5E3 Type Circuit with 6L6's, best OT impedance?

[Wombaticus]

Great. So now you've taken what I've said completely out of context. I hope that the fact that you've completely missed the point is down to a mere misunderstanding. I'm not going to waste any more of your time. Thank you to everyone who has had something constructive to say.

The other's comments here about the ultimate subjectivity of ideal primary Z for an OT are certainly well justified. The observation that the real world load line is better described as an ellipse rather than a straight line as a consequence of the reactive nature of the speaker load also gives some theoretical justification as to why the exact impedance isn't always critical.

However, from an engineering standpoint, the fiction that the load line is straight is still a useful one. First, note that the family of ellipses constituting the real-world line still converges (nearly) to a line at two critical areas - near cutoff, and more significantly, near grid current limiting.

Now, for every operating point for a pentode (plate, screen, bias voltages) there *will* be a load line that will give minimum total harmonic distortion distortion, and (a different) one that will give maximum power. Manufacturer data sheets will often give recommended impedances that will result in one or the other, and it was often one of these that early amp manufacturers adopted. Often, these turn out to be pretty close to passing through the knee of the characteristics curve. Note, however, that the assertion is made only regarding best power or lowest distortion...Neither of these points are guaranteed to give best tone - they are just reasonable and safe starting places.

Note also that despite the generally forgiving nature of tubes, it is quite possible to end up with an unsafe load line if one picks OT impedances willy-nilly, and, again, a traditional straight-line load analysis will make it apparent if this is likely to be the case. Randall Aiken had an interesting thread awhile back here demonstrating that the load lines of even some commercial amps (old 100W Marshall was the example) are fairly marginal, and that this goes a long way towards explaining some reliability issues in the originals, and some show stoppers in clones that tried the same load line with a less saggy PT..

[Chuck H]

Great analogy. But this also raises another issue... You imply that ideals are more critical than we suppose. And since the actuality changes with speaker impedance curves, and this changes with frequency, does this mean that we should also be choosing the primary Z for our instrument OT's based on :

What instrument (bass guitar, six string guitar, harmonica, etc.) and what playing style (lead, rhythm or both) as well?!? Not to mention that some speakers are MUCH more non linear than others. So should there also be a limitation to speaker parameters based on the OT ideals?!?

It seems to me that the art of making tube amps distort is becoming as stringent as making tube amps NOT distort.

[Wombaticus]

Great analogy. But this also raises another issue... You imply that ideals are more critical than we suppose. And since the actuality changes with speaker impedance curves, and this changes with frequency, does this mean that we should also be choosing the primary Z for our instrument OT's based on :

What instrument (bass guitar, six string guitar, harmonica, etc.) and what playing style (lead, rhythm or both) as well?!? Not to mention that some speakers are MUCH more non linear than others. So should there also be a limitation to speaker parameters based on the OT ideals?!?

It seems to me that the art of making tube amps distort is becoming as stringent as making tube amps NOT distort.

Well, sure. People do this all the time, don't they? ...picking OT impedances based on trying to get more headroom, or earlier breakup, etc. Usually it's easier to swap speakers than OTs, but the interaction of the speaker reactance with a particular OT impedance must be an important tonal factor in addition just to the frequency response of the speaker.

A few further points are worth making .. first, after the resonant peak, there's a lot more upside in the impedance relative to nominal impedance than downside, so effective load generally increases with frequency, which you might wish to take into account when choosing a load line. Also, it's worth noting that a fundamental difference between true pentodes and beam terriers lies in how well the knee of the characteristic curve is defined...pretty sharp in beam terriers, and more ill defined in pentodes...so it's a little more involved to decide on "ideal" load lines, and a bit more care is required to avoid issues with screen current. ( although lots of guidance exists on how to go about this, naturally. The point is that this probably explains a lot of the tonal differences). One final factor to consider is the influence of NFB around the OT, and how NFB will make a power amp output relatively less sensitive to the loudspeaker impedance.


Edit: just noticed that autocorrect rendered "beam tetrodes" as "beam terriers". Good one!

[pdf64]

NFB will make a power amp output relatively less sensitive to the loudspeaker impedance

Power amp NFB in tube guitar amps is almost always arranged as a voltage amp configuration; to my thinking, that will make the power output more sensitive to speaker impedance, due to the voltage output tending to be more constant (as frequency is varied) across a varying load impedance. Therefore power output will tend to increase as the load impedance value falls below nominal, and will tend to decrease as the load impedance value rises above nominal, when NFB is applied, compared to open loop.

[Wombaticus]

Power amp NFB in tube guitar amps is almost always arranged as a voltage amp configuration; to my thinking, that will make the power output more sensitive to speaker impedance, due to the voltage output tending to be more constant (as frequency is varied) across a varying load impedance. Therefore power output will tend to increase as the load impedance value falls below nominal, and will tend to decrease as the load impedance value rises above nominal, when NFB is applied, compared to open loop.

Sorry, no, you have this backwards. Voltage feedback *decreases* output impedance and therefore makes driving the load less variable with frequency dependent impedance... It is current feedback that *increases* output impedance and therefore makes output levels more dependent on impedance. There are many references, but Aiken discusses this in his tutorial on feedback.

[pdf64]

Voltage NFB will make the amp output (as in power output, as it's a power amp) more dependant on the load impedance; because the voltage can't rise to accommodate higher load impedance values, so as the impedance increases, the power output will fall (because the voltage stays the same).
Therefore an open loop tube amp may put out a more consistant power level to a speaker than one with voltage NFB.
I think the above is in accordance with Aiken's page?

[Wombaticus]

Voltage NFB will make the amp output (as in power output, as it's a power amp) more dependant on the load impedance; because the voltage can't rise to accommodate higher load impedance values, so as the impedance increases, the power output will fall (because the voltage stays the same).
Therefore an open loop tube amp may put out a more consistant power level to a speaker than one with voltage NFB.
I think the above is in accordance with Aiken's page?

Contrary to intuition, *lower* load impedances are *heavier* loads and harder to drive than are high load impedances, which represent a relatively lighter load.......

[pdf64]

And a NFB voltage amp will attempt to put the same voltage across a light or heavy load, resulting in a lower power to the light load etc.
Whereas I think a high output impedance, open loop tube amp will tend to put a more consistant power to a load whose impedance value (or weight) varies.

[makepeace]

loving the recent replies.. really helpful to my understanding. thanks!

[pdf64]

Wombaticus and I are probably looking at the same thing from different perspectives and so analysing/summarising it differently; if the discussion has been helpful then that's fantastic.

[loudthud]

It should be noted here that many of the early tweed designs specified 6L6 {metal}, 6L6G, or 6L6GB. Pin 1 will be grounded, it was connected to the metal envelope on the 6L6 {metal}. These tubes were rated at 19W plate dissipation and 360V plate voltage. The RCA tube manual RC-19 (1959) lists for class AB1 service the maximum output power in the examples as 26.5W. Curiously enough one example for fixed bias has a load of 6600 ohms (26.5W) and another with identical voltages has a load of 3800 ohms (18W). It's safe to say that a wide range was acceptable. The 5881 had slightly higher ratings of 23W and 400V. The 6L6GC (not in the 1959 book) is an up-rated version with 30W dissipation and 500V plate ratings.

[tubeswell]

I heard a really nice 5E3 for 2 x 6L6s that used 6k6 reflected load on the OT running into a greenback posted here a few years back. (I think it was an amp by 'mcinku' IIRC) 2CW

(Edit - here is the post; http://music-electronics-forum.com/t3610/ )