# power amplifier circuit design

Discussion in 'The Projects Forum' started by salih, Oct 29, 2009.

1. ### salih Thread Starter New Member

Oct 20, 2009
4
0
hi there. i've been working really hard on this circuit for very long and i think i won't be able to do it without any help. any help is appreciated.

i'm to design a 2-3 stage class ab amplifier with no feedback. signal input is 1Vp-p 30KHz. supply voltages are ±12V. my aim is to amplify this signal upto 24Vp-p (24 gain). as i use a small load (20ohm) i have to use a push-pull stage (or maybe darlington pair) to drive this load. i have to use bc237 (beta=100) and bd911-912 (beta=60).

my first problem is that i have never reached max swing at 1st stage output. there is clipping at higher gains. i've actually wrote a matlab code for calculating the resistor values. one assumption is that i set Vc=0 as i don't want any dc offset at output.

re=1000 ;(i assumed re to be 1k)
rc=24*re
ic=12/rc
rth=10.1*(re)
vth=ic*((rth/100)+re)-11.3
r1=(24*rth)/(vth+12)
r2=(rth*r1)/(r1-rth)

the first problem here is that, the theorethical values never matches up with simulations. Vc never turns out to be 0 V. i have to change bias resistors (mainly R1) to set for 0 (so there may be something wrong with the code?). another point is that, gain is roughly rc/re=24, but at this case clipping occurs. so i drop the gain in order to prevent clipping. when i set rc/re=16, i get (+9.7) - (-10V) max swing without clipping which is also biased a little bit. with increased gain more biasing and clipping occurs.

the second problem is, i don't know how to connect the first stage output to the push-pull stage. when i use the push-pull alone with 24Vpp input, it gives the same output (which is what it should do). but when i connect them directly (after c2 to c4-c5 connection with rc/re=16) i get the peaks of first stage (+85mv) - (-225mv) and the peaks at output (155mv) - (-160mv). biased and much smaller.

using a current amplifier (common collector) as a second stage may be necessary. or there might be an additional transistor used for bias current stabilization (for diodes) for output stage. i don't really know what's the problem.

well this circuit looks like a simple one but indeed a very problematic one for me.

2. ### studiot AAC Fanatic!

Nov 9, 2007
5,003
523
C3 is way too small, C4 and C5 are probably too large.

Stages, at the moment you only have one voltage gain stage, plus a compound emitter follower, which has no voltage gain. You probably need a 3rd stage as noted. Conventionally this would be a PNP driver for the ouput pair, operated in CE mode, with the bias chain included in its collector resistance.

I suggest you replace the diodes with a 100Ω pot, to set the amp into class AB. You cannot guarantee this with diodes alone.

You should be able to then arrange the DC levels to avoid the need for C4 and C5 altogether.

BD911 /912 seem rather overkill for this application and probably have pretty low gains at low current, so yes you may need darlingtons to overcome this.

3. ### hobbyist AAC Fanatic!

Aug 10, 2008
887
88

When checking for clipping on the output, of this stage, ALSO remember to check if the input signal is being clipped, then youll know that the input impedance is so low as to drop the signal and cause distortion, this is very important to see where the clipping is happening at. Sometimes an output clipping is not nessary a transistor being overdriven, but rather the input signal being distorted on it's way into the into the amp. Due to amp input impedances.

Last edited: Oct 30, 2009
4. ### bountyhunter Well-Known Member

Sep 7, 2009
2,498
512
If you learned nothing besides that in college, it would still be well worth the cost of tuition. That will serve you well in life.

5. ### salih Thread Starter New Member

Oct 20, 2009
4
0
firstly, i checked my input signal and it's not clipped or distorted.
i set my capacitor values to some working ones. first improvement is that with biasing and such, i set first stage output level to (+10.6V) - (-12.2) unclipped (max unclipped i can get). well i don't know if it will be accepted as a +/- X value as it is pretty much distorted. however capacitor can also takes the center to 0. i'm not really sure about this.

secondly, i tried to use CE driver with 100Ω pots instead of push pull. but i didn't manage to calculate the necessary resistor values and capacitor values. so i added a common collector as a 2nd stage for current amp. i didn't spend much time on bias resistor values of common collector. at the output of 2nd stage, 1st stage's output level drops to (6.5v) - (-7.5). i guess it might be because of impedance matching. how could i do this matcing in order to reach again to 1st stage's original swing?

also there is clipping after 2nd stage, output is the same of 2nd stage and they're both distorted very much.

latest circuit schem.

there is no need to be rude dude. i'm just trying to be formal.

again thanks all btw.

6. ### hobbyist AAC Fanatic!

Aug 10, 2008
887
88
I just read over your first post again, I realized your going to have trouble getting signal swing of 24v. pk-pk, because to get a good undistorted swing, the supply should be chosen to be around 10% higher than the maximum pk-pk output voltage swing.

For the class A amp.

Either raise the VCC to around +/-14v.
or lower the output swing..

If you keep the +/- 12v. for VCC, then experiment with a output swing of around 20v. pk-pk.
and see if it can be realized without distortion. And then take it from there to see how high of a swing you can get to find the limit on this stage with that particular VCC, value.

Last edited: Oct 30, 2009
7. ### studiot AAC Fanatic!

Nov 9, 2007
5,003
523
It's gone midnight here but I will post a couple of circuit suggestions tomorrow.

8. ### bountyhunter Well-Known Member

Sep 7, 2009
2,498
512
I was not being rude, I was being truthful.

It sounds like the specs you are trying to achieve:

could easily be accomplished by any typical audio amplifier slightly modified. There are literally thousands of such designs posted on the net, along with application information. If you research them, understand how they work, you will accomplish this design easily and gain design skills applicable to general analog circuit design. Then you will be skilled enough in design to not need computer simulation to design the circuit. Just my opinion.

Last edited: Oct 31, 2009
9. ### hobbyist AAC Fanatic!

Aug 10, 2008
887
88
I think that your supply being +/- 12 v. needs to be higher to obtain desired gain with no distortion.

I usually breadboard my circuits, but I only have a DC bench supply with single + voltage output.
So I had to simulate this. Using Multisim.

Here is the results I was able to get with the supply your using.

for the class A input stage only.

I had a 1K for R4 but it showed clipping on the neg. peaks, due to VC being to neg. with respect to ground. so I raised it to 1.1k to put the bias in a region of nondistortion.

Without the C2 cap. the output swung around 20v. pk=pk.
With C2 a little over 21 1/2 v. pk-pk.

10. ### salih Thread Starter New Member

Oct 20, 2009
4
0
i know there are many examples, i've looked at every one of them, i couldn't manage the circuitry otherwise i wouldn't post here.

hobbyist;

i have to use +/-12 supply. distorsion is unavoidable i guess. i obtained a gain of 22, that's acceptable. so there's only one problem left: how to connect these stages? i'm not even sure what i need to get the last stage working.

11. ### hobbyist AAC Fanatic!

Aug 10, 2008
887
88
Looking at your first schematic, you can lose the C2, since you have C4 and C5 in there.

Your whole design of this last stage depends on input impedances.

You have a 20K output impedance of the class A stage. And your using 1K bias for the class B stage. That's a lost in signal right there.

May I suggest a systematic way of doing this.

If this is not a computer simulation but a real breadboard design.

First You need to determine the generator output impedance.
This is done using resistor divider and oscilloscope, to measure when volt. drops by 1/2.

Once that is established, then plan on making your Zin, of the class A stage at least 10 times that. (for starters)

Now with a 12v. +/- VCC, plan on a 20V. output swing, NO distortion,(for starters).

Now build your class B stage.
Input a 20v. pk-pk, into this stage and adjust the base bias resistors until you have no cross over distortion.

Now this bias resistance will be the minimum value needed to transfer a NONdistorted 20v. pk-pk signal to the 20 ohm output.

So now you have to make the output impedance of the class A stage, at least around 10 times less then the value of a bias resistor on the class B stage.

Now you have established the input impedance needed to the class A stage as well as the output impedance needed.

From there you work your values for RE , RC, RB1, RB2, to put around 0v. or something close to there at VC, remebering to use a value of RC that is determined to be the output impedance. And the combo of RB1//RB2 to be close to your imput impedance determined. Also remember with a 500mV. pk signal coming in you need your Vb, to be high enough to be able to handle it without driving the transistor into cutoff and saturation.

Then test and adjust values until satisfied with design.

Last edited: Oct 31, 2009
12. ### studiot AAC Fanatic!

Nov 9, 2007
5,003
523
So let's start at the beginning.

I assume this is a teaching project, designed to give understanding of the action of circuit building blocks through design experience.

Why is there to be no feedback?
Perhaps you are meant to appreciate the impossibility of achieving low distortion without it and come to the conclusion that it is necessary?
I assume you realise that there is considerable local feedback around Q1?

What calculations have you done on the implications of the statements that you require 24 volts peak to peak output using a +/- 12v supply.

Using a 20 ohm load this equates to a 3.5 watt amp. Say 3 watt if realistic output voltages are used.

For some reason you have a 30KHz signal to amplify, although you haven't stated the low frequency limit, which must also be considered.

Since this is a teaching website and you are putting in lots of effort it is worth expanding the detail.

Why have you chosen the circuit values I see?
Do you understand how a CE stage works? What for instance is the purpose of RE and C3?

If you need help understanding what's going we can discuss this in detail.

13. ### salih Thread Starter New Member

Oct 20, 2009
4
0
i've been working and modified the circuit. i've changed the output stage with complementary darlington pair and it worked i guess. i haven't finished the circuit, need to increase the gain (if possible), and try to achieve max swing by changing resistor values in accordance with impedance matching also.

as seen from sim i'm able to increase 1 Vp-p upto [8.5 | -9.5]. well it might be hard to see from the graph but the transistors in darlington needs to be biased with diodes i guess. i couldn't manage how to connect them in order to get a continuous sinusoid.

it's quite late here. going to apply this model on my circuit in the morning. sure hope it will work.
thx again for everything!

Last edited: Nov 2, 2009
14. ### hobbyist AAC Fanatic!

Aug 10, 2008
887
88
Thanks for taking the time to report back on how you did with your project.

Looks like your getting a lot of progress with it. It takes patience, but that's all part of getting a good understanding of design work.

15. ### Audioguru Expert

Dec 20, 2007
11,249
1,350
Replace R27 with a common-emitter driver transistor then re-bias it like this:
(Reduce the value of RE for enough voltage gain).

File size:
96.7 KB
Views:
31