Distortion in darlington amplifier

[crutschow]

I meant I did not see any _visual_ distortions on the scope screen.

That can still mean a high harmonic distortion of several percent, well above a typical HiFi standard of <<1%, but may be tolerable for listening to AM radio.

 

[Stan2020]

That can still mean a high harmonic distortion of several percent, well above a typical HiFi standard of <<1%, but may be tolerable for listening to AM radio.

Totally down with that, cpt.Wally, I am just wondering what it has to do with a bold statement "There's still a LOT wrong with that circuit" of our esteemed pointy-haired boss?

 

[Stan2020]

You did not see distortion because C3 was cutting all the high frequency distortion harmonics. C3 also cut high frequency audio making the amplifier muffled like an old AM radio.
I fixed it and got an output power 5 times more than before. The output power is still very low because the power supply voltage is very low.
Bootstrapping is when the output reaches down with C7 ands pulls up the bootstrap voltage on R8 and R14 allowing a higher output signal swing.

Your simulation is a bit disconnected from reality.
Original curcuit, no modifications, 1 mV 1000Hz input.
Output (on 8 ohm speaker) p-p ~1v
I don't see any obvious distortions:

 

[audioguru]

My simulation of the "horrible" circuit shows that its maximum positive output is only to +2.9V but with the output emitter resistors value reduced and with bootstrapping the positive output is much higher at +4.5V.

Instability when the LM386 is properly on a pcb and with gain less than 9 is in the datasheet section 9.2.1.2.1 Gain Control. A gain of 20 is close to a gain of 9 so the problems caused by a Mickey Mouse solderless breadboard adds to the instability.

Instability is also caused by a missing supply bypass capacitor that the designers of the LM386 assume that you have, but is not shown in the circuits on the datasheet.

 

[Stan2020]

My simulation of the "horrible" circuit shows that its maximum positive output is only to +2.9V but with the output emitter resistors value reduced and with bootstrapping the positive output is much higher at +4.5V.

Instability when the LM386 is properly on a pcb and with gain less than 9 is in the datasheet section 9.2.1.2.1 Gain Control. A gain of 20 is close to a gain of 9 so the problems caused by a Mickey Mouse solderless breadboard adds to the instability.

Instability is also caused by a missing supply bypass capacitor that the designers of the LM386 assume that you have, but is not shown in the circuits on the datasheet.

section 9.2.1.2.1 only mentions limitations related to the presence of bass boost RC, which is not the case here.
Your first simulation shows gain factor of 200 on 3 mv input with some distortion (original circuit) and a second one - gain factor of 17 ('improved' one)
In fact, original circuit shows gain 1000 of 1 mV input without any distortion. If the initial conditions mismatch that big, there is no reason to trust any further calculations.
And you still did not respond - so, are you saying emitters should be just connected? 0.22ohm is ridiculous.
And just noticed - you simulation has an error.
C3 is NOT connected to the ground in original circuit. So dump your results and start over.

 

[audioguru]

1) The "bass boost" for an LM386 amplifier IC does not boost low frequencies. Instead it cuts the gain of high frequencies which can result in instability if the reduced gain is less than 9 times.
2) Most amplifiers produce low distortion at low output levels My simulation of the original "horrible" circuit at maximum output level shows the top of the waveform is squashed which is horrible distortion.
3) Look at many of the amplifiers at Elliot Sound Products to see 0.22 ohm emitter resistors.
4) Yes, C3 was in the wrong location in my simulation. In the original location where it cuts high frequencies it causes much more distortion.

 

[Stan2020]

1) The "bass boost" for an LM386 amplifier IC does not boost low frequencies. instead it cuts the gain of high frequencies which can result in instability if the reduced gain is less than 9 times.
2) Most amplifiers produce low distortion at low output levels My simulation of the original "horrible" circuit at maximum output level shows the top of the waveform is squashed which is horrible distortion.
3) Look at many of the amplifiers at Elliot Sound Products to see 0.22 ohm emitter resistors.
4) Yes, C3 was in the wrong location in my simulation. In the original location where it cuts high frequencies it causes much more distortion.

1) As I said, Bass boost is not the case here. Stop wondering about it, IT IS NOT HERE
2) Most amplifiers of non- hifi level produce high distortions at maximum output level. ANd this is why every amplifier has an optimal output level (or, rather optimal output levels range). Your suggested "improvements" increased greatly output level but totally killed amplifier gain. For that I'd rater us any power amplifier, while this one supposed to work with AM detector stage as an input. We are talking about small AM radio amp, if you have not noticed yet. Please try to make your simulations relevant and useful.
3) Sorry, Australia (or is it Austria) is a bit too far. Amp for radio amateurs must have easily available parts only. Preferably from a standard IEC series. We have standards on resistors, you know. Lowest starts with 1 ohm. If your circuit uses 0.22 ohm resistors then it is waaay more horrible. This is the basics of circuits design.
4) Proof, please?

 

[Stan2020]

So my next question is - can you please explain the purpose of R9&R10?
Most example diagrams of this type show direct connection of emitters.
Do they have anything with limiting output current and power consumption or is there something more sophisticated to it?

 

[audioguru]

R9 and R10 should be 0.22 to 0.47 ohms to help match the output transistors for lowest distortion. For an old AM radio then they are not needed.
The value of 4.7 ohms is much too high since it throws away a lot of output power.
My fixed circuit has a max output level of 1.7V peak into the 8 ohm speaker which is a peak current of 213mA. The output transistors have a max rating of 800mA so the current does not need to be limited as much as the 4.7 ohm resistors did.

 

[Stan2020]

I love your calculations, they have only one quite minor issue: they are a bit disconnected from real life.
After I removed R8&R9, Ic current of Q4 is 320ma without any input, and both Q4 & Q3 are getting hot as hell. Just sitting at rest, Carl!
Did it ever happened to you to assemble anything using your hands, or you just enjoy the bliss of simulations?
BTW, Imax for 2N2907 is 600 mA, not 800. 800 is for its neighbor. Heard about bottlenecks?
From 2N2907 specs I can see that optimal Ic is around ~100 mA.
Now I do understand what R8 and R9 are for. Reasonable compromise between gain and consumption/thermal conditions.

 

[Nigel Goodwin]

Forgive me my little knowledge of electronics...what is 'bootstrapping R8'? Were you trying to say in this highly professional way 'I'd connect a capacitor in parallel with R8'? Please use simple words, if possible, I'd appreciate that. Why do we need this capacitor? Something related to thermal stability?
Can you give a link to ' four transistors (and less passive components) ... used in commercial radios ' schematics _without transformers_ that would have sensitivity of 10mV, output to 8 ohm speaker and satisfy you in regards to the scale from 1 to 10?
Please point out an example of the 'negative feedback from the output' in 4 transistors push-pull amplifier (no transformers)?
How I can see 'high distortions' you've mentioned? I applied up to 10 kHz 20 mV sine input, did not notice anything distorted on 8 ohms speaker output.
How I can see the mentioned 'unstability'? It works every time I turn it on, no self-generation, no spontaneous output change.

You can't see distortion very well on a scope, you can hear far more clearly though.

Bootstrapping is a standard electronics term and technique, dating back to long before transistors.

Here's an example of a four transistor amplifier - actually from an HMV radiogram from 1969 - so it's a little more complicated than a radio one would be, but essentially the identical basic circuit. Because it's so old it's germanium and PNP, so it's 'upside down' compared to yours.
Your R8 is 42 and 44 in this circuit, and is bootstrapped via capacitor 22.

 

[Nigel Goodwin]

So my next question is - can you please explain the purpose of R9&R10?
Most example diagrams of this type show direct connection of emitters.

Most have emitter resistors, it's crucial to stop thermal runaway, it would be a rare (and really poorly designed) one which didn't.

Do they have anything with limiting output current and power consumption or is there something more sophisticated to it?

As above, they are crucial components to prevent thermal runaway and keep the amplifier working. In this poor example they are much too high, as a consequence of all the other poor design. Personally I'd never use a pair of diodes to try and set the quiescent current, I'd always use a Vbe multiplier (another transistor) which allows you to accurately set the quiescent. The 1969 circuit I've just posted shows a 'resistor' doing the job - but I suspect it it's actually a thermistor, which was commonly used for that purpose back then. 'Some' old circuits did use diodes, but these were special diodes selected for the job, and not a bog-standard IN4148. Because of that, they are using excessively high emitter resistors.

Essentially you're looking for thermal stability, and the diodes/transistor/thermistor should be mounted on the same heatsink as the output transistors to ensure that.

View attachment 126018

 

[Stan2020]

Thank you!
Tell me the truth please, do you seriously deep in your heart believe that this blurry tiny fragment of some exotic radio interception from 50 years ago - does it match you brave words '... Using four transistors (and less passive components) you can make a 'proper' push pull amplifier, as used in commercial radios etc. for decades'??? really?
This is just hilarious. I appreciate you efforts, though, it took you a whole day to dig out this piece of Roswell blueprints.
It is funny to see how are you criticising all circuites easily available in the net, but you are obviously to shy to publish a link showing your own impeccable designs?
BTW, one of your local 'guru' here (he really calls himself a guru) gave me a great advice to get rid of R10&R9 and I barely managed to turn it off before the magical blue smoke showed up. I so love theorists!
I am sorry for tons of sarcasm...it is definitely not your fault.
I was just trying to find a group of people who has actual practical knowledge of analogue electronics.
Oh well, will keep looking
Thank you for your efforts again, I know I was a pig.

 

[Nigel Goodwin]

3) Sorry, Australia (or is it Austria) is a bit too far. Amp for radio amateurs must have easily available parts only. Preferably from a standard IEC series. We have standards on resistors, you know. Lowest starts with 1 ohm. If your circuit uses 0.22 ohm resistors then it is waaay more horrible. This is the basics of circuits design.

0.22 ohm is a standard value and easily available resistor, and is commonly used in high power amplifiers as the emitter resistors.

However, in a small low power amplifier, you don't need to go that low at all - and 2.2 ohms would be a more common value, but such amplifiers are usually intended for feeding 8 or 16 ohm speakers, the lower the speaker impedance the smaller the resistors need to be.

 

[Nigel Goodwin]

Thank you!
Tell me the truth please, do you seriously deep in your heart believe that this blurry tiny fragment of some exotic radio interception from 50 years ago - does it match you brave words '... Using four transistors (and less passive components) you can make a 'proper' push pull amplifier, as used in commercial radios etc. for decades'??? really?
This is just hilarious. I appreciate you efforts, though, it took you a whole day to dig out this piece of Roswell blueprints.

I don't know where you are, but I've been in bed!. The circuit is a typical example of how amplifiers are built - and from 1969 shows how long ago the correct way was developed - just as valid today as it was then (I presume the original design probably came from a Mullard application note?).

BTW, one of your local 'guru' here (he really calls himself a guru) gave me a great advice to get rid of R8&R9 and I barely managed to turn it off before the magical blue smoke showed up. I so love theorists!

You can't just 'remove' those resistors, as they are crucial to the design, as poor as it is - and I doubt AG said to 'remove' them - to improve power you need to reduce them, but also alter the bias settings - the two diodes are useless (hence the excessively large emitter resistors).

As I've said all along, and as anyone with a slight knowledge of electronics will say. it's a REALLY, REALLY poor circuit, created by someone without any clue as to what they were doing.

I am sorry for tons of sarcasm...it is definitely not your fault.
I was just trying to find a group of people who has actual practical knowledge of analogue electronics.

You've already got that, but aren't prepared to listen.

 

[Stan2020]

0.22 ohm is a standard value and easily available resistor, and is commonly used in high power amplifiers as the emitter resistors.

I am aware of IEC standard where the minimal value starts with 1 ohm. Do you have in mind some Australian standards? I was looking at all kind of resistors for 50 years and never heard of '0.22'.

You've already got that, but aren't prepared to listen.

I am obviously too low for your shining wisdom, or great gurus of magical blue smoke! )

Don't flatter yourself.

BTW, your self-proclaimed 'guru' said literally -' For an old AM radio then they are not needed '. Please keep doubting, you will be more comfortable this way.

I am a simple man, I ask simple questions and I see a wise man when I am getting concise and simple answers. Not a chain of excuses or some mumbo-jumbo only produced to make you feel insignificant.

 

[audioguru]

Antique 2N2222 and 2N2907 transistors in the old metal case might not match the thermal characteristics of modern 1N4148 diodes. Use them with modern P packages.

Do you know why 1N4148 diodes are used to bias the output transistors instead of using resistors? Because the diodes match the voltage and temperature changes of the output transistors.

The output transistors have "thermal runaway", as they warm up they conduct more which makes them hotter which makes them conduct more which makes them hotter which makes them conduct more .....
The forward voltage of the diodes should match the Vbe of the transistors and if the diodes get hot then they reduce their forward voltage which reduces the current in the output transistors if the diodes are bonded to the output transistors. Then there is no thermal runaway.

Red Circuits is another website full of amplifier designs. Many do not have emitter resistors for the output transistors.

 

[Nigel Goodwin]

I am aware of IEC standard where the minimal value starts with 1 ohm. Do you have in mind some Australian standards? I was looking at all kind of resistors for 50 years and never heard of '0.22'.



I am obviously too low for your shining wisdom, or great gurus of magical blue smoke! )

Don't flatter yourself.

BTW, your self-proclaimed 'guru' said literally -' For an old AM radio then they are not needed '. Please keep doubting, you will be more comfortable this way.

I am a simple man, I ask simple questions and I see a wise man when I am getting concise and simple answers. Not a chain of excuses or some mumbo-jumbo only produced to make you feel insignificant.

If you're not prepared to accept simple answers, and are incapable of understanding those simple answers, why bother asking in the first place?.

Why your Australian obsession?, it's no different to anywhere else in the world (as far as I know, I've never been) - the only reason AG mentioned it is as there's an excellent audio site hosted there. RS Components list 32 0.22 ohm through hole resistors, I didn't check SM ones.

 

[gophert]

AFAIK the minimal, built-in gain of LM386 is 20.

You are correct but... the MINIMAL design was set at 20 but that does not mean additional components cannot be added to get lower. The datasheet explicitly says the device is only internally compensated for gains down to 9.

 

[unclejed613]

You know, resistors 0.22 Ohm are not kind of realistic.

actually they are used a lot in audio amplifiers for output emitter resistors, and they're more common these days than back in the 1970s and 1980s when most amplifiers used 0.47 ohm or even 1 ohm emitter resistors.