Oscillation, eh?

Nah, it doesn't satisfy Barkhausen's Criteria.

Out of curiously, I hit it with a 1V 1kHz square wave in the simulator. There is not the slightest sign of overshoot or ringing, just a slightly damped square wave output.

I think the idea rather than risk adding another mistake, is to sort out the fading. An analogue meter on B+ could be useful. A slow motorboat causing AGC voltage to change can happen. However, we look at the PSU first and any series resistances in the "B" rail first. Also keep an eye on the heaters one could be faulty.

As before it is normal for pentagrids to drift high until they stabilise (RCA). So be careful that you are not chasing a "Red Herring".

There are 2 issues here, Marc.

1. The frequency drift - may be solved adequately with a replacement 6BE6. We don't know yet.
2. Inadequate audio gain to get satisfactory volume from the FM conversion which it seems only outputs 1 or 2 volts.

Hence my suggestion of adding some audio gain in the least disruptive way possible, and do so without degrading the AM performance.

You haven't seen the circuit I'm suggesting, it hasn't been put up yet. But the OP has.

Photo uploaded to Post 35.

I will check the valves and voltages as is, especially the 6BE6 screen grid voltage which Robbert reported can be high giving a hot valve.

Having a look at the circuit with 5Y3. That is back biased via R6, Grounding C17 to chassis will cause hum & voltage issues. And hooking things up inappropriately can destroy the sets overall bias.

I note on a Phillips set here from 1947, its back biased and has a magnetic pick up: It floats and is DC blocked, so it does not wipe out the bias of the first AF tube. So it really comes down to sorting out the set before adding anything to it. That way you are not chasing your tail, wondering which one is causing issues.

I don't understand what you are talking about, Marc.

I'm not suggesting anything that has anything to do with the rectifier, the back bias or anything of the sort.
Just adding some gain between the volume control and G1 of the output bottle. Like most other contemporary radios did.
So the FM feed will be loud enough.

Not a problem!

My interest here is how its coupled, & turntables and other foreign devices, can like the Philips need special attention when its back biased or an inductive source.

I would assume you are feeding into the output tube across the volume control. For full volume you will need about 1.6 or more swing volts (at the grid) to drive it. In the MA527 which has a pickup they added another tube.

You may need to get some appreciation of the new devices output, as you may have to DC block its earth as you are loading down its output and isolate its power by supplying it from a transformer type wall wart, or initially a battery & don't ground its PSU or battery, directly to the radio.

I note on US forum, that with some couplings of devices to tube radio they isolate the headphone outputs which often need to be inductive, using transistor radio output transformers, especially to recombine stereo.

You can use an oscilloscope to measure the PP amplitude (Voltage) preferably from a signal generator to establish just how much voltage swing you have to put into the grids divider (Grid leak & stopper) To get the volume you want, then see what you have to do to achieve it.

OK, Marc and Graham:

First, and I had assumed this was obvious, you'll need an "AM / FM" switch.

This will break the circuit at the top of the volume control.

When switched one way, it connects the AM detector, the other, the FM receiver module.

That gives you an AM / FM radio with probably not enough volume on the FM.

This model radio working on AM is unable to drive its output valve to maximum, i.e. clipping, even at maximum volume pot setting. So the AM would benefit from a bit of extra audio gain, too.

The bottom of the volume control is ground referenced, so there is no problem introducing an extra gain stage, also ground referenced and powered from the 2nd B+ line, that is, the junction of R10 and C20 on the circuit.

The circuit, as posted here (see post 35), simply interposes between the volume control slider and G1 of the output valve, replacing the cap that's there now.

It is a classic series voltage feedback common emitter amplifier using a high voltage NPN transistor. It has a voltage gain of 10, determined by the ratio of the (unbypassed) emitter and collector resistors, and is biassed for half the B+ voltage. I'm assuming a B+ of about 200 volts, but that's not critical, the circuit will work on anything between about 50 volts and 350 volts (for the 2N6517)

A bootstrap cap has been added to raise the AC load impedance above 1 megohm so as to not add an excessive AC load to the AM detector at high volume settings, which would cause distortion on AM.

None of this affects anything else in the radio. The circuit will draw about 1mA from the B+ line. This will drop the B+ at the junction of R10 and C20 by about 2 volts.

The 2N6517 will run very slightly warm, dissipating 100mW, but it is working well inside its 300mW rating.

Quite a debate, thanks for putting so much thought into this Marc and Ian.

My only reaction is, if the output valve isn't operating to its full potential in the original circuit, is it possible that the output transformer and speaker were taylored to this reduced output. If this is the case then adding the transistor preamp may over-drive them.

I have the radios now but the good weather has made outstanding gardening jobs a priority.

They would be the same speaker and transformer that were used in the 5 valve version. Changing them would not have increased the audible output and they are not "weaker" or more vulnerable in any way.

So no, not a problem.

Of course, now you will be able to turn it up to the point where the output valve clips, the sound will be very loud and the distortion will be unpleasant. You won't leave it there for more than a few seconds!

But if you are listening to a very weak AM station, or a quiet passage of music on ClassicFM, you'll find the extra gain is handy.

You may find you could even afford to lose some gain and improve the radio's distortion figure to approach HiFi standards by adding some negative feedback - only 1 resistor is needed to do that.

Ok thanks, so no risk to the output tranny and speaker.

Thw 6AQ5 circuit already has negative feedback to the 6AQ5 cathode from the output transformer secondary. How would that go with your suggestion re negative feedback from an additional transistor?

Still writing up the drift assessment, but getting late so will post tomorrow.

Conducted a listening test on these radios and one (white, for the record) has 6AQ5 output, with feedback from the output transformer; the other (brown) N78 output.

No instruments used, just by ear and eye.

The aim was to review their drift characteristics during the day to avoid fading. (Using the term drift to describe the loss in volume due to drift in the oscillator frequency {presumably}, rather than fading which is due to atmospheric affects at night).

The station used was ABC RN on 1431 the transmitter for which located at Windang on the eastern side of Lake Illawarra. So nice and close to me to the north and not subject to fading even at night - so volume change has to be due to drift. About a metre of aerial was used.

Drift Assessment

The white 6AQ5 radio did not drift as severely as I first remember, in that the effect was complete in about 8 minutes, and an off-tune output remained, the peak signal having drifted to a higher indicated frequency on the dial, by about 5-7kHz, as can best be estimated from the dial.

The brown N78 radio drifted also but only taking about 2-3 minutes, and again an off-tune output remained, the peak again at a higher frequency, again by about 5-7kHz. This is more manageable, as if the radio had been tuned to a particular station, then recovery from fading could perhaps be included in the warmup time.

The drift is about the width of the tuning indicator.

In summary -drift in frequency about the same.

- drift took 3-4 times longer with 6AQ5 white radio than with N78 brown radio.

The front ends of these radios are nominally identical, so not surprising that the drift frequency is the same. Issue is why take longer with the white radio?

Another observation was that with just the stub of the aerial wire, about 150mm ie low signal, the white 6AQ5 with feedback radio was audibly distorted, while the brown N78 without feedback radio was not. Something going on here with low signal. AGC issue? Feedback circuit issue?

So some things to sort out to start with.

Re the distortion on a weaker signal (i.e. little or no AGC being developed) is probably due to the lack of standing bias on the front end or the IF stage. It's a design flaw. An oversimplification of the design.

I'll have a look at it and suggest a way of improving it. Later AWA models using the 6BV7 did just that.

It's been suggested that the 6BE6 could benefit from a lower screen voltage / current, causing it to run cooler and probably mitigating the drift.
I agree with that suggestion, it makes a lot of sense.
I'd try doubling the resistance of the screen dropper for a start.

The distortion wasn't a real practical problem, it being a city radio. A metre of wire and there was no distortion apparent.

I do recall 6BE6 screen voltage was high and and plate voltage a bit low originally. Component values were OK. Tried cleaning and tightening up the pins and valve base back then, which made very little difference. I didn't consider increasing screen resistor. A new 6BE6 might make the difference if that fails.