TIP31/32 aren't particularly high standard transistors, so I'm sure there's a lot of variation between different parts. Therefore it shouldn't really matter much what brand you use, the results will be equally "poor".
In my case I will use a AD161 Siemens and a AD162 Tungsram in A Philips Stereo amplifier.
This amplifier have a nice sound because use germanium transistors. Have problems in this combination ?
Silicon transistors have very different Vbe (0.7V compared to about 0.25V for Germanium) so the bias voltage, bias control circuit and emitter resistors all have to be redesigned and modified considerably to correct the increased crossover distortion and different thermal compensation needs of Silicon transistors.
The circuit can probably be modified and a suitable bias control circuit could be designed and added to the amplifier to suit but the real problem is that the sound will never be the same as the original product.
I suggest that if you can buy similar germanium types, the best solution is to restore the amplifier to original condition. Otherwise, replace the power amplifier section with a suitable Chip amplifier kit or simply move on to a newer product that uses available parts (even though that is now becoming impossible too).
The circuit can probably be modified and a suitable bias control circuit could be designed and added to the amplifier to suit but the real problem is that the sound will never be the same as the original product.
I suggest that if you can buy similar germanium types, the best solution is to restore the amplifier to original condition. Otherwise, replace the power amplifier section with a suitable Chip amplifier kit or simply move on to a newer product that uses available parts (even though that is now becoming impossible too
).In my opinion the best output push-pull circuit use two NPN or Two PNP transistores. The use of a complementar pair is poor, because the NPN transistor do not have the exact characteristics of a PNP transistor, aand each semicicle can sound different.
A tube output push-pull stage is best because use two tubes of same type.
A tube output push-pull stage is best because use two tubes of same type.
But the feedback loop takes care of any differences.
The bias setting takes care of the switch on of the transistor pair.
I have often used complimentary pairs and never had any problems.
I quite often use irfp240/9240 and they are quite different devices.
If it was a problem then a lot of amps out there simply wouldn't work or sound good.
I agree with your approach 100%.
However, I don't agree with your last sentence. My power amps feature 2 X IRFP9140 in push pull (choke loaded). You don't need valve amplifiers to use same type devices in the output.
That's why everything I design is differential. Much simpler than complementary design. And has has better PSRR and CMRR.
The crossover or rather the transfer characteristics of the upper and lower transistors in a Quasi-complementary amplifier is very different - even feedback can't remove the distortion as easily as it can with complementary output stages. That is the reason for using them. See figs. 2,3 here: http://www.keith-snook.info/wireles...orld-1968/30-Watt High Fidelity Amplifier.pdf
Even so, matched devices in close thermal contact can achieve very low distortion, as you find with Chip amplifiers.
Even so, matched devices in close thermal contact can achieve very low distortion, as you find with Chip amplifiers.
Last edited:
I'd say usually yes in typical circuits, assuming basic specs are the same. Complementary transistors tend to be hardly very complementary to begin with - match beta, current capability and voltage ratings, and input capacitance will be noticeably different between npn and pnp. Of course different manufacturers' ideas of safety margins may still vary. Other than that, only distortion may suffer a bit if you get unlucky.
This is one of those cases where having a curve tracer around may prove pretty handy.
This is one of those cases where having a curve tracer around may prove pretty handy.
It's already been answered.
Short answer: Yes depending on what your definition of 'matched' is.
Long answer:
In the case of the TIP31/TIP32, look at these datasheets from fairchild:
https://www.fairchildsemi.com/datasheets/TI/TIP31C.pdf
https://www.fairchildsemi.com/datasheets/TI/TIP32C.pdf
Note the graph "DC current gain" vs collector current. The gain of the TIP32 (PNP) is nominally slightly lower than the NPN. It also has a different shape curve. Note there there is no guarantee about the gain matching the curve. In the graph at 3A collector current the DC gain is 40, however you look up at the Electrical Characteristics table and the typical value is quoted at 50 with the minimum at 10. Therefore you could buy two TIP31/32 transistors and one could have more than 5 times the gain of the other, and neither would be considered to be defective. The gain is allowed to exceed the quoted specification infinitely, so if the spec calls for a typical gain of 100 for certain conditions, there is no reason they can't sell you a transistor that gives a gain of 300 or 10000 (though, gains this high for a power transistor are uncommon or impossible). For that reason, the same part number from different manufacturers and different manufacturing dates can perform differently.
The difference is easily compensated for by negative feedback but if you wanted to 'match' a pair of NPN and PNP then you need to obtain a large number of transistors, preferably all from different manufacturing batches so there is some variation in their performance. Then you test every single one of them with a transistor curve tracer.
Perhaps you may be lucky to find a "poorer than average" TIP31 which more closely matches the gain of a "better than average" TIP32.
TIP31/32 are general purpose transistors - they are not specifically intended for audio circuits where the gain curves are ideally exactly the same.
There are power transistors which the intended application are audio power amps so the NPN/PNP devices have closer nominal* gain curves - e.g. 2SA1943/2SC5200
*again, the guaranteed spec is pretty wide so be the absolute best possible performance you need to match.
Last edited:
Brand is unimportant, my doubt is whether they are fakes or not.
It must be very difficult to get AD161/162 these days, let alone match them, you'll have to use what you get.
EDIT: I saw there is a ton of them being offered in Brazil for a mere 5R$ each (1.25U$).
Probably somebody found a humid dusty cardboard box in some old shop, they have not been made for decades, I haven't used them for some 45 years, go figure.
Don't overthink it, just get a pair and repair your amplifier, I very much doubt anybody matched them in the old days, those Philips amplifiers were mass made for home use, nothing critical.
It must be very difficult to get AD161/162 these days, let alone match them, you'll have to use what you get.
EDIT: I saw there is a ton of them being offered in Brazil for a mere 5R$ each (1.25U$).
Probably somebody found a humid dusty cardboard box in some old shop, they have not been made for decades, I haven't used them for some 45 years, go figure.
Don't overthink it, just get a pair and repair your amplifier, I very much doubt anybody matched them in the old days, those Philips amplifiers were mass made for home use, nothing critical.
I get this AD161 and AD1 in Brazil and are fakes.
The junction voltagem is 0,6V and the gain is low.
Are you using the TIP31/2's as drivers or outputs? It makes a difference. At a power level for which the TIPs are suitable (20 watts or so) I wouldn't even worry about matching them. Use them in CFP's with nice complementary small signal drivers and call it good. Even as EF2's they wouldn't be bad. Maybe not state of the art, but useable. At a higher power where you might consider using them as drivers I might look for a better grade of transistor (higher ft and flatter gain). In most output stages, it is really the driver characteristics that dominate the distortion. That's where you need a high fT, flat beta over current, and gain matching.
Ahem! .... aldovan is NOT using TIPs, he just mentioned them in a generic philosophical question "can NPN and PNP be matched if different brands?"
Do they *really* measure 0.6V BE ?
FDP !!!!!!!!!!!!!
Then they were made recently, on purpose, to cheat people.
Not easy because they needed to get relatively expensive and definitely hard to find metallic TO66 cases , buy, hey, anything is possible to make a fast buck.
I could not understand why is Mercadolivre do Brasil *full* of them, now I know.
10 X Transistor Ad161 + 10 X Transistor Ad162 / Kit C/20 Pçs - R$ 110,00 no MercadoLivre
They have so many that they are selling them in lots of 20
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.