SE amps and B+ voltage

[EFK]

I'm new to the whole SE thing and I'm probably looking at it with more of a p-p mentality. Is there a limit as to how high you can run the B+ comparative to a p-p amp? I've been messing with a SE EL34 'mini marshall' which I set up essentially like a JMP with fixed bias but the PT I had on hand to use is putting the plate voltage around 430V. This is perfect for the preamp as it is maintaining all the preamp voltages exactly where I would want them to be, and frankly I'm happy with the plate/screens being where they are BUT I'm a little confused as to what I should be doing with the bias. Single ended should be biased at 100%, correct? However, at this plate voltage into a 5K OT, I put the meter on the cathode to watch the bias and even when I set it to @ 70% at idle, it almost doubles when I crank the amp and hit it with a signal. So, 100% is obviously going to be a problem and the current certainly is not remaining approximately static under load. (Just for the moment, I'm not considering the screen current, although maybe I should and subtract that?) So if it's single ended but not running in class A, what IS it doing?

Just looking for some thoughts, advice etc. Anything!

[Steve Conner]

A SE amp is supposed to be biased so that the idle point is about halfway between cutoff and the tube conducting as hard as it can.

Now, as you increase the B+, the maximum current the tube can supply goes up. But the maximum idle current you can safely run goes down. So, you necessarily move away from symmetrical clipping.

You can restore the symmetry by reducing the screen voltage, which cuts down the tube's peak current capability.

[EFK]

Hmmmm. Got a formula, preferably easy on the math? Well I don't want to start sticking too much resistance in the screen feed, as all that sag will basically kill the whole "tight plexi" idea. I've got a big ass pile of 5W zeners that would probably do the trick if I stick them inline, but what are we talking here? 20V? 100V? Is there some way to determine an approximate voltage drop that allow me to run the bias back up somewhere *in the vicinity* of 100%, other than just going the trial/error route? I have a suspicion that we're talking more than just 20V here, given the 430-ish plate voltage.

[Steve Conner]

I've got a formula that's very easy on the math. Adjust the screen voltage till it sounds good. It may need to come down quite a bit, maybe to 250V.

For each screen voltage adjustment, you may need to change the cathode resistor too.

[EFK]

Fixed bias! Nice and easy adjustment. Well I'll break out the zeners and see how it goes.

Out of curiosity, what is happening to the output stage when a SE amp is *not* running at 100%? Because honestly, this does not sound bad. When the volume is lower, there is a bit of that trailing distortion that sounds like crossover/cathode bias shift in a p-p amp when the bias is too cold (cathode bias) but in a SE amp, there can't really be 'crossover' distortion can there? If I didn't hear a hint of that clacky crap at lower volume (that's how it sounds to me anyway) I wouldn't even be worrying about it.

[loudthud]

I think the root of the problem is that the primary impedance of the OT is too low. Can you try it with a higher impedance speaker? If you want to experiment, use a dummy load that is variable and find where you get maximum power. A quick calculation will tell you what the primary impedance makes the tube happy. Then, get one of those Hammond universal OTs with all the taps and you can get close to whatever load suits you.

[EFK]

Actuall the OT I'm using can do 5K or 8K. I set it up for 5K as I don;t think I've ever heard of anyone going higher than that with SE EL34, but maybe I ought to try 8K? I could probably just initially try running one of my 16 ohm cabs off the 8 ohm tap, for 10K. I'm a bit worried about flyback though, as I already had one tube arc over between pins 3 and 2. Might have just been a freak occurrence, I don;t know.....

[Steve Conner]

Well, you can look at it two ways.

Reducing the screen voltage increases the optimum load impedance that the tube wants to see.

Increasing the load impedance requires lowering the screen voltage, to prevent the screen from burning up under overdrive.

[EFK]

Been working with it. Can someone double check some math for me? I want to make sure I'm checking bias in an accurate enough manner. FIXED bias.

(1) I'm running it into a fixed load, running 1 Khz test signal into input.

(2) I run the volume up to the point where the most current being drawn and make note of current in mA (across 1 ohm resistor).

(3) Check plate voltage at this point and make note of it.

(4) Check drop across the 1K screen resistor at same point.

(5) I then subtract the screen current from the plate current.

(6) Then I do the math to calculate the plate dissipation (w/ screen current subtracted) at that point of most current draw. I'm shooting for 100% there.

Anyone see any problems with this? So far I have not seen any redplating.

I tried the 8K and while it sounded like a gain monster, it was WAY too compressed sounding. Went back to 5K, now working with zeners to see how high I can keep the screen voltage while still keeping the plate dissipation within limits. The lower screens are definitely affecting current draw, but they are also adding to compression - I have no doubt of this as it is audible. I'm shooting for as little compression as possible while keeping everything leashed!

[Steve Conner]

The compression is probably because you're using a resistor to drop the screen voltage, try the zeners instead.

In the end, what sounds best is right, as long as it doesn't blow up. :-p

[EFK]

Admittedly, I do like that philosophy. Very much!

I only have a 1K screen R right on the tube. That shouldn't effect anything other than normal for a Marshall? I had two 33V zeners inline, was getting about a 68V drop across them. I know the zeners are not supposed to create compression, and maybe they are not in and of themselves, but possibly the much-reduced screen voltage is, because it definitely made an audible difference. Sounds like a compressor got kicked on to me.

[g1]

Out of curiosity, how does the no signal dissipation compare to your full power with signal method?

[EFK]

I was able to set it up so that I had it idling at 100% (calculating by subtracting screens at idle) and when going all out, it was actually dropping back to @ 95%, I suspect because (1) the screens were pulling pretty good current, @ 45 to 50 mA measured over the 1K screen resistor on the tube, and (2) the zeners were dropping the screen feed about 66-68V, so the much reduced screens were apparently limiting the plate current. What I'm working on now is to try to actually flip flop that, because I personally am not after much compression here so I'd rather it idle around 90 to 95% and then just creep up a bit to @ 100% at full bore. I'm slowly swapping out zeners to see how much voltage I can give back to the screens before the plate current starts to get away from me.

[Zozobra]

Some interesting reading in here. I was idly thinking about building a KT66 based SE amp but the spare PT I have is of the order of 375 VDC (taken from a reissue vox AC15 IIRC) but that voltage is way off the mark for anything looking even remotely center biased!

[Gingertube]

Guys,
For heavens sake look at the published data sheets and application data. You can vary operating points etc. a bit but use the published values as a start point.

Here is what is published for SE EL34

Note that for SE you want MUCH lower B+ than you would see in a PP Amp. Also note that 5K already higher than the recommended primary impedance AND look at the recommended screen resistor value of 2K

Class Va Vg2 Vg1 Ia Ig2 Ra S Rk Zout Pout THD Notes
A S/E 250 265 -14.5 70 - 73 10 -15 18,000 9 Fixed bias 3,000 8 10 With Rg2 = 2000 ohms
A S/E 250 265 -13.5 100 - 105 14.9 - 29.0 15,000 11 Fixed Bias 2,000 11 10
A S/E (triode) 375 - - 70 - 73.5 - 370 3,000 6 8

Damn all my carefull spacing of data in rows went west - look at it at the source.
Source:
TDSL Tube data [EL34WXT]

and for KT66
TDSL Tube data [KT66]
You could certainly run SE Triode mode at that 375V or even higher.


Cheers,
Ian