I see, I was trying something more simplified. But looking at what you did i guess you need a differential amplifier to feed back in to the output. It really shows at what levels you can take amplifiers. Thank you very much for your response. There is a lot of work in the first stage I must say despite the differential amplifier.

@AHX123 I just simulated another idea that gives about 0.02% THD. But there's a redesign of the output stage and without lots of extra design and construction work that's about the limit for discrete parts, I think.

You can share I’ll take a look. Creating this class AB amplifier is a part of my project and I’ve been itching my brain for weeks. And here you are doing it in minutes. I’m very impressed. They asked for a simple 2 stage amplifier with 3-5 transistors

@AHX123 Are you allowed to use other than 12 V rail?

We can use anything. As long as it’s a class AB amplifier. With a good signal output and low distortion.

@AHX123 Then I can recommend a two-stage system that yields about 0.04% into 16 Ohms with output of +/- 4 V. But I'd use +/- 8 V rails if that's okay. I probably could lower the rails a bit. But if THD is the goal, more excess headroom is better than less.

For my project the parameters don’t really matter. But I’ll take a look on what you are saying. As long as I present a decent enough working Class AB circuit they’ll be happy. THD is one of the main things I am measuring so is linearity

Have you got a circuit diagram that can show me?

@AHX123 I used BJTs instead of diodes in the above. Do they cound diode-connected BJTs as BJTs? Or diodes? Also, how much gain are you looking to achieve. That matters.

They would prefer diodes. And they are looking to amplify 30-50x

@AHX123 This uses \$\pm 8\:\text{V}\$ supply rails, gets \$A_v\approx 34\$, and has a THD (with \$\pm 4\:\text{V}\$ output into \$16\:\Omega\$) of 0.048%. That's simulated, keep in mind. But it gives an idea. Using your transistors would probably improve the THD. I just tried that change and it goes down to 0.03% THD.

@AHX123 So don't use the D44H and D45H power transistors... Though you can, if you want.

Wow this does give me a good idea where to go from here again. I’m looking for something way more simpler and I will definitely use these schematics to progress. You have taught me alot about supply rails and differential amplifiers. My supervisors are asking for a 2 stages amplifier that’s contains 3-5 transistors overall in both stages

@AHX123 Then you are stuck with just two BJTs for the output stage, plus one VAS BJT, plus two more BJTs for the long tailed pair front end. (You can tell them that the VAS BJT is part of the output stage. So 3 BJTs in the output, 2 BJTs in the input. You can get that down to about 0.3%, maybe 0.2%, THD. I hate using diodes and would prefer using diode-connected BJTs. But if they count them as BJTs and not diodes (sad, if so), then you are stuck with diodes.

Yes that’s what they are looking for. Most likely. It’s in my briefing and presenting something more complexed will be breaching. This is only a simple project for starters in electronics design like me

@AHX123 Well, just replace those diode-connected BJTs I show in my answer with diodes, then. And I guess you have to replace that current source of 2 BJTs with a resistor. Since you are allowed dual rails, though, I'd definitely take advantage of that fact.

Which answer are we talking about?

@AHX123 This one. Oh. Forgot. The R13 R14 divider won't be needed since you have access to ground. They can be replaced with one resistor, instead.

Wow thank you very much I really appreciate it. I’ll get to simulating soon and I’ll see what I come up with

@AHX123 I start at the end. What's my load? What's my peak voltage at the output? What does that mean for my peak current? What does that mean for my peak base current? Etc. I work backwards towards the input stage in that way. Unless I have a known input source (like an electret.) In that case I have to start with the input stage because I am not using an ideal signal generator anymore. But then, I'm not limited to just 5 BJTs, either.

I see so I start off with my end and work backwards ensuring the values and components I have are meeting the criteria

what would I do with the r3 resistor in the middle of q3 and q4?

@AHX123 R3 is added so that you can adjust the quiescent (class-A) current in the two emitter degeneration resistors. If you make it 0 Ohms, then perhaps there is a little current. But if you make it larger, the quiescent current in the output transistors will increase. You want it to be about 10-20% of the peak output current. It's wasteful, but it helps with cross-over distortion. You do need to understand how and why this works. Or else you cannot really design it. This works backwards to the base current "curtain" that R5 and R4 supply. Etc.

As I’m removing the transistors do I place this resistor inbetween the 2 diodes

@AHX123 It can go in the middle or on either side. They are all in series. So it doesn't matter. The resistor isn't all that good a choice for making adjustments like this -- it's too soft. A VBE multiplier would be much better. But that's another transistor. And you are short on those. I can't even remember how many answers I've given on designing these things. You might use user:330261 in your search here. See what comes up for amplifiers. I'm sure there's more than a few with added details. Or break down your questions, new ones starting, say, at the output stage. I can work with that.

@AHX123 Can you go to 6 transistors??? And is class-A eliminated as a choice? (Better THD!) Is there any output power requirement? 100 mW? 500 mW? 10 mW? 5 W? Etc. Any requirements regarding delivered output power to the load? Specs matter. And you should write everything you can down about them. Everything!!! That means what you know for sure. And that means what is NOT specified and is allowed to be anything we want it to be.

I will give you the outline of my project. These will be low power (around 1 Watt rms) but very high fidelity amplifiers. Class A is suitable for low power and low distortion amplifiers but it suffers from low efficiency. Class AB is providing a trade-off between an improved efficiency and a slightly increased distortion. A design based on discrete BJT NPN and PNP transistors is desirable. The number of transistors used will be approximately 3-5, and there will be two stages:

a voltage amplification stage, and an output power stage that in one case will be Class A and in the other case Class AB, with at least two transistors in both cases. Negative feedback will be used to reduce distortion. Please use small-signal transistors BC547/BC557 and medium power transistors BD137/BD140 or equivalent

My project is comparing Class A and AB I’m currently doing my class AB circuit

@AHX123 Do you have to build one from real parts in real life? Or is this just a simulation exercise? Because low distortion comes from very very high open loop gain coupled with lots of NFB. The VAS itself is very high gain since it is grounded-emitter. The long-tailed pair will, itself, have some more gain. But if you use real parts you will be forced into using degeneration and that lowers open loop gain (a lot.) So that increases distortion. Simulation would make this easy. Real life will make this a fair bit harder.

Yes I will have to go ahead and build this in real parts

@AHX123 Well, that's annoying. It means you will have to craft this carefully. Is there a temperature range over which this must work well? Like in the winter air at the north pole and in the summer with the sun bearing down on it in a desert at the equator? Or can you use a nice even-temperatured classroom?

This is why I’m here. This project is fairly hard and annoying but simple at the same time due to it not being a very technical circuit

I might be able to get my hands on the class room otherwise the North Pole it is: anyways would I be able to get your aid in this project?

@AHX123 The crap of all this is -- you might experience as much as \$\pm 90\:\text{mV}\$ variation over temperature for the base-emitter junction from -20 C to +55 C. And between identical family parts another \$\pm 30\:\text{mV}\$ variation. Coping with that is ... not easy. And it ruins the open-loop gain when you try to add degeneration to cope with that mess. Which increases distortion. Etc. And limited to just a few BJTs??? Seriously? This is going to be a compromise of many factors.

I’m an undergraduate student and I have done maybe 40-50 hours of basic electronic learning. I have no idea…I’ve just been attempting to bias class ABs and differential amplifiers seeing if I can achieve a low distortion with at least some amplification

@AHX123 What year? Hopefully not 1st year. Would this be 3rd year? Maybe? By the way, I'm just a hobbyist. Physics and math are my life. Not electronics, which is just a hobby. I haven't even had a single class on so much as DC electronics and Ohms Law. Not so much as one hour of class time. So keep that in mind. I'm just an idiot in electronics.

Yes this is 3rd year. But I got given my last choice in my project. Which was very very unlucky

Is there any way I can talk to you in depth about this project?

@AHX123 Yes. In chat. I forget how to do that. But I'm pretty sure some admin type will jump in and force that onto us... soon. They hate seeing this kind of chat going on. Like the plague.

I don’t have enough “reputation” to chat the rules are there…what can we do…

So the circuit where you told me to input diodes that would be impossible to build with real parts?

@periblepsis - And "some admin type" now swoops in and will move this discussion into chat :) Please wait a moment... (It's not that we hate discussion in comments, but that is specifically prohibited as a use of comments in the commenting policy).

Okay. We've dropped into the chat. Are you there?

AS I was saying

Yes I’m here

Go ahead.

There isn’t any way for me to build this circuit with the diagram you showed me earlier with the diodes

The diodes I implement?

And the negative supply rails

You can use real diodes if your instructor doesn't count them as transistors.

They are fine. Just a bit of a pain because they don't respond to temperature the same way as a transistor does.

I gather you are allowed negative rails. Is that true?

The circuit will be running for a couple mins max at a time for testing and showing purposes

Yes im allowed negative supply

Yes. But if the room is cold that's one thing. If the room is hot, that's another. But if you can count on testing your circuit at the temperature that you will be tested on, then that's good. It means you don't have to cope with temp variation. Good!

That just leaves part variation. Which is less to worry over.

Good stuff. I’m really glad you could help. I’ve been under a lot of stress with this project and other modules

Are you permitted to use Darlington transistors and just call them one apiece?

I don’t see why not

Or would they count as two?

What's the output power requirement?

All they’ve said to me is that they are looking for an amplification 30-50x

And is the speaker a 16 Ohm one? Or will something else be used?

Yeah. But 30-50X times 10 uV is just 300-500 uV. And I can't imagine they will accept that.

A speaker will be used but it can be any type

Can you use a headphone with 600 Ohms?

Oh I don’t know. I will have to ask my supervisor

Does the output have to be audible to a human? Or will it be scoped with an oscilloscope?

This is the process of finding out your specifications!!! This is IMPORTANT!

he made me change the frequency to 1Khz from 60hz so I assume audible by himan

60 Hz is harder. Means bigger capacitors for one thing.

The headphone thing I don’t know but if you give me questions you want to find out I can ask

Again thank you for helping me

1. How will the output be evaluated? Via oscilloscope or by human perception? How will the THD be measured? What range of speaker/headphone load is allowed? Are there any limitations regarding power supplies? Can a dual tracking supply be used? Or must it be single-rail?

2. What's the input source? What it's source impedance? What will the signal peak voltage be?

3. The gain range is from 30-50X. Is anything in that range okay?

, audio measurements with oscilloscopes and spectrum analysers will determine the performance achieved (e.g. output power vs linearity, T.H.D Total Harmonic Distortion)

Yes that gain range is oksy

So an output of 1V peak is enough? Or not?

Will it be able to power a speaker?

Of 16 or 8 ohms

For headphones, you only need a few tens of milliwatts to hear. But for a speaker in open air you may require 100's of milliwatts or more.

All these things are important.

You need to know exactly how the evaluator will be making their evaluation.

You need to target that.

Yes I see

So find out exactly what their procedure is. No guesswork. Exact details.

They said 1 watt rms a generally low powered amplifier

Okay. That's a good benchmark!!

I would say do whatever is easier either with headphones or a speaker

So, sqrt(2*power*impedance) gives you the peak voltage needed.

Example, sqrt(2*1 watt * 16 ohm) is about 5.7 V peak.

This would mean 11.4 V peak to peak.

That’s very hight is it not?

Given overhead voltages of about 2+ volts on either side , this means about 16 V supply rail difference.

High

Maybe change the speaker to 8ohms

Or +/- 8 V. If possible.

Yes. 8 Ohms would mean peak 4 V. But that still means about +/- 4 V rails or 8 V single supply.

Or else you would need to use a bridged arrangement of two duplicate amplifiers. But that's off the table given the transistor limits.

Whatever is easier I will choose that path

Dual +/- supplies is easier.

Dual rails?

Go ahead and use it I guess I don’t see why not

A + rail, a - rail, and ground. Most power supplies can provide tracking outputs like that.

In your case I'd probably want +/- 6 V or else +/- 8 V. The more I get, the better it is for THD. So if I'm allowed even more, I'd take it.

That’s fine if it makes life easier then go ahead

Let's go with +/- 10 V. That allows you more latitude with the output impedance. So you can use even a 32 Ohm output in that case.

get a power supply that can provide both +10 and -10 V, plus ground.

Ah I see what you mean that will be acceptable

Since the temperature is stable, you can only have to deal with about +- 30 mV part variation. This means using about a factor of 5 of degeneration, or about 150 mV across those degeneration resistors in the output stage.

That will swamp out part variation, easily.

Yes hopefully the temperature is stable. On the off chance the circuit is unstable I can still get a very decent mark to without working circuits

The problem now is gain. The power into 4 Ohms at 1 watt means a peak current of 700 mA. The power into 32 ohms at 1 watt means 250 mA. So the worst case situation is 700 mA.

So what will this mean

That's a lot. So you need to select your output BJTs on the basis of this peak current as well as the average power involved.

I can'

I can't say what BJT should be selected. But you have to figure that each BJT will be dissipating about 1/2 watt each.

Wow this is racking my brain

So make sure they can handle it.

For example, a TO-92 package is about 200 C/watt. If you have 1/2 watt each that means about 100 C increase in temperature.

That would be difficult.

Wow

So you may want a different package. And this means, often, a different transistor.

So perhaps a TO-220 package?

want me to use these certain transistors

Are you getting the idea about how much goes into design?

It's not just simulation!

Yes I am

My university underestimated this project heavily

And we haven't even gotten through the output stage, yet.

I might file a complaint

We have to consider so many things.

So this is why I had selected the D45H11 and D44H11 devices in a circuit I showed you.

These devices can handle LOTS of power, easily.

But they have terrible beta values.

And that will reflect backwards onto what you do behind the output pair.

Design involves lots of details.

Electronics is a pain sometimes

Yes I can see that now

Reality is the pain.

Nature is a pain.

I’m in pain

We imagine things simply. We exclude so much from our thinking in order to simplify. But reality is quite complex.

We cannot easily keep track of all the details that reality throws at us.

So keep in mind that when you think simply about things, the reality will be vastly more complex than you imagine.

Our brains are limited.

I understand

we exclude things.

You are quite the teacher I must say

So, the output stage has two transistors. As you already know. You need degeneration resistors at their emitters because two BJTs aren't the same. So you need to provide about 150 mV of degeneration to swamp out the differences between them.

You know you need 700 mA maximum for a 4 Ohm load (if you want to accept such a thing.) And so for class-AB this means about 100 mA (or so) with 150 mV (or so) which means about 1.5 Ohms for each degeneration resistor. This is guess-estimate work. You can go either way from there. But it gets you close.

I would say that 2.2 Ohms would be more than enough.

I trust your judgement

But 1.5 Ohms would be fine, too.

Something in that vacinity.

I see

This provides you with the two transistors and now also the two resistors for the output stage.

That's a start.

Oh yes we are working backwards

The next problem is to work out two other details. The voltage difference between the two output BJTs and also the necessary current. The current flowing near the base should be about 10X the necessary base currents. If the output transistors have a beta of about 70 (let's say) then we are talking about 700/70 = 10 mA of base current. That means we need about 10X or 100 mA for the "curtain" of current near their base.

If the base-emitter voltages for the output BJTs might be about 800 mV and we are already considering about 150 mV for emitter degeneration, then that's 950 mV for each side or about 1.9 V difference between the bases.

Oh no

At 100 mA through the diodes and using 1N4148 we are talking about 780 mV each. That's two of them, plus a little. The "plus a little" means a resistor that we can adjust.

That will be hard to achieve?

So I'd use two 1N4148 diodes plus a resistor.

Yes

yeah. And you'd need to make adjustments to the resistor value. As you increase its value, the base-to-base voltage increases and this means more quiescent current through the emitter degeneration resistors.

This will be a process for you -- tweaking the resistor value to get the 150 mV across each of the emitter degeneration resistors.

Meanwhile, you need to work out a way to get the right 100 mA current operating.

Ah okay that madness sense

Makes^

The VAS BJT (Q3?) cannot have less than 1 V at its collector. With the output swing at maximum, you have to account for the degeneration resistor drop plus the BJT drop of one of the output BJTs. So this is 1 V + about 900 mV + about 700 mA*2.2 Ohm or about 1.5 V + your maximum voltage swing. If this maximum

is about 4 V, just to pick a number, this means 7.4 V as the midpoint.

Add another 4 V plus another 1.5 V plus another 900 mV and we are already at 13.8 V above the negative rail. If that's -10 V then we have about 6.2 V to the top +10 V rail for R4 and R5.

Anyway, this is the kind of process that one goes through.

It's a lot.

My head is hurting looking at it

What you want to find is the target center voltage. In this case, it might be about 3.8 V, given two rails. Not zero. But a little above zero. This may be a problem and a question. Can we reduce this so that the output should be zero, not +3.8 V? Maybe. So this means some more thinking before proceeding.

I’ve been awake for 20 hours doing this circuit

There is an advantage if we can make it zero. But that's an open question at this point.

What would that be?

You should, I think, study more the LM380 an LM386 designs!

Anyway, I need to leave at this point. Hopefully, some thoughts are in your mind to consider more.

Have you started a schematic?

Not for you, yet. No.

But I know what I'd do.

Would you be able to help me start one?

And I haven't discussed the dominant pole capacitor. It's also important for the VAS transistor.

Ah okay

perhaps. But perhaps it would be best for you to break this problem up and post separate questions for each stage, providing as much

info as you can about the requirements.

I’ll text on this chat later i need to sleep. The good thing is this project isn’t due till april

It's usually called Cdom. It's value sets the frequency roll-off on the high end and it is very important.

Thank you for your help

Okay. Post questions as you go. I do want you to succeed. As I age and die, everything I know disappears. I must pass it along (pay it forward) or else our society loses knowledge. And that will have disastrous results in the long term.

You matter. Keep that in mind.

I appreciate it. Speak soon

Best wishes

I’ve been itching my bruising on this circuit for a month at this moment in time

Brain^

I have updates

On my circuit

I should be able to use the schematic you gave me

He said I can use more than 5 transistors. The output can be headphones and I can use transistors instead of diodes

That schematic you gave me could that be built in real life?

Hi I built the circuit exactly the same but I’m getting 7% THD