Can anyone explain the apparent preference for PNP transistors as the input devices in a differential amplifier? I'm designing an amp using 2N3904/2N3906 in a differential input circuit with current mirror collector loads and current source tail. I can't decide which polarity to use for which, though - SPICE simulations haven't helped me clear up the issue either.
I see PNP inputs in many published designs; is there a reason they are considered superior to NPN in audio applications?
Thanks
I see PNP inputs in many published designs; is there a reason they are considered superior to NPN in audio applications?
Thanks
It does not matter which you use.zapo said:
What matters is the quality of the input pair.
This is the important. Not if they are P or N.
--------------------------
But often you first decide what output transistors
and Output stage configuration you will use.
And the second stage, the voltage amplifier,
what transistor will you use there?
Have you found some good NPN for voltage amplifier.
Well then you better use PNP in the input pair to drive that
VAS-transistor.
So the later stages can sometimes decide what you use for input.
Sometimes the other way.
Most important is that the total works good together.
And that your design is adjusted to take advantage of
different available transistors.
------------------------
If you have PNP as voltage amplifier,
it is most suitable to have NPN in input.
Quality is more important, than the polarity of the transistors.
/halo - has built amplifiers with either kind of input pairs.
If you design a full complementary amplifier you will need both transistor polarities i.e. NPN and PNP.zapo said:
See f.a.:
http://home.kimo.com.tw/skychutw/ampzilla/schematics/sonAmpzilla_sch.jpg
Re: Re: NPN vs PNP input stage
Not with todays transistors.
We can come close to left and right symmetry. (compare Pass designs)
But not Positive Negative symmetry.
The best amplifier designers, have realized this fact.
JFETs & MOSFETs have even worse Complementary Performances.
/halo - almost never designer of comlementary amplifiers,
at least not in the input stage.
Fully complementary, symmetric amplifiers do not exist.Elso Kwak said:
If you design a full complementary amplifier you will need both transistor polarities i.e. NPN and PNP.
See f.a.:
http://home.kimo.com.tw/skychutw/ampzilla/schematics/sonAmpzilla_sch.jpg
Not with todays transistors.
We can come close to left and right symmetry. (compare Pass designs)
But not Positive Negative symmetry.
The best amplifier designers, have realized this fact.
JFETs & MOSFETs have even worse Complementary Performances.
/halo - almost never designer of comlementary amplifiers,
at least not in the input stage.
Re: Re: Re: NPN vs PNP input stage
The example I gave was clear enough I think. It is a excisting design!
If you prefer a single frontend not complementary like the JLH design be my guest.
As early as in the seventies I remember a French designer claiming that complementary transistors never could be complementary.
You understand all his schemes were NOT complementary. But James Bongiorno designer of the example in the link claims that all audio signals are bipolar and ineed to be amplified by a full complementary design. John Curl, who I know you admire, also uses complemenatary design in his poweramplifiers. John uses complementary JFET's at the inputs: 2SK389 & 2SJ109.
Hi Halojoy,halojoy said:
The example I gave was clear enough I think. It is a excisting design!
If you prefer a single frontend not complementary like the JLH design be my guest.
As early as in the seventies I remember a French designer claiming that complementary transistors never could be complementary.
You understand all his schemes were NOT complementary. But James Bongiorno designer of the example in the link claims that all audio signals are bipolar and ineed to be amplified by a full complementary design. John Curl, who I know you admire, also uses complemenatary design in his poweramplifiers. John uses complementary JFET's at the inputs: 2SK389 & 2SJ109.
Okay ELSO - We differ in likings.
John Curl is a very experienced and great designer.
So I have to think again.
I state my views with whole
my heart
good then, that i am not the only voice around.
And that my friends can help to broden the view
as seen from their horizon.
thanks elso
---------------
/halo - takes a stand, when he has one
- and doesn't feel bad for doing so on the contrary!
-------------------------------------------------------------------
PS. you mention one pair of transistors, you claim to be
"fully complementary" in their qualities.
This is what is needed for a full complementary working amp.
I do not know any such N+P, judging from the curves and data.
Maybe there are more fully complementary pairs I do not know of?
I want to know ... I always do
- as far as I know. DS.
John Curl is a very experienced and great designer.
So I have to think again.
I state my views with whole
my heart good then, that i am not the only voice around.
And that my friends can help to broden the view
as seen from their horizon.
thanks elso
---------------
/halo -
takes a stand, when he has one - and doesn't feel bad for doing so
on the contrary!-------------------------------------------------------------------
PS. you mention one pair of transistors, you claim to be
"fully complementary" in their qualities.
This is what is needed for a full complementary working amp.
I do not know any such N+P, judging from the curves and data.
Maybe there are more fully complementary pairs I do not know of?
I want to know ... I always do
- as far as I know. DS.
quote
Fully complementary, symmetric amplifiers do not exist.
Not with todays transistors.
I totally disagree. A fully complementary amp can be built
as long as it is fully balanced input and fully balanced
(i.e. H bridge) output.
Every signal ends up going thru the exact same amount
of pnp and npn transistors. The resulting rise and fall
slew rates are absolutely identical.
Fully complementary, symmetric amplifiers do not exist.
Not with todays transistors.
I totally disagree. A fully complementary amp can be built
as long as it is fully balanced input and fully balanced
(i.e. H bridge) output.
Every signal ends up going thru the exact same amount
of pnp and npn transistors. The resulting rise and fall
slew rates are absolutely identical.
Full complementary amplifier
I agree with you. You mean something like Ampzilla III?. Full bridged amplifier. Still asking myself if the Ampzilla 2000 is using this scheme. Mr Bongiorno is very reluctant in giving circuit details.
http://home.kimo.com.tw/skychutw/ampzilla/schematics/ampzillaIII_sch.gif
Hi kevin,kevin gilmore said:
I agree with you. You mean something like Ampzilla III?. Full bridged amplifier. Still asking myself if the Ampzilla 2000 is using this scheme. Mr Bongiorno is very reluctant in giving circuit details.
http://home.kimo.com.tw/skychutw/ampzilla/schematics/ampzillaIII_sch.gif
elso_kwak, your ampzilla link does not work...
Here is what i'm talking about
http://gilmore.chem.northwestern.edu/uberpage5.jpg
also most of the rest of it
http://gilmore.chem.northwestern.edu/uberpage1.jpg
http://gilmore.chem.northwestern.edu/uberpage2.jpg
http://gilmore.chem.northwestern.edu/uberpage3.jpg
http://gilmore.chem.northwestern.edu/uberpage4.jpg
Here is what i'm talking about
http://gilmore.chem.northwestern.edu/uberpage5.jpg
also most of the rest of it
http://gilmore.chem.northwestern.edu/uberpage1.jpg
http://gilmore.chem.northwestern.edu/uberpage2.jpg
http://gilmore.chem.northwestern.edu/uberpage3.jpg
http://gilmore.chem.northwestern.edu/uberpage4.jpg
Ampzilla schematics
Hi Kevin,
I am sorry the link did not get through.
Just try
http://home.kimo.com.tw/skychutw/ampzilla/
and click on schematics of Ampzilla III for the balanced amplifier and on Son of Ampzilla for a simple example of a amplifier with PNP & NPN inputtransistors. I have downloaded your uberpages and have a look at it.
My first impression is your scheme is the same general picture as the Ampzilla III. Quite a achievement Kevin! Did you actually build the amplifier in the uberpages?
Hi Kevin,
I am sorry the link did not get through.
Just try
http://home.kimo.com.tw/skychutw/ampzilla/
and click on schematics of Ampzilla III for the balanced amplifier and on Son of Ampzilla for a simple example of a amplifier with PNP & NPN inputtransistors. I have downloaded your uberpages and have a look at it.
My first impression is your scheme is the same general picture as the Ampzilla III.
Quite a achievement Kevin! Did you actually build the amplifier in the uberpages?
Folks, complementary symmetry is NOT perfect, because the comp. input devices are not perfectly matched, BUT look at the alternative. Single differential designs are OK on their input, BUT what about the drive for the second stage? Comp. Differential gives you almost perfect push-pull drive for the output stage. This criticism of the lack of perfect mirror image is lost in the advantage of one drive device turning on, while its complement is turning off from the opposite rail. This is invariably better as far as open loop distortion is concerned.
I aggree, complementary symmetry is NOT perfect, because the comp. input devices are not perfectly matched.
However, Can I get my 2 cents in on part of the original question
"PNP" vs "NPN" ??
The origins of Transisitors, PNP was a very easy latice structure to acheive, the P doped portion then the N, then another P doped section, "Tada!" PNP, HOWEVER, NPN, ain't so easy to make (not then, and still is harder than pnp now) see, the bonding junction wants to be P biased, therefor the process we must use is WAY more complex to build an N bonding junction (Something to do with atomic latice or something) anyhow, so back in the beginnings PNP prevailed simply because they where a cheaper transistor to mass produce (VHS vs BETAMAX ???) since these where what everybody used back in those days, the progress continued on the prefection of the PNP junction, because NPN is needed in medical equipment (don't go there) the progress on manufacture of them has kept pace to a lesser degree, but NPN units rivial the cost of their PNP counterparts nowdays. but are still percived by "the old hands" in designing power sections, as being a "poor transistor" simply by design.
Some guys might say "P to N to P, that's the same as N to P to N ??? (NO it's not, as the N becomes the substrate, instead of P).
However, Can I get my 2 cents in on part of the original question
"PNP" vs "NPN" ??
The origins of Transisitors, PNP was a very easy latice structure to acheive, the P doped portion then the N, then another P doped section, "Tada!" PNP, HOWEVER, NPN, ain't so easy to make (not then, and still is harder than pnp now) see, the bonding junction wants to be P biased, therefor the process we must use is WAY more complex to build an N bonding junction (Something to do with atomic latice or something) anyhow, so back in the beginnings PNP prevailed simply because they where a cheaper transistor to mass produce (VHS vs BETAMAX ???) since these where what everybody used back in those days, the progress continued on the prefection of the PNP junction, because NPN is needed in medical equipment (don't go there) the progress on manufacture of them has kept pace to a lesser degree, but NPN units rivial the cost of their PNP counterparts nowdays. but are still percived by "the old hands" in designing power sections, as being a "poor transistor" simply by design.
Some guys might say "P to N to P, that's the same as N to P to N ??? (NO it's not, as the N becomes the substrate, instead of P).
Yes, in many cases, PNP transistors and N channel fets have an advantage. Why, because of the difference in mobility between holes and electrons. PNP has N in the base (sensitive region), and N channel fets use it for the channel. This is why, I am told. However, SOME NPN transistors have very high betas, better than pnps for the most part, and there must be some reason for this. The usual difference that can be easily measured between devices is the rbb' or intrinsic base resistivity. PNP's usually have about 1/2 the resistance compared to NPN's.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.