Paired transistors for audio amplifier - do pairing matter?

[balage]

Hi,

I am repairing a Pioneer audio amplifier. I have found the transistors on Ebay: 2SA1939 and 2SC5196. I need them for the same channel.

There are paired ones that cost higher (12EUR): https://ebay.to/2YRRFnD
The individual ones costs much less.

Do you think paired ones worth the prize? Or is the difference not that big to hear?

Thanks

[David Hess]

I cannot read German so it is not clear to me that it says they are matched.  Are they just selling them as a pair?  Many failures will destroy both output transistors.

Mismatch of the output transistors in a complementary design contributes to second harmonic distortion to a level which depends on the topology, but I doubt you will hear any difference.

[magic]

I kinda suck at German but still got an impression that the description only says about datasheet translations, eBay feedback and problem resolution. No claims appear to be made about matching.

They also list country of origin as China. I would start with asking the seller if they are genuine Toshiba/Sanyo/Hitachi/whoever or modern Chinese "second sources". The latter are very lousy, may barely meet the "minimum" specs listed in the datasheet and there are stories of them blowing up altogether. Those Chinese vendors never show SOA in their datasheets.

[David Hess]

I would not even bother buying originals as replacements.  There are plenty of modern ring emitter transistors for audio amplifiers which have practically the same specifications or better.  80 volts, 60 watts, 6 amps, 30 MHz ...

https://www.onsemi.com/products/discretes-drivers/audio-transistors/njw1302
https://www.onsemi.com/products/discretes-drivers/general-purpose-and-low-vcesat-transistors/njw0302

[bob91343]

Matching is seldom important.  Staying with original part numbers is a good thing, as the circuit has been designed around them with regard to stability and saturation.  Using parts with too high or too low a transition frequency can cause oscillation.  If beta is too different, other problems can arise.

[CaptDon]

If the circuit is complementary symmetry it is best to have
matched-pair output and pre-driver transistors. This way
everything tends to stay centered as per the original design
and the harmonic distortion will remain as low as the designer
intended. I would NOT under any circumstances by discreet
semiconductor devices from China. Most of it is recycled junk
and re-surfaced or relabeled as a higher rated part than what
it actually is. I have been burned enough times to not buy from
China and the company I work for has incurred thou$and$ of
dollar$ of field failures and lost customers due to the fake
parts getting into even reliable source streams. Some of these
Chinese fakers are even set up with U.S. business addresses.
Beware!!! My company and I learned the hard way!!!!
 

[Kleinstein]

I would not worry so much about matching. From China and Ebay in general the question is more if they are real or fakes. The question is if there is a reliable source for these, of if a substitute from a more reliable source may be the better.  Even if the same number but a a different manufacturer, there may be differences.

[Labrat101]

If its an help I have used this sell and all his stuff is original   
And got quite a few transistors from him and all have passed on a 'curve tracer' as per manufacture specs.
So don't worry . He also has several sites all in different languages .

 The only way to match transistors if they are true matched is by using a curve tracer.
 

[bdunham7]

Unless you have an amplifier with an unusually sensitive design or multiple parallel output devices, per-device selective pairing is not needed and you probably won't get it even if you pay for it.  If you are replacing the output devices the usual practice would be to use modern equivalent devices that are complementary--so don't just replace one--but not specially selected. 

[magic]

If you buy anything from sites like AliBay you have no one but yourself to blame.

What I'm warning about is not straight-out counterfeits, but supposedly legit "second source" parts from companies like InChange Semiconductor / ISC or SavantIC. These look like serious manufacturers of discontinued parts, with their own datasheets and with factories churning out shiny new chips, but in reality their products are crap. And their datasheets have not a single "typical performance" plot too, hint hint.

[JohnPi]

The importance of pairing depends on the circuit design.

If it is a class AB output stage and there are 2 diodes or some other bias circuit to bias the output pair (back-back emitter followers), then a) it's a bad design if it depends heavily on matching; b) if matching really is important, the diodes and power stage transistors will be thermally coupled (often diodes bolted to the transistors); c) there may be a low value (10-100 Ω typ.) adjustable resistor to compensate.

Matching also used to refer to beta and fT, but thats less important, and 'good' transistors will have plenty of beta and BW.  For your case, matching probably refers to the VBE needed to bias the devices and how it matches the diodes (or they could also have been transistors). Modern transistors are all quite identical, so searching for 'matched' devices is fruitless. You'd be better off combining devices from different manufacturers

[magic]

A matched pair should be just that: a matched pair, matched to each other. How could the vendor know if they are matched to your drivers, if each amplifier has its own set of drivers?

As David said in the first reply, beta match will have some influence on even order distortion, which will be largely suppressed by feedback in common amplifier topologies.

[flolic]

Instead original 2SA/2SC combo, just put in NJW3281G / NJW1302G bought from reputable distributor and you will be fine. Doesn't need to match them.
Those Japanese audio power transistors from ebay/aliexpress sellers are quite often fakes. Don't ask me how I know...

[Kleinstein]

The NJW3281 / NJW1302 are nice powerful transistors with the same fT, but much higher power and higher capacitance (about 2 x collector-base capacitance). So they may work, but not for sure.  So Ideally there is a smaller brother.

[flolic]

Yes, they are overkill and I think there is less powerful version from the same series.
I have used those together with MJE243G / MJE253G drivers to restore my old Technics SU-V5 amplifier. It works just as good as with original ones (2SA1106 / 2SC2581).

[JohnPi]

In the end, matching is probably not that important and won't make much difference if you are not using oxygen-free cables for your mains input and hand-selected auto quality decoupling capacitors in the power supply.

[Kleinstein]

It depends on the circuit. In some circuits matching of hfe is important, in other it is not and other factors like having a large or linear hfe or a constant C_CB is more important.

[balage]

Thank you all for the help. And sorry for my late response.

I was just one click away from ordering the Ebay pair of BJTs, but I though I should ask here. It worth. 

I think I am going for the NJW0302G/NJW0281G or smaller power ones. I have never bought semiconductors on Ali/Bay yet and will not do so ever.

Once the BJTs arrive and I test the Pioneer I'll let you know the results.