Attached is a schematic of an audio amp that I've built, which was originally described at http://ludens.cl/Electron/audioamps/AudioAmps.html

I'm testing it by feeding it with 1Khz signal from a soundcard. I'm not getting any output from the amp. I lifted the connection from U1-6 and confirmed I'm getting good output from that first stage. As soon as I connect U1-6, I don't see much of anything from that point outward. I've double checked all the wiring.

I've got an o-scope and voltmeter. I don't have much experience with circuits like this output stage. I'd appreciate tips on what/where I should check and what I should be seeing.

TIA,
Barry

 

What do you see for voltage (DMM) and signal on base of Q3, collector of Q3 (which is also the base of Q1 and collector of Q4)?

Make sure Q4 is conducting, measure voltage between base and emitter, should be around 0.6v There should be 0.6v or so between base and emitter on ALL of the transistors. The collector voltage should be quite a bit higher than the emitter voltage for NPN transistors.

Voltage and signal at emitter of Q1 as well.

Once we have these numbers, a good start will have been made.

When I ask about signal, don't need a screen cap, just describe it as "1khz sinewave with xx.xx Volt peak-peak" or "clipped sinewave" etc.

 

Quick rundown to help you understand how it works:

Q4 is set up as a current source by the 2 diodes and resistor to provide a constant current to the base and collector of Q3.

Q1 and Q2 are a push-pull amplifier, pretty straightforward, the top transistor amplifies the positive going part of the input, while the PNP amplifies the negative going swing.

The "heart" of the amplifier is Q3, which is fed both the input signal and the voltage from the constant current source. This keeps the push-pull amplifiers biased to the working range of the transistors, as well as feedback (inverted from collector to current source) to prevent oscillation. There is also feedback from output to input via R4, which prevents oscillation.

The 10k variable resistor sets the quiescent current point for the outputs. If you measure voltages/currents you should be able to see the fault by following the "how it works" above.

Audioguru will also have some good input on this, I don't specialize in audio amps, and this one appears to be missing a few parts, but we'll worry about that later.

 

I see the 10 k resistor as the adjustment for idle current, not a volume control. It's an old circuit called "Vbe multiplier" or "rubber zener".

Do not try to use it as a volume control. Use it to set the idle current of the output stage as measured across either of the 1 ohm resistors. If you zero out the 10 k control, the natural results will be "no output". If you max it out, the natural result will be smoke.

 

The article you posted has an error:
the LM386 amplifier has a distorted output because it oscillates because the author did not use the RC network at its output to ground that is on ALL schematics in its datasheet.

The schematic you posted looks fine. Maybe the pins on its opamp or transistors are backwards. The output should be 0.6W at clipping (less than a cheap clock radio).

 

I see the 10 k resistor as the adjustment for idle current, not a volume control. It's an old circuit called "Vbe multiplier" or "rubber zener".

Do not try to use it as a volume control. Use it to set the idle current of the output stage as measured across either of the 1 ohm resistors. If you zero out the 10 k control, the natural results will be "no output". If you max it out, the natural result will be smoke.

Thanks, corrected.

 

Piece by piece, we eventually get it figured out.

Datasheet attached.

Datasheet shows a way to limit high frequency by adding a capacitor so the chip doesn't try to work at 200kHz. Optional: If oscillation, try capacitor.

 

What do you see for voltage (DMM) and signal on base of Q3, collector of Q3 (which is also the base of Q1 and collector of Q4)?

Q3-B 13.75V
Q3-C 13.75V

Make sure Q4 is conducting, measure voltage between base and emitter, should be around 0.6v There should be 0.6v or so between base and emitter on ALL of the transistors.

I measure .71V drop from emitter to base of Q4. I don't see a drop on Q3. Q1 and Q2 each have base at 13.75V and emitter at 6.34V

I'm not seeing much of a signal anywhere, other than the input.

I goofed and failed to set the bias with the 10K pot initially so might have fried something (Q3?). I wrongly thought I was supposed to apply signal first and adjust the pot for minimum distortion.

GREATLY appreciate the help!

 

It looks like the emitter circuit of Q3 is open, or...Q3 is fried, or...wiring error...

Back up and check for fried transistors. Then set the idle current starting with the 10k pot with zero resistance between base and collector of Q3. That will saturate Q3 so the bases of Q1 and Q2 are close together.

If you unplug the amp chip (disconnect the amp chip from Base of Q2) and put 620 ohms from BaseQ2 to ground, you will get a DC level near 6 volts there. With Q3 saturated, the base of Q1 should be about 3 tenths higher than 6 volts and Q1 emitter should be 3 tenths below 6 volts.

Now, Q2 shold be 5.7V emitter and 6V base. It's actually reverse biased. No current. Start moving the 10k pot adjustment. The voltage on the collector of Q3 will start rising. When it gets to about 7.2V, Q1 qnd Q2 will start conducting. Watch the voltage across a 1 ohm resistor. Try for about 10 millivolts to show 10 milliamps idle current.

Unplug the power.

Take out the (dummy) 620 ohm resistor and plug the chip back in.

It should work...but then I've said that so many times....

 

Thanks for the great help #12. I checked the transistors and Q1 looked bad so I replaced it. I put 600 ohms from base of Q2 to ground and disconnected the opamp. As I was slowly increasing the pot (and still hadn't seen much drop across 1-ohm resistor), I smelt something burning and quickly turned off power supply. The smell went away when I turned it back on but I notice the most voltage drop I'm seeing across the 1-ohm resistors is 1.8mV with collector of Q3 at 13.6V. I went back and rechecked transistors and it now looks like Q2 and Q3 were both fried in that "event".

 

I replaced Q2 and Q3. I'm concerned that I'm not seeing the voltages/biasing you predicted and I want to be cautious going forward. Would like to not repeat the transistor frying event.

With the 10K pot backed all the way off and with the 600 ohm resistor in place, I see Q2-Base 4.6V (not 6V) and Q2-Emitter 5.2V so it is FORWARD (not REVERSE) biased.

Q1-Base is at 5.3V , Q1-Emitter 5.2V , Q2-Emitter 5.2V

So is there any rethinking?

(fwiw, if I remove the 600 ohm resistor and reconnect the opamp, leaving the 10K pot backed off, I do see some amplification when I apply a 200mV test signal)

 

The output transistors are supposed to have an idle current of about 40mA. But adjusting the pot too far you probably had hundreds of mA. If the current was 800mA and the supply is 13.6V then the total heat in both output transistors was 0.8A x 13.6W= 10.9W.

 

Thanks very much AG. I got a milliamp meter in line with the supply and a millivolt meter across the 1-ohm resistor. As I slowly adjusted the pot, there was a surprisingly tiny transition region where the current just shot up. This adjustment requires great care (duh).

I was forced to substitute ZTX449 and ZTX549 for the output transistors. Does the 40ma idle current setting hold for these as well? I don't see it on the spec sheet. Is it necessary to adjust idle current for lowest distortion or just set it at 40ma and leave it? I haven't really been able to visualize the distortion on a scope or soundcard spectrum, as the original article implied.

Thanks!

 

The little transistors you are using have a low max current rating and also might get too hot.
With an output of about 1V RMS into a 4 ohm or 8 ohm load, adjust the idle current for the lowest crossover distortion. The idle current will be somewhere from 20mA to 60mA.

Usually the adjustment pot and fixed resistor have lower resistances and the pot has another resistor in series with it to reduce its sensitivity.

 

Thanks!

I don't have the greatest scope, but I'm able to slightly see the crossover distortion on a sine wave with the bias pot backed all the way down. As I slowly increase the bias pot and just start to see the current rise (5ma), the crossover distortion seems to disappear. So I'll just set it at slightly above that point.

 

Glad you got it working! I knew somebody else had the answers I failed on.

How does it sound, compared to an LM386 audio amp?

General Info: DigiKey and Mouser sell nifty little heatsinks for TO-92 packages, great for those times when the transistor does the job perfectly but gets warm. Just add some thermal grease and slip on, usually around $0.85 each, I use them with TO-92 H-Bridge pager motor drivers with excellent results, compared to the cost/bulk of going to a TO-220 package.

 

Haven't given it the final test yet - hopefully later today. Very grateful for all the help! I'm going to need a volume control for the test. Can a volume control be inserted between the opamp and base of Q2 or does it need to be upstream of the opamp?

Thanks for the heatsink tip.

 

Haven't given it the final test yet - hopefully later today. Very grateful for all the help! I'm going to need a volume control for the test. Can a volume control be inserted between the opamp and base of Q2 or does it need to be upstream of the opamp?

If you add a volume control between parts of the circuit then the DC biasing will be all messed up and cause no sound, smoke or severe distortion.

A volume contol is always connected at the input of an amplifier.
The volume control is an adjustable voltage divider of the signal source to the amplifier.
But it is too bad that this amplifier uses the low impedance inverting input as its input instead of the high impedance non-inverting input as its input. Then its volume control must also have a fairly low resistance, 3.3k to 10k. The volume control should have a logarithmic audio taper, not linear.

Since the emitter resistors have the fairly high value of 1 ohm each then they use up 1/5th of the output power. If the supply voltage is 13.8V then the amplifier will be clipping when its power into 8 ohms is only 0.85W (like a cheap clock radio).

For a comparison, A TDA2003 power amplifier IC produces 2.4W into 8 ohms with this 13.8V supply.