Push Pull Load Lines - revisited

LT spice, Curve Captor, PSUDII and whatever other sims you can think of.

Post Reply
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Push Pull Load Lines - revisited

Post by sbench »

Spice lets you do a voltage sweep. Ideal for plotting tube curves. But you can also alter it slightly to allow you to manipulate push pull operation, thus allowing you a really easy way to create and manipulate a push pull load line. The only thing you need to remember is that as the anode voltage on one tube goes up, the voltage on the other goes down. Likewise when sweeping different grid voltages, one goes up while the other goes down.

Thus, if, while you're setting up the sweeps, you also create a separate non swept power or bias supply, the sweep can be set up for one tube, and connecting a controlled voltage source (E) with a gain of -1, the complementary source can drive the other tube. Likewise with a bias source, and the grid step source connected to the grid (and to the other grid with a gain of -1) the entire complementary set of curves are available. Simply subtracting the plate currents give you a beautiful set of push pull tube curves.

If you place a load resistor across the plate stepped source, plotting current thru that resistor gives you the plate loadline.

Spice also allows for multiple plot panes. One plane can plot the plate current differences (the push pull plate curves) and the load (which you can then manipulate for your desired load line). A second plane can plot one of the tubes plate curves (without the subtraction) to allow you to monitor idle current. If you're using a tetrode, beam power or pentode, a third plot pane can show you screen current.

Here's an example with a 6L6GC. I have chosen to use a 5.2k plate-to-plate load on a 450V plate supply, 362V screen supply, -36v bias voltage. All these can, of course, be manipulated. Overall this seems to be a really interesting operating point, producing 55 watts at .7% distortion or less, and really low distortion at 1 watt.

The source file, as usual is attached, and a typical run .GIF is also presented.

Enjoy,
Stephie
Attachments
pp load line.asc
(3.25 KiB) Downloaded 87 times
pploadlines.GIF
pploadlines.GIF (396.49 KiB) Viewed 3147 times
dave slagle
Posts: 2037
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey Stephie !!!!

The composite load lines has been of interest to me over the last few years since I have been 'gasp' been listening to PP amps to 'direct drive' the quad esl '57 panels. It always struck me that the composite lines appear to be much closer to the parallel even spaced lines of an ideal tube and the "gotcha area" of the high voltage low current single tube operation disappears.

I have had endless battles in the past about elliptical loadlines and disortion and it is my belief that if the ellipse is in the part of the curves where the lines are parallel and evenly spaced, having finite inductance is not a major issue. It only becomes a problem when the ellipse pushes the operation to the bunched up area in the lower right hand corner.... or worse yet into cutoff.

I could type all day on this but my question is... can I simply switch from pentode to triode in your model?

dave
Get Your Fix
www.hifiheroin.com
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Quick answer, YES. Note that since triode curves are lower Z (current increases rapidly with voltage) the LIMIT function is really helpful. I indicated that on the schematic notes.
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

The classic example would be 2A3 operated at PP with 3k load, 62 volt bias. Notice this produces 15 watts, following the datasheet operating conditions. It looks like this...

Stephie
Attachments
pploadlinestriode.GIF
pploadlinestriode.GIF (403.09 KiB) Viewed 3127 times
dave slagle
Posts: 2037
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Very cool!

any thoughts on how this would handle an elliptical loadline due to the capacitive nature of ESL panels?

For some reason I have yet to hear a SE amp drive ESL panels as well as PP and being a SE guy who despises all things PP this really bothers me.

Years ago Lynn Olson insisted to me the entire trick on driving an ESL is how the amp responds to the discharge of the panel capacitance back into it which lead me to think about the composite curves and elliptical loadlines.

dave
Get Your Fix
www.hifiheroin.com
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

I would agree that reactive loads need to be handled correctly by the amplifier in order to sound good. If an amplifier presents a low impedance inherently, it's probably reasonable. Varying impedance as part of the overall operating cycle is problematic. Quads sound wonderful when driven with an amplifier that can handle them.

Having said that, establishing a true, manipulate-able load line gets tricky. Since you have frequency dependent loading, along with need for large signal analysis (read: non-linearities) forces you into doing the load line in a transient analysis, not a DC sweep analysis. The tricky part is that you would LIKE the analysis that is simultaneously part of an active circuit (with interaction between the differential halves) AND artificial to show you the curves as well. That gets tricky.

A quick run shows I can do a elliptical load without the curves, or a set of curves without the ellipse easy enough. I'll look into the true curve+elliptical line.

Do you have any non-linear L or transformer models you were thinking of using?
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

OK. You asked for it. Here's a Spice Transient analysis load line analyzer that supports triodes, pentodes, single ended, push pull, AC and DC load lines. The AC analysis (elliptical load lines) supports different drive levels and frequencies.

Enjoy,
Stephie
Attachments
eliptical load lines3.asc
(9.68 KiB) Downloaded 83 times
eliptical load lines.gif
eliptical load lines.gif (387.43 KiB) Viewed 3112 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Some additional notes...

The SE and PP curves have the grid bias lines set up so that the middle (5th of 9 lines) is at the bias point. So, it you pick a bias of -40 volts, the grid lines are 0, -10, -20, -30, -40, -50, -60, -70, -80.

In running the program, if you see VERY slow execution times, change the GMIN and ABSTOL limits in the Spice tools box to 1e-009.

Feel free to rescale the output transformer inductance and/or coupling.

Here's some example plots...
Example1: Shows elliptical load line at a frequency of 2kHz. In this example I'm stepping the AMP parameter (.STEP PARAM AMP 2.5 20 2.5) from 2.5 to 20 in steps of 2.5. You can see how the non-linearities further affect the ellipse.

Example 2: Same but at 500Hz. The ellipse has narrowed considerably.

Example3: Single ended. Notice how lopsided the ellipses become. Notice further that the DC bias point shifts with amplitude. A lot of texts point this out, but it's kinda nice to see this "in action".

Example 4. Removed the reactive load, but notice that the bias point shifts with amplitude, even on a pure resistive load.

Example5: Changed tube to the 2A3 triode example.

Have fun....
Attachments
eliptical example5.GIF
eliptical example5.GIF (189.55 KiB) Viewed 3110 times
eliptical example4.GIF
eliptical example4.GIF (170.28 KiB) Viewed 3110 times
eliptical example3.GIF
eliptical example3.GIF (173.05 KiB) Viewed 3110 times
eliptical example2.GIF
eliptical example2.GIF (172.03 KiB) Viewed 3110 times
eliptical example1.GIF
eliptical example1.GIF (171.19 KiB) Viewed 3110 times
dave slagle
Posts: 2037
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Stephie....

words cannot explain how awesome you are! I was simply asking a general question and you go and do this???

I promise I'll be careful what I ask for in the future and all I can say is I am not worthy!

dave
Get Your Fix
www.hifiheroin.com
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Awww. Thanks! Enjoy. There's a lot to be learned in those pictures. Although I didn't explicitly point it out, the inductance of the coupling transformer, of course, can add phase shift at lower frequencies. You'll see it by running a set of curves at 20Hz or so. If you "reasonably" approximate the inductor non-linearities via Spice's non-linear inductance model, you can get some "interesting" loadlines.

Have fun!
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Now, what load line plotter would be complete without Ultra Linear operation?

Here's a mod to allow you to do either single ended or push pull ultra linear load lines. You can program any value of insertion of the plate signal from 0 (Pure pentode) to 1 (pure triode operation). You may use separate windings if you want the screen voltage to idle at a different voltage than the plate. Attached is an example stepping thru 3 different couplings and their effects on the effective tube curves.

Also the Spice .ASC file is attached. I slightly modified the attached .ASC file for better convergence during SE simulations.

Note that it is very instructive to change the proportion of G2 signal fed in. Interesting how you can change the effective plate resistance and thus amplifier output impedance.

Enjoy,
Stephie
Attachments
eliptical load lines3ul.asc
Updated version
(11.46 KiB) Downloaded 81 times
elipticalloadline_ultralinear.GIF
elipticalloadline_ultralinear.GIF (425.1 KiB) Viewed 3097 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Finally, here's a progression of the lowly 6L6GC in push-pull Ultra-Linear mode from pure pentode (beam power) operation to pure triode operation (screen voltage tapped so that the screen swings from 0 to 2x the quiescent voltage. Notice how rapidly the curves linearize in push pull mode as you increase the swing on the screen.

Operating point chosen is Va=475V, Vsg=285 volts, Bias = -24 volts. Plate load = 8.8k plate-to-plate. Plate idle current is 48ma per tube.

Condition Power output
pentode (0%) swing 41 watts
16% 34 watts
33% 25 watts
50% 19 watts
66% 15 watts
83% 12 watts
triode (100%) swing 9.5 watts


Enjoy,
Stephie
Attachments
el_6l6_tri.GIF
el_6l6_tri.GIF (190.77 KiB) Viewed 3091 times
el_6l6_83.GIF
el_6l6_83.GIF (187.71 KiB) Viewed 3091 times
el_6l6_66.GIF
el_6l6_66.GIF (191.78 KiB) Viewed 3091 times
el_6l6_50.GIF
el_6l6_50.GIF (186.39 KiB) Viewed 3091 times
el_6l6_33.GIF
el_6l6_33.GIF (190.74 KiB) Viewed 3091 times
el_6l6_16.GIF
el_6l6_16.GIF (186.27 KiB) Viewed 3091 times
el_6l6_pent.GIF
el_6l6_pent.GIF (186.46 KiB) Viewed 3091 times
Post Reply