Home Built TRF radio

[davidgem1406]

This is the third posting in the series for this radio, links to the first and second postings are shown below.

Link to first posting August 2011: Radio Identification Required

https://www.vintage-radio.net/forum/...ad.php?t=73641

Link to second posting July 2019: Unknown Home Built TRF radio

https://www.vintage-radio.net/forum/...d.php?t=157759


Third Posting May 2020: Home Built TRF radio

I have again (another of many times) picked up this radio in a further attempt to discover the reason for the distortion that is present on the audio and is suspected to be RF coming through with the audio.
This was first mentioned in the second posting.

This time I made some changes to the HT rails as I could see no reason why the HT for V2 and the screen of V3 should need to be be going up and down with the changes of the gain pot.
I therefore added R10b and C16 and moved the HT feeds to V2 and V3 screen to that point. Leaving R10a to feed V1. (See the attached modified circuit diagram with voltage data added).
That change made no difference whatsoever to the working of the set, nor to the distortion we are looking to cure.

I also removed V1 and connected an audio source in the way of a Sony Walkman to the grid of V2 to check that audio reproduction for both stages was satisfactory, it was.
Also, I used a modulated signal generator as the source but it was rather difficult to decide if the RF breakthrough was present although I thought that it was. It was inconclusive.

At a later time I decided to use the signal generator but modulate it from an external audio source from the Sony Walkman, therefore simulating something closer to that of a radio station.
It actually worked very well. The radio audio was much improved but still had the background noise, and some audio distortion, that seems to be caused by RF breakthrough.

My assumption has always been that the combination of C8, C9, C10, R8 are or should be responsible for removing the RF content. But playing with the values of those capacitors results in either an increase or decrease of the HF end of the audio, but no change in the suspected RF content.

I am now right out of ideas as to what this problem could be, or what to try next. I have tried every option I can think of.

However, I do wonder about the change in HT voltage to V1 with a 125V change with use of the gain pot, that’s a big change.
Maybe the pot should be of a lower value that holds the HT rail at a lower value anyway so that the change was smaller. That then raises a possible current rating problem for the pot, and that maybe could explain why the original was a wire wound type, but that could equally well be just what was available at the time of building.

All the component values used are those that I recorded when stripping down the radio, other than the value of the gain pot. The original was a dual track wire-wound but the tracks were damaged and a
value could not be read. (The equivalent circuit of the pot is on an earlier posting).

I have the circuit for the Premier AC TRF, that is identical, although no values are given, in all respects other than one in that C10 in my unit does not exist in the Premier.
I cannot imagine that the Premier had the problem that mine has, so I guess it is something pretty obscure I am looking for.

Attached are:
Updated wiring diagram
Chassis underside layout drawing
Photograph of the chassis underside (To show the actual layout)

[PJL]

It's an anode bend detector and these do cause distortion particularly with low signal levels. I am not too familiar with pentode versions but yours seems to be running with quite high V2 cathode voltage and anode current. Below is an extract from a Champion circuit diagram. Your design appears to be lowering the anode current by fixing the g2 voltage with a potential divider but this does have the disadvantage that it will vary with HT volts so first thing I would try is the champion g2 bias arrangement.

[davidgem1406]

Thank you PJL,

I will certainly try that, then post the result back here.

Regards Dave.

Quite popular in the days of "valve tax, so much per valve socket" you got gain and detection all at once. The days before multifunction valves, diode triode for example.

[davidgem1406]

Hi PJL,

I looked up the circuit for the Champion radio it looks to be the Champion Model 784.
What I did was to rehash the V2 circuit to match that of the 784 as best I could without replacing all the components.

I disconnected g3 from the cathode and grounded that, g2 removed the lower resistor of the divider chain and replaced the upper resistor. Shunted the 25k gain pot to 10K, also bridged R1 to 33K and other resistors to near the Champion values. Added the 10pf capacitor between g1 V2 and the anode V1.
Also, I removed C10 and shunted R8 to 500K (Ref my circuit) to look like the Champion 748.

The sum total of that was just a reduction in gain, no change to the distortion problem at all.

Any ideas on what to try now?

Regards Dave.

[PJL]

What are the cathode and anode voltages with the new circuit? The idea is to run the valve close to cut-off so only the top half of the RF signal is seeing any significant gain. It may simply be that you need a much larger signal to get rid of the distortion.

[davidgem1406]

Hi PJL,

I am using the strongest signals obtainable here that are normally loud and clear.

These are the voltages for the modified V2 circuit.
1/. With the HT circuit as original (same as Champion)
Anode full gain 90V - zero gain 140V and Cathode full gain 1.5V - zero gain 2.2V

2/. With the modified circuit as per my diagram.
Anode full gain 150V - zero gain 155V and Cathode full gain 2.3V - zero gain 2.4V

With a change from signal to no signal (aerial or no aerial) there is negligible change.

The loss in gain that was found with the champion circuit is due to the HT rail for V2 going up and down with the gain control. Returning it to my modified circuit increases the overall gain and lost stations are returned.

The distortion we have is considerably reduced with speech only, and is almost acceptable. Whilst on music it is considerable. This is the case with either circuit.

Hope that helps,
Dave.

[PJL]

You seem to have gone a bit overboard with the Champion changes as the idea was to concentrate on V2 and try to get the operating conditions right. Go back to your circuit attached above, The only change I suggested was to copy the Champion bias arrangement for g2 but you should also remove C10.

[Sideband]

Question, does the valve type matter? I understood that some valves are better as anode bend detectors than others. Some time ago I used an EF80 with great success but had trouble with other valves.

[Keith]

Have you tried injecting audio (possibly from your Walkman) into the grid of V3 (junction R7/R8) to check that the distortion is not coming from the output stage?

[PJL]

Audio was tested by injecting into V1 (see post #1).

The 6SJ7 is one that is commonly used and even includes 'biased detector' as one of it's uses in the datasheet. The Champion uses a 6AM6 so it is different.

However, it is possible this particular 6SJ7 is not behaving itself.

[Sideband]

Originally Posted by PJL

The 6SJ7 is one that is commonly used and even includes 'biased detector' as one of it's uses in the datasheet. The Champion uses a 6AM6 so it is different.

However, it is possible this particular 6SJ7 is not behaving itself.

If it's an old used one, the grid base (the point of cut-off) may have changed. Might be worth trying another.

[Keith]

Originally Posted by PJL

Audio was tested by injecting into V1 (see post #1).

OK, didn't spot that (although I think you meant V2). I guess the only scientific way to proceed is to inject a very low level sinusoidally modulated signal into the input and trace it through with a scope to find where the distortion is being introduced.

[davidgem1406]

Hi all,

I did check for distortion in the output stage on its own at V3 g1, also from g1 of V2 using the Walkman, in both cases the audio was fine.
Later I used the Walkman to modulate the signal generator and injected it into the aerial socket it was better but still showed signs of the distortion.

The 6SJ7 was replaced using another with a better M/AV and emission, the result was no change.

PLJ,

I did change just the bias arrangement to g2 of V2 to start with, that having made no difference except for a reduction in gain I made the other changes.
I will however put it back as the original, except the bias for V2 g2, and then check the V2 voltages again.

Regards Dave.

[PJL]

I did go looking for a detector circuit that used the 6SJ7 but could only find ones that were regenerative. I assume when you injected modulated RF that the signal level would have been higher than when receiving via the aerial? Modulation depth will also play a critical role in distortion.

Keep the Champion 3M g2 feed and try raising the anode load to 1M. Any attempt to move the bias towards cut-off to reduce the distortion is likely to reduce the V2 gain but that may be a good thing as it will demand a stronger RF signal.

[davidgem1406]

Hi PJL,

I have now returned to my original circuit, except the V2 g2 is as the Champion and C10 remains removed.
The voltages at V2 are as follows:

Anode full gain 70V - zero gain 76V.
Cathode full gain 4.6V - zero gain 4.7V
There is only a small variation due to my HT mod to stop the V2 and V3 HT from having a large swing from gain pot adjustment.

I will try changing the anode load to a 1M but as the gain is already reduced if there is a further reduction it may be a step to far even with the strongest stations.

I will return with the result.

Thanks, Dave.

[davidgem1406]

I have now changed the anode load for V2 to be 1M, the gain remained OK, or maybe slightly more. However, the 6SJ7 became very micriophonic so it was changed but the replacement was also suffering with it but not to the same extent. Neither had that previously.

There was no change in the distortion.

Thanks, Dave.

[PJL]

4.6V across 27K is 170uA which sounds far too much.

My last suggestion is to put back the 330K and replace R6 with 100K.

[Herald1360]

Originally Posted by Sideband

Question, does the valve type matter? I understood that some valves are better as anode bend detectors than others. Some time ago I used an EF80 with great success but had trouble with other valves.

EF80 is a "straight" valve not a vari-mu type. This will make its curves near cutoff more suitable for anode bend detection linearity.

[PJL]

The 6SJ7 is also a sharp cut-off valve.

I wonder if there is a problem with the 22uF cathode decoupler (C7)? My last+1 suggestion is you put a 0.1uF in parallel with C7.