Problem with oscillation on receiver audio stages

[neazoi]

I have tested this AF amplifier after the mixer stage of a direct conversion receiver that uses a diode mixer.

It does not have the output volume I seek (not too loud). But When I increase the gain of the tda7052 bad audio oscillation happen.
What could the problem be and how to overcome this?

 

[betwixt]

Why a 50 Ohm preamp? There is already a 50 Ohm load on the mixer (which probably causes huge volume drop) so you should really follow it with a high input impedance pre-amp. The 3.3mH/1uF LC filter will only work properly (~3KHz LPF) feeding a high impedance anyway.

Why the 1M potentiometer on pin 4 of the TDA7052A? Pin 4 has no internal connection in the IC.

[Edit - if it is TDA7052B rather than TDA7052A, pin 4 can be used as a volume control in which case you should remove the other control completely and bypass the potentiometer with a 1uF capacitor]

Brian.

 

[neazoi]

Why a 50 Ohm preamp? There is already a 50 Ohm load on the mixer (which probably causes huge volume drop) so you should really follow it with a high input impedance pre-amp. The 3.3mH/1uF LC filter will only work properly (~3KHz LPF) feeding a high impedance anyway.

Why the 1M potentiometer on pin 4 of the TDA7052A? Pin 4 has no internal connection in the IC.

[Edit - if it is TDA7052B rather than TDA7052A, pin 4 can be used as a volume control in which case you should remove the other control completely and bypass the potentiometer with a 1uF capacitor]

Brian.

It is the 7052A variant. Pin 4 is used as a volume control by setting an appropriate resistor or by a DC (I think).

The post mixer preamp is taken from this document www.arrl.org/files/file/Technology/tis/info/pdf/9208019.pdf I am also not sure why he used this kind of preamp. Note that I have not used his full diplexer approach because I did not care about terminating frequencies below 300Hz properly.

Can you propose me an intermediate stage between the tda7052 (or the lm386) and this diode mixer in this load/LPF configuration? This would be really helpful.

 

[betwixt]

The article explains why it was done that way but I have reservations about how sensible it is to mention noise introduced by resistors when they are followed by a noisy and rather low gain audio amplifier anyway.

You probably do want some DC leakage to ground (fig 2, R2) after the mixer to prevent the coupling capacitors charging up and blocking the mixer diodes but I would be inclined to add a simple single stage common emitter amplifier before the main audio amp if you need more gain.

I did a little research and it seems there are three types of TDA7052, the original with no letter suffix, the 'A' version and the 'B' version. The 'B' was a revised version with the volume control added but during a transition period, some manufacturers (or all at some date) renumbered the original or new version as 'A'. If the control doesn't do anything you will know why! It has been out of production for 16 years in any case.

You must bypass pin 4 if you want to control the volume that way.

Brian.

 

[neazoi]

You probably do want some DC leakage to ground (fig 2, R2) after the mixer to prevent the coupling capacitors charging up and blocking the mixer diodes but I would be inclined to add a simple single stage common emitter amplifier before the main audio amp if you need more gain.

You must bypass pin 4 if you want to control the volume that way.

Brian.

You mean to connect a 100nf from pin 4 to the ground, parallel to the gain setting resistor, on the tda7052a?

Ok a bc547 with it's emmitter to gnd. Then a 1k from the collector to vcc. then a 470k from the collector to the base. and DC coupling caps. |Where shall I add it. before the shown 50R amplifier or after it?.

Why isn't the tda7052 be able to cope with this gain needed? The setting resistor is set to low value and there is lot of headroom for more gain (up to 1M). but it oscillates when I set the gain that high. Won't it be also the case with the extra AF stage you propose?

 

[betwixt]

Where shall I add it. before the shown 50R amplifier or after it?.

Instead of it.
I would:
1. Add a resistor of about 1K after the 3.3uH choke to ground. I think there should be a DC path there or charge on the audio coupling capacitor could bias the mixer module.
2. Add the single stage amplifier as you described.
3. Take the output from the collector, through a capacitor to the top of the volume control.
4. Leave pin 4 of the TDA7052 disconnected - you don't need two volume controls and isolating it will eliminate ground current ingress.
5. I can't see clearly but there should be 220uF and 100nF capacitors across the TDA7052 supply pins close to the IC.

Brian.

 

[neazoi]

Instead of it.
I would:
1. Add a resistor of about 1K after the 3.3uH choke to ground. I think there should be a DC path there or charge on the audio coupling capacitor could bias the mixer module.
2. Add the single stage amplifier as you described.
3. Take the output from the collector, through a capacitor to the top of the volume control.
4. Leave pin 4 of the TDA7052 disconnected - you don't need two volume controls and isolating it will eliminate ground current ingress.
5. I can't see clearly but there should be 220uF and 100nF capacitors across the TDA7052 supply pins close to the IC.

Brian.

Change 1 had no effect
Change 2, 3, 4, 5 Did not improve things at all.
I guess there is sufficient gain, both on the previous solution and with the new one, but the power amplifier does oscillate so the gain has to be limited, which of course has as a result lower output volume.
So the thing is how to keep the gain high and the output volume high?

I have not found a stable enough circuit on my experiments.
Any ideas?

- - - Updated - - -

Change 1 had no effect
Change 2, 3, 4, 5 Did not improve things at all.
I guess there is sufficient gain, both on the previous solution and with the new one, but the power amplifier does oscillate so the gain has to be limited, which of course has as a result lower output volume.
So the thing is how to keep the gain high and the output volume high?

I have not found a stable enough circuit on my experiments.
Any ideas?

Surprisingly this circuit between the mixer and the speaker http://www.techlib.com/electronics/induction.html worked as well as the much more complex circuit of post 1
interesting

 

[betwixt]

I think 1. should be there anyway, not for stopping oscillation but to make sure the mixer works properly.

What kind of oscillation is it?
If you disconnect the signal from the top of the volume control does it still oscillate?

All the gain is in audio stages so I would suspect a layout or power wiring problem is allowing amplifier current to pull down the supply on earlier stages. Things to double check are that the speaker is not grounded on one side, for example if you use a headphone socket and it is mounted down to grounded chassis. The speaker wires must both float for the TDA7052 to work properly and it will draw excessive current if one output is grounded. Make sure the speaker wires are twisted, it helps to stop magnetic coupling. Ensure the power supply is fed in at the amplifier end of the circuit so drops along the supply wiring are minimized.

Perhaps a full schematic would help.

Brian.

 

[neazoi]

I think 1. should be there anyway, not for stopping oscillation but to make sure the mixer works properly.

What kind of oscillation is it?
If you disconnect the signal from the top of the volume control does it still oscillate?

All the gain is in audio stages so I would suspect a layout or power wiring problem is allowing amplifier current to pull down the supply on earlier stages. Things to double check are that the speaker is not grounded on one side, for example if you use a headphone socket and it is mounted down to grounded chassis. The speaker wires must both float for the TDA7052 to work properly and it will draw excessive current if one output is grounded. Make sure the speaker wires are twisted, it helps to stop magnetic coupling. Ensure the power supply is fed in at the amplifier end of the circuit so drops along the supply wiring are minimized.

Perhaps a full schematic would help.

Brian.

When I disconnect the top of the potentiometer there is no oscillation. The oscillation is a loud 1KHz or so tone.
I have also tried the last circuits on this page https://www.sm7ucz.se/TL431/TL431.htm and the results are similar, when I increase the AF volume from the pot oscillation happens although the tone is much higher.
I suspect there is a limit on the gain of the AF amplifier after the mixer. You see I am using the RF input port of the mixer directly on a wire antenna. It may be that the AF gain is at the max allowed and it the AF amplifier needs just more signal, which could normally come from an RF amplifier.
Could it be the case?

- - - Updated - - -

I have also built an LM386 amp, with high gain (10uF across pins 1 and 8) and feedback (120R in series with 10uF from the speaker to pin 7).
This configuration, connected directly at the mixer output (after the LPF) gives the most AF volume of the ones I have tried.
In this configuration something weird happens, Oscillation occurs either I set the pot to maximum or close to minimum. There is just a point at the middle that oscillation does not happen. Isn't that weird? How can it be?

 

[betwixt]

Your basic problem is that you are trying to bring RF signals of 1uV or less into several hundred ohms impedance up to maybe 1V to drive a low impedance loudspeaker with all the gain in one block, the audio amp. You probably need a power gain of 70dB or more. Adding an RF amplifier stage might help to some degree but as it is un-tuned you run the risk of overloading it with strong signals. The only good solution is not to do a direct conversion but to add an IF stage so you get a good combination of gain and selectivity in one block.

With such high gain, even at low frequencies you have to be very careful with the layout, in particular the signal grounds. When disconnecting the top of the volume control stopped the oscillation, it tells me there is audio reaching the mixer and/or oscillator stages. Bear in mind you only need maybe 1uV of audio signal to start the feedback and once the volume has increased, so does the amount of feedback.

All of the amplifier stages should work although I would be cautious of using the TL431 as a pre-amp, because it isn't low noise or intended to be used that way. My preference would be the LM386 for it's simplicity and single ended output. When you have a differential loudspeaker signal in a high gain amplifier, new problems of signal coupling start to arise and you have to be very careful with the route the wires take. Accepting the simplicity of the design, your underlying problem has to be the layout, use a ground plane under the audio amp and keep the power rails closely decoupled. Be especially careful of the speaker and volume control grounding to minimize any voltage drop between them.

Brian.

 

[neazoi]

My preference would be the LM386 for it's simplicity and single ended output. When you have a differential loudspeaker signal in a high gain amplifier, new problems of signal coupling start to arise and you have to be very careful with the route the wires take. Accepting the simplicity of the design, your underlying problem has to be the layout, use a ground plane under the audio amp and keep the power rails closely decoupled. Be especially careful of the speaker and volume control grounding to minimize any voltage drop between them.

Brian.

Thanks!
Should I use "star" grounding scheme even on grounds spaced a few cm away?
Assuming the LM386 solution, where should I ground the potentiometer and the speaker?
Right now the speaker and potentiometer are connected with untwisted wires of 3-5 cm length to the circuit.
Does it play a role that the speaker unshielded magnet is very close to the whole receiver circuit? It is below a copper pcb and it can magnetize metals above that copper sheet.

Hints on the above points would greatly help

 

[betwixt]

Your layout makes it difficult to be sure but without using a PCB, my approach would be to make a single 'star' point for the bottom of the volume control, the ground pin of the amplifier, the RC network (if you use one - recommended you do) across the amp output and the loudspeaker. Use that point also as the negative supply input point.

Feed the positive supply directly to the LM386 and add a 100nF and 100uF capacitor across the supply pin and the star ground. The idea is that the star point is as near to a common 'zero signal' point as possible. Your next task would be to give the star point as good as possible connection to the PCB as a ground plane. I'm not sure what the other parts in your photograph are (the ones before the mixer) but they look like ICs of some kind. I would take the power feed to them from the audio stage through a small resistor (10 - 100 Ohms) and decouple each IC to the ground plane close to them to reduce signal conduction along the supply wires. If you can use a screened cable from the volume control to the LM386 it would help, use the screen to ground the bottom of the potentiometer (and connect nothing else to it) and at the amp end connected to the star point as I described above.

Brian.

 

[neazoi]

Your layout makes it difficult to be sure but without using a PCB, my approach would be to make a single 'star' point for the bottom of the volume control, the ground pin of the amplifier, the RC network (if you use one - recommended you do) across the amp output and the loudspeaker. Use that point also as the negative supply input point.

Feed the positive supply directly to the LM386 and add a 100nF and 100uF capacitor across the supply pin and the star ground. The idea is that the star point is as near to a common 'zero signal' point as possible. Your next task would be to give the star point as good as possible connection to the PCB as a ground plane. I'm not sure what the other parts in your photograph are (the ones before the mixer) but they look like ICs of some kind. I would take the power feed to them from the audio stage through a small resistor (10 - 100 Ohms) and decouple each IC to the ground plane close to them to reduce signal conduction along the supply wires. If you can use a screened cable from the volume control to the LM386 it would help, use the screen to ground the bottom of the potentiometer (and connect nothing else to it) and at the amp end connected to the star point as I described above.

Brian.

Thanks for all that help. I will follow closely your instructions. I also prefer no RF preamp prior to the mixer for this ultra broadband design. I attach you the complete schematic up to now (DSB) but not drawn the LM386 amp and the AF preamp is not removed, so it is as initially posted here.
If you notice at the bottom right, there are a few regulators there for the different TX/RX voltages. Should I still use these small resistors and decoupling caps you propose? After or before the regulators?

 

[Nik_2213]

I'm a bit rusty, but this looks like an 'AofE' case where even 'good practice' does not suffice, and you must put ferrite beads on several strategic components...

First, though, please do the other stuff, and figure how that TDA version prefers to be stabilised...

 

[betwixt]

I agree Nik_2213 but this seems to be more of a low frequency instability problem due to ground loops and maybe being too ambitious with high gain amplifiers. I think ferrites may come later when sorting out RF issues.

Neazoi, adding resistors at the inputs of the regulators might help. Do not add them after the regulator or you lose the regulation!. As long as there is no risk of the voltage dipping below the regulator minimum input voltage you can use as high a value as convenient. Add the regulator input capacitor as close to the regulator as you can and directly between the Vin and GND pins. The idea is that the resistor and input capacitor form a low pass filter but the output side of the regulator stays at low impedance and constant voltage. Any voltage dropped across the resistor is just less dropped across the regulator but it forms a signal barrier along the supply that otherwise would be conducted to ground where you least want it.

Brian.

 

[BradtheRad]

I also got similar oscillations when I built an audio amplifier (maybe LM377 or 383). It occurred at a certain range when adjusting the volume control potentiometer. I had to test various ways to hook it up. (Maybe I installed an extra resistor or capacitor.) Finally I got the oscillations to stop. I never figured out what caused them in the first place.

 

[neazoi]

I also got similar oscillations when I built an audio amplifier (maybe LM377 or 383). It occurred at a certain range when adjusting the volume control potentiometer. I had to test various ways to hook it up. (Maybe I installed an extra resistor or capacitor.) Finally I got the oscillations to stop. I never figured out what caused them in the first place.

I had similar oscillations in previous direct conversion designs as well. I also never figured out what the problem was.

- - - Updated - - -

I agree Nik_2213 but this seems to be more of a low frequency instability problem due to ground loops and maybe being too ambitious with high gain amplifiers. I think ferrites may come later when sorting out RF issues.

Neazoi, adding resistors at the inputs of the regulators might help. Do not add them after the regulator or you lose the regulation!. As long as there is no risk of the voltage dipping below the regulator minimum input voltage you can use as high a value as convenient. Add the regulator input capacitor as close to the regulator as you can and directly between the Vin and GND pins. The idea is that the resistor and input capacitor form a low pass filter but the output side of the regulator stays at low impedance and constant voltage. Any voltage dropped across the resistor is just less dropped across the regulator but it forms a signal barrier along the supply that otherwise would be conducted to ground where you least want it.

Brian.

Ok I did all the things you said. The only thing I have not yet tried is to shield the potentiometer signals. Guess what, it did not work. I just can't make a simple lm386 (and other amps) produce enough volume out of the mixer. And I have also tried a simple BJT preamp prior to the lm386. Oscillations always start at a certain speaker volume.
Why is it so difficult? I start to get dissapointed, this thing is even more difficult than the extreme smd prototyping of the DDS chip!

 

[betwixt]

Well, don't forget that the output of the mixer will be no higher than (signal input - mixer losses) which will only be a few uV at most. This is why distributed amplification stages are normally used rather than lots of gain in one place. Remember that in conventional receivers there is generally 10dB or more gain ahead of the mixer then two or more 10dB gain stages after it before the detector, you don't have them in your design.

I still think you should be able to run an LM386 with a single stage pre-amp without any problems though, it is what is done in many domestic products without problems. Try dropping the LM386 gain a little, you might find you get more audio with the volume turned up higher and the gain reduced. With more than one million times amplification in a tiny space you really have to be careful.

Brian.

 

[neazoi]

Well, don't forget that the output of the mixer will be no higher than (signal input - mixer losses) which will only be a few uV at most. This is why distributed amplification stages are normally used rather than lots of gain in one place. Remember that in conventional receivers there is generally 10dB or more gain ahead of the mixer then two or more 10dB gain stages after it before the detector, you don't have them in your design.

I still think you should be able to run an LM386 with a single stage pre-amp without any problems though, it is what is done in many domestic products without problems. Try dropping the LM386 gain a little, you might find you get more audio with the volume turned up higher and the gain reduced. With more than one million times amplification in a tiny space you really have to be careful.

Brian.

I do not quit
Ok I will try other designs with more stages. I have also tried the preamplifier from this one http://qrp.gr/allbander/index.htm but the preamplifier oscillated itself (it is a 100db gain!). This is mind blowing, because in that page with the regen receiver the amplifier worked flawlessly.

It is very weird that this one http://www.techlib.com/electronics/induction.html is currently the most successfull. I thought it was the lm386 one but this provided more volume and it did not oscillate, but only on the very ends of the potentiometer settings (high and low).
And it is more weird that I tried to boost the output by one of these designs http://qrp.gr/allbander/index.htm and it did not work, no more output volume at all despite the added output stage.
This thing is giving me hair pulling problems!

 

[neazoi]

Your layout makes it difficult to be sure but without using a PCB, my approach would be to make a single 'star' point for the bottom of the volume control, the ground pin of the amplifier, the RC network (if you use one - recommended you do) across the amp output and the loudspeaker. Use that point also as the negative supply input point.

Feed the positive supply directly to the LM386 and add a 100nF and 100uF capacitor across the supply pin and the star ground. The idea is that the star point is as near to a common 'zero signal' point as possible. Your next task would be to give the star point as good as possible connection to the PCB as a ground plane. I'm not sure what the other parts in your photograph are (the ones before the mixer) but they look like ICs of some kind. I would take the power feed to them from the audio stage through a small resistor (10 - 100 Ohms) and decouple each IC to the ground plane close to them to reduce signal conduction along the supply wires. If you can use a screened cable from the volume control to the LM386 it would help, use the screen to ground the bottom of the potentiometer (and connect nothing else to it) and at the amp end connected to the star point as I described above.

Brian.

Ok I found the problem. It is this 50R from the mixer to the ground. When I remove it, along with it's 1uF capacitor. I can set the volume all the way up and no oscillation occurs. Currently I have built the LT031 amplifier so I cannot test it with the lm386.
But they say these mixers need to be terminated with 50R. I do not know what to do, the 50R there just ruins everything.