Audio Amplifier Help

Audioguru

Joined Dec 20, 2007
11,248
Your 1st opamp has a gain of 10k/1k= 10 so its output is 1V peak.
The 2nd opamp is not needed (the 1st opamp can easily have a gain of 50) and has a gain of 5 so its output is clipping at about 3.8V peak.
The 3rd opamp is not needed (the 1st opamp can have a gain of 50 and drive the output transistors) and has no voltage gain.
The output transistors might overload the 3rd opamp since some transistors have a high current gain and other transistors have a low current gain.

I think the output will be clipping at 1.5V to 3V peak when there is a 4.5 ohms load. The transistors are not biased so there will be awful crossover distortion.

Your simulation program is stupid because it doesn't know that the opamp output cannot go as high as the power supply voltage and that the output transistors also have a voltage loss because they are emitter-followers.

You could have the same results if you used 1 opamp, two resistors to set the gain and 2 transistors. A very, very simple circuit.
 

Georacer

Joined Nov 25, 2009
5,182
Your simulation program is stupid because it doesn't know that the opamp output cannot go as high as the power supply voltage and that the output transistors also have a voltage loss because they are emitter-followers.
Heh, funny you would say that. You 're right, but that's a couple thousands of dollars of stupidness we are looking at. Multisim is considered one of the best general simulation software out there.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
Your 1st opamp has a gain of 10k/1k= 10 so its output is 1V peak.
The 2nd opamp is not needed (the 1st opamp can easily have a gain of 50) and has a gain of 5 so its output is clipping at about 3.8V peak.
The 3rd opamp is not needed (the 1st opamp can have a gain of 50 and drive the output transistors) and has no voltage gain.
The output transistors might overload the 3rd opamp since some transistors have a high current gain and other transistors have a low current gain.

I think the output will be clipping at 1.5V to 3V peak when there is a 4.5 ohms load. The transistors are not biased so there will be awful crossover distortion.

Your simulation program is stupid because it doesn't know that the opamp output cannot go as high as the power supply voltage and that the output transistors also have a voltage loss because they are emitter-followers.

You could have the same results if you used 1 opamp, two resistors to set the gain and 2 transistors. A very, very simple circuit.
Well there's nothing I can do about the simulation program...that's what we're given to use at university and to be honest I don't know if its displaying correct results or not because I don't want to build the circuit and have it burn out my transistors and whatnot.

As for the schematic, I tried using one opamp for the gain stage but it was better to use 2 opamps with gradual gains than have one do all the gain. And as for the output, I told you it is displaying 2.790 Watts and I believe you're right that it is nearly impossible for the opamp to output rail-to-rail Vcc as its output. It should be around 4V peak max as voltage output goes and I'm not sure if the 2 power transistors are functioning correctly either.

Here's my updated schematics

1st design has no negative feedback from output to opamp
2nd design has negative feedback from output to opamp
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
Now the output transistors correctly show a base-emitter voltage loss of 0.99V each when the output current is too high at 0.88A peak.
I think the Sim program will mess up again if you connect the R10 negative feedback resistor to the output.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
So which one is correct? The first design or the second one? The only difference between 1st and 2nd is that negative feedback wire I have from the output of the two power transistors back to the output of the opamp...
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
Since your output transistors have no bias (class-B instead of the usual biased class-AB) then they produce crossover distortion (look in Google). When the output transistors are included in the negative feedback loop (the feedback resistor connected to the output of the amplifier instead of to the output of the opamp) then the extremely high open-loop gain of the opamp at low frequencies cancels the low frequency crossover distortion and other low frequency distortion of the output transistors.

All audio amplifiers (but maybe not guitar amplifiers) have the negative feedback resistor connected to the output.

Some guitar players like as much distortion as is possible but many other people do not want to hear any distortion.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
Ok so I connected the negative feedback resistor to the output and I'm still getting 2.790 Watts at the output and my distortion just went down from 10% to 0.03%...See I don't get it AudioGuru, in different configurations I'm getting higher power output and almost 0% distortion (eg. When you said to connect negative feedback resistor of opamp to output of amplifier, I did both: I connected it to output of opamp AND output of whole amplifier and I got 0% distortion)...should I actually build these? I mean, my voltage follower opamp is getting higher gain than the rails!!! How in the world....
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
You are mixed up with "voltage gain" and "maximum output level". They are completely different.

I am confused when you said that you connected negative feedback to the output of the opamp and to the output of the whole amplifier. You can't do both at the same time.
Negative feedback must be from the output of the whole amplifier.

I said that your Simulation Software has many errors.
10V peak-to-peak into 4.5 ohms is (10/2.828 squared/4.5=) 2.78W.
But the stupid software doesn't know that the output is no where near 10V peak-to-peak.

Now your stupid software shows an output that is higher than the supply voltage? Hee, hee it is extremely stupid.

Parts can have minimum spec's, typical spec's or maximum spec's. If you build one circuit then it might not work, might work poorly, might work well or might work very well depending on how you designed the circuit so that spec's variations of parts affect or do not affect performance. I designed all my circuits for "worst case" spec's and all of them and thousands of copies work perfectly.

Your amplifier might work well only if the transistors are matched and have very high gain. Then you might need to buy thousands of transistors and measure them then select only the best ones for your circuit. Instead I would add emitter resistors to match them and add driver transistors to increase their gain.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
What I meant by the feedback to both output of opamp and output of whole amplifier was that I ran a wire from negative feedback resistor to output of opamp and I also ran a wire from same negative feedback resistor to output of whole amplifier. Is this ok to do? When I did that my distortion went down to 0% for some reason...

As for the design itself, the simulation program is correct in displaying 2.790 Watts but for some reason the opamps gain is too high, higher than the actual rails. That's what's bothering me now...I guess its that particular opamp which is causing problems so I'll change the opamp to something else. Do you know of another opamp I can use in power amp designs?

Here is my most up-to-date design:
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
Now you have a wire shorting the bases of the output transistors to their outputs so the transistors do NOTHING.
The output of the opamp is driving the load to 10.1V p-p which is a peak voltage of 5.05V and the output current from the opamp is 5.05V/4.5 ohms= 1.12A. The stupid SIM program doesn't know that the max output current of the opamp is only about 30mA. The stupid SIM program doesn't know that when the output current from the opamp is 20mA then its typical highest voltage is only 3V, not 5V.

I simulated an opamp and your amplifier circuit in the free SIM program LTSpiceIV from Linear Technology. It performs correctly since the unloaded opamp clips at an output level of +4V and -4V.
Its clipped output level is very low when it is overloaded with a 4.5 ohm load.

I also simulaterd your amplifier. Its clipped output level is plus and minus 3v and it has horrible crossover distortion because the transistors are in class-B instead of class-AB.

OOps. My amplifier inverts but yours is non-inverting. Their outputs will be the same.
 

Attachments

Thread Starter

zero_coke

Joined Apr 22, 2009
294
So my second design is incorrect due to the transistors being shorted, but once I remove the short, I get an output of about 2.9 Watts with 0.003% distortion in design 1 and 2.878 Watts with a distortion of 3.58% on design #2.

Oh and with regards to the buffer in design #1, is it necessary to have it? I made two different designs (attached) and one has it and one doesn't. There's a big difference in terms of distortion apparently, but not output power much.
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
Your SIM software is completely wrong so why do you use it?
Didn't you see the correct Simulations I posted?

If you analyse the first circuit and build it then you will see 8.0V p-p at the output of the first opamp (not 10.2V p-p) and 6V p-p at the output of the circuit (not 10.2V p-p).

You will see that the buffer opamp is not needed because the 1st opamp can do it.
The distortion measurement is useless since the amount of clipping is wrong.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
Oh I'm sorry, it was a typo Joe. Same output doh, I don't know what to do. See here's the thing: NI Multisim was given to us by the faculty and that's what they're going to use to mark us on this project. They're basically going to run the simulation and mark me based on it. I can't just switch software and I don't know why the stupid opamp is giving a higher gain than the rails...
 

Audioguru

Joined Dec 20, 2007
11,248
You are still confusing "gain" and "max output level".

Gain is the amount of voltage amplification. You have the negative feedback resistors with a ratio of 50 so the gain is 50 times.
1mV input gives 50mV output.
10mV input gives 500mV output.
100mV input tries to make the output peak at 5V but it can't go that high so the output has clipping distortion.

The max output level is supposed to be plus and minus 3V but your stupid SIM program shows the peak output level higher than the rails.

I think Multisim makes errors because it doesn't have detailed models of the opamp and for the transistors. I have never used Multisim so I don't know if it has a model library.
The program I use has many opamps and transistors in its library and there is a forum where more models are available.

Maybe you should look for detailed models of the opamp and transistors or look for an opamp and transistors that have models available.

Why don't you add a few parts to make the output operate in class-AB like real audio amplifiers so there is no crossover distortion? You will get extra marks for it.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
Ok my most updated design is attached. However, you said to connect feedback resistor of opamp to output of entire amplifier and my distortion gets worst. Why is that??

And regarding the class AB, how do I convert my class B into class AB??
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
Why do you have 3 opamps when only one or two are needed?

It is good that you reduced the gain to 40 so that the opamps do not have clipping.

But the output level from the transistors is wrong because it is the same as their input level (8V p-p). The transistors are emitter-followers so their output level is supposed to be almost 2V p-p less than their input level.

In post #6 you said you learned about a class-AB transistor circuit or look for it in Google then do it.
 

Audioguru

Joined Dec 20, 2007
11,248
I am glad to see that your stupid software shows a normal lower output level at the output of the opamp and the normal base-emitter voltage drop for the output transistors.

If you use only one opamp instead of two then the results will be almost the same (the two opamps reduce the crossover distortion a little).

Many old 741 opamps do not work with a supply as low as plus and minus 5V but your Multisim software doesn't know that.
The gain is 39 so the output is clipping fairly badly which is causing the high distortion. If you turn down the gain a little then the output peak level will be the same but the amount of clipping and amount of distortion will be less.
Since the output transistors operate in class-B then the crossover distortion is horrible.
 

Thread Starter

zero_coke

Joined Apr 22, 2009
294
Yeah when I used two opamps instead of one opamp my distortion went down from 12% to 10% but my output power also decreased so I guess it's a tradeoff.

As for the distortion, I'm trying to convert my class B into Class AB but I am having no luck in doing so...I added the 2 diodes and the resistors but no promising results ...


EDIT: Ok I converted my circuit to class AB and it's fine now. I'm getting about 900mW output at a voltage of 2.7V peak which is expected (actually, you can get around 3V but I had to compensate lower gain for lower distortion) using an LM741. For any other readers interested in the topic, here's what I learned:

1) Using LM741CN OpAmps (what I had to use), you should set your gain to no more than 27x than the input if your input is around 100mV. Having this gain prevents clipping and distortion at the output. Otherwise, if you can use MC34071 OpAmps, use those instead. Note that some OpAmps like the 741 work better with +/- 15V DC , anything lower may cause problems or have the opamp not function properly.

2) Use a ceramic capacitor in parallel to your speaker load to reduce distortion dramatically

3) Use a Class AB amplifier instead of Class B for better distortion and stability

4) Try to get matched power transistors from manufacturer if you can, this helps with distortion and power amplification

5) Have fun, ask questions!

P.S If you're really wondering, my gain was 2.7V peak and distortion of 0.3% which is not bad. However, I had to further reduce my voltage gain from 27 to 22 using a +/- 5 Vcc DC source to my LM741 OpAmp because it was clipping really badly during lab session today.
 
Last edited:
Top