New build advice

[Phil_S]

There is a lot for me to digest here. It's the 80/20 rule; 80% planning 20% doing. That tends to give the expected result. I am really liking this idea of a SE EL34. I just don't know yet if it will get traction.

Here's a 2-fer on the test rig. I think voltages are about +20 over what will actually occurred and mA are high by 20-25mA, but it is consistently like that for all the real test loads I've used. Not shown here, with 5U4GB the results for 2.5K at R1 are 322V 136mA. I understand, this is not the power tube's operating point, but it is a confirmation of whether current will exceed the capacity of the iron. IOW, it appears to be in the safe range and 322V could be used as the assumption for B+. I will have to see if I can work the load line calculator.

[martin manning]

Phil, do you have data from a series of test points from your rig, load current vs. load voltage?

[didit]

My SEs are SS rectified 50uF cap followed by 100R followed by 50uF

Dare you to swap in Hammond 159Q or equivalent 7H 100R choke for 100R power resistor.

Best .. Ian

[didit]

I should re-run the tests using the 179Ω MIL/CET 8H 150mA choke. Thoughts? Am I headed in the right direction? Thanks.

This was my initial suggestion, so gotta agree. I would recommend putting series of 4007s in your test rig as silicon really handles transient over voltage badly. As you drop the resistance of the load put your power resistors in parallel to avoid smoke.

Best .. Ian

[Firestorm]

My SEs are SS rectified 50uF cap followed by 100R followed by 50uF

Dare you to swap in Hammond 159Q or equivalent 7H 100R choke for 100R power resistor.

Best .. Ian

I don't know why not, but this was too costly already with two OTs, fairly large PT, relays, relay power supply, custom footswitch, etc. Maybe penny wise pound foolish since the 100R is 25W and has to be mounted on a heat sink. Also no room in mine.

[Phil_S]

Here are all the actual test results I have. There is a jpg for the voyeurs. There is an Excel file for those who want the numbers to play with. The rig is exactly as pictured in the above PSU snapshot with 1N4007 FW rectifier.

I am not inclined to rig up the 5V supply and a socket to test with a tube. The difference between the two rectifiers in a static test like this should be around 11% on the B+. If B+ goes down 11%, I suppose current has to go up by a similar amount. I judge the difference to be not so important in obtaining an understanding of the effect of the choke. If you feel I am wrong about that, I'm all ears, so please say something.

I visited Patrick Turner's website earlier (dense read on SE) and am about to make an attempt to draw a load line for SE EL34.

[martin manning]

Phil, I think what matters is shown in the plot below. I calculated the DC load current and plotted it against B+. The idle point will be on this line. I'll pick 355V and 80mA. Looking at the Vg2 = 250V curves for EL34, you find at Vak = 340, Vg1 = 15, Ia = 70mA Pa = ~25W (~100%). I found this match by trial and error. The screen node voltage you will need is 265 to account for the 15V on the cathode, plus whatever the drop across the screen resistor is. The extra 10mA is for the screen and the preamp. For this point and screen voltage, a load of 5k looks pretty good, as load impedance for center bias can be estimated by (Edit) Vak^2/Pa max = 4k6.

[Firestorm]

Wow. Martin has good software. (And is way smart as I often say). What he's done is not estimation, but successive approximation. If Newton hadn't beat him to it, he would be Newton.

The tubes actually don't care. They'll put out what they can into whatever load is presented. We can manipulate that number by about 50‰. I dare you to hear the difference.

So as Martin so elegantly showed, and I suggested Cowboy-style, I'll bet you like 5K.

Can the PT keep up? Probably.

[Phil_S]

Martin, thank you very much for taking time to do this.

I spent part of the afternoon schooling myself on pentode load lines. I found Patrick Turner's stuff was too dense for me at first, though later I realized I had comprehended more than I thought. The second source was Steve Bench. He started talking about the Q point but gave no obvious information on how to determine Q -- at least I couldn't comprehend it. Finally, I landed at Merlin's site, where he posted a good piece done by Iain who hangs out (or at least used to) at 18watt.com. It was Merlin's post that hit the spot, partly because he chose to work with the EL34 as his example. How convenient.

It is more than a point of curiosity, after attempting to draw load lines for quite a while, I arrived at Ia=65mA and suspected that was a bit on the low side. This came off a load line drawn from 180mA to 500V, which is the 2.5K load line. When I stuff your numbers into the equation 368v/130mA (from the graph), that sets the load line at 2.8K.

I am not questioning your statement that (I think you wrote it backwards) Vak^2/Pa Max = 340V^2/25W = 4.6K is the way to estimate center bias. I am just trying to understand 2.8K vs. 4.6K. (I have a deep seated feeling these are apples and oranges.)

Based on the read from Merlin, I will figure 42% efficiency. If that's the case and you are on the 25W power line, output will be about 10W.

I'm kinda hopeing I can use the 10W 5K SE OT that I've got. I'm inclined to try it and see how it sounds.

[martin manning]

Yes, I inadvertently inverted the RHS of the equation. The load line power is around 11W, so I'd estimate something approaching 10W at the output. You can plot a 5k load line directly on the data sheet curves since the target Vg2 is 250V. I think that 5K OT will be fine. Unfortunately the screen current isn't easily found for that point. At Va=Vg2=250, it's about 10mA, so I'd expect a bit less than that at Va=340, Vg2=250.

[Firestorm]

Load lines are a way to plot a dozen different tube characteristics (which are inherently 3-dimensional) onto a page (which is inherently 2-dimensional). Useful as a starting point, even an approximation, but the curves have a z-axis, too.

Fortunately, it doesn't matter all that much. The tubes do the math for you.

Don't confuse P-P AB loadlines with P-P A loadlines with SE loadlines. All different (and just a snapshot, even then).

Start with raw numbers (these define the curve and are easier to manipulate).

For SE, primary impedance is plate voltage divided by plate current (technically peak numbers, but it's Class A so who cares). AB is different because it sees only a fraction of the load).

If you get it wrong, you might lose a little power and a little bandwidth, but you won't notice. Or you could get it horribly wrong and be Ken Bran. Didn't work out so bad for him.

[Phil_S]

All good. Now that I've got an OT off the bone pile too, all I need is the EL34 and some inner parts. As someone else recently noted, this one is practically free since I already had the parts before I had the idea.

Anyone wanna trade some used but not used up USA made 6L6's or 6V6's (been tested on my home built tester) for an decent EL34? I might have other good stuff too!

[Deric]

I have a couple of JJ EL34s laying around - leftovers from matched pairs where one died a premature death....

Gladly send you one of those for free if you're interested. Just PM me an address.

[Firestorm]

Take what you can get. But once I built these amps I started using them as a high voltage check against my Hikock. To see what current they wanted at full voltage.

Damn. XF2s are a whole different animal, soundwise. What did those ladies in Blackburn do?

[Firestorm]

You won't be disappointed with a JJ if you don't have Mullard to compare it with. Nowadays we must not compare with unobtatium. Sad.