Compound Pair vs Darlington Pairs

Lucky-Luka · Dec 26, 2019

Hi all
I've read that Sziklai pair are more thermically stable than Darlington ones.
I've read that the influence of the output device is considerably less than that of the driver using the Sziklai configuration.
Unfortunately I haven't really understood what it means...

Happy holidays

gophert · Dec 26, 2019

Note that the bias on Q5 only needs to be 1.2 to 1.4v range and Q10 must be double. The power dissipated by Q10 is, therefore, double as well. The thermal effects on the bias will be much greater as well so either too much current or not enough. Also, this transistor is typically mounted on the same heat sink as the two output transistors (typically Darlington) to maintain the same bias throughout the temp swing but does not track as well with Darlington as a single NPN or pnp.

on the other hand, streaming is the wave of the future and the quality of amplifier is the least of one's concerns if they are looking for quality listening experience. Vinyl, take me away!

audioguru · Dec 26, 2019

You cannot stream directly to your brain, yet. Wideband low distortion power amplifiers and wideband speakers are still needed.
Vinyl produced click, pop, click, pop, rumble, hiss, wow and flutter.

crutschow · Dec 26, 2019

audioguru said:
Vinyl produced click, pop, click, pop, rumble, hiss, wow and flutter.

Yes, I totally fail to understand the fascination with vinyl (along with tube amps).
I listened to vinyl records for may years, and all those problems you mentioned (along with reduced frequency response at the inner grooves) were a very large annoyance to me.
I was really happy when I listened to my first CD recording, and all that disappeared, with the music being perfectly clear.

Lucky-Luka · Dec 26, 2019

So, is the voltage bias on the Vbe multiplier the main culprit of the difference in thermal stability?

crutschow · Dec 26, 2019

Lucky-Luka said:
I've read that Sziklai pair are more thermically stable than Darlington ones.

That's because the base-emitter voltage change with temperature is generated by two base-emitter junctions for the standard Darlington pair, and only one junction with the Sziklai pair.

Lucky-Luka · Dec 26, 2019

I've also read that Szikai configuration should behave better against distortion... Is it true?
I've tried a comparison but it looks like that Darlington behaves better. Am i wrong?

audioguru · Dec 26, 2019

How can you talk about distortion when you do not have a complete amplifier? Your amplifier is missing lots of open-loop gain and negative feedback.

If the voltage gain of the front of the amplifier is 10,000 times then the 0.2% distortion caused in the output stage will be reduced to about 0.002% with negative feedback and a closed loop gain of 100 times.

crutschow · Dec 26, 2019

Lucky-Luka said:
I've tried a comparison but it looks like that Darlington behaves better. Am i wrong?

That small difference in distortion may be just do the slight bias differences between the two circuits, as much as the difference in the two configurations.

Nigel Goodwin · Dec 27, 2019

I agree with the comments about vinyl, I can't understand the sudden fascination by a small percentage of the population for old low spec technology?.

However, as for the actual discussion - both techniques have been used in countless amplifiers over the decades, and essentially there's no 'best' method, some of the best sounding amplifiers of all time were relatively crude, and long pre-dated Sziklai pairs.

Technically the Sziklai pair has a number of advantages, and the only reason it wasn't used more was that high power PNP transistors weren't available, and then those that were cost a great deal more, so weren't used. The standard technique was quasi-complimentary, so as to use NPN output devices, and many such amplifiers are still well regarded today.

unclejed613 · Dec 27, 2019

many call that a Complementary Feedback Pair (CFP)

Compound vs Darlington

ESP - Compound Pair Vs. Darlington Pair

sound-au.com

this is good explanation of the differences between CFP and darlingtons.
most articles i have seen about CFP agree that the CFP has lower distortion and better thermal stability. the thermal sensor (bias transistor) doesn't need to be on the heatsink near the output devices, but thermally bonded to the drivers instead.

simulation tip: LTSpice has data compression activated by default, which tends to add artifacts that show up as distortion in the sine wave source. turning the compression off in the options menu will remove the distortion from the sine wave source... with compression on, you could have a circuit that only causes 0.05% THD, but your results will never show below 0.5% with the compression off you will see a big difference. if you need to verify that, just get a trace from V3 or V4, and open the FFT window, and you will see what i mean.

Lucky-Luka · Dec 27, 2019

unclejed613 said:
the thermal sensor (bias transistor) doesn't need to be on the heatsink near the output devices, but thermally bonded to the drivers instead.

In the schematic above, is Q5 the bias transistor and Q2 and Q4 the drivers, right?

gophert · Dec 27, 2019

Oh God, I didn't mean to say vinyl was better, just old-school and easier to understand that the varying bit-rates and various noises, buffering and other crap of streaming that greatly overshadow the very slight differences between the various amplifier designs the OP is discussing (as does vinyl).

gophert · Dec 27, 2019

Lucky-Luka said:
In the schematic above, is Q5 the bias transistor and Q2 and Q4 the drivers, right?

Yes on Q5 but, Technically, Q1 and Q3 are the drivers.

Nigel Goodwin · Dec 28, 2019

gophert said:
Yes on Q5 but, Technically, Q1 and Q3 are the drivers.

If he means the schematic in post #1 (the only schematic I can see?), there are no drivers, just a bias transistor and darlington outputs (of various types).

Lucky-Luka · Dec 28, 2019

Nigel Goodwin said:
If he means the schematic in post #1 (the only schematic I can see?), there are no drivers, just a bias transistor and darlington outputs (of various types).

I thought the biasing transistor and the drivers were the ones pointed in the attached pic.

audioguru · Dec 28, 2019

The latest schematic was in post #7. The driver transistors were Q2 and Q4.

Nigel Goodwin · Dec 28, 2019

audioguru said:
The latest schematic was in post #7. The driver transistors were Q2 and Q4.

Not really 'drivers' as they are simply part of the compound output transistors (Q2/3 is essentially one device) - the 'driver' would be where V4 is.

audioguru · Dec 28, 2019

Thermally, the transistors in a compound pair (Sziklai pair) are completely separate unless they are bolted close together in a heatsink..

Lucky-Luka · Dec 28, 2019

audioguru said:
Thermally, the transistors in a compound pair (Sziklai pair) are completely separate unless they are bolted close together in a heatsink..

I've read that it's better not to do it. Do you agree?

Compound Pair vs Darlington Pairs

Member

Attachments

Well-Known Member

Well-Known Member

Well-Known Member

Member

Well-Known Member

Member

Attachments

Well-Known Member

Well-Known Member

Super Moderator

Well-Known Member

Member

Well-Known Member

Well-Known Member

Super Moderator

Member

Attachments

Well-Known Member

Super Moderator

Well-Known Member

Member

Similar threads

Privacy & Transparency

Privacy & Transparency