Using Many Small Signal Transistors Paralleled in place of one output Transistor

Hello everyone,
As you may have guested, I have some unusual ideas when it comes to electronics.
Well hear's another one.
A few months ago i decided to see if instead of using the usual power / driver transistors in my amplifier, I could Parallel small signal transistors (example BC546) together with low value resistors ( current sharing ) and see what the effect would be.
so i parallel 20 BC446's and 20 BC556's together to take the place of the normal output power transistors and 3 each paralleled in the same manner for the driver transistors to see how it would sound.
As i decided to run the amplifier at 16volts rails and only a single + and - supply with a output eletro capacitor mainly to see if it would work.
It did and the sound was ok. it sounded a bit tinny and i used it for a couple of days before going back to my old setup.

In the upcoming months i plan to build a set up with 100 signal transistors in place of one power transistor and run it with a split supply of 28v / 28v and see what the effect would be on the sound.
i am interested in seeing how hot each transistor gets and i may even try the idea with the JLH 10Watts CLass A Design and see how hot the transistors get in a Class A.

Has anyone else already tried this idea and if so how did it go?
 
What you are doing is commonly done on modern power output transistors with high Beta and Ft. On the die, they are actually many small transistors in parallel.
Don't even think about using 100 parallel small signal transistors in class A as the maximum power dissipation of 100 transitors under IDEAL conditions would be about 25W, under real conditions it's more like half that. You could make things better with cooling but not that much better - unlike power transistors that distribute heat from the crystal through a copper slab, small signal transistors do this through plastic, which is a hell of a lot worse.
 
LOL!

You could use just a fan blowing on the PCB to cool them instead of using heatsinks. :D

Instead of using small signal transistors, why not use TO-220 devices? They are small and can deliver current to loads much better, with less heat made due to losses. They would run cooler and dissipate heat better than TO-92 especially with airflow.
 
ilimzn said:
What you are doing is commonly done on modern power output transistors with high Beta and Ft. On the die, they are actually many small transistors in parallel.

Really??? Are MJL4281/4302 like this. I noticed they have a LOT of gain, and a fairly high FT for such a high power transistor. Also what makes me wonder is that the 350V/15A rating.

Is this also done for things like power diodes and other semiconductors to have multiple parallel junctions on one die?
 
I used this approach for the output amp of an signal generator,
which had to work up to 1MHz.
10x BC550 & 1x BC560 biased with about 7mA each transistor
and operating from +/-12V.
It worked fine, but I have no idea how it would sound.

The attached schematic show the output amplifier, but please be
aware that the input stage is not good for HiFi, because the emitters
are referencing to the rails and any noise in the rails will be found
in the output signal too. For the signal generator it was OK.
For HiFi I would prefer some diamond input stage, which is referencing to signal ground !!!
 

Attachments

  • signal_amp.jpg
    signal_amp.jpg
    35.6 KB · Views: 2,476
  • Like
Reactions: 1 user
EWorkshop1708 said:


Really??? Are MJL4281/4302 like this. I noticed they have a LOT of gain, and a fairly high FT for such a high power transistor. Also what makes me wonder is that the 350V/15A rating.

Is this also done for things like power diodes and other semiconductors to have multiple parallel junctions on one die?

For me the data sheets look typical for an internal parallel structure
like ring emitter BJTs...
The combination of high gain AND high tf is indicating this.


BTW:
Sorry for the typing error in my last post. Of course there
are also 10x BC560... (not just 1....)
:rolleyes:
 
Four TIP41 and four TIP42 can turn a very good output pair

Using 0.47 ohms resistor to each emitter, and having the care to buy them with not too much different gain, as emitter resistors will not make miracle, you will be happy with 260 Watts of maximum dissipation and 24 ampéres maximum capacity, around 100 Volts maximum voltage.

I have my suspections that parallell transistors, made this way, can produce more distortions than a single unit.

But this is my "feeling"...i am sniffing something but have nothing written on books to be the base of my worries.

I use to do that, when needed, and result good for ears.

I never asked the Scopes, or distortion analisers, what they think about the audio produced...and i have strong suspections that Scopes and Analisers will not answer me.

regards,

Carlos
 
ChocoHolic said:
I used this approach for the output amp of an signal generator,
which had to work up to 1MHz.
10x BC550 & 1x BC560 biased with about 7mA each transistor
and operating from +/-12V.
It worked fine, but I have no idea how it would sound.



Me too, just build some sort of power stage for function generator, mine went straight to 10Mhz. AD817 and emitter follover stage inside feedback loop. Stability, no problems.
I had strange experiences when i tried to bias emitter follower made of 10x bc337 and bc327 in parallei each having 5 ohms emitter resistors. Without any sort of biasing it was working fairly well with large signal, but even couple of milliamperes bias resulted weird waveforms. Has something to do with base charge "suckout"??
 
sa:
:nod: :nod: :nod: :nod:
...but powerwise not really satisfying? Right?


mzzj:
"base current suckout" ?
Sorry, I never observed something like this.
But if the Ube of your transistors were quite different, and
you tried to bias with low voltage drops across the emitter
resistors it could look like like this.....
The BJTs with low Ube will carry the idle current while the others are
still passive. When paralleling BJTs you should select for identical
Ube, or work with with high values of emitter resistors.
 
Yeah Choco, you are rigth, and i feel good reading all that knowledge.

Yeahh!.... you can capture.... very nice.

And my daugther lonely.... pretty girl, sad.... alike a ship encraved in a sand bank...so good girl she is.

And a lot of handsome single guys in our forum.... oh!..life is not fair.

Cheers,

Carlos
 

Attachments

  • thata.jpg
    thata.jpg
    21.9 KB · Views: 1,201
ChocoHolic said:



mzzj:
"base current suckout" ?
Sorry, I never observed something like this.
But if the Ube of your transistors were quite different, and
you tried to bias with low voltage drops across the emitter
resistors it could look like like this.....
The BJTs with low Ube will carry the idle current while the others are
still passive. When paralleling BJTs you should select for identical
Ube, or work with with high values of emitter resistors.

in order to turn-off BJT faster it is usefull to reverse-bias base-emitter junction to "suckout" base charge. I Thought that it may be the reason to my output stage's strange behaviour biased vs. unbiased case. Dunno, maybe i should start new thread about sucking the base out...
 
....you mean that you had issues with the storage time
of your BJTs... ?
Ok, storage time has heavy tolerances, but I would not expect
trouble with this when you use a BC327 in linear mode (not switched to saturation) at HiFi frequencies....

any scope pics or description of the effects that you had observed?