Little opamp/darlington amplifier

[b_force]

I was looking for a amplifier with great THD results, but also not that complex.

Because I will only use it for measurements (THD, tests, etc etc).
5W-8ohm/10W-4ohm would be enough.
(maybe I will go to 10W @ 8ohm)

The idea is very simple, use a decent opamp (opa134/lm4562 etc etc) and put a current amplifier behind it.

I first tried a Mosfet follower but found out that the currect was very big.
So I tried a darlington pair, and I think it's better.



This is simulated with the ZTX605/705. I think the ZTX651/751 would be better. (because of the current)

I have a few questions.
Is this going to work this way ?
What does C2 (1uF) do exactly ? (in the simulation it reduces the THD enormous ! )
Are there any other improvements ?

[Stuart Easson]

I may be wrong, but I think your diodes are backwards, and you may find that a third one will lower the distortion further...

IIUC they are to provide a some bias voltage for the output transistors, and as they are drawn they are reversed biased by R4 and R5, and therefore not having the desired effect. I think perhaps you can see the effect of the excess bias voltage in the idle current of 635ma, shown on the meters for the voltage rails.

If you connect the output of the opamp to the junction of the diodes, reversed, you can avoid loading the opamp at idle and depending on it's characteristics things will be different. Adding a resistor from the output of the opamp to one rail or the other will force it to operate with the output in class A. Not all opamps like this, and you may get different results depending out which rail and the value of the resistor.

The cap across the diodes helps discharge the base current from the output transistor that is turning off, without it the output is much slower in transition from full one direction to the other.

HTH

Stuart

[b_force]

Thank you for your comments, let's change it and look at the results

I thought that the ZTX651/751 where also Darlington,s but I think that's a mistake.

are there other Darlington pairs that can easy deliver 2-3A ?

I think it is not necessary to operate the opamp in class-a because the THD of the opamp is also very low.
Maybe a improvement later


edit:
The diodes are connected the right way. If I connect them backwards the THD is much higher.
There is also no difference between 2 or 3 diodes in the simulation. (but that's not the real world)

What kind of heatsink can I use ?

[Stuart Easson]

Unless I am missing a key point...

The distortion may be higher with the diodes reversed, but the output transistors will not be passing >600ma of current. I think what you have right now is a very improperly biased classB amp. It only works because the models for the transistors are unrealistic and the bases are fed from 100k resistors.

What is the idle current with the diodes reversed, I think it will drop to almost zero, indicating a 'normal' class B amp, then you have to bias the output transistors properly to minimize the crossover distortion. If you want to make an amp with a classA output stage you need to arrange the bias conditions with more control and less dependence on the sims transistor model.

Good luck

Stuart

[jaycee]

What you are doing here is similar to this:

http://sound.westhost.com/project76.htm

and this

http://sound.westhost.com/project113.htm

In your diagram, the diodes are the wrong way around - the circuit would not work in real life.

I've had good results with using a diamond buffer on the end of an opamp. See for example Per-Anders Sjöström's QRV05 circuit.

[AndrewT]

Eline output transistors that can dissipate 1.5W, if you can keep them at 25degC, will not be man enough for an 8W output.

With ClassAB amplifiers it is usual to have the output devices able to dissipate 2 to 3 times the maximum output power. This would indicate you need 15 to 30W devices.

[mlloyd1]

the diodes are the wrong way around.
think about the actual function of the diodes.
you can certainly build it (instead of simulating), but we would recommend a current limited power supply and some spare output devices.
also, don't hook up your good speakers just yet either ...


mlloyd1

[jerluwoo]

Those Zetex devices would die a horrible death as output devices. The low distortion you see with the diodes wrong way round is because the outputs are operating in class A. Those poor Zetex darlingtons are cooking with 9+ watts of heat on them. Think they are only good for 2 watts at most, haven't looked them up but I think thats right. With the diodes the correct way round the outputs idle at cutoff which will show as very high crossover distortion, thus the reason you see a huge increase in your distortion sims. You need 4 diodes to get a correct forward bias using darlingtons. Remember a darlington is two transistors stacked together so they have two BE drops per device. Attached is a circuit that would be a much better starting point for your experiments.

[Stuart Easson]

...an echo chamber here or something...

Your idea is sound, but you need to understand the limitations of the simulation environment vs a real construction. There are physical boundaries the simulation is not honoring, in this case, the output transistors being cooked etc.

Even if you can find a huge set of darlington output devices, Motorola have 300w devices I think, they will not behave the way the simulation suggests they should. The pair will have dissimilar gain, Vbe etc that will seriously compromise the real world behavior of the amp.

If you are willing to look around a little more you can easily find a stable output stage onto which you can graft the opamp input. If simple is the goal, Class A will probably get you the best bang for your part count, but you need more power and heatsinks. Class B will need more attention to biasing detail, but has lower power needs and smaller heatsinks.

From the Passlabs arena you might try an output mosfet loaded by a CCS; very easy to get stable, simulates well and adds very little distortion as long as you set the CCS properly...

HTH

Stuart

[ostripper]

This amp will most likely work, but having built a similiar one,
it will sound like garbage.

After having a heated discussion with our beloved carlos (DX)
about how crappy a darlington amp sounded, I decided to build
this:elector darlington amp
on a veroboard using 100w sanken devices from a dead sony
HT unit. The elector amp has a real Vbias generator and
LPT, so I figured it would sound OK.
Once it was built I almost conceded to carlos's opinions...
crossover distortion, harsh sound..
I was discouraged.

But, later, still wanting to give the sanken devices a home,
I came across the LM4702 and this circuit:LM4702 darlington amp
Much to my surprise, it worked and sounded very good!!
playing around, I subbed tip120/125's, a Real EF OP
section, and various other junkbox outputs into my prototype
and they all sounded better than the HT receiver.(same speakers,
same music.)

In conclusion, for the same price as a couple of HQ op-amps
the LM4702 (7-8 US$$) will give you stereo you can use and
enjoy while giving a home to all your orphaned darlingtons.

BTW.. In a future DIY project I'm thinking of a 4 OP device EF or
CFP (maybe even Thermal trak) variation of the 4702.
OS