I had this old transistor coat pocket radio that died on me so I used some of the parts to build a copy of its audio section to use as a bench test amp (the original audio section worked well enough so why change things?). I wanted to build another test amp because my "Radio Champ", as I built it, isn't really all that safe (I really need to find a power transformer for it). It will deliver about 1/2W into 8 ohms.

The transistors and transformers are very common and can be found in virtually any Japanese-made radio from the 1960s, and the audio sections of these old radios are all essentially the same. Most of them also included the thermistor, which will be about the same value as the resistor it is in parallel with. The 2 ohm resistor and the thermistor are thermally coupled. The thermistor has a negative temperature coefficient (its resistance drops as it heats up); if the 2 ohm resistor gets too hot the thermistor will sense it and shut down the power amp until it cools off (the amp will start to make a terrible sputtering sound and fade out). The NTC is not absolutely necessary, though. If you don't use it, change the 220R in parallel with it to 100R or so.

Here is the schematic and layout. Transformer sizes may vary.

And some rather large pics. The scope pics are using a 1x probe.
Scope trace-- amp volume at max, input signal level set to just before the amp starts clipping (~45mv peak)
Scope trace-- amp volume at max, input signal level at max (~1.4v peak)

Turning down the amp volume by about 1/3 from max allows it to faithfully reproduce the full test signal voltage.

Have fun.
Last edited by Invader Jim at Jan 14, 2015,
The thermistor/resistor combo is pretty cool, I've normally just used NTCs to measure temperature with MCU based circuits.

The circuit seems fairly simple, do you have an idea of how much current is drawn? Off hand I'd guess it would be about 250mA or so to get 3W of power output. If I can find good transformers I might have to try and build one of these.

Any thoughts on modern replacement parts? I don't like working with unobtanium parts so much anymore
Broken Japanese-made radios are still quite common at flea markets and such (they were so good at making things cheap they flooded the market and millions of these things were sold). Pretty much all Japanese audio gear used the same transistor types (2SB175 are extremely common) and the audio sections from the radios are all essentially the same, give or take an extra gain stage or two.

You can build such a circuit with modern transistors though. Here are a couple of links worth checking out.
http://www.angelfire.com/planet/funwithtransistors/Book_TS_Contents.html (scroll down to Chapter 7-Audio Amplifiers for a working circuit using modern silicon devices)

As for the current draw, it uses about 63mA at full-tilt (max volume and max input signal; see OP) and 8mA idle at 9v.
Last edited by Invader Jim at Jan 4, 2015,
The parts wouldn't be too big of an issue to find, but the place I work at puts in orders for parts almost daily for our projects so we get good prices on almost any parts. Because of this I usually just get guys to add the parts I would need for projects at home into the orders sometimes and give them cash to pay for the parts.

I could probably get all of the parts to build one of these excluding what I already have for $20 or less. I'm guessing a lot less if the transformers aren't too expensive.

I'll have to find some transistors and adjust the circuit for those transistors so I can build one of these. It looks like a nice low power home practice amp. I need to simulate it in LTSpice and see what the frequency characteristics of the amp look like.

EDIT: If you get a chance would it be possible for you to measure the voltages at the three terminals of the transistors in the circuit? If its too much work it isn't a big deal, but I'm doing some calculations and the only data sheet I've been finding isn't in English so I might just back calculate some parameters if you can get the voltages.
Last edited by XgamerGt04 at Jan 3, 2015,
It's no trouble at all. With -9.00v at the board, -8.33v at the first filter cap and -7.34v at the second filter cap, no input signal, the voltages at the transistor pins are as follows:

E -0.32v
B -0.44v
C -5.96v

E -0.91v
B -1.10v
C -6.96v

Q3 &Q4:
E -0.0073v
B -0.179v
C -8.32v

These voltages were taken with a 10M DMM. Note: the component designators DO NOT refer to the layout pic. I made the layout "backwards" because the power section was the easiest to do.

Anything else you need while I've got the hood up?

And I just did some Googling and found out that my output power measurement is completely wrong (feel kinda stupid right now ) but yeah, the amp will deliver about 1/2W into 8 ohms...
Last edited by Invader Jim at Jan 4, 2015,
You can do output power as you mentioned, but you have to use the RMS voltage. In your case with the square wave your RMS voltage is ~1.4V so your power output is ~250mW if I did the calculations correctly.

So this amplifier spends most of its life clipping. I'm a bit surprised by that and would have expected it to be cleaner being a solid state design.
Well the amp volume and input signal were both dimed. Turning down the volume about 1/3 from max cleans it right up. Diming the amp can give some grit from a guitar or phono but the thing is pretty loud so you wouldn't often have it turned full-up.

I found another method for measuring output power and it made a lot more sense.


Ep is the peak voltage across the load and R is load resistance. It's only valid for a sine wave into a purely resistive load though but people seem to use it as a "close enough" approximation for the real world. Idk.
Last edited by Invader Jim at Jan 7, 2015,