Hi Guys:
I had posted this at the END of Badrestorer's Philco-rebuild..where this info came from.. but I've just removed it and re-posted here to start a new thread and where it may find more informed replies.


Hi you guys:
I've been trying to figure out this screen-inversion thing.
Can anyone explain the circuit?
It looks like 165v on the upper screen grid and the full 180v on the lower screen grid.. is that right?
So the 3900 ohm resistor drops 15 volts on the way to the upper tube?
Why did they do this? To isolate the feedback signal to the lower tube as coming ONLY from the upper tube?
Seems odd to me.
Why not feed the two screen grids separately using separate dropping resistors? Wouldn't that keep the lower tube output from ever feeding back via the .01 cap to it's own grid?

Also.. why if this circuit works and since it's so simple and easy.. why wasn't it used everywhere for push pull instead of feeding the two output tube grids from the conventional preamp inverter stage?

There has been quite a bit of discussion about the screen-grid inversion circuit on Philco Phorum. It works like this:

1) Audio signal is applied to the grid of the upper output tube.

2) The 3900 resistor is a load resistor for the screen grid of the upper tube, which serves "double-duty" -- a) as a screen grid for the plate, and b) as a second "plate" for the cathode and grid.

3) Because of the load resistor, the voltage on the screen grid will vary inversely with the voltage on the grid. This voltage is capacitor-coupled to the grid of the lower tube.

The purpose of the circuit was cost-reduction, for which Philco was famous. It eliminated the need for a phase inverter tube. Back in the day, these guys would kill to save a tube.

The disadvantage was that it impacted the symmetry of the push-pull circuit, imposing a degree of imbalance. The distortion reduction and common-mode hum rejection advantages inherent in a push-pull output stage would not have been as good compared to a traditional circuit using a PI stage. That's why the higher-level sets didn't use it, employing a PI stage instead. For example, the 41-250, 41-255, 41-280, and 41-285 use screen-grid inversion, while the 41-290 uses a PI stage.

As to why the circuit wasn't used by other manufacturers, I suspect that Philco protected it very strongly via patents.

I have a 41-250 using this circuit, and it does sound very good.

-David

Excellent description, David. I believe you are right about the patents.

Here is a description regarding this and other inverter circuits you run across. http://www.radiomuseum.org/forum/phase_inverters.html

I don't see anything that would cause a screen grid inverter to be harder on the audio output transformer. Maybe more likely, since these were not top end radios, Philco did a little cost cutting on the audio output transformer too which led to a bit of a reliability problem.

Curtis Eickerman

Agree, half a**s design and quickly falls apart when output tubes age differently.

Our Family set was a Philco 37-650, and possibly one of the fist models to have the screen grid inversion. I couldn't figure it out, because in my youthful eyes, there was no PI....but one of the the kids my classes pointed out this unique inverter circuit. We never had a problem with the output transformer failing, so I'm not convinced that the inverter circuit was the culprit.

Stretching the envelope here a tad, a bad imbalance in 6F6 plate current MIGHT cause failures in output transformers. Philco has a good reputation for few, if any transformer failures. Remember that their circuit design was done by Hazeltine Corp., for the most part, and that the funny inverter here could have been their idea.

But say as any of us will, the damned things sounded good, especially if we never knew about that inverter in the first place!

The only disadvantage that I can see with the screen grid inverter is that the output stage must run Class A rather than Class B. So, efficiency is not as good. On the other hand, I would expect fidelity to be better for that reason.

Philco came up with many ingenious designs over the years. I wouldn't dismiss this one as "half a**s" without first making an attempt to understand what their design goals were.

You know, that's an excellent point that didn't occur to me. Since both output tubes would have to be in class A (i.e., drawing current throughout the full audio cycle) as opposed to class B (each tube only drawing current during it's respective half cycle), there really IS somewhat more stress on the audio output transformer.

Maybe there really is something to the idea that these radios lost output transformers at a higher rate.

Class B push pull stages should still have good fidelity, but with slight bias problems it would be easy to pick up cross over distortion (distorting the audio at the zero crossings). If they kept the tubes closer to AB1 or AB2 rather than B that was minimized greatly.

Curtis Eickerman