Load Line with EL84@325v

[surfsup]

So here is a conceptual load line of an amp with EL84s and 325 on the plates.

[img:961:635]http://chicagocadcam.com/ChrisHahn/EL84_PA_2.jpg[/img]

RED - 8k OT
BLUE - 12k OT
YELLOW - 15k OT

The 8k is fairly standard. In fact I've seen it recommended at 6k as well. But notice the load line in read through the bias point at around -11v.

The load line passes quite a bit through the max dissipation area, as well as terminating at -0v at the edge of max dissipation.

I see amps with 350 on the plates, or even 400v for some with EL84s. So why wouldn't one use a 12k or 15k OT to rotate that load line down to more reasonable levels?

[katopan]

You've got it right that the Class B section load line is 1/4 of the OT primary impedance, and that's your zero bias line going through 325V at 0mA. But the load line around the bias point up to the point of one valve reaching cutoff, or the Class A region of operation, is actually 1/2 the OT primary impedance. You need to halve the slopes on the lines you have going through the bias point for the Class A region. Then the actual operating load line is the 1/2 impedance line from bias to intersection with the 1/4 impedance line, then the 1/4 impedance line the rest of the way up to grid clipping. This composite line doesn't spend as much time above the max power line.

Also keep in mind you'd only bias to 100% for a cathode bias amp with a small cathode bypass cap. In this case the bias cools off with signal, dynamically dropping your load line away from the max power line.

[surfsup]

katopan thanks. So I assume this is right then? The load line still has quite a bit of area in the max dissipation. Also, running actual values of the amp while on, I measure 10.25v cathode bias with a 125R = 41mA current, which according to the graph the bias point is inside the max dissipation area. Are the graphs that inaccurate?

[martin manning]

Two things to remember: First, you are looking at one of two tubes, which only has to conduct high current every-other cycle so in theory it could take 200% of max dissipation if it only had to do it half the time. Second the whole map is scaled up and down with screen voltage so the grid curves won't lie in the same place if your screen voltage doesn't match the 300V reference value. Then there is variation, which is on top of the differences between the manufacturers (old and new production tubes).

[surfsup]

Martin, I am starting to understand the signal is not max 100% of the time (being a sine wave) so the max dissipation line is not real-world accurate. Thanks.

I also realize the screen curves compress but I have not figured out how to draw them, nor have I found a graph with the screen at 325. 300 is as close as I could find, and I figured the 25v wouldn't make a huge difference. Anyway, this thread has been a good learning tool for me, thanks.

Now back to figuring out the buzz problem in my other thread.

[martin manning]

I also realize the screen curves compress but I have not figured out how to draw them, nor have I found a graph with the screen at 325. 300 is as close as I could find, and I figured the 25v wouldn't make a huge difference.

To find the bias point for your voltages you have to interpolate on the mutual characteristics curves, or if your screen and plate voltages are essentially the same you can use the triode-connected plate curves.

[surfsup]

now that you mention that I seem to recall something in Merlins preamp book that showed how to get screen bias curves. I'll look into that after I experiment a bit to reduce some of the ac signal into the el84s. The power amp section is basically the same as many other amps (aside from exact plate voltage) so I assume it is okay. The tubes are not plating or anything either...

[katopan]

Lastly from your other thread it looks like your plate voltage to ground is 325V? The chart is based on plate to cathode voltage, so you need to subtract the 10V and draw your bias point from 315V on the x-axis.

The plate curves need adjusting for actual screen voltage, but of course the max power curve is correct.

[surfsup]

katopan, true and thanks for that reminder. The 10v will shift it left some. I am 100% convinced the problem is not the PA, it is in the PI somehow - or the interaction between the two.

I cut out the Ruby mod diode/zeners and I get more headroom before I hear the buzz now. I think they were clipping the signal too hard perhaps. I used 12v zeners with 10.25V on the cathode. I thought that would be okay. Maybe I need higher v zeners...

Anyway, when the PI clips or goes into cutoff, the buzz starts. The moment the scope trace shows the PI out signal round off the tops, the buzz instantly disappears.

So I need more voltage swing on the PI. ANd to do that I think I will change out the 47k tail resistor for something a little less but I am going to do some reading tonight first. If someone does not mind commenting I am listening....