Mystery fault in IF stage - any ideas?

[kestrelmusic]

I'm building a valve radio to my own design and have come to a really strange problem - the IF valve is oscillating like heck.

I've tried all the obvious things - more decoupling in the anode and screen-grid supplies (this actually makes it worse!), swapping over the connections to the IF transformer, physically moving one of the IFTs (it still does it even when they are at right angles to one another), swapping the valves etc. The only thing that seems to stop it is earthing the IF valve grid, which isn't really very helpful. I'm mystified! Can anyone make any helpful suggestions?

Circuit diagram of the relevant parts attached. THe IFTs are Denco IFT11.

[MichaelR]

I would decouple that 10K screen grid resistor that you are using to joint feed the two valves , ideally I would feed those seperately.

Mike

[AlanBeckett]

100pF as a screen decoupler isn't very generous, and as Mike says, seperate feeds, with decouplers, would be ideal.
Alan

[Skywave]

Two thoughts:

1. Ensure you are using 'single point earthing';
2. Ensure that you have pins 1 & 6 of the EF89 wired to gnd. for that stage.

Al. / Skywave.

[kalee20]

I take it there's a decoupling capacitor for the AVC line? It's not shown on the circuit diagram.

Also, what frequency is it oscillating at? And, try to find what affects the frequency of oscillation. For example, does tweaking the IFT trimmers make any difference?

As MichaelR and AlanBeckett say, get your screen grid decoupling good and substantial. Without it, the circuit can easily form a Tuned-Anode-Tuned-Grid oscillator if there's just a smidgen of capacitive feedback from anode to grid.

[f.cooper]

Hi,
Suggest you use 0.05mfd to 0.1mfd for the screen decoupling capacitor, also the value of the screen resistor might be a trifle low in value, I would suggest something more like 22k or higher. If you split up the two screens then you could probably double the value yet again.

Best Regards

Fred ( G0BYU )

[f.cooper]

Hi,
I would further suggest you need a demodulator reservoir capacitor about 100pf connected between junction of secondry 2nd IF transformer and 50k IF filter resistor
to ground.

Best Regards

Fred ( G0BYU )

[ukcol]

Originally Posted by kalee20

I take it there's a decoupling capacitor for the AVC line? It's not shown on the circuit diagram.

I also take it there is an AVC line, that's not shown either. As it is the control grid of the IF amplifier is floating.

[MichaelR]

The AVC feeds the controlgrid of the IF amp via the mixer /oscillator Anode IF secondary

Mike

[Michael Maurice]

I would also use a larger HT decoupling capacito, 0.1µF is not really high enough, ideally it should be about 5 - 10µF.

[ukcol]

Originally Posted by MichaelR

The AVC feeds the controlgrid of the IF amp via the mixer /oscillator Anode IF secondary

Mike

Yes clearly, but the bottom end of the IF TX secondary goes to a point marked AVC line, but no AVC circuit is shown. IF a valid AVC circuit is present the fine, but if it is as shown in the diagram the grid is floating which would cause all sorts of problems.

[MichaelR]

Originally Posted by ukcol

Yes clearly, but the bottom end of the IF TX secondary goes to a point marked AVC line, but no AVC circuit is shown. IF a valid AVC circuit is present the fine, but if it is as shown in the diagram the grid is floating which would cause all sorts of problems.

Point taken..

Mike

[jim_jobe]

Solder pin 6 to central spigot and then solder a piece of tinplate as a screen from the spigot and between pins 1 and 9.
Jim

[G8HQP Dave]

Try putting some decoupling in the HT feed to the ECH81. At present it shares the EF89 HT - better to keep them separate. Ensure the detector circuit can't leak signal current back into the IF amp or freq changer - pay attention to grounding. Is there any EF89 cathode decoupling? You don't show any, so you should be getting cathode degeneration which will help but should not be necessary. If you make the AGC line go negative does the oscillation stop?

Finally, is it possible that the IFT11 was originally designed for lower gain or lower anode impedance valves such as EF92 or the B8A series? If so, you might have to put a damping resistor across one or both IFT primaries - try 100k. I once built an IF strip with two EF85s and IFT11s and they needed this damping - but the EF85 has higher a-g capacitance and higher gm than EF89 so is less stable.

[G8HQP Dave]

More questions: is it definitely the IF stage which is oscillating? What happens if you remove the ECH81? Is the ECH81 functioning as a freq changer, or as an additional IF amp (i.e. is the ECH81 triode oscillating as it should)?

[kestrelmusic]

Originally Posted by G8HQP Dave

More questions: is it definitely the IF stage which is oscillating? What happens if you remove the ECH81? Is the ECH81 functioning as a freq changer, or as an additional IF amp (i.e. is the ECH81 triode oscillating as it should)?

It makes no difference whether the ECH81 is there or not. If the C.G. of the ECH81 is shorted to earth the amplitude of the IF oscillation is slightly less, but it's still there. The oscillation is at around 465kc/s and the frequency varies if I adjust the tuning slugs of IFT2.

The centre spigot of the EF89 is already earthed, but I will try soldering in a bit of tinplate as suggested, also decoupling each anode and screen-grid feed separately.

BTW, I've not shown details of the AVC line as I wanted to keep the diagram clear. However, it is quite orthodox: it is fed from the anode of the EF89 via a 100pF coupling capacitor to the second diode anode of the EB91. AVC voltage is built up across a 500k resistance and 0.1uF capacitor.

Many thanks to everyone for the helpful suggestions. I'll keep you posted!

[G8HQP Dave]

I know these are daft questions, but all the easy ones have been done by now:
1. is the AGC negative going? i.e. have you got the diode round the right way?
2. is the AGC line the feedback path? What happens when you short the AGC to earth? Have you tried more decoupling on the AGC line? You need to be careful not to make the AGC lumpy, but a bit of extra filtering might help.
3. is the AGC diode referenced to ground, or to the IF cathode for AGC-delay purposes? If so, the IF cathode needs to be well-decoupled otherwise it becomes the feedback point for a grounded-grid oscillator.

[PJL]

A few more tests to identify the mode of oscillation:
1. Short the AGC to earth.
2. Short across the previous IF output

[MichaelR]

It might be opportune to pull the other valves out one by one and see if the oscillation stops.

Mike

[kestrelmusic]

Originally Posted by G8HQP Dave

I know these are daft questions, but all the easy ones have been done by now:
1. is the AGC negative going? i.e. have you got the diode round the right way?

2. is the AGC line the feedback path? What happens when you short the AGC to earth? Have you tried more decoupling on the AGC line? You need to be careful not to make the AGC lumpy, but a bit of extra filtering might help.

3. is the AGC diode referenced to ground, or to the IF cathode for AGC-delay purposes? If so, the IF cathode needs to be well-decoupled otherwise it becomes the feedback point for a grounded-grid oscillator.

It still oscillates with a) the AGC shorted to earth and b) with a fixed negative bias of 6 volts on the IF amp grid. And yes, the diode is the right way round! THe AGC diode is referenced to ground, i.e. the diode cathode is earthed. I haven't been able to do anything with it for a few days as I have been at work, but hopefully I may get another go at it tomorrow.

I'll keep trying. I know IF stages generally do work without oscillating, and other people have used Denco IFTs with EF89s, so there must be a way to get it working!