What Do You Think Of This Power Amp Design?

wayneh

Joined Sep 9, 2010
17,496
I think you’ll get more feedback if you attach the schematic as an image. Folks are reluctant to open attachments.
 

Marley

Joined Apr 4, 2016
502
What do I think of it?
Well there are better circuits - just Google.
My first thought is that the quiescent current in the output transistors are not controlled so you will get cross-over distortion or possibly thermal runaway. The output transistors are not very well driven with their base current being sourced through the 100ohm resistors.
 

Marley

Joined Apr 4, 2016
502
Here is the classic circuit. Long-tailed pair input. Reasonably high input impedance. Class A stage in the middle where most of the amplification happens. Miller feedback capacitor on the class A stage for HF roll-off and stability. The overall audio voltage gain of the complete amplifier is set by Rfb and R4. The 100uF capacitor in the feedback loop gives 100% feedback (so unity gain) at DC, centering the output at 0V.

The output transistors could be darlington pairs. The 100uF capacitor is "bootstrap" feedback and gives better drive to the NPN output transistor on positive peaks. The network of the transistor and two resistors between the bases of the output transistors sets the bias of the output transistors. This transistor should be in thermal contact with the output transistors so that the bias is reduced when they warm up. R2 is normally variable. Set this to zero when first powered-up. Increase slowly to give a small quiescent current through the output transistors.

I have just guessed the resistor values. Try this out in the simulator!

Just noticed a mistake! there should be another 100ohm resistor between the bootstrap feedback capacitor and the base of the NPN output transistor.
 

Attachments

BobTPH

Joined Jun 5, 2013
8,813
The simulation shows it to be quite good. You can get 10V p-p output with very low distortion, the worst harmonic on FFT is at -40dB.

This is a discrete variation on amps I have seen that use an op amp with resistors in the V- and V+ paths providing the base driver for a Darlington push-pull output stage.

Bob
 

Thread Starter

newbie2019

Joined Apr 5, 2019
95
Here is the classic circuit. Long-tailed pair input. Reasonably high input impedance. Class A stage in the middle where most of the amplification happens. Miller feedback capacitor on the class A stage for HF roll-off and stability. The overall audio voltage gain of the complete amplifier is set by Rfb and R4. The 100uF capacitor in the feedback loop gives 100% feedback (so unity gain) at DC, centering the output at 0V.

The output transistors could be darlington pairs. The 100uF capacitor is "bootstrap" feedback and gives better drive to the NPN output transistor on positive peaks. The network of the transistor and two resistors between the bases of the output transistors sets the bias of the output transistors. This transistor should be in thermal contact with the output transistors so that the bias is reduced when they warm up. R2 is normally variable. Set this to zero when first powered-up. Increase slowly to give a small quiescent current through the output transistors.




I have just guessed the resistor values. Try this out in the simulator!

Just noticed a mistake! there should be another 100ohm resistor between the bootstrap feedback capacitor and the base of the NPN output transistor.

Thanks for your reply. It doesn't work in the sim.
 

Attachments

Jony130

Joined Feb 17, 2009
5,487

Attachments

crutschow

Joined Mar 14, 2008
34,285
The simulation shows it to be quite good. You can get 10V p-p output with very low distortion, the worst harmonic on FFT is at -40dB.
That may may an artificially low value due to the long minimum timestep.
If you set it to a shorter value, like 1μs, the distortion value will increase due to the apparent better resolution of the harmonics.

Another factor is using a simplified model for Q5 and Q6, which have a constant Beta value.
More complex models will have a change in Beta with current, which adds to the distortion.
 
Last edited:

BobTPH

Joined Jun 5, 2013
8,813
Where is a minimum timestep set? I see a setting for maximum timestep ar 10ns. Setting this back to default does not change the result.

Bob
 

BobTPH

Joined Jun 5, 2013
8,813
Just changed the models to ones I got from ON Semi website. Same results.

Edit: the efficiency, however, is dreadful. 17.75W used to get 4W out.
Bob
 
Last edited:

crutschow

Joined Mar 14, 2008
34,285
Still the same result. I have never entered anything there for any LTSPICE simulation I have done.
My simulation of a slight modified circuit did give a high total harmonic distortion when I used 1μs.

Occasionally I have had simulations show odd distortion in a sinewave, and adding a maximum timestep cleans it up.
 

BobTPH

Joined Jun 5, 2013
8,813
My simulation of a slight modified circuit did give a high total harmonic distortion when I used 1μs.

Occasionally I have had simulations show odd distortion in a sinewave, and adding a maximum timestep cleans it up.
Is there a way to get THD in an LTSPICE simulation?

Bob
 

Audioguru

Joined Dec 20, 2007
11,248
This power amplifier circuit with many parts produces only 10W into an 8 ohm speaker with considerable distortion. Then why not use an inexpensive higher output power and lower distortion IC amplifier instead?
Texas Instruments and others have class-D amplifier ICs that do not waste much power by getting hot.
 

crutschow

Joined Mar 14, 2008
34,285
Is there a way to get THD in an LTSPICE simulation?
You do a .four (Fourier) SPICE directive (below).
The declared frequency is the same as the input source SINE frequency, and v(Vout) is the output node voltage.

upload_2019-4-18_23-13-47.png

To see the results look at the SPICE Error Log under View.
It gives you the value for the first 9 harmonics plus the total harmonic distortion.

upload_2019-4-18_23-17-23.png
 
Last edited:
Top