

Close, but not quite. Jean Hiraga did this in about 1979 in an attempt to make a more affordable SE amp that was as linear as the 300B. From the measurements, pretty close. It included a UL tap and mild GNFB of 6dB, IIRC. A little bounce back up the higher harmonics, but not bad at all. I heard it, liked it, but still preferred the DHT. A nicely thought out approach to the design goals.

 

[wa2ise]

Looking at the curves and specs for the 2A3 http://www.mif.pg.gda.pl/homepages/frank/sheets/127/2/2A3.pdf, a tube that has a 2.5V filament, and wants to have -45V on the grid, you could run the filament on DC and the tube looks to have enough linearity (spacing of the lines representing grid voltage and plate current) that the shift of 2.5V seen at one end of the filament and grid vs the other end shouldn't matter, as the grid voltage is varied (looks like you could safely feed the grid with 14Vrms of audio, 40V p-p), the change of plate current from one portion of the filament vs another same size portion would be the same. This variation of plate current is what the audio signal is, and the output transformer in turn feeds the audio to the speaker. As the filament voltage is way smaller than the range of the grid voltage, so there's no danger of a portion of the filament being cut off by excessive negativeness of the grid, or hitting grid current with the grid going above 0V compared to a portion of the filament.

Looks like DC would be better on the filament, as any nonlinearities caused by differences from one end of the filament to the other vs the grid would be constant, vs with AC on the filament, these nonlinearityes would vary every 60Hz waveform. Distortion that comes and goes at a 60Hz rate could be more annoying than constant distortion.