simplest amplifier possible with BJT's?

[rmgvs]

simplest amplifier possible?

The question how to build the simplest possible amplifier using transistors intrigues me. I will report here my recent findings. To share the information with you and to hope for still further improvements.

My design criteria were:

* output power no real issue, around 1-5 Watts RMS is okay (translates to: 5 Watts would be great).
* input impedance above 33kOhm, sensitivity of around 1 Volts RMS or higher.
* two gain stages max.
* no fets or mosfets, just BJT's.
* class-A
* single-ended
* no global feed back

I came to this set after visiting Rudi Stor's website (www.rudistor.com/sound-lab). Here was the AR-3 head phone amp designed for simplicity and 8 ohms loads. I built this amp for fun and it sounds (contrary to my expectation) very very good with whatever phones you like (expensive Grado's, Sennheisers, AKG's, Sony's). I have several tube head amps for comparison (musical fidelity, Morgan Jones, Futterman). Actually, this is the first transistor amp that really intrigued me. The amp delivers around 50 mW RMS @ 8 Ohm. Just for fun I hooked up my Klipsch speakers: not bad at all.

So ....

Why not beef this amp using only 3 transistors (of which 2 2N3055's) to a few watts? I used around 1,25 Amps of idle current and an emittor resistor of 8,2 Ohms (50 Watts, becomes very hot of course) and a voltage of around 20-25 V DC.

The output power now is around 3/4 (ie 0.75!) Watts RMS. This is the best transistor amp I ever heard in my house. It betters the Hiraga and the John Linsley Hood, my references in simple class A transistor amps. Actually it rivals the better tube amps.

Now I started the quest for even more power. The power is limited in the design by the input stage that can not give enough swing. I changed some resistor values to get more swing and ended with around 3 Watts RMS (but a lower input impedance). The magical sound was gone however. I fiddled around using several other input transistors that can deal with more current. I did not succeed in getting the sound as good as I wanted it.

After more than one month of listening, soldering, experimenting, I finished the result ending up with one of the first trials, leaving me with around one (1) Watt of power.

If only this design could be brought to about 5 Watts of output ...

Who wants to take over the fiddling from me, I think this is it for me at this moment given the design criteria...?

Rudy

 

[Nelson Pass]

I do like a challenge.

Select both devices for high beta. I assumed 50
on the power device, and 100 on the driver.

 

[John Law]

NP, will your circuit work?

With a +30V supply rail and the collector of the output transistor at +10V, the output stage quiescent current will be 2.5A. If the output transistor has a gain of 50, its base current will be 50mA. The base of the output transistor will sit at about 1.9V so there will be a current flow through the 220R resistor of nearly 9mA. The total current through the input transistor should therefore be 59mA.

The maximum available voltage across the 1k collector resistor is somewhat below 28V so the maximum current through it will be under 28mA, less than half that required.

If output transistor collector voltage is increased to 20V, which will halve the quiescent current (this will still be sufficient for 6Wrms into 8ohm), the required current flow through the input transistor will be reduced to 31mA, still greater than that available through the 1k collector resistor. Also, the peak positive output excursion becomes less than that required to meet the power specification. Adjusting the values of the output stage emitter and collector resistors to reduce the quiescent current, whilst retaining the same gain and collector voltage, will overcome the voltage excursion limitation but will produce a similar result with regard to the input transistor current.

The conclusion? The 1k collector resistor must be significantly reduced in value (or omitted completely).

Also, the input impedance of this circuit will be around 2k5, considerably less than the 33k Rudy was seeking.

 

[Gabevee]

Mr Pass,

I totally agree with the "high beta" statement, but I think that a power device of a beta of more than 100 would be in order, since the first stage could be the gain device, where the power out could be a follower for better load handling. The base current from the output device would load down the output of the first stage, IMHO.

I made a voltage regulator once and needed to add a transistor as a darlington pair in order for the load, which was high current, not to bring the regulator into a low voltage condition, since the base current was high enough to change the voltage at the base.

But there are power transistors with betas higher than 100 without going Darlington.

Now, if the first stage were PNP and the second NPN, then maybe he could get away with it.

According to your schematic, tho, he could use a single Darlington transistor with a beta of 2000 or more. With a load of 8 ohms, that would get about 30K or so impedance at the base. Real simple, no?

I have made them, btw. Mostly as a headphone amp.

My 2 cents.

Gabe

 

[Nelson Pass]

Dear John,

You are right. It just shows I can't get away with anything
around here.

 

[lohk]

Rudy,

did you also try the ARS20 for the same site ?
Looks a bit like the the wellknown JLH amp, but is different nevertheless.
New site to me - all those Italians designing intelligent and idiosyncratic amps...

I would parallel the output transistors using small emitter resistors.

What's your opinion, Geoff ? ( - are you still there btw ?)

Klaus

 

[rmgvs]

Lohk,

Yes, I did see the other design of Rudi Stor. I did not try a build however. Mainly because it is not easier or simpler to build than the JLH and it is not apparent from the start that it outperfoms the JLH on sonical grounds (the lack of global feedback is for one thing a possible motivator).

I like to compare both amplifiers very much though. But you can not build everything you see and like (why not?). I did a job before by comparing the Hiraga and JLH (both available in the home, side to side).

The JLH has a unique smooth and 'flowing' or 'natural' sound, I think it is very special indeed. There are a few buts:
* I needed a 0.25 Ohm/5Watt resistor in the line between the upper 3055 and the positive side of the power supply. This takes away some hardening of the sound. We can not explain why this should make any diffence at all.
* I tweaked the circuit at great length (about 3 years of experimenting): resistors brands, transformers, types of 3055, caps etc etc. So I am pretty assured, after comparing it to several JLH's of friends, that the sound is okay.
* I miss some speed and clarity in the sound at times, it can sound a tad 'thick' and 'slow'.
* The Hiraga sounds more modern and open in comparison, but hasn't the beautiful midrange (vocals, piano) and flowing nature of the JLH.

Take it into the right perspective: do not compare the JLH nor Hiraga with Nad, Cyrus, Marantz, Sugden (A48 and the like), Musical Fidelity (not all types by the way) and what do you have. Just after a few seconds the differences are so clear.

I owe a lot to Geoff Moss, who helped me all the way with simulations, suggestions etc.

And yes, a very good tube amplifier (even with push-pull el84 topology) defeats both JLH and Hiraga in almost all respects. But we do not do this discussion all over again.

Rudy

 

[Christian]

Hi!

I've been wanting to build a headphone amp ( or even any amp based on transistors) for a long time, because I've only made chip-designs until now, mainly with the LM3886 chip.

So I came across this circuit at rudistor, http://www.rudistor.com/sound-lab/ar3eng.htm
which seems to be the simplest possible. To gain a little power, (so I could drive small speakers) could I use some other output transistors that I have, instead of those 2n3055?

I'm not out for hi-fi or anything, just a real easy-made transistor based amp, made out of things i already have.

Btw, the transistors I want to use are BD829, 1A max

 

[rmgvs]

headphone amp

Christian,

the headphone amp design you mention is very fine. I also wanted to use this design to make a small amp for normal speakers. You can by just building it and making it draw more current, around 1 A. Even then, the available power into 8 Ohms is a mere 500 mW or so. Several attempts to change the circuit to make this available power higher did not succeed. The 3055's are okay, this is not the problem, the problem is in the driver stage.

 

[PMA]

simple BJT amp

How about this one? Only 3 BJT's and will probably do better work than any of the ZEN amps..

 

[PMA]

In fact the amp shown above has a serious flaw - very limited output voltage swing, especially in a negative half wave. It can be improved by adding three diodes in series with Q4 emitter lead, but anyway it is impossible to get more than some 2Vrms from this circuit.

 

[halojoy]

Simple BJT power amps - nice thread!

In fact the amp shown above has a serious flaw - very limited output voltage swing, especially in a negative half wave. It can be improved by adding three diodes in series with Q4 emitter lead, but anyway it is impossible to get more than some 2Vrms from this circuit.

nice thread. I am busy downloading the schematics provided.
RudiStor was a headphone intended amp.

This last is also simple.
I would sabotage this by adding a constant current source
instead of the 8 ohm resistor.

There is no need to have the input at half the voltage supply, 12 V
You can have it much lower (4-6V) , to give the output
more "space" to operate. Gives more output voltage swing.

But nice, very nice.

/halojoy - Beauty can be Simplicity

 

[Nelson Pass]

"Death of Zen" is a workable circuit, but constitutes 3
gain stages. Personally I think BJT's are do-able with 2 if
you select high beta devices and don't get too fussy.

The fact is, though, that it's a lot easier with Mosfets
due to their high input impedances.

 

[LennyK]

funny thing:

I often read threads with no care about authors. I downloaded simple.bmp, looked and thought that I would do this in a different way. Why he did it that? And then from Gabevee: "Mr Pass,... ". Scroll up .... simple.bmp author - The one and only .



regards

Lenny

 

[Circlotron]

How's this for simple???

Never built it though.

 

[PMA]

simple BJT amp already built

I was very curious how the circuit that I had shown here yesterday would work . So I have built it and it runs quite happilly, exactly according to my yesterday's description. Negative output halfwave is very limited, because of the Darlington 2Vbe influence to Vce voltage of Q2. But add 4 diodes (I used 1N5408) in series with emitter of Q4 and you have a working amp. It will give you some 1W into 8 Ohms, then there is a very very soft limitation (mostly second harmonic) to some 10 Vp-p voltage swing. The amp is very quick but not very stable, so put a 470pF cap between base and collector of Q3 and everything runs well. You can increase R5 to some 15 Ohms/10W (if R5=8 Ohm it must be 20W) not to produce so much heat .
Pavel

P.S. to see much better amps please visit my homepage

 

[halojoy]

I will soon present an even Better and Simple Bipolar Amplifier
have to learn how to use these damndish
cicuit drawing programs.
Seems they are anything but
simple to use and Good I really hate them.
They can do spice and all exotic things
but to draw a simple circuit - without swearing abd coursing
seems only in my dreams

This was a good contribution Circlotron, old cardealer

But I think we have to beat it!

/halo

 

[Nelson Pass]

This circuit is seriously wasteful of the supply voltage, as
even with perfectly ideal transistors, the maximum output
per speaker will be 1 V rms, which is 1/8 watt for 8 ohm
speakers. A more ordinary output stage would give you
1/2 watt.

 

[PMA]

This one will be workable and will give reasonable output o several watts....

 

[halojoy]

yes, and still only 4 transistors.
And almost 4 Watts Out/ 8 ohm
Needs V+ 35 Volts for 4 W.

Will have a Idle power consumption: 35x2A= 70Watt
Effiency: 4/70= 5.7 %

But this one is not bad ...
I would put Collector of Q4 to 0V-line.
Q4 will work more linear 30V all the time.

I lose a little Current from Q4 into load,
but that is anyway not so much, that it is worth having
such a darlington coupling.
I say let Q4 get some REAL working space.
I promise it will be noticed in performance (distortion)

/halo needs at least 4 T