Help with conversion of Heathkit AA21 from germanium to silicon outputs

Thread Starter

MagicBlueSmoke

Joined Jun 20, 2022
7
Greetings.

This amplifier has 8 output devices. My goal is to replace all 8 PNP germanium output devices with equivalent PNP silicon devices.
I realize changes to the circuit will be necessary to achieve correct bias. I need help with that, and also with selection of suitable silicon devices.

I am uploading a circuit description, along with a schematic of the output stage for one channel.
My thoughts are to replace all of the output devices with MJ21195G. Way overkill, but readily available, relatively inexpensive, and likely more tolerant of mistakes I might make during the process of re-biasing.
CircuitDescription.jpgLeftChannelOutput.jpg
One thing I am concerned about is the series-arranged circuit, where X10L (2N2147) drives X9L (34425).
Can the MJ21195G be used in place of both the 2N2147 and 34425?
TN2147
34425: PNP 50V 5A 13W 4MHz (This is the only data I could find.)
MJ21195G

Also, being only casually knowledgeable about circuits, I am not confident on which resistors I need to replace.
My thinking is R75 (270 ohm) and R69 (100 ohm) to bias X9L. R71 (270 ohm) and R73(100 ohm) for X11L in the feedback loop. But I am at a complete loss as to which resistors are used for biasing X10L and X12L.

I would be very grateful if you would take a look at the photos and offer some guidance.

Thanks.
 

MrChips

Joined Oct 2, 2009
26,537
I would not do that. You cannot easily change from germanium to silicon transistors in an audio circuit. All the bias voltages will have to be changed.
 

MrChips

Joined Oct 2, 2009
26,537
On second thoughts, after taking a closer look at the schematic, perhaps there is another solution.

Since the output stage is transformer coupled I would eliminate the entire output stage and use a balanced pair of NPN/PNP transistors in common collector push-pull configuration.

Hang on while I search for a ready-made circuit off the web.
 

Thread Starter

MagicBlueSmoke

Joined Jun 20, 2022
7
Is it worth all the effort, when there are ready made amplifiers of better quality than this one, you're going to have to reverse the diodes and electrolytic caps as well as the supply..?
Actually yes, very much so. This a personal project I took on to become proficient with the principles of circuit design. It was a rare kit in a sealed box that I assembled. Since it is 50+ years old, I replaced all of the caps and resistors with equivalents or better. Only later did I find that some of the germanium outputs were bad or poorly matched.

As far as reversing the diodes and caps - since I will be using equivalent PNP devices - why would this be necessary?
Thanks for joining in on the conversation.
 

Ian0

Joined Aug 7, 2020
5,571
The voltage across D2 and D3 sets the bias along with R77 and R79.
Are D2/3 silicon? If so they set the bias as (Vdiode-Vbe)/R77
You might need two diodes in series or a silicon junction diode in series with a schottky.
By the way, the cascoded output stage is not superfluous, because you have transistors with sufficient voltage rating, so X9 and X11 can be deleted.
 

Thread Starter

MagicBlueSmoke

Joined Jun 20, 2022
7
The voltage across D2 and D3 sets the bias along with R77 and R79.
By the way, the cascoded output stage is not superfluous, because you have transistors with sufficient voltage rating, so X9 and X11 can be deleted.
That's actually a pretty great idea. I believe the diodes are silicon.
Thanks for joining in on the conversation!
 

Ian0

Joined Aug 7, 2020
5,571
That's actually a pretty great idea. I believe the diodes are silicon.
Thanks for joining in on the conversation!
I‘ll correct that - The cascoded stage IS superfluous. (Too busy persuading the iPad that “cascode” is a word to notice what else it is doing)
 

ronsimpson

Joined Oct 7, 2019
2,172
Because you moved to PNP & NPN you will need to reverse the phase on the drive. See red dots.
Next time draw with +35V at the top and -35V at the bottom.
1655741598731.png
 
The two-diode biasing idea is a good one, but you will require more, as it would barely bias the transistors in class-B, but you really want class-AB, with some idling current, which itself should be adjustable.
Search “Vbe multiplier” for an adjustable bias circuit.

One last thought, the bias circuit should be in intimate thermal contact with the output transistors
 

MrChips

Joined Oct 2, 2009
26,537
Because you moved to PNP & NPN you will need to reverse the phase on the drive. See red dots.
Next time draw with +35V at the top and -35V at the bottom.
View attachment 269873
I intentionally drew the A and B supply voltages that way because that is how the original circuit is drawn. I did not want to confuse the TS.

Check the phasing of the interstage coupling transformer windings. I believe my drawing is correct.

Agreed, the node between D1 and D2 may need a resistor to COMMON plus one or two added series resistors for Class AB biasing.
 

Thread Starter

MagicBlueSmoke

Joined Jun 20, 2022
7
Would it be possible to convert to a class A output stage?
Perhaps using larger filter caps (currently 3000uf/50v) and some additional heatsinks?
I don't need it to play loud, but improved fidelity is always nice.
 

Ian0

Joined Aug 7, 2020
5,571
Would it be possible to convert to a class A output stage?
Perhaps using larger filter caps (currently 3000uf/50v) and some additional heatsinks?
I don't need it to play loud, but improved fidelity is always nice.
Just give it more bias voltage - that's all that is required.
However, the lack of fidelity is due much more to the reduced phase margin because it is transformer coupled, than to the biassing.
 
Top