How to Avoid the Drudgery of Drawing Push-Pull 6L6 Composite Load Lines?

[bob p]

I need to draw-up a composite load-line chart for a push-pull 6L6 setup that will run at plate voltages of about 530 VDC. Instead of doing things the old fashioned way (cutting and pasting printouts of the tube characteristics) or a more modern way (photoshop), I'm wondering if anyone has gotten really high-tech and converted the 6L6 characteristic data into a machine-readable format.

Steve Bench has a spreadsheet > here < that does the calculations for a theoretical tube, but it contains no data for a real tube like a 6L6. This spreadsheet looks like it would be a very handy tool once it was populated with the right data.

So I've been wondering...Has anyone converted the 6L6 characteristic curves into a machine-readable format? If so, that would make the process of drawing composite load lines a whole LOT easier.

[jazbo8]

Prof. Fawcett just posted this recently over at AX84.

Jaz

[FredB]

I use LTspice to generate plate curves from tube models, which is useful because then you can generate curves for tetrodes and pentodes at screen grid voltages other than the ones in the published curves. Next, I just draw load lines on those curves. This all can be done in LTspice.

[loudthud]

Don't the curves need to overlay such that the bias point is the same spot on both graphs? this would mean you would need to redo it each time bias or B+ changes.

[jazbo8]

I use LTspice to generate plate curves from tube models, which is useful because then you can generate curves for tetrodes and pentodes at screen grid voltages other than the ones in the published curves. Next, I just draw load lines on those curves. This all can be done in LTspice.

Provided that the pentode (or tetrode) Spice model has relatively accurate screen current vs. plate current characteristic. Otherwise, the curves will not be scaled properly. Many of the models that you find on the net, do not meet this criteria.

Jaz

[FredB]

I just graph out one tube, and yes if you change B+, screen voltage, or bias have have to run the sim again, although that's as easy as a mouse click once you have things arranged.

[FredB]

Yes there are limitations concerning the screen grid current and function, especially as one moves away from the screen voltage, which was used when the models was created. I hold the result of the screen models with particular skepticism, however until someone comes up with better models that's all we have do use.

[J M Fahey]

I'd rather print, cut and paste and just use it the old fashioned way.
Much faster than the alternatives (Photoshop, etc.) and much more trustable than SPICE models simulations.
Personal opinion, others may differ.
But I have been doing it for over 40 years and it's widely known that old dogs hardly learn new tricks.

[jazbo8]

Yes there are limitations concerning the screen grid current and function, especially as one moves away from the screen voltage, which was used when the models was created. I hold the result of the screen models with particular skepticism, however until someone comes up with better models that's all we have do use.

Yes, good pentode models are hard to find, the ones from Ayumi seem to be the most accurate I have found so far, and is based on the physical model (unlike Koren's or Duncan's), take a look at his methodology, and try a few of his models to verify for yourself. They are the only models that I use now.

Jaz

[bob p]

Don't the curves need to overlay such that the bias point is the same spot on both graphs? this would mean you would need to redo it each time bias or B+ changes.

That is *exactly* the "drudgery" that I was referring to. Cutting and pasting a new chart every time that you make a significant change can be somewhat of a PITB. I guess I fall into the category of old-dog willing to learn a new trick. Like JMF said, cutting and pasting is guaranteed to provide the right answer. I'm just worried about trusting modelling software that may or not be accurate.

The idea of Spice models is interesting. Being an old-dog, I don't know how to use Spice. Tried it once and never got it working for lack of a good tutorial. I'm not sure that it's even worth the effort of learning Spice if the models are in any way less accurate than paper, scissors and tape.

[jpfamps]

To find the optimum loading, presumably you also need to calculate the HT and screen supply under full loading. Is this included in the pSPICE models mentioned above, or would you have to add this to the models?

I now determine optimum loading experimentally.

Firstly I "guestimate" the loading.

I then employing a very large wirewound resistor with a terminal that can be clamped along it's length to vary the resistance to find the optimum loading for maximum power.

Of course this approach is not ideal for one off designs as if your 1st estimate is a long way off you end up having to buy another output transformer.

[jazbo8]

The pentode/tetrode models only model the tube itself, any external components still need to be added to complete the circuit analysis, just like in the real thing. Of course, nothing beats actual bread-boarding which needs to be done even SPICE is used in the initial stage of the design process - but it can help you eliminate some of the guesswork involved.

Jaz

[FredB]

To find the optimum loading, presumably you also need to calculate the HT and screen supply under full loading. Is this included in the pSPICE models mentioned above, or would you have to add this to the models?

I now determine optimum loading experimentally.

Firstly I "guestimate" the loading.

I then employing a very large wirewound resistor with a terminal that can be clamped along it's length to vary the resistance to find the optimum loading for maximum power.

Of course this approach is not ideal for one off designs as if your 1st estimate is a long way off you end up having to buy another output transformer.

How do you determine which loading yields maximum power?

Would you need to have two variable resistors to determine the optimal load for a push pull output stage?

[J M Fahey]

Yes, that's the way.
No need for two resistors, just a variable one.
I took the SS trail *long* ago, around 1972, but still make the occasional tube amp for those who *insist* on the orange glow.
And experimenting as you say I found that the standard run of the mill 6L6 I can buy here over the counter are happier with 5500 ohms plate to plate instead of the clasic 4000/4500 .
Suppose they are somewhat lackluster in the emission area, plus I don't beat them to death, use moderate 420/430V +V and 1K screen resistors instead of 470r.
End result?: 40 W RMS , quite real by the way, not cheating by allowing some clipping but *just* reaching the edge and then backing a little.
Perfect power for regular Club work with 1 or 2 12" without drowning everybody else and enough for Heavy Metal headbangers in large stages *if* driving a good 4x12".
Of course, the audience gets the miked PA version of this sound.

Can you get more power?
Yes, by using NOS or premium/selected tubes, raising +V, loweringscreen resistors, the works.
For my reality, that increases costs *a lot* for meager 5/10W extra.
Simply not worth it (here).

By the way, I wind my own transformers, so this impedance correction is achieved simply by winding the original, Fender specs primary, but lowering secondary from 60 turns to 50.
Works like a charm and , as I said, run of the mill tubes last a lot.

EDIT: I see you intend to use 530V +V
All I can comment is "OUCH !!!!"

[bob p]

EDIT: I see you intend to use 530V +V
All I can comment is "OUCH !!!!"

Yes, OUCH! I wouldn't have designed the amp that way, but the power supply iron is what it is, and the creeping rise in wall voltage here in the US isn't helping either.

I have given thought of things like bucking the wall power to vintage levels, and zener trickery, but before doing things like that I'd like to crunch some numbers for 530 VDC.