LM386 low noise alternative?

[neazoi]

The 10pF capacitor in the anti-hiss circuit seems to be waaaaaaaaaaaaaay too low to be an effective. Its turnover frequency appears to be around 1.5 Mhz.

Substitute it for 1,000pF and then 10,000pF and let's see what happens.

I have found another IC, which requires no external components (great!) and is has more gain. It is the TDA7052.
However the LM386 can be boosted to 46dB, whereas the 7052 is 39db.
I wonder if a hack can be done (feedback?) to the 7052, so that it can achieve the higher levels of gain needed?

 

[chemelec]

I Doubt that TDA7052 is any better for noise especially using headphone, But it does have More Power and a Higher Supply voltage rating.

 

[Audioguru]

The TDA7052 is also cheap and noisy. It is bridged so you need to wire it differently from the way shown in its datasheet to drive headphones that have only 3 wires, not 4 wires.

 

[betwixt]

Perhaps if you told us the impedance of your headphones it would help. If you are unsure, even a resistance measurement might give a clue.

All these amplifiers are fine for their intended purpose - driving small loudspeakers with reasonable volume, noise and distortion. They are for use in consumer equipment where cost is more important than quality. It isn't really fair to condemn them, they serve a pupose admirably, it just isn't the right one for Neazoi's needs.

Brian.

 

[chemelec]

I use the LM386 for Numerous Headphone Amps.
If Built Correctly using the Schematic Example on the Data Sheet and using a Suitable Value of the Series Ohm Resistor, The Noise Level should be Quite Acceptable.

For 8 Ohm Headphones, 100 -200 Ohms is good
For 16 Ohms, Double this Value.
For 32 Ohms, Quadruple this Value.

You should NEVER Connect Headphones direct to the Output of Any Power Amp.

 

[neazoi]

Perhaps if you told us the impedance of your headphones it would help. If you are unsure, even a resistance measurement might give a clue.

All these amplifiers are fine for their intended purpose - driving small loudspeakers with reasonable volume, noise and distortion. They are for use in consumer equipment where cost is more important than quality. It isn't really fair to condemn them, they serve a pupose admirably, it just isn't the right one for Neazoi's needs.

Brian.

The resistance of the stereo phones is 32R each (32R left to GND and 32R right to GND).
The attached thing worked nice and with less audible noise for me. However I would like more gain. Maibe a low noise single transistor amplifier at the input solves the problem once and for all. The schematic as it is has about 30db of gain. It might seem more complex than the 386 but if you count the numbet of components it is simpler, not to mention that the components are possible to be found in a junk box.
I just need more gain

- - - Updated - - -

The TDA7052 is also cheap and noisy. It is bridged so you need to wire it differently from the way shown in its datasheet to drive headphones that have only 3 wires, not 4 wires.

Interesting point, Indeed!

- - - Updated - - -

You should NEVER Connect Headphones direct to the Output of Any Power Amp.

Why is that?

 

[betwixt]

Why is that?

A power amp, even as small as an LM386 can seriously damage your hearing, either by simply producing too much volume or by becoming unstable and sending loud (possibly beyond audible frequency) oscillations down your ears!

The problem with all the proposed designs so far is they are intended to push high currents through low impedance loads. That is exactly what is needed in a power amplifier but your needs are different, you want lots of amplification (high sensitivity) and reasonably high voltage output but into a higher impedance load. 32 Ohms resistance implies the impedance is at least 32 Ohms so it moves the target more towards a voltage amplifier than a power amplifier.

I would propose you use a dual op-amp and see if that is sufficient. You can use the first stage to give say 30dB gain and the second stage to give say 20dB gain. The second stage connecting to the headphone through a resistor to isolate the reactive effects of the load and provide the essential safety feature. I am thinking of an amplifier something like a TL072/TL082 or equivalent. There is at least one commercial product doing just that, using a TL072 and costing around 1,000 Euros (~$US 1,000). Of course the manufacturer claims they use specially selected resistors to justify the price!

Brian.

 

[neazoi]

A power amp, even as small as an LM386 can seriously damage your hearing, either by simply producing too much volume or by becoming unstable and sending loud (possibly beyond audible frequency) oscillations down your ears!

The problem with all the proposed designs so far is they are intended to push high currents through low impedance loads. That is exactly what is needed in a power amplifier but your needs are different, you want lots of amplification (high sensitivity) and reasonably high voltage output but into a higher impedance load. 32 Ohms resistance implies the impedance is at least 32 Ohms so it moves the target more towards a voltage amplifier than a power amplifier.

I would propose you use a dual op-amp and see if that is sufficient. You can use the first stage to give say 30dB gain and the second stage to give say 20dB gain. The second stage connecting to the headphone through a resistor to isolate the reactive effects of the load and provide the essential safety feature. I am thinking of an amplifier something like a TL072/TL082 or equivalent. There is at least one commercial product doing just that, using a TL072 and costing around 1,000 Euros (~$US 1,000). Of course the manufacturer claims they use specially selected resistors to justify the price!

Brian.

Yeah right, the "golden ears" guys hit again...
Only because I like discrete schematics more, do you think the discrete schematic in post #26 is ok?
I have built it and it is quite stable. In fact it was recommended in another forum I asked. You mentioned a series resistor, is it enough to place it in series with the headphones and what value?

I wonder if an AGC could be easily applied in this one instead.

 

[KlausST]

Hi,

You mentioned a series resistor, is it enough to place it in series with the headphones and what value?

--> read post#25

Klaus

 

[neazoi]

Hi,


--> read post#25

Klaus

Of course, yes, thank you.
Will this also attenuate volume much, or it just protects you from sudden loud "cracks" and "noises"?
Just place it in series with the headphones right?

 

[Audioguru]

Your discrete amplifier in post #26 has a very low input impedance of only 1k ohms (because it is an inverting amplifier) that kills the audio level from the high impedance radio output and 10k volume control. If you use an input transistor with higher hFE like a BC549C (it also has low noise) then you can increase the values of the 1k and 680k resistors and reduce the loss of level.

 

[neazoi]

Your discrete amplifier in post #26 has a very low input impedance of only 1k ohms (because it is an inverting amplifier) that kills the audio level from the high impedance radio output and 10k volume control. If you use an input transistor with higher hFE like a BC549C (it also has low noise) then you can increase the values of the 1k and 680k resistors and reduce the loss of level.

Excellent! This will also save me from having to buy another type of transistors, since my radio uses bc549C transistors as well.
Ok so I will replace the 2n2222 ones with bc549 and the 2n2907 with a bc559. Then I will increase the input resistor to 10k and the 680k to 6.8k?
How does it sound to you?

 

[betwixt]

Note that the common connection to the headphones is supply, not ground!

Also note that in the circuit in post #26, if you add a series resistor in the loudspeaker connections it will upset the bootstap action slightly so you may not get quite the results you expect. It might be better to replace the existing loudspeaker connections with a resistor of around 100 Ohms and capacitively couple to the headphones (through a second resistor) from across it.

Personally, I would still go with the dual op-amp idea but its your project!

Brian.

 

[Audioguru]

Excellent! This will also save me from having to buy another type of transistors, since my radio uses bc549C transistors as well.
Ok so I will replace the 2n2222 ones with bc549 and the 2n2907 with a bc559. Then I will increase the input resistor to 10k and the 680k to 6.8k?
How does it sound to you?

Completely wrong.
You need to change only the input transistor Q1 to a BC549C, not a BC549 because only the "C" has high hFE. The other two transistors are fine as a 2N2222 and a 2N2907.
Then increase the 1k resistor to 4.7k and increase the value of the 680k resistor to 2.2M. Then the input impedance will be 4.7k and the gain will be closer to 2.2M/4.7k= 468 times.

 

[neazoi]

Completely wrong.
You need to change only the input transistor Q1 to a BC549C, not a BC549 because only the "C" has high hFE. The other two transistors are fine as a 2N2222 and a 2N2907.
Then increase the 1k resistor to 4.7k and increase the value of the 680k resistor to 2.2M. Then the input impedance will be 4.7k and the gain will be closer to 2.2M/4.7k= 468 times.

Ok I tried it.
I changed the input transistor to BC549C. The volume was much improved.
Then I changed the 680k to 2M (I did not have 2.2M). The volume was a bit more improved.
My detector already has a series 10K resistor so that it is isolated from the audio amplifier, and adding another 4.7k in series only degraded the audio volume.
I also changed the regenerative detector transistor from BC549b to bc549c and the sensitivity was more improved. The regeneration effects was ok like before.
Finally I changed the rf preamplifier transistor from BC549b to bc549c and the sensitivity was even more improved.

With all these changes the volume is nice, sometimes ear-splitting when a good signal is received.

Overall success!

I was wondering if I could increase the 2.2M even more, to 3.3M say, even for more gain?
Also if changing all the audio amplifier transistors to bc549C and bc559C if it will improve the gain/volume even more?

 

[Audioguru]

The 680k, 2.2M or 3.3M resistors are for negative feedback (gain) and for biasing. 680k provides a bias current to the first transistor of 5.6uA and a 2N2222 transistor in that circuit needs about 17uA so it was designed wrong and some 2N2222 transistors might work and many will not work. A BC549C transistor needs a bias current of about 3.3uA so the 2.2M is good and some BC549C transistors will work with 3.3M ohms and have a little more gain.

The output transistors are emitter-followers that have a voltage gain of 1 so changing the 2N2222 and 2N2907 to BCxxx will make no difference.

"Sometimes ear-splitting"? See, a real AM radio circuit has automatic volume control so that a weak distant station sounds as loud as a strong local station. Your simple regenerative circuit does not have it.

 

[neazoi]

The 680k, 2.2M or 3.3M resistors are for negative feedback (gain) and for biasing. 680k provides a bias current to the first transistor of 5.6uA and a 2N2222 transistor in that circuit needs about 17uA so it was designed wrong and some 2N2222 transistors might work and many will not work. A BC549C transistor needs a bias current of about 3.3uA so the 2.2M is good and some BC549C transistors will work with 3.3M ohms and have a little more gain.

The output transistors are emitter-followers that have a voltage gain of 1 so changing the 2N2222 and 2N2907 to BCxxx will make no difference.

"Sometimes ear-splitting"? See, a real AM radio circuit has automatic volume control so that a weak distant station sounds as loud as a strong local station. Your simple regenerative circuit does not have it.

Thanks for the info.
I was thinking to build this project as a KIT (if I manage), so having to purchase only one type of transistors (well apart from the PNP one) is more economical. That is the only reason for changing the audio amplifier transistors to BC549C types, since you say there will be no difference I think I will do it, unless you instruct me that this this not good.
Do I need to change any resistors when using the BC types?

Ok I will leave the resistor to 2.2M there just in case.

This radio does not have an AGC indeed and it should be quite challenging to do so by limiting the RF gain in a regenerative receiver.
However, this can be done on the audio amplifier. A JFET that acts as a variable resistor at the audio input, fed by some feedback from the audio output?
Or....?????
It would be interesting to hear some ideas.

 

[betwixt]

The principle difference between audio AGC and RF AGC is that one works on the recovered audio and the other works on carrier level. The carrier level will be fairly constant (allowing for fading) so the AGC can keep the perceived level constant. If you use audio AGC it will adapt to the sound level and adjust to maintain equal volume. RF AGC is good because it help to keep the signal level within the best range for the detector to work and gives best sensitivity on weak signals while preventing overload on strong ones. When you use audio AGC you bring up the background noise to the same level as normal audio and compresses the sound peaks, although it does give a degree of protection against loud sounds, it also distorts them considerably.

You ideally want to use RF AGC to optimize the receiver and have a peak limiter on the audio. The problem is that in a regenerative receiver it is difficult to derive a suitable signal for AGC and even if you were sucessful, applying it would change the operating parameters. For example, imagine your regeneration was set too low and the recovered AGC voltage was 20mV, you increase the regeneration to best point and the AGC rises to 50mV, you go a little too far (as for receiving CW/SSB) and the AGC suddenly jumps to 2,000mV because it is now based on the circuit working as an oscillator. The increased AGC kills all the preamplifier gain and probably (depending on the time constant) turns the whole receiver into an unstable, pulsed oscillator.

Brian.

 

[neazoi]

You ideally want to use RF AGC to optimize the receiver and have a peak limiter on the audio. The problem is that in a regenerative receiver it is difficult to derive a suitable signal for AGC and even if you were sucessful, applying it would change the operating parameters. For example, imagine your regeneration was set too low and the recovered AGC voltage was 20mV, you increase the regeneration to best point and the AGC rises to 50mV, you go a little too far (as for receiving CW/SSB) and the AGC suddenly jumps to 2,000mV because it is now based on the circuit working as an oscillator. The increased AGC kills all the preamplifier gain and probably (depending on the time constant) turns the whole receiver into an unstable, pulsed oscillator.
Brian.

Can a simple peak limiter be made on this audio circuit?

What if the AGC is not applied to the regenerative detector, but to the untuned RF preamplifier prior to it?
To talk specifically here is my RX so far including the audio amp mods.
An element (jfet?) used at the emitter of the preamp or as a base input attenuator could do that. I have no idea about the circuit details though.

- - - Updated - - -

Now that I am thinking of it better, AGC cannot be easily applied to this circuit. Most of the times you would receive DSB, so in crowded frequencies the AGC might respond to an unwanted signal in the opposite sideband, and attenuate the wanted one as well.
It will only have a meaning to do it, only in the special case where you attenuate the LSB and set it "out of puff", by fine tunning the regeneration (special case with this receiver). However, in that case the wanted signal is attenuated by this setting by some degree as well, and you would want more amplification, not AGC, unless the station is next to you.

So a simple audio peak limiter would be the best just to protect from excessive audio levels I think and it yould be interesting to know how to do it in simple means.

 

[betwixt]

Can a simple peak limiter be made on this audio circuit?

Two parallel diodes wired opposite ways around across the headphone is the simplest solution but you MUST add a series resistor or risk the amplifier dumping high current into them.

The trouble with RF AGC is:
1. changing the pre-amp characteristics will alter the detector stage so you need to keep an uncontrolled buffer stage between them. In other words add AGC to a new stage before the existing pre-amp.
2. where do you get the signal strength measurment from? You might be able to extract some RF from the emitter of the detector stage but you would again have to buffer it to reduce it's loading effects then rectify and probably amplify it as well.
3. consider what happens to the perceived signal strength when you deliberately increase the regeneration to receive CW or SSB. The oscillation would be thousands of times stronger than the signal you were trying to measure. The AGC would be the oscillator level, not the incoming signal level.

If you are still there right now, I am monitoring 14.255MHz at the moment but only for a short while!

Brian. (GW6BWX)