Designing problem...

Hi,

I've just tried for the first time to design and calculate a class AB amplifier. It's just a first beginning try-out, but I'm glad with what I've reached by now. I've set up a schematic on pc with the program circuitmaker (www.circuitmaker.com) and did some tests. But I've to deal with one measure problem: the amplitude of the positive alternantions is smaller than the negatives. I know that's the reason of a voltage-drop over the resistance of the cross-over compensation. So I've connected a 100µF capacitor between the middle of Rc and the point where the speaker is connected to the output-devices to compensate the difference, but this seemed not to be enough.

You can see the picture of the simulation at:
http://hugobross.tripod.com/amp.htm

Does anyone known how to solve this? I'm just trying further know, maybe I'll find a solution this evening or maybe you can help me. I'm just thinking of increasing the value of Rc (=R1+R2) and reducing the current through this resistor, so that the voltage-drop over the cross-over compensation can be reduced.
Or can this problem be solved if I connect the feedback to the differencial amplifier (Q6 & Q7)?

Any suggestions welcome,

Hugobross.
 
Did you consider trying to decrease the value of R1. Also, connecting the feedback should help because that should help to cut off Q1 and subsequently Q5. It is interesting if you were able to get it to work at all without the overall feedback connected.

Alternately replace R1, R2, and C1 with a constant current source. you can actually build a simple one with a single PNP transistor with a base resistor connected to ground, saving one part. The resistance is chosen to provide just enough collector current to balance the circuit because of the transistor's fairly stable beta. It should work well since the current source for Q7 and Q8 also varies with the power supply. Therefore, the two variations' effects tend to cancel in the overall circuit. You also get the added bonus of the same VCE at clipping for this current source as Q5.
 
designing amps

Hi Hugobross,
It is interesting to see how others work on the design .
I must say I have a totally different approach; combining known circuit elements. Not simulating but immediatelely building and listening to my designs. Maybe this is because I am a chemist . much paper chemistry does not work if you try it in the reaction vessel.
As for you non complementary differental frontend: this is analysed in Horowitz and Hill, the Art of Electronics page236 in my edition.(fig 4.72 transistor power amplifier with negative feedback) .This typing I realise I missed the negative feedback in your schematic.
You can further develop this schematic to a full complementary frontend like the Son of Ampzilla schematic I draw attention to earlier on this forum.
I lke the diode string, as it is simple and easy to understand. I don't like the amplified diode detail as I had to repair too many ampliers with this structure totally blown up.
Hope this helps.:)
 
Hugobross

There is something not quite right about your schematic, which is particularly of concern since it is a captured screen of the simulated circuit.

Q1 is shown as an npn device but is labelled TIP2955, which is a pnp device. Conversely, Q4 is shown as pnp but labelled TIP3055. If Circuitmaker has automatically labelled the diagram, there is something wrong with the models you are using.

As a matter of interest, I have simulated the circuit in SIMetrix Intro using your component values and find very little difference betwen the +ve and the -ve peak voltages (in fact it shows the +ve slightly higher than the -ve). This further suggests that there may be a problem with your models.

Geoff
 
I'm sorry,

I forgot to mention I was trying to simulate for the first time the idea I've had in mind. So I just started with placing some transistors, diodes and resistors. Then I calculated some values of the resistors that would make the design "work"; and indeed there's a high current through the diff amp, I didn't noticed that because of the fact that a non-simetric signal appeared on the output. Indeed Nelson, I'll try to replace R7 by a 22K resistor, than I can also decrease the total value of R8 and R9 and make increasing of the amplitude of the input signal possible.

Geoff, I'm sorry for that,
There has been something wrong with filling out the specifications of the transistors, thanks, I'will check that out. hmm, it's weird that you've reached a higher +ve than -ve with SIMetrix, I've also SIMetrix here on my HDD somewhere, can you mail me your version ( [email protected] )??

Thanks for replying, but after starting this thread I've also tried a lot and I decided to design another amp to "prevent" the difference between the positive and negative amplitudes. So here's a link to the other simulation, again the resistor-values and the transistor specifications might not be perfect!!, but I liked the result very well.

amp design 2

There must be made some modifications to this design, such as adding a diff amp (or replacing the class A pre-amp by the diff-amp stage). If use the "class A pre-amp" I've to deal with the 180° phase shift when adding the diff-amp. I found this design having better resuts than previous one, but it will be a little harder to add the diff amp. Also C1 is now used to hold the DC voltages and currents. I'm going to try soon to replace the class A pre-amp by some diff amp design, and to try to obmit C1 by making Vc (Q5) 0V (reffering to the mass).

I would like to know your opinion about this design,

Hugobross
 

djk

R.I.P
2001-02-04 4:23 am
USA
As regards your first amp.Other than the wrong polarity vs part numbers on the TIP stuff, you realize that in real life you need a 100V rating on the drivers and the 2955/3055 are only 60V when forward biased? With R7=22K R6 should probably be 1.5K because the part you have chosen for Q5 is a darlington. Bootstrap current sources don't work unless the output stage is actually conducting, either stick another diode in your bias string or add a pot. Add a bias/flushout resistor between the emitters of Q1 and Q4, start with around 100ohms. And add the feedback loop.
 
I say it again, I'm sorry for the bad selection of the parts but I was just trying to simulate that kind of design. As I'm now, I was probably too tired to think clearly. I've just finished the examinations at highschool and I'm very very tired at the moment, so I will take some rest the following days. So I guess to see you in a couple of days.
I've just this question now: if you don't look at the values or the types of the components (because I know they're wrong), which kind of design do you prefer most, and which one do you think will have the smallest distortion if right components are used?

Thanks for replying, I'm going to bed right now...


HB....