PY 81 as power rectifiers?

[ppppenguin]

If that method was really good and useful it makes me wonder why it wasn't more widely used. There are huge numbers of valve radios and amps with SE output stages but very few have this refinement. I know that cutting cost matters a lot but this could be implemented inexpensively. Certainly without the extra valve and phase splitter needed for a PP output.

Perhaps you'll find out.

[pwdrive]

Here's some stuff on the hum cancelling audio output transformer from one of my books, that method was very common back in the day and was used by most of the radio manufactures at one time or other, the extra number of turns required are small compared to the rest of the primary winding.

Lawrence.

   

[Kalee20]

Hi Amie,

Theory predicts what you suggest, won't work.

You can indeed get currents flowing in opposite directions in the two primaries, which will at least partly cancel out the standing magnetisation in the core. However, the output valve is trying to put audio voltages across one primary, but the HT capacitors across the other primary are trying to keep the voltage here constant. So the poor little transformer would be beset by two fighting forces.

There is a technique as Lawrence above has posted, where the windings are unequal, and the 'extra' winding has a resistor in series (most important as it allows some audio voltage to appear across this winding). By getting the value right, it is possible to completely cancel any hum on the HT rail from appearing in your speaker.

Jeffrey asks why it is not used widely? I'm thinking that it was a Philips patent, and they only allowed a few other companies to use their idea.

Amie, if you start experimenting, good luck, trying something in practice and proving or disproving in ten minutes is sometimes quicker than sitting with pencil and paper with a wet towel round your head for an hour. My own technique is to use whichever method is likely to get me the answer fastest!

[Mike Watterson]

The audio current is irrelevant as that is AC.
The DC and AC voltages are also irrelevant.
All that matters is the DC currents (especially as it's Class A). So feeding rest of set from one winding and output anode from the other does work. Obviously it works best if the ratio of the windings equals the ratio of the currents.

I'm sceptical that the idea was patentable, though lots of things that should not have patents did and do. I've seen this on a number of brands older radios. It's arguable if it saves much money or reduces hum much compared with suitable R & C after main cap. I'd think it fell out of favour when Electrolytic capacitors got bigger & cheaper, resistors replaced chokes and suitably gapped SE transformers were cheaper. Earlier there were Field coil Electrodynamic speakers with hum bucking windings, so only saturation of O/P TX core rather than hum would have been the issue.

I'd think it wasn't used widely because there was little advantage to it then compared to now.

It's a very useful technique today for SE class A as it allows use of a mains transformer. The DC current to rest of set vs the Class A quiescent current can be balanced with a shunt HT to 0V resistor or additional taps/windings on the O/P primary or rest of radio primary. The CT goes to HT as the DC current of the rest of set is in anti-phase to O/P valve, as in case of the twin quiescent currents of Push Pull Class AB or Class B. Class A on Push Pull is wasteful.

[Kalee20]

The audio current is irrelevant as that is AC.
The DC and AC voltages are also irrelevant.
All that matters is the DC currents (especially as it's Class A). So feeding rest of set from one winding and output anode from the other does work. Obviously it works best if the ratio of the windings equals the ratio of the currents.

I think Amie meant something like in the sketch.

Then as you say, the currents flowing in opposite directions will cancel out the magnetisation in the core. And if the numbers are right, complete cancellation will occur.

But as soon as you put a signal into the output valve, you will get audio voltages across winding A (which is the whole point of the stage). Transformer action will then induce audio voltages in winding B, which will be fed to the rest of the circuit.

This could result in weird feedback effects. Or, the 'rest of circuit' could load the transformer leaving precious little remaining for the speaker.

[Amie]

Thanks for the comments everyone.

So, leaving the power supply, (and the topic), for the moment, I've cobbled together a reasonable copy of the Grundig set's audio stage, but using a PCL80 and the mains transformer, (6W 0-115 0-115. 0-6 0-6. From RS.) for output.

Boy, did it oscillate well!

Got the NFB wrong way, also had to play with the pentode biasing. Grundig arranged for negative grid bias, I had to provide it separately.

Without anything on the other winding it did work, OK, but the problem was poor frequency response, and clipping at low power, couldn't do much to improve that.

Putting a load on the other winding to draw simulated "balance" current improved things remarkably, what surprised me was how little it took, about 4mA....as against the 16mA standing pentode anode current in the circuit.... You would think optimum would be Ia=Ib....

I don't know what tradeoffs are with the effect....

I was doing this with a resistive sink load on the second winding.

What happened with a 47uF capacitor from transformer balance winding to chassis?

It made it worse.... It does seem to be loading the output of the valve, with increased Ia, and increasing the low frequency "roll off" frequency, by killing the inductance no doubt.

....A well deserved "told you so" from Kalee 20.

I haven't tested for hum cancelation yet, its running on my stabilised supply at the moment....

Seems like a good way to utilise a cheap mains transformer in an SE amplifier though, as Mr Watterson said....

It was giving 600 to 700mW clean output from the rats nest lash up.

Soon have to start the presentable version 

Thanks,
Amie.

[Kalee20]

There's no valve as the PCL80, do you mean ECL80? (It does actually draw 300mA heater current so justifiably could be labelled PCL80, but the 'voltage' letter seems to dominate in the numbering system).

If so, it's inconvenient to bias it, with its common cathode. A somewhat awkward valve!

Any chance of a sketched circuit diagram? Would help lots!

Congratulations so far BTW!

[Amie]

Yea, ECL80, that's what I meant to type!

Will do a diagram and post later on, weather is too nice for indoors here at the moment!

Thanks,
Amie.

[Terry]

The ECL80, like the ECL113 in your Grundig, is unusual in only having one common cathode for two sections with entirely different characteristics. It is common practice in Triode-Heptode and Triode-Hexode mixer/oscillator but otherwise these are oddities.

If you have an ECL82, ECL83 or ECL86 to hand, these all have separate cathodes for the two sections and can be separately biased as appropriate.

The method used by Grundig is a hangover of the days of battery valves. Here, the filament is also the cathode and can't be separated, so grid bias was the only solution. Usually, this was done with a separate battery of 6 cells, each with a brass tube to accept a wander plug. These were marked 0V, -1.5V, -3V and so on up to -9V and it was common to have several leads coming from the radio. One, of course, was ground and connected to 0V and the others were plugged into various negative voltages to suit the bias requirements of the individual stages.

Then someone came up with the idea of 'auto bias' which was a chain of resistors in series with the negative return to the battery to provide these voltages without the need for a third battery. This continued throughout the life of the  battery valve radio right up to the time the transistor reigned supreme.

Some manufacturers, though, continued to use this scheme for many early mains valve receivers rather than the 'new' method of cathode bias. The HMV radio I grew up with used this method despite being a relatively mature design from about 1938.

Many valves you encounter will use the Pro-Electron coding system which tells you a lot about each valve type. You may well be aware of this already but may still find this look-up table useful.

There is one omission I've spotted which you should be aware of which is an 'O' for the first letter. It means zero volts, aka a cold cathode device such as a voltage regulator but it was later applied to transistors when they came on the market, which was quite logical.

Diodes started with OA, which is obvious and, because a transistor has three connections, it was treated as a triode - hence OC. The numbers were then allocated to indicate the type of device such as AF, RF, Power, etc.

Eventually semiconductors were given their own numbering system as the types and uses proliferated.

[Terry]

Returning to the experiment with the output transformer, I think having an rqual number of turns on boyh halves of the primary are likely to bring about its downfall, especially judging form what the effect of the 47uF capacitor was. On the other hand, perhaps inserting a normal resistor at this point (as if it was a conventional output transformer and smoothing arrangement) might turn this experiment into a miracle.

We will have to wait and see.

Looking at the rain radar, unless you are well to the north and east of the county, your nice weather seems to have been rudely interrupted - we've had quite a downpour here in Lincolnshire as well, we just didn't get the 'nice' before it!