Push/Pull Plate Load Choke - how to figure Inductance?

[rongon]

Question:

If I have a choke with a center tap, I can use that for a plate load choke for a long-tailed pair stage used in class A only, as below.



Correct? OK, so...

In class A operation, should I look at the choke load for each triode from the choke center tap to the plate of that tube? In other words, if the inductance measures 100H from plate to plate, does that mean each tube in the p-p pair sees a load of 50H?

Thanks in advance for any guidance...

[Yvesm]

No, inductance ratio (when coupled together) are like impedance ratio: they change as the square of the turns ratio.
So, each half winding (1/2) shows only 1/4 of the total, that is 25H.

Yves.

[bob91343]

Rather than call it a choke, a better term might be autotransformer. The trouble with a circuit like that is the fact that the impedance drops linearly as frequency drops, so for the bass it's a rather low impedance. You have to make sure it's high enough at the lowest frequency of concern. At the other end of the spectrum you also have to make sure the distributed capacitance is small enough to push the self resonant frequency above the highest frequency of interest. That's why the details of winding techniques are important. Also the core losses will vary with frequency, another factor to annoy the designer.

[Michael Koster]

Originally Posted by Yvesm

No, inductance ratio (when coupled together) are like impedance ratio: they change as the square of the turns ratio.
So, each half winding (1/2) shows only 1/4 of the total, that is 25H.

Yves.

Ah but that's not the whole story... There's something to be gained by having the 2 coils on one core...

There are 2 ways to calculate it. One is that with the 2 coils in parallel, the inductance is indeed 1/4 of the 2 coils in series. NB That's with either one coil or both coils in parallel. Since they're wired and driven out of phase, the 2 tubes are also driving in parallel. So even though it's 1/4 the inductance, the 2 tubes Rp are driving in parallel. Each tube in effect sees an inductive reactance of 1/2 that calculated for the 2 coils in series.

Another way to calculate it is with the inductance of the 2 coils in series, driven by the Rp of the 2 tubes in series.

Either way, you get each tube seeing 1/2 the inductive reactance of the series connection. Both methods yield the same Fc.

All this applies of course to class A operation only.

Cheers,
Michael

PS Something in the above may not be obvious to everyone. Why does putting 2 identical coils on the same core in series result in 4X the inductance and putting them in parallel result in no change of inductance? According to Faraday, Inductance is blah*N^2/blah where N is the number of turns, IOW the inductance increases as the square of the number of turns. Putting the coils in series is 2x the turns, 4x the inductance. Putting 2 coils in parallel neither adds or subtracts from the number of turns, same inductance as one coil. Because the coils are on the same core.

[SpreadSpectrum]

I have never used a plate choke in an application like this, but it seems to me that you would almost need a power tube to get decent distortion at low frequencies, since reactance will be so low at low frequencies. Do people get good results with these?

[Michael Koster]

Originally Posted by SpreadSpectrum

I have never used a plate choke in an application like this, but it seems to me that you would almost need a power tube to get decent distortion at low frequencies, since reactance will be so low at low frequencies. Do people get good results with these?

This is actually a good application for a plate choke, because the DC current can be balanced and the choke can be on the order of 300H. This works really well with triodes having 2K or less plate resistance (ECC99, 5687, 7119, D3A, 5842, 6C45pi).

Low frequency performance is outstanding, corner frequency of something like 2-3 Hz, which means that distortion at 20Hz is not starting to increase much yet due to finite inductance. Each tube sees 17K ohms inductive reactance at 20 Hz for 270H total inductance.

High frequency performance is excellent, with the driver going out well past 50 HKz.

This performance is in part due to the high quality of the Lundahl LL1667 choke.

[SpreadSpectrum]

My current project amp requires around 200Vrms to drive the output stage into clipping. I am currently using CCS plate loads for the driver with ~800V B+. I'm getting more than 1% second harmonic at 1W due to the fact that a differential stage with CCS plate loads is hard to balance AC-wise.

I was thinking maybe a plate choke where the two sides are coupled together might help balance this out, but I can't think of a way to make this work well. A tube capable of putting 200Vrms into 17k (with comparable distortion to what my 6SN7s are putting out now) would be monstrous and probably doesn't exist, plus it looks like gain would vary quite a bit with frequency due to changing load impedance. I think I am gravitating towards an adjustable first stage(differential) load imbalance (a pot) even though I would love to have the circuitry be self-balancing.

[astouffer]

One industrial model McIntosh used a center tapped choke to drive the output tubes. The model number escapes me now but the output tubes were hi MU types like the 572B.

[rongon]

Originally Posted by Michael Koster

This is actually a good application for a plate choke, because the DC current can be balanced and the choke can be on the order of 300H. This works really well with triodes having 2K or less plate resistance (ECC99, 5687, 7119, D3A, 5842, 6C45pi)...

Wow, I like it! I like how you used the negative supply so that you don't have any dangerously high voltages around. But the input needs to be blocked from DC, right? That's why the input transformer...

That amp's a lot like a two-stage, DC-coupled, push-pull 2A3 amp I built. It's my current favorite, but only makes 6 watts class A each side. I was thinking of making another one but with 300B outputs.

My amp has the output stage power supply "stacked" on top of the driver stage power supply (say if the driver stage is +200V, then the output stage 'ground' is referenced to that, so floats at +200V). I think the output stage B+ is about +300V from the 2A3 plates to floating ground, so the whole stack has a total of +500VDC to earth ground. It would be quite a bit higher than that if 300Bs are used. Driver is a 6H30Pi long-tailed pair with an unidentified PP OPT (designed for PP 6BQ5) as the PP plate load. There's a -30V supply to feed the tail of the 6H30Pi. I think there's a 4kohm resistor in the tail, which should definitely be made larger in value. Perhaps an LM317T or some kind of FET as a CCS...

What kind of line stage or source do you use to drive that amp? I have two line preamps I'm using, both common-cathode stages with input selector and volume control; one has a 5687 running at 20mA per triode, the other a 6DJ8 at 8mA per triode.

I had made a Mullard 5-20 style amp and found that it was way too sensitive for my use, so I figured that if you're going to have three stages anyway, why not take the voltage amp out of the 3-stage power amp and make it a medium-mu triode line amp, and have that drive the volume control and (2-stage) power amp input? After DC-coupling the 2 stages in the power amp, that leaves only the one cap on the output of the line amp.

Anyway, I didn't know anybody else was thinking along those lines. It doesn't seem like a popular way to go. I like the 'nothing wasted' simplicity of the approach, and I think it sounds good. Have you found advantages to your setup that you couldn't get before?

Best,

[rongon]

Another way to calculate it is with the inductance of the 2 coils in series, driven by the Rp of the 2 tubes in series.

So if the plate choke measures 100H from plate to plate, and I use a 5687 with 2500 ohm rp, I'd do my figuring based on L=100H and rp=5kohms?


--