Germanium transistor amplifier improvements ?

[Jolly 7]

Here is a push pull amplifier I put together using three vintage Hitachi transistors. It works ok but is not going as loud as another one using the same transformers and silicon transistors. Also it is consuming a lot of current, 86 mA :-(
Any suggestions for improvement please ?
I should add that the output transistors are also getting warm to touch soon after being powered on.

[Herald1360]

Unless that lower resistor in the bias chain is effectively turning the 1N4148 off, the bias will be far too high for Ge devices. You may get away with just resistive bias control, but there're good reasons why NTC thermistors often featured in the better circuits.......

Also- your lower output transistor is drawn as NPN!

Edit: The bias voltage with the resistor values shown is around 540mV- way too high. Around 150-200mV would be better, but this is one area where a twiddlypot in the upper resistor position is a good idea. Upper so that poor wiper contact doesn't overbias the output stage.

[livewireless123]

I think it's taking more current than it should because the bias is too high,
The bias diode should be ge as well. I have been playing with a similar unit
using old 2sb56 types in push pull and am using a pot' in the bias network to
adjust the IQ setting it so that the crossover distortion just goes away on
a signal. you should be able to get away with a standing current of only about 5 or 6 mA if the transistors are fairly well matched.
It may get a bit louder if you give it a bit more voltage, say 6V or so.
Hope that helps, Mike.

[Refugee]

You will need to replace that silicon diode with a germanium type as it looks like it is supposed to be a thremal regulator. There will be some value in fitting the diode on the heatsink too.

[livewireless123]

The junction of the 220R resistor and the driver tranny should be de-coupled to deck with a cap' in the usual way, this may improve the performance.

[Peter.N.]

I vastly improved my 'Berec Demon' audio quality by swapping the standard OC78s for 81s but that was in the very early transistor days when they were still developing quite rapidly.

Peter

[Argus25]

I agree, germanium transistors need a germanium bias diode that you can make from a 2SB77, just connect its base and collector. Likely as it stands there is far too much output stage bias current with the germaniums and explains why the silicons seem better. With the correct diode, normally you would adjust the value of the 1K resistor for the correct output stage bias current, and not have a resistor in parallel with the diode. Sometimes a bypass capacitor can be used in parallel with the diode, but mostly that is not required.As noted in post 4 the bias diode should be near the output transistors, on the heatsink (if there is one) but there might not be for a low power 3V amplifier like this.

Also, there is a missing component, an electrolytic bypass cap say 100 to 470uF after the series 220R resistor, without this the gain of the amplifier will be substantially reduced.

The collector current in the driver should be in the range of 1 to 2 mA and maybe about 8 to 10 mA for the output stage, overall current(with no signal) should be far less than 20mA and probably less than 15mA and ideally around about 10mA.But the quiescent output stage current in some transistor radios may only be 5 or 6mA, you just need enough to eliminate the cross over distortion.

[paulsherwin]

You can get a rough indication of required component values by looking at any 1960s transistor radio circuit where transformers and OC81 or AC128 transistors are used.

[Jolly 7]

Thanks everyone for the ideas. I will use the B-C junction of an ASZ21 transistor as the germanium diode instead of the 1N4148 as I have quite a few of these and make other modifications as suggested.

[livewireless123]

This might be useful, An eagle circuit for a fairly standard 60s transistor radio.
good luck with your efforts Mike.

[Jolly 7]

Thanks for the circuit diagram ☺

[Herald1360]

Note the NTC bias stabilizer....

[Jolly 7]

Originally Posted by Herald1360

Note the NTC bias stabilizer....

Noted, thanks. Since I don't have this thermistor, I have used a Ge diode instead made from an ASZ21 transistor, with no connected resistor parallel to it.
I have also added a 220uF cap in between the driver transformer primary and the series 220 ohm resistor. I will next look at changing the bias resistors. I think the 15K in my diagram will need to be increased to around 27K and the 12K one halved to around 5.6K, as per the Eagle diagram. Then I will use a 1K pot instead of the fixed resistor. Hope I have got this right.

[Argus25]

One quick way to measure and adjust the bias current in the audio is to put a volt meter across the emitter resistor, and from the voltage measured across the resistor and the resistor value, you can get the current (ignoring base current) without having to break the circuit to insert the current meter.

So adjust the 15k resistor on the driver transistor until there is roughly 0.7V across the 470R emitter resistor and adjust the 1k (with the germanium bias diode in place) until there is about 15 or 20mV across the 2.2 Ohm resistor.

[Jolly 7]

The 15K resistor needed to be increased to 22K to get a reading of 0.69 V across the 470 ohm resistor. The 1K resistor had to be increased to as much as 18K to get 21 mV across the 2.2 ohm resistor. But previously where the amplifier was amplifying a little (like picking up a hum while touching one of the audio inputs), it's almost stopped doing so after these changes. Is there any other modification I need to make ? Do I also need to change the 12K resistor near the volume control potentiometer ?
I should add that I can barely hear some white noise only when fully turning up the volume and putting the speaker next to my ear.

[Jolly 7]

I will need to look at the 220 uF capacitor connections, they may not be correct. The Eagle diagram seems to omit an electrolytic in favour of a 200 of one to ground.

[Argus25]

The supply filter (bypass cap) is C2 on the Eagle diagram.

It is worth injecting a 1kHz tone from a sine wave signal generator to the input of the amplifier or volume control, to see that the amplifier is working with a solid output to the speaker.

Normally, with these sorts of tests either an AC voltmeter or scope would be put across a dummy load resistor (instead of the speaker) to see the output sine wave and then easily calculate its power. That way you don't drive other people up the wall with the loud tone with the speaker connected.