What is the theoretical advantage of direct heated triodes?

[TheGimp]

But have you spent as much time and effort with the 6SN7 or 2C22 amp as you did with the 26?

That is, does the 26 sound better because you made it sound better?

 

[Rod Coleman]

The Directly Heated Triode has serious, and measurable advantages over triode-connected pentodes and beam-tubes.

I confine this assertion to power output stages, driving a loudspeaker, since here, the advantages are clear and obvious.

To avoid any charge of bias, I will call on independent measurements, taken with the same equipment, at roughly the same time. The examples I give here are used with the kind permission of their owner (and driver of the measuring equipment) - Audiomatica SRL:

Audiomatica Srl

I have compared their curves for the KT88 beam tube against the directly-heated VV30B (a 300B variant).

In both cases, a load approximating a loudspeaker corresponding to a nominal 3500 ohm anode load is shown, and the idle point for both is 360V 60mA [Pa = 21,6W].

The load line is an elliptical envelope to represent the reactive nature of the speaker. As levels and frequencies vary, the instantaneous operating point can be anywhere within the ellipse, or even further out!
The use of elliptical load lines to model speaker loads is widely discussed, please look it up, or find a library with good books in:

Oxner, Edwin S. "Power FETs & their Applications" Prentice Hall, 1983
Moir, J. : "High Quality Audio" Chapman & Hall 1958
van der Ven, Jonker & Heins: Philips Technical Review, 5, 61-68, 1940.

The pictures tell the tale. The green markers are just lines of equal width to help judge the spacing.

As Thomas mentioned above, the triode-connected beam tube shows much more variation in mu than the DHT.

More seriously still, the KT88 curves become seriously compressed high voltage.

Altogether, the linearity across the required working range is worse for the KT88. The comparison may be made between others, with similar results.

But the fame of the 300B for linearity in SE speaker driving stages is perfectly justified.

Provided the power supplies are top notch (HT AND Filament), the difference is easy to hear, in SE amps for sure.

 

[Magz]

But have you spent as much time and effort with the 6SN7 or 2C22 amp as you did with the 26?

That is, does the 26 sound better because you made it sound better?

That's a fair question, and one I'm not sure can be answered, as the different tubes require different surrounding electronics and structure. I suppose I could build another version using exactly the same surrounding components and 6SN7s just to answer it, but short of that (which is not going to happen), some assumptions need to be made.

I did spend quite a bit of time and money on the other preamps as well, but the best I could get out of them was less than what I hear from the 26.

I admit that I did enjoy the challenge of working with the 26s, and overcoming their disadvantages was part of the fun. Plus they DO look cool...

 

[MrCurwen]

I have tested 6SN7, 2C22, 2C26, 27, 56, 76, 6SL7 and some DHT including 26, 71A, 46, 47 and some russian DH-tubes in the same conditions; MOSFET gyrator as plate load, filament or heater actively filtered.

My main interest is headphones - they really let the details (or lack thereof) and all the tiniest of differences show thru in a way that speakers usually don't (well at least my speakers).

So my experience has been very clear cut that DHT's when properly implemented simply offer a greater transparency than IDHT's. Unfortunately that doesn't always mean making everything sound better; with my reference amp I've become painfully aware of the inadequacy of my source material. On the upside, if the source is good (vinyl for example), there's nothing that has the realism of DHT's in my experience. With some well recorded movies it can get spooky at times.

 

[rongon]

SY challenged my assertion that DHTs are more linear than triode-wired beam tubes (6L6GC, KT88, etc.). So I built a PP 6L6GC-triode amp to test that, in a Dyna ST70 chassis using the original iron.

I've had a PP 2A3 amp for years, with Tango XE-45-5 OPTs. The 10kHz square waves are pretty amazing. Those are great OPTs. Far better than the Dyna OPTs. That amp was my baseline reference. I also have a Hafler P1000 in good working order, for comparison.

In the end, I was able to get the PP 6L6-triode amp w/ 5dB of global NFB to sound every bit as good as the PP 2A3 amp w/ ZNFB, at least to me, in my listening room, driving my Snell J/III's. I had to put a shelving cap in parallel to the feedback resistor to tame a ringing (audible as a "sizzle" in the highs).

What is left to find out is whether the PP 6L6GC-triodes can be run with better iron, without NFB, and still sound as good as the PP 2A3's.

As a sanity check, I put in the Hafler P1000. The difference was subtle, with the SS amp sounding predictably "colder" and with tighter bass reproduction (higher damping factor, probably). It's just not as much fun to listen to as either tube amp. But it ain't bad, really.

I wish I had a PP pentode or ultralinear-wired amp to compare. They require a lot more NFB than the triodes.

My end result: Either DHTs or triode wired IDHTs can sound really good. It took more work (adding NFB and compensation) to make the PP 6L6GC-triode amp sound good enough for me than it did with 2A3s (no NFB, but far more expensive OPTs). Maybe -- all other things being equal -- it really is the quality of the OPT that dictates the end results.

 

[Chris Hornbeck]

Could it be that all the attention to detail in the surrounding circuitry which may be required for a DHT is part of the explanation? DHTs may need this, while 'lesser' valves can manage reasonably well without it so don't get it.

I'd have to agree that this is at least 90% of the positive side of DHT/single-ended/ yada yada amplifiers. Physically large valves like the 50 watters (211 and 845, etc.) have some advantage in intrinsic linearity, and the mandatory class A operation helps with real world reactive loads, but everything else becomes so much more demanding that one's designing focus is sharpened and less effort and expense spared.

What I'd tried poorly to say earlier is that the AC heating causes an amplitude modulation of signal at twice AC heating frequency. It doesn't help to heat at supersonic frequency because modulation products will still fall into the audio range. Here's an interesting guy's take on things:

On Correlation Between Residual DHT Filament Hum and AC Frequency. Distortion-induced hum in directly-heated triodes.

Thanks,
Chris

 

[rongon]

...The pictures tell the tale. The green markers are just lines of equal width to help judge the spacing.

As Thomas mentioned above, the triode-connected beam tube shows much more variation in mu than the DHT.

More seriously still, the KT88 curves become seriously compressed high voltage.

Altogether, the linearity across the required working range is worse for the KT88. The comparison may be made between others, with similar results.

But the fame of the 300B for linearity in SE speaker driving stages is perfectly justified...

Interesting stuff!

The picture shows the KT88 at operating points of plate volts = 360V and plate current = 60mA. That's only 21.6W plate dissipation. KT88 is capable of >30W plate dissipation. Perhaps there is a more linear operating point possible for the KT88-triode?

 

[Merlinb]

Interesting stuff!

The picture shows the KT88 at operating points of plate volts = 360V and plate current = 60mA. That's only 21.6W plate dissipation. KT88 is capable of >30W plate dissipation. Perhaps there is a more linear operating point possible for the KT88-triode?

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.

 

[rongon]

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.

And then there's the EL34, which is even easier to drive...

That should be "driver circuit" no?

 

[Rod Coleman]

Interesting stuff!

The picture shows the KT88 at operating points of plate volts = 360V and plate current = 60mA. That's only 21.6W plate dissipation. KT88 is capable of >30W plate dissipation. Perhaps there is a more linear operating point possible for the KT88-triode?

You can tinker with it but the problems do not go away.

The variation in mu with operating point, and the compression of curves with decreasing current/ high voltage remain, and make a wholly unnecessary compromise in performance - if you are driving a speaker.

The EL 34 is as bad, if not worse.

But don't take my word for it - head over to Audiomatica and try your own operating points on the curves.

I compared the KT88 and 300B becuase they are both 35-40W (depending on manufacturer) anode rated. We should be careful to compare things on on equal footing.

 

[Rod Coleman]

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.

Only if you insist on using tired out 1950s driver circuits.

Achieving high linearity across a swing of +/- 80V is something of a challenge, but if we move the design standards onward, and use modern transistor-supported triode drivers - we can drive this swing wholly within the high-linearity range of the triode.

 

[cotdt]

Hi!
With all plate curve measurements I did, DHTs have been more linear than IDHTs. Especially in comparison to triode strapped pentodes. There had been some experiments done by S. Bench, comparing IDHTs to DHTs and DHTs with oxide and thoriated tungsten filaments. He measureed the change of mu with signal ampltitude. In this test the DHTs had mroe constant mu and among DHTs those with thoriated tungsten filaments had been better

Best regards

Thomas

Thank you very much. Now I know that 300B is more linear than indirect triodes like KT88 (wired as triode).

 

[rongon]

You can tinker with it but the problems do not go away.

The variation in mu with operating point, and the compression of curves with decreasing current/ high voltage remain, and make a wholly unnecessary compromise in performance - if you are driving a speaker.

The EL 34 is as bad, if not worse.

But don't take my word for it - head over to Audiomatica and try your own operating points on the curves.

I compared the KT88 and 300B becuase they are both 35-40W (depending on manufacturer) anode rated. We should be careful to compare things on on equal footing.

Well, looking at the curves, I do see your point.

In the end, I suppose it comes down to budgetary concerns. You can get a quartet of new 6L6GC's for the price of one E-H 2A3 (least expensive 2A3 I can find on short notice).

By "head over to Automatica" did you mean "buy a CLIO setup and test for yourself"?

Achieving high linearity across a swing of +/- 80V is something of a challenge, but if we move the design standards onward, and use modern transistor-supported triode drivers - we can drive this swing wholly within the high-linearity range of the triode.

But distortion is proportional to level, so an easier to drive load will result in lower distortion from the same driver circuitry, no? In other words, I take a driver circuit that is capable of driving +/-90V. If I take that driver circuit and drive a load that requires only +/-40V, won't it create less distortion than if I take that same driver circuit and make it drive a load that requires +/-80V?

--

 

[Rod Coleman]

I'd have to agree that this is at least 90% of the positive side of DHT/single-ended/ yada yada amplifiers. Physically large valves like the 50 watters (211 and 845, etc.) have some advantage in intrinsic linearity, and the mandatory class A operation helps with real world reactive loads, but everything else becomes so much more demanding that one's designing focus is sharpened and less effort and expense spared.

What I'd tried poorly to say earlier is that the AC heating causes an amplitude modulation of signal at twice AC heating frequency. It doesn't help to heat at supersonic frequency because modulation products will still fall into the audio range. Here's an interesting guy's take on things:

On Correlation Between Residual DHT Filament Hum and AC Frequency. Distortion-induced hum in directly-heated triodes.
Thanks,
Chris

Chris,

It's also true that the directly heated output stages respond to the extra care in the design & implementation. The 300Bs are not easy to use, and if you are in any way careless about HT or filament power, or the driver design - you could end up wondering what the fuss is about.

 

[Rod Coleman]

Well, looking at the curves, I do see your point.

In the end, I suppose it comes down to budgetary concerns. You can get a quartet of new 6L6GC's for the price of one E-H 2A3 (least expensive 2A3 I can find on short notice).

By "head over to Automatica" did you mean "buy a CLIO setup and test for yourself"?

--

Well CLIO would tell you the answer to any question - yes!

But if you are impatient, please take a look at the other curves measured and ready for us:

eg

The EL34 Tube


.

 

[cotdt]

See post #56

The 300B is clearly more linear than the KT88, and sounds better too, but what about the 45 versus EL84?

 

[DF96]

Rod's curves show that a particular DHT is more linear than a triode-connected KT88. In itself this does not show that DHTs are necessarily more linear. It could be that low mu valves are more linear - there are good physical reasons for this. The curves show that the DHT has half the mu of the KT88.

 

[Pano]

Thanks Rod, interesting stuff!

I have always assumed that people like DHTs because they are unusual and big and expensive.

All people? What about people like me who thought that SE-DHT were funny and cute, but not a real amp. Until I heard one. (circa 1984). I was not ready to have my ideas about amplifiers changed.

They seem to be preferred by the sort of people who proudly display their valve amp, as though it were an art object.

There are those who do, but that is not limited to SET amps - many other fancy amps get set up as an objet d'art. Let's not confuse fashion with sound.

It is conceivable that some people prefer them because the net result is more distortion.

Conceivable, but I like the sound because to me, they have far less audible distortion. They seem cleaner and clearer than others amps. Not all SETs, of course, but the good ones do.
Could it be that those who prefer pentodes or solid state amps simply like the distortion signature of those amps better? Perhaps it's what they grew up with, or just like better. It's an equally valid argument.

 

[rongon]

Chris,

It's also true that the directly heated output stages respond to the extra care in the design & implementation. The 300Bs are not easy to use, and if you are in any way careless about HT or filament power, or the driver design - you could end up wondering what the fuss is about.

Yes, that is probably some of what is happening to me. I should rebuild that PP 2A3 amp and do it better this time. I paid attention to the HT supply, but the heaters are plain old AC. I would be much more careful about them this time, maybe use your heater supply.

Re: EL34, etc -- If you take an EL34-triode with plate voltage = 270V and plate current = 80mA, it looks really linear, but since that's within a smaller window, will yield little power, probably only as much as a pair of 2A3's (5 or 6W rms). But much cheaper. Also, insanely easy to drive -- only like 20V peak needed. Or am I doing that wrong...

On second thought... It looks like I was doing that wrong. Very little swing out of an EL34-triode biased like that. 2A3 sure looks nice, though.

--

 

[Wavebourn]

I believe it is all about area of cathode in respect to the distance to the control grid and distance between wires in it. As you may see, the most benefit DHTs have on small currents when negative voltage on the control grid is maximal. It creates so called "island effect", when the tube seems like a lot of triodes in parallel, each with different curves. The closer their characteristics to each other, the better is linearity. Look for example at 4П1Л and 12П17Л -- almost the same lamp, different cathodes. And exactly the difference I explained above!