I was able to get some suggestions here on my initial design for an audio amplifier and I then came up with this one.


I made it dual supply and had the transistors to be more robust like the 2N5551 and 2N5401. The design is pretty much the same with what is shown on Bob Cordell's book. The problem is I am not reaching the 15W output requirement. Another is that I am thinking to change the output transistors Q12 and Q14 to TIP2955 and TIP3055, much cheaper and I don't think I need a very big output transistor like the MJL's as seen on the circuit for my 15W requirement. If I change to the TIP pair, what driver transistors should I use? is BD139 and BD140 a good one?

I am also open for what I should change on the design. I attach the LTSpice file.

 

at first glance:
C7 is too small - 50 picofaraf, usually it's a fraction of microfarade
R13, R14 -33 Ohm too big, usually it's about 1 Ohm
Substitutes TIP2955 and TIP3055 can work, although, it's better to find a quicker transistors. Check current gain bandwith product of the substitute. You was right, a lower power transistors can be used

 

at first glance:
C7 is too small - 50 picofaraf, usually it's a fraction of microfarade
R13, R14 -33 Ohm too big, usually it's about 1 Ohm
Substitutes TIP2955 and TIP3055 can work, although, it's better to find a quicker transistors. Check current gain bandwith product of the substitute. You was right, a lower power transistors can be used

I see thanks, my R13 and R14 are actually 0.33's. As for C7's value, I'll consult the book again. Cordell's book said that is to be:
C_miller = 1/(2*pi*f*R_LTP) in which the R_LTP is the re' + RE of the LTP.

re' at 1mA per leg of the LTP would be 26ohms so 10:1 degeneracy would be 234 as the book showed. Should I still make C7 higher though?

 

C7, R23 typical values are 0.1uF and 1 Ohm. In most cases those components affect amplifier's performance a little. It can work without them, in most cases.

 

C7, R23 typical values are 0.1uF and 1 Ohm. In most cases those components affect amplifier's performance a little. It can work without them, in most cases.

Typical values are 47nF and 10Ω, not 1Ω and 100nF.
Without them, the amplifier is likely to be unstable with inductive loads, and all loudspeakers are inductive loads. The resistor should be the same value as the nominal impedance of the amplifier it is supposed to drive, so the overall load impedance looks mostly resistive at high frequency. 1Ω would overload the amplifier at high frequency, so would adversely affect the amplifier’s performance. Using a better choice of resistor value does not affect the amplifier‘s performance

 

I am also open for what I should change on the design. I attach the LTSpice file.

Have you checked it for stability? (Gain margin/phase margin)
@ericgibbs told me of a way of doing this by inserting an AC voltage source between the base of Q2 and the point you have labelled NFB, then running an AC analysis and plotting Vout/Vnfb.
You can check then effects of using TIP2955/3055 vs. Something quicker such as 2SC5200/2SA1943, and the effect of different values for C2.

 

Why is your simulation at 20kHz? I usually test at 1kHz.
I could not find your "trans" command. The simulation shows millions of cycles instead of only a few detailed sinewaves.

An LM1875 amplifier IC costs $6.54 at Digikey and produces 25W at very low distortion. It also has protections in it.

 

Have you checked it for stability? (Gain margin/phase margin)
@ericgibbs told me of a way of doing this by inserting an AC voltage source between the base of Q2 and the point you have labelled NFB, then running an AC analysis and plotting Vout/Vnfb.
You can check then effects of using TIP2955/3055 vs. Something quicker such as 2SC5200/2SA1943, and the effect of different values for C2.

Thanks! I'm gonna try this one once I get back home. I decided to go for the 2SC5200 and 2SA1943 transistors. My drivers would be the same, MJE243 and MJE253. I do not know if this one's a good combination though.

 

Why is your simulation at 20kHz? I usually test at 1kHz.
I could not find your "trans" command. The simulation shows millions of cycles instead of only a few detailed sinewaves.

An LM1875 amplifier IC costs $6.54 at Digikey and produces 25W at very low distortion. It also has protections in it.

I am testing both cases acutally, I test at 1kHz and the other is 20kHz, my previous designs has oscillations at greater than 1kHz, this time, I am testing even the extreme 20kHz just in case.

As for my trans command, this is what I hate about LTSpice if you use the interface for the Run command and fill in the boxes for the needed time. The trans command is below where I defined my transistor models. It gets placed there automatically for some reason.

 

Your Tran setting causes the simulation to show millions of cycles and take a long time to settle.
Here is an earphones amplifier also at 20kHz like your amplifier but Tran is set to show one cycle with details that show no clipping but showing a little crossover distortion.

 

An LM1875 amplifier IC costs $6.54 at Digikey and produces 25W at very low distortion. It also has protections in it.

30 W amplifier kit for $1.59 only:
LM1875T mono boom level power 30W amplifier board speaker power amplifier PCB production DIY kit LM1875
US $1.59
free delivery

The Chinese amplifier board shows "LM1875 and TDA2030A". The TDA2030A is obsolete so I bet that is what you get.
None of the kits show any writing on the IC.

Especially for @Audioguru again:

 

The Chinese amplifier board shows "LM1875 and TDA2030A". The TDA2030A is obsolete so I bet that is what you get.
None of the kits show any writing on the IC.

 

I still have a product in production that uses a couple of TDA2030's. At US 39¢ each in single quantity and 25¢ in 1,000's they are a great (inexpensive) way to make medium power amplifiers for audio frequencies. They may be obsolete to some people but the distributor my client is using has 4,685 in stock. Electronic Source Not all parts designed last century are obsolete, they are classics!