Invicta 33...volume won't reduce to zero

[Tim]

So something else is doing the detecting

That's more or less the conclusion I came to. But What?
Could that be the valve base? I wonder if there may be some stray copper oxide somewhere?

I will connect another base and valve to the radio, and isolate the original. it's about the only thing left!!

2nd IF coil both halves check out for continuity.

[Tim]

That's more or less the conclusion I came to. But what??

BTW. Sorry, C20 on the diagram shown.(Different number on the trader sheet.
Yes, I disconnected that. No discernible difference.
A few PF as I remember.

[Tim]

Hi PJL
I shorted C20 in YOUR diagram today(22pF), and signal disappeared. No signal at all though, even at full volume.(Trader c21)

I connected a 6Q7 instead of the EBC41 and problem still there, so the holder isn't to blame. 6Q7 was a metal valve, and signal went v quiet when I touched the outer casing of valve. Does the EBC41 have a screen?

[Neil Breward]

Originally Posted by Tim

Hi PJL
I shorted C20 in YOUR diagram today(22pF), and signal disappeared. No signal at all though, even at full volume.(Trader c21)

Er, shorting out C20 will put HT on the AGC diode - not really a good idea, and will bias off the EBC41 by raising the cathode voltage, hence no signals!

Isn't this circuit simply a flawed design? With the negative feedback loop attached to a non-decoupled cathode, both the detector and AGC diodes will produce AF signals at the EBC41 cathode, irrespective of the volume control position, and give detection efficiency which varies with the AF signal - again not a good policy! Surely better to bypass the cathode, connect the detector load to cathode, not chassis, and connect a 470 ohm resistor or so in series with the botton end of the volume control. Then connect the feedback loop to this junction rather than the cathode. Not original, but it should work!

Cheers,
Neil

[Tim]

Thanks Niel
I might do that.
The audio is quite loud though, clearly audible across the room. Has anyone else got one of these sets, and if so is it the same. I know Invicta were a budget model, but this is pretty poor.

[Steve_P]

I wonder if it's getting into the HT Rail and being detected by the Output Stage.

If you haven't changed it already, change C40. Then, try fitting a choke between R25 and the link to the Control Grid of the O/P Valve.

Just an Idea. Using Trader Sheet 1093.

Cheers,

Steve P.

[Tim]

Thanks for the suggestion Steve. I had considered that, and checked with a scope. HT has 20 volts p-p ripple(first smoothing cap), 2 volts p-p ripple(2nd smoothing cap) and more or less flat DC on the junction of the triode anode resistors(smoothed by 2uF). No sign of any RF, but I get what looks like some local oscillator.

This is why it's driving me nuts!

I fear Neil and others may be right, and this is simply poor design. I think we have tried everything else!!

[PJL]

Tim, the 'oscillator noise' is probably IF injected by the AGC diode...same design issue as the detector.

Can I suggest the following with the volume set to lowest:
1. Short the AGC diode to chassis.
2. Short the detector diode to chassis

Can you also confirm that when you turn up the volume the sound output does not initially reduce then increase...I am thinking it might have had a special volume pot that didn't start at zero.

[Steve_P]

It wouldn't take much signal to generate a listenable sound! If you set the volume to zero and take the valve out, (It's E range so should work!) do you still get signal?

If so, fit the choke and try it.

Cheers,

Steve P.

P.S. Have you tried a new smoothing cap?

[paulsherwin]

Originally Posted by Tim

I fear Neil and others may be right, and this is simply poor design. I think we have tried everything else!!

Quite a few radios of this era did make it into production with design faults. I'm thinking of the Pye P43 which hums quite badly even with all the valves removed because of the location of the mains and output transformers

Paul

[Tim]

Hi PJL

Can I suggest the following with the volume set to lowest:
1. Short the AGC diode to chassis.
2. Short the detector diode to chassis

Doing either reduces signal slightly(and introduces sibalance in the first case) but does not eliminate it completely. The set is completely silent with the EBC41 removed.

Volume increases smoothly, I have tried another control in any case.

If you haven't changed it already, change C40. Then, try fitting a choke between R25 and the link to the Control Grid of the O/P Valve.

Hi Steve
What sort of value would be suitable?
I have some RS RF chokes somewhere.

[PJL]

So the AGC is also serving as a detector...not all that surprising as it is putting IF on the cathode.

My favourites are:
1. The AGC is not working properly - this design will be very dependent on the correct signal levels at the AGC diode and detector.
2. There's a wiring error somewhere.

[ukcol]

As has been suspected this design is likely to be prone to this symptom.

Originally Posted by Tim

Shorting out either diode makes no difference

Note that in the following text component ref numbers are as Trader Sheet 1093.

The signal fed to the AGC diode is with respect to chassis, which puts R13 (the AGC diode load) in series with R20 (V3 cathode resistor). The current in these resistors will have an audio component, just as with the detector circuit proper. I suspect this is where your rouge audio signal is coming from.

Of course, if you short circuit the detector diode this signal source will still be valid.

If however, you short circuit the AGC diode the rouge signal cannot come from this source but there will now be no AGC and the signal at the detector diode will be larger. The detector diode load is returned, as it should be, to V3 cathode and not via R20 but the audio current flowing in the volume control track DOES flow in R20 and this current is now bigger due to the loss of AGC.




You might like to try this small modification to the circuit (if like me your not a purest).

Disconnect the bottom end of R9 and R13 from R20. Connect the ends of R9 and R13 to the positive plate of a 25uF 25v electrolytic capacitor (value not critical).
Connect the negative plate of the capacitor to chassis.

Take a 22k resistor and connect it between the junction of the 3 components above and the top (V3 cathode) end of R20, from which you have just disconnected R9 and R13.

The value of this added resistor (22k) is chosen to be large enough not to upset the negative feedback circuit, and low enough not to upset the diode loads.

If you draw out the circuit of this modification I think you will agree that it removes any audio current in R20 from the detector and AGC circuits whilst maintaining the correct bias conditions for the diodes and correct negative feedback via the tone control circuit.

[hondacxrider]

Hi, didn't I read somewhere that a dry solder joint can act as a detector/diode. Worth checking maybe?

[Neil Breward]

Hi - Sorry - with my suggested modification to the negative feedback circuit you would need to change the phase of the feedback!

Cheers,
Neil

[Tim]

Hi all

Solved!! Well mostly. I tried UKCol's modification but it had no effect. Even changing the value of the 22k resistor didn't help. In the end I disconnected the "bottom" of the volume control from chassis and connected it instead to the junction of R9/13/20. The audio is now down to a whisper, the volume control works fine and the tone control has some effect. I haven't tested it for very long but it seems fine. I suppose the AGC may affect the cathode voltage on the valve, but it does that anyway. I may include a capacitor (1n?) between the volume control wiper and the grid of the EBC41, to prevent any DC flowing into the grid of the valve. Can anyone spot any disadvantages to connecting the volume control this way?

Thanks to everyone for their help, but I am going to quit while I am ahead!!
And thanks for making an interesting thread. I have certainly learnt a few things!!

[ukcol]

Originally Posted by Tim

Hi all

In the end I disconnected the "bottom" of the volume control from chassis and connected it instead to the junction of R9/13/20. The audio is now down to a whisper, the volume control works fine and the tone control has some effect.

Hi Tim

I think your set must be different to the published circuit in some way, connecting the volume control as you have should remove the grid bias for V3 and produce distortion.

But, as you say, if it's OK quit while your winning.

[Neil Breward]

I've just checked on the circuit for the Invicta 33 in the Newnes Radio and TV Servicing books (Volume One, Part 1 - Molloy and Poole).

This version shows the EBC41 cathode decoupled by a 30 microfarad capacitor and the negative feedback loop connected to the junction of the earthy end of the volume control and a 4.7K resistor to chassis - a much more sensible arrangement!

However, both the AGC and detector diode loads are connected to the EBC41 cathode - perfectly sensible for the detector, but the AGC diode load should really go to chassis as this would provide about a volt of AGC delay.

Odd that this set seems to have been wired up wrongly? Invictas may have been relatively low-cost sets but they were part of the Pye empire and I wouldn't expect too many design faults to have slipped through.

Cheers,
Neil

[Tim]

Thanks Neil
Evidently there is more than one version of this circuit. It seems as though I have the version in the trader, whereas my modification is nearer the version in R&TVS. Maybe the circuit was changed during production. Anyway, this seems to be the version that works properly.

[PJL]

Evidence suggests either the signal at the detector is too high or there is too much audio gain.

It's difficult to believe the radio was sold with the volume control inoperative. That implies something has changed...perhaps the valve characteristics changed during the production run? or maybe it needs a few leaky paper caps?

I'd be surprised if the tone control works still!...Peter