Effect of Changing Output Tube Loads

[HotBluePlates]

Took this statement from another thread, as it relates to a common misconception folks have about the "correct load" for a given output tube:

... Looking at the [6V6] data sheet, and extrapolating higher voltages because it doesn't go to 400+V.
It seems I am at the sweet spot (~285v plate & screen) for THD.  As voltage rises, the 3rd harmonic climbs with it to the end of the graph.
Around my voltage level is where that 3rd drops down to it's lowest level. ...

I assume the attachment below is the graph you referenced from the G.E. 6V6GT data sheet.


How to interpret this graph:
-  Plate and screen voltage ("E_b" and "E_c2") are fixed at 250vdc.
-  Assume cathode is grounded, and negative bias to the control grid ("E_c1" or pin 5) is -12.5vdc.
-  Signal applied to the grid is 8.8v RMS. But that's also 8.8v * 1.414 = 12.44v peak.


The above means we're evaluating power output & distortion while plate load impedance is changed, with a peak signal input just a touch below the standing bias voltage. As a result, we'll get all the clean output power from this operating point, though there will be some distortion. The graph shows how that distortion changes (as well as how output power and screen current draw changes).


Look for the point with the lowest THD. It's at ~6kΩ. You can assume a SE 6V6 with the specified 250v plate & screen will have the lowest total distortion at ~6kΩ. You can also use the common "cheat" and notice that you could run push-pull 6V6's with the same plate, screen and bias voltages at 6kΩ plate-to-plate.


What happens at loads other than 6kΩ? Well, I guess looking closer you must have taken the graph from a different sheet, as this one only specifies THD. However, all beam power tubes typically show the same pattern:
-  At low plate load impedances, 3rd harmonic (and other odd-harmonic) distortion is low, while 2nd harmonic (and to a lesser extent other even harmonic) distortion is relatively high.
-  As plate load impedance rises, even harmonic distortion drops and odd harmonic distortion rises. There is some median load impedance where the total of all harmonic (THD) reaches a minimum.
-  As you continue to raise plate load impedance above that median point, 2nd harmonic continues to drop but 3rd harmonic rises, and THD again rises from its minimum.


Across the range of plate load impedances (on this graph, from 2kΩ up to 9kΩ), lower plate load impedance provides less power output (but more 2nd harmonic), but there is a fairly broad range of loads for which power output is near maximum. This graph shows something between 4.5w and 4.7w output for any load from 5kΩ up to 9kΩ.


So you should hopefully see that a precise plate load impedance is not critical for your output tubes. There is some ideal plate load at which distortion will be a minimum and clean output power will be a maximum. On either side of that point, you get either less output power, more distortion, or both.


I said "ideal plate load" above. But the distortion figures are for THD, though we should know that for loads below the minimum-THD point, the distortion is mostly even harmonic (and mostly 2nd harmonic out of that number).


If you also know that push-pull operation cancels even harmonic distortion produced by the output tubes, you really might have your cleanest, loudest output at a "lower than ideal" plate load impedance, when running push-pull.

[HotBluePlates]

Graphs like this were developed with a guy plotting each load (probably 2kΩ, 3kΩ, 4kΩ, etc) for a single operating point on a set of plate curves. That guy would then carefully calculate output power, 2nd & 3rd harmonic distortion, etc for each load, and plot just those points to the graph. Then he'd use a french-curve to connect the dots.

So you could re-derive all this stuff for a different plate & screen voltage, a different bias point and grid input for other conditions. But the general rules will still apply... That is, lower loads than the "optimal" may yield less output power but will be heavy in even-order harmonic distortion, while higher loads will have predominantly odd-order harmonic distortion.

So what about some other "rules of thumb" where you quickly estimate an ideal load by assuming the tube is a perfect device (no load-line plotting)? We'll assume the plate dissipation rating is 12w for our "ideal 6V6" and the same 250v plate-to-cathode (and so we infer this is class A operation).

An ideal Class A-operated output tube would idle at 100% dissipation, and swing its plate from the supply voltage to zero volts, as the plate current rises from the idle value to double-idle value.

12w/250v = 48mA idle plate current.

So the plate load will see a change of voltage from 250v down to 0v (a ∆ of 250v-0v=250v), while plate current rises from 48mA to 96mA (a ∆ of 96mA-48mA=48mA). The implied load impedance/resistance is (following Ohm's Law):
∆volts/∆current = 250v/48mA = ~5.21kΩ

So pretty dang close to the ideal low-THD point shown on the graph, and we plotted nothing!