Attempt on a 7591 Se design

[bea]

Hi,

here my first steps toward a 7591 SE amp:



And here its .ac analysis:



This one is based on an Epiphone Valve Junior and a Jan Wüstens ATRA0211 output transformer with a primary impedance of 5.2 kOhm and a max current of 60 mA. Within the limits of the simulation it looks as if the ideal anode voltage of 315 V of a 7591 at 5.2 kOhm quite exactly corresponds to the 60 mA of the OT. The G2 current meets the expectation as well.

I have yet to estimate the power and to check the dynamic behavior of the amp - a nice opportunity to learn how to do that with LTSpice. The preamp is still work in progress. It is based on the preamp i am using in the G2000 but with a different tone stack (AMZ control vs. Bone Ray control).

Suggestions / comments welcome.

[loudthud]

What is the power rating of the OT? 5.2K is on the high side and the 7591 can take a lot more Voltage.

RCA suggests 3K OT with B+ 300V for 11Watts.

[bea]

RCA suggests 3K OT with B+ 300V for 11Watts.

I know, and at Ra=3.3k. But i do not believe. Class A means 50% of the max anode dissipation which is 19 W. So the 11 W would mean driving the tube beyond its specs. If the specs are sufficiently conservative that might be still ok. (What about 7591 from recent production... would You dare?)

The Epi Valve Junior is an amp with a single ended EL84, hence the 5.2 kOhms. But its power supply delivers more that 300 V (which would have been perfectly suitable for the 7 kOhms used in the 1st release of that amp, but Epiphone changed that later to 5.2 kOhms to to pressure of the users). My main motivation for this project is to utilize the energy which is wasted to reduce the anode voltage to reduce the driving voltage by 50 V and obtain more output instead. This will obviously require a larger valve than the EL84 such as the 6L6 or the 7591.

Back to physics:

The 5.2 kOhms are only optimal for an EL84 if the anode voltage is ~250 V - according to
Z = Ua^2 / Pa
where:
Ua = Anode voltage.
Pa = Maximum anode dissipation.

Applying this to a 7591 yields Ra=4.7 kOhm - and Ua=315 V as the optimum for Ra=5.2 kOhm. It is obvious that i cannot expect anything near to the 11 W. I did a rough estimate of 6.5 W from the curves for Ua=300V/Ra=6.5W. Extrapolating this via P=U^2/R would yield 7.2 W. BTW: taking the 300V/11W as given and extrapolating down would yield larger values.

Now to the power supply. There is not much known about it except that it is reasonably oversized for an EL84 in SE mode and can successfully drive a 6L6 and even an EL34. It is said to be have the following specs: Usec=260V~ (known), Isec=100 mA (speculative). I did an estimate of the effective internal resistance from the voltages without tubes and in operation of the amp and found an effective internal resistance of 509 Ohms.

Th OT: i did already change the original OT to the ATRA0211 which can do with 60 mA. Measured primary resistance: 365 Ohms.

My simulation uses the following values: Ri=509 Ohms, V=376 V which will appear after the rectifier if i use the amp in its 220V setting but with 230 V from the wall. Due to lack of actual data i use "reasonable estimates" for the OT except of the measured primary Ri.

I am aware that i am exploring the limits of both transformers.

[jazbo8]

I know, and at Ra=3.3k. But i do not believe.

Please note the 11W output in the datasheet has a distortion figure of 11%, i.e., it's running on the ragged edge. But 8-9W would be easy to manage with much less distortion. If the goal is to get the maximum out of the existing PT, then using a higher Zpri seems to be counter-intuitive with the 7591. If you must use the 5k OPT, then another output tube might be more suitable... some of the more experienced builders can probably make some good suggestions.

[bea]

As i mentioned i have a fully working EL84 amp (actually two, but one still requires some work in the preamp anyway...), and i am exploring to which degree it can be tweaked. What i am showing here is currently just a feasibility study. Therefore i do have the power supply and i do already have installed the fat OT.
(And i know of several successful conversions to 6L6 and EL34.)

A note to the data sheet: if You do the math of designing an SE amp with the 7591 with the standard approach You will end up with a much larger Ra than the suggestion of the data sheet. Something You should be aware of that. The designs shown in data sheets should be taken as "just proposals" which might or might not fully exploit the possibilities of a tube. It might always be meaningful to try something different.

BTW: i noticed need to reduce the input grid resistance; it is a bit large for fixed bias operation.

[jazbo8]

A note to the data sheet: if You do the math of designing an SE amp with the 7591 with the standard approach You will end up with a much larger Ra than the suggestion of the data sheet.

Kindly elaborate.

[bea]

i need to leave for work.

[jazbo8]

i need to leave for work.

No worries, whenever you get a chance.

[bea]

Kindly elaborate.

It is about minimizing THD - it reaches a minimum for Ra=Ua/Ia for given values of Ua and Ia where Ia is controlled by biasing.

The abovementioned formula Ra = Ua^2 / Pa also takes this argument but implies that the anode current is chosen to make Ua*Ia = Pa_max, the maximum anode dissipation.


I am aware of another rule of thumb: Ra ~ 10...15% of the internal resistance of the valve which for the 7591 is given as 29 kOhms. That would explain the suggested load of 3 kOhms - but apparently on cost of increased THD even at smaller signal levels.

It is also clear that in my simulation the max dissipation of the 7591 is very closely approached - it is 18.9 W and therefore really close to the limit of the tube.

[jazbo8]

It is about minimizing THD - it reaches a minimum for Ra=Ua/Ia for given values of Ua and Ia where Ia is controlled by biasing.

Hmm, not sure where that came from... in any case, in guitar amps, seldom is minimizing the distortion a concern, usually maximizing the output power is the top priority.

It is also clear that in my simulation the max dissipation of the 7591 is very closely approached - it is 18.9 W and therefore really close to the limit of the tube.

That may be true, but the distortion is still rather high (before NFB is applied) even when compared with the datasheet example. As you mentioned earlier, there are some successful 6L6 and EL34 conversions, is there a reason that you are not following those examples?

[bea]

As it can be seen, the 7591 matches pretty nicely into the framework of the amp, so why should i not use it?

The 7591 is closest to the EL84 with about 50% more possible anode dissipation.
I'll also simulate the 6L6, but it's amplification is smaller, and it is quite obvious in advance that i will not be able to obtain more power from it (i actually did the analysis on paper). Same fort the EL34.

A valve larger than the 7591 will give me not advantage except of the lower prices of 6L6 or EL34.

Something else i would like to mention: heater current. The 7591 draws a tiny bit more current than the EL84, the 6L6 even more, and the EL34 needs 1.5 - twice as much as the EL84. And that IS a concern to me.

[bea]

Hmm, not sure where that came from... in any case, in guitar amps, seldom is minimizing the distortion a concern, usually maximizing the output power is the top priority.

But the output power is determined by the capabilities of the power supply and the load resistance. The valve acts like a voltage controlled resistor, it must be sufficiently large to handle it. The 7591 is sufficiently large for the given hardware, and it will not help if a larger valve is used. I actually expect similar values in the simulation if i insert a larger valve and adopt the biasing accordingly.

(And if i want even more power, i'll use different hardware - i have some other stuff lying around: an EL84 PP powerstage (again with the question if it is possible to achieve a bit more by using larger valves, but i need to collect more data before i can evaluate that), and a few Dynacord transformer sets which allow to drive either 2 EL34 at 750 V. The would be a pretty straightforward project: restoring the power stage and adding two of the preamp stages i just developed for the G-2000. No math at all...

[jazbo8]

Perhaps I did not make my point well enough... yes, 7591 can indeed handle the output power (at least for the NOS, not sure about the current production), but its plate characteristic does not appear to be the best choice for the PT and OPT that you plan to use, unless of course, you just prefer how the 7591 sounds compared to the others.

[Jazz P Bass]

From what I have gathered the current 7591 tubes do not hold a candle to the NOS design.

Not even close.

[ThermionicScott]

From what I have gathered the current 7591 tubes do not hold a candle to the NOS design.

Not even close.

Yep, generally best practice to design around the tubes that are being made now, not those of yesteryear, unless you have a stash of the good stuff. And at the risk of sounding unromantic, the 7591 was basically a refined 6L6: a slightly cooler filament, but higher transconductance and voltage ratings to squeeze more juice out of fewer tubes in an amplifier (economical!) Drive voltage isn't really hard to come by, and the high voltage ratings are of much more use in a Class AB push-pull design than a single-ender.

Remember that with an SE amp, the maximum plate dissipation happens at idle. The 8-11 watts of output power that we're talking about goes to the load, actually reducing the heat generated at the plate.