nelsonvandal said:
If I may be very honest, and I am not trying to be a smart guy.
I do not match at all anymore (I also use a lot of NFB). If all electronic manufacturers had to match transistors for every application then we would be throwing away 99% of all components, or we would be employing thousands of people matching semiconductors to keep up with the high speed pick and place machines, and things would cost considerably more than they do now.
Modern manufacturing technologies used in producting semiconductors have improved considerably since their incepion in the 50's. If transistor matching or their models were so bad from manufacturers, they may as well print random numbers on them.
Nico
PMA said in this thread http://www.diyaudio.com/forums/showthread.php?postid=1690588#post1690588
However, parallel devices do show benefits in careful close matching.
An LTP benefits even more with precise matching.
whereas many other builders/designers say that matching between N & P is not particularly important.
However, parallel devices do show benefits in careful close matching.
An LTP benefits even more with precise matching.
Hi Nelson...when we assemble double complementary differential
those ones use a PNP Differential and a NPN differential...alike Leach amplifier...if you do not match you perceive bad sound.
It is hard to match them.... maximum i could, my whole life was to mach PNP and NPN into a 12.5 percent range... they had always difference in gain.
Now a days i do not care about anymore... i accept that is hard and them i avoid double differential and things that needs such kind of matching.
But i know that i am not doing the best possible job accepting the mismatching.
And if this sounds or do not sound is not more important...if you feel that something is not correct you will always feel unsafe about the amplifier performance...you will always ask yourself:
- "And if was matched?... would be better than that?"
So... the problem goes inside our intimate psychology... the problem goes from technicall to psychological... you do not feel safe to say it is good or bad...not perfeclty assembled!
regards,
Carlois
those ones use a PNP Differential and a NPN differential...alike Leach amplifier...if you do not match you perceive bad sound.
It is hard to match them.... maximum i could, my whole life was to mach PNP and NPN into a 12.5 percent range... they had always difference in gain.
Now a days i do not care about anymore... i accept that is hard and them i avoid double differential and things that needs such kind of matching.
But i know that i am not doing the best possible job accepting the mismatching.
And if this sounds or do not sound is not more important...if you feel that something is not correct you will always feel unsafe about the amplifier performance...you will always ask yourself:
- "And if was matched?... would be better than that?"
So... the problem goes inside our intimate psychology... the problem goes from technicall to psychological... you do not feel safe to say it is good or bad...not perfeclty assembled!
regards,
Carlois
Nelson,
you have no control over most characteristics. Matching as a result of die design and manufacturing precision is much more important. PNP has necessarily poorer characteristics. And again, you cannot cure the main problem of the signal waveform being amplified by essentially two different parts.
you have no control over most characteristics. Matching as a result of die design and manufacturing precision is much more important. PNP has necessarily poorer characteristics. And again, you cannot cure the main problem of the signal waveform being amplified by essentially two different parts.
Re: Hi Nelson...when we assemble double complementary differential
I agree with you totally. I am getting too old to sit and measure transistors. I rather build it and listen. If I like it then I may make some efforts. But like you I want to get a result quickly with my ears.
I think the human ear cannot tell whether a wave is symetrically placed around 0V.
Nico
Hi Carlos,destroyer X said:
I agree with you totally. I am getting too old to sit and measure transistors. I rather build it and listen. If I like it then I may make some efforts. But like you I want to get a result quickly with my ears.
I think the human ear cannot tell whether a wave is symetrically placed around 0V.
Nico
Using complementary transistors from different Mfgrs
Assembling 10 device Leach clone boards now, and have a question about the 2nd stage and driver transistors.
Out of the several MJE340/350 transistors that we have in our group, some are made by Fairchilds and some by ON Semi.
I notice some fairly close matches were found(using a multimeter) between a few ON semi JE340(NPN) and Fairchild MJE350(PNP) devices.
So, my question is:
In this application (Leach clone at locations T8,T9,T13,T14), would it be better to have:
a) Complementary devices (PNP and NPN) with widely different hFE from the same manufacturer,
or
b)Complementary devices(PNP and NPN) with similar hFE, each from a different manufacturer?
Is it worth it to mix and match brands like that, is it a bad idea, or is all this a complete waste of time?
Thanks for tolerating such questions from us "noobs"
-Chas
Assembling 10 device Leach clone boards now, and have a question about the 2nd stage and driver transistors.
Out of the several MJE340/350 transistors that we have in our group, some are made by Fairchilds and some by ON Semi.
I notice some fairly close matches were found(using a multimeter) between a few ON semi JE340(NPN) and Fairchild MJE350(PNP) devices.
So, my question is:
In this application (Leach clone at locations T8,T9,T13,T14), would it be better to have:
a) Complementary devices (PNP and NPN) with widely different hFE from the same manufacturer,
or
b)Complementary devices(PNP and NPN) with similar hFE, each from a different manufacturer?
Is it worth it to mix and match brands like that, is it a bad idea, or is all this a complete waste of time?
Thanks for tolerating such questions from us "noobs"
-Chas
Different brands of the same BTJ "should" be within the same tolerances (if not fakes).
And as also said in this and many other threads here, there are major differences between complementary NPN and PNP devices from the same manufacturer.
As one of the main factors is Hfe matching, I go for this and mixing brands if needed.
my input
about this is
that in modern amplifiers and well constructed , featuring things like constant current sources, curent mirrors , sophisticated vde multipliers and so on matching is not that critical given as a fact
( as nico said ) the use of high quality semi's that are supposed to be...." give or take" with in the same tolerance .....
on the other hand simple designs bootstraps and cfps might be seriously effected if pnp and npn are missmatched (thats just recentlly tested )
regards sakis
about this is
that in modern amplifiers and well constructed , featuring things like constant current sources, curent mirrors , sophisticated vde multipliers and so on matching is not that critical given as a fact
( as nico said ) the use of high quality semi's that are supposed to be...." give or take" with in the same tolerance .....
on the other hand simple designs bootstraps and cfps might be seriously effected if pnp and npn are missmatched (thats just recentlly tested )
regards sakis
Re: Using complementary transistors from different Mfgrs
Of course in input stages maching is key.
Specifically in ouput push pull stages, this is what I believe:
If you have a good class A driver with enough current ( like in the Leach amplifier), teh assymetry of Hfe should not be a problem because the emitter follower is a voltage amplifier with gain 1.
This of course if the driver has enough current to follow linearly the demand in base current.
At high levels of current, the Vce sat can be different and the start of clipping can be different. The answer is : do not allow to clip by having enough headroom.
The real question is crossover distortion at collector currents under 100ma and crossing zero.
There, the idea is to keep the output resistance of the amplifier constant while you pass from NPN to PNP transistor.
This is impossible while the two are conducting in class AB and an optimum in biasing has to be found.
The optimum is for a collector current Ic so that gmRe=1
Re beeing the resistor in the emitter
gm = Ic/Vt
This is equivalent to the well known rule of thumb: bias for about 26mV on Re.
This is why we should have a symetrical Vbe vs Ic in the low current region around (zero Ic). That is what I would check.
The Leach output stage is the best if properly biased.
JPV
clm811 said:
Of course in input stages maching is key.
Specifically in ouput push pull stages, this is what I believe:
If you have a good class A driver with enough current ( like in the Leach amplifier), teh assymetry of Hfe should not be a problem because the emitter follower is a voltage amplifier with gain 1.
This of course if the driver has enough current to follow linearly the demand in base current.
At high levels of current, the Vce sat can be different and the start of clipping can be different. The answer is : do not allow to clip by having enough headroom.
The real question is crossover distortion at collector currents under 100ma and crossing zero.
There, the idea is to keep the output resistance of the amplifier constant while you pass from NPN to PNP transistor.
This is impossible while the two are conducting in class AB and an optimum in biasing has to be found.
The optimum is for a collector current Ic so that gmRe=1
Re beeing the resistor in the emitter
gm = Ic/Vt
This is equivalent to the well known rule of thumb: bias for about 26mV on Re.
This is why we should have a symetrical Vbe vs Ic in the low current region around (zero Ic). That is what I would check.
The Leach output stage is the best if properly biased.
JPV
Re: Using complementary transistors from different Mfgrs
Of course in input stages maching is key.
Specifically in ouput push pull stages, this is what I believe:
If you have a good class A driver with enough current ( like in the Leach amplifier), teh assymetry of Hfe should not be a problem because the emitter follower is a voltage amplifier with gain 1.
This of course if the driver has enough current to follow linearly the demand in base current.
At high levels of current, the Vce sat can be different and the start of clipping can be different. The answer is : do not allow to clip by having enough headroom.
The real question is crossover distortion at collector currents under 100ma and crossing zero.
There, the idea is to keep the output resistance of the amplifier constant while you pass from NPN to PNP transistor.
This is impossible while the two are conducting in class AB and an optimum in biasing has to be found.
The optimum is for a collector current Ic so that gmRe=1
Re beeing the resistor in the emitter
gm = Ic/Vt
This is equivalent to the well known rule of thumb: bias for about 26mV on Re.
This is why we should have a symetrical Vbe vs Ic in the low current region around (zero Ic). That is what I would check.
The Leach output stage is the best if properly biased.
JPV
clm811 said:
Of course in input stages maching is key.
Specifically in ouput push pull stages, this is what I believe:
If you have a good class A driver with enough current ( like in the Leach amplifier), teh assymetry of Hfe should not be a problem because the emitter follower is a voltage amplifier with gain 1.
This of course if the driver has enough current to follow linearly the demand in base current.
At high levels of current, the Vce sat can be different and the start of clipping can be different. The answer is : do not allow to clip by having enough headroom.
The real question is crossover distortion at collector currents under 100ma and crossing zero.
There, the idea is to keep the output resistance of the amplifier constant while you pass from NPN to PNP transistor.
This is impossible while the two are conducting in class AB and an optimum in biasing has to be found.
The optimum is for a collector current Ic so that gmRe=1
Re beeing the resistor in the emitter
gm = Ic/Vt
This is equivalent to the well known rule of thumb: bias for about 26mV on Re.
This is why we should have a symetrical Vbe vs Ic in the low current region around (zero Ic). That is what I would check.
The Leach output stage is the best if properly biased.
JPV
Well, the question is if it is worth the effort. I do not really think that 0.002% or 0.0002% of only 2nd and 3rd, nothing else in spectrum, makes any audible difference. BUT, this is related to just one circuit, the other circuits may have much higher distortion, and then it might make a sense.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.