6EM7 Push-Pull

[sorenj07]

I know there are a few of these but I was wondering if this one, which I just came up with, might be viable.

Each triode in the LTP is biased at -1.7V 1.2mA, after dropping 50 of 250VB++ in the tail resistor. For section 2, I started at plate voltage of 260 and drew a line to the 130mA mark (260/2000). I picked the -32V bias point, with about 35mA of current for each tube. The unlabeled coupling capacitor is also .47uF 400V. Could I direct-couple any of these stages?

How does it look? Can this LTP splitter swing+-32 clean volts?



Thanks!

[JohnAtwood]

This circuit looks good, but the resistor values in the driver stage seem a bit off. If each section #1 of the 6EM7 draws 1.2mA, and the plate voltage is about 150V (0.6 times the plate voltage - near the lowest distortion), then the plate resistors should be about 82K (R = E / I or 100V / 0.0012A). The grid voltage would be just under -2.0V, giving a cathode resistor of 1.9 / (2 x 0.0012) or 790 ohms. Under these circumstances, the undistorted plate should swing between 75V and 250V or 175V p-p or about 61Vrms, plenty enough to drive the output stage.

With the low value cathode resistor, you won't get much long-tail phase splitting. You might want to consider the paraphase inverter. The grid of the bottom 6EM7 #1 would come from the center of the grid resistors of the output stages. I've used this, and it works pretty well with high-mu driver stages like you have here. Overall, this should sound pretty good.

[sorenj07]

Thanks! I'll take a look at those results and try to figure out what you mean, but I suspect that you are right. I chose an anode load the same as Section 1's plate resistance for no real reason, and I do recall seeing figures that are quite a bit above that. I'll take another look at the curves and see what I can do about increasing the cathode resistor.. Glad to see that it's a mostly working design, though.

[Dave Cigna]

Concerning the input section, I agree with John for the most part. You probably want to run somewhat larger plate resistors.

I've attached a drawing that shows how to use a negative supply rail for the bottom of the tail resistor. This does a number of things. (1) It provides the full 250V B+ for the input tubes. (2) It allows a larger tail resistor making the LTP a better diff amp. (3) It allows you to eliminate the input coupling caps. Look at what Poinz does with his Musical Machine http://www.audiotropic.net/Projects/machine1.html

The 150k tail resistor acts as a CCS and sets the operating point of the input tubes. As drawn it drops 375V through 150k, so sets the current at about 2.5mA or 1.25mA per tube. The voltages shown are approximate, but should get you in the ball park.

With the 150k tail resistor I don't think it will be necessary to use different value plate resistors for good balance.

The amp will be fairly insensitive to the negative supply, so you don't need to go nuts filtering it. What I've shown should work fine.

Concerning the output stage, I don't see your selection of cathode resistor(s), B+, and operating current jiving, but maybe you're looking at different curves. When looking at plate curves, don't forget that you need to subtract the cathode voltage (drop across the cathode resistors) from the B+. I might suggest aiming for an operating point around 220V @ 40mA, which, according to the plate curves, requires around -35V bias. The required cathode resistor turns out to be just about what you have...

-- Dave

[sorenj07]

Originally posted by Dave Cigna
Concerning the input section, I agree with John for the most part. You probably want to run somewhat larger plate resistors.

OK, sounds good.

As for the negative bias supply, that seems like a very elegant way to provide some negative voltage. I keep the whole 250V on the B+ then, right? I'd assume then around 3mA total. In order not to have to drop down to 2V from 420VDC rectified, could I instead rectify from the heaters? 8.9VDC seems a bit more reasonable.

Originally posted by Dave Cigna
Concerning the output stage, I don't see your selection of cathode resistor(s), B+, and operating current jiving, but maybe you're looking at different curves. When looking at plate curves, don't forget that you need to subtract the cathode voltage (drop across the cathode resistors) from the B+. I might suggest aiming for an operating point around 220V @ 40mA, which, according to the plate curves, requires around -35V bias. The required cathode resistor turns out to be just about what you have...

-- Dave

Yeah, I forgot to do the subtraction. On a similar note to the above, would fixed-bias derived from a little 24V 1A toroid work OK? I'm guessing that it'd need to supply ~40mA to each cathode at -35V?

Thanks for the ideas!

[Dave Cigna]

Originally posted by sorenj07
In order not to have to drop down to 2V from 420VDC rectified, could I instead rectify from the heaters? 8.9VDC seems a bit more reasonable.

You want the tail resistor to be as large in resistance as possible. The more negative voltage the better.

would fixed-bias derived from a little 24V 1A toroid work OK? I'm guessing that it'd need to supply ~40mA to each cathode at -35V?

You could use the toroid, but if you already have a negative supply I don't see why you should bother...

Negative fixed bias would be applied to the grids and draw very little current. The cathodes would be connected directly to ground. Look again at Poinz's circuit.

--- Dave

[sorenj07]

Originally posted by Dave Cigna
You want the tail resistor to be as large in resistance as possible. The more negative voltage the better.

OK, that makes sense. The more resistance in the tail, the more it acts like a constant current source or something, right? I read that somewhere..

Originally posted by Dave Cigna
Negative fixed bias would be applied to the grids and draw very little current. The cathodes would be connected directly to ground. Look again at Poinz's circuit.

--- Dave

Yeah, I've since reviewed a Poinz's circuit as well as a bunch of others after searching "fixed bias schematic" on google. Fixed-bias might well be a good approach but I figure that cathode bias with a bit of a higher-voltage PSU (a 273BX choke-input, 350V 240mA) should work OK after dropping 45 to the choke and 40ish for the cathodes.

[planet10]

Originally posted by sorenj07
OK, that makes sense. The more resistance in the tail, the more it acts like a constant current source or something, right?

Exactly

dave

[sorenj07]

OK, I finally knocked together a schematic. I already have the 270FX, 193J and 156G's, and the toroid. The PSU splits between channels after the pi filter. The OPT is 6K6 a-a. The Amveco wired backwards should put out between 68 and 72VDC, I calculated the higher one on a house AC of 122V, especially since it's got virtually no load on it. I assumed 2mA of current to bias each tube, I saw that in another 6EM7 schematic I think.

[mach1]

Alternatively, consider using an LM334 current reg http://www.national.com/pf/LM/LM334.html under the long tailed pair. They work very well at these low currents without the need to provide a negative voltage rail.

pm