NPN on both sides of the output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
NPN etc

Hi Pedro,

I think people will have trouble answering your question because it doesn't really say what your problem is. Do you have a specific schematic in mind that you want to change? Do you just have this idea or what? You do of course realise that just changing a PNP for an NPN ends in disaster.
Maybe you should look up the thread on the circlotron amp with power fets, that is using like-polarity devices in the output stage.

Jan Didden
 
Most designers tend to avoid N-channel or NPN output stages. I think the main reason for this is assymmetry (spelled right?). You get emitter out (low output impedance) for positive signals and collector out (high output impedance) for negative. You can reduce the negative aspects with "compound" transistors and/or feedback. I would say that there are no high performance amp with this kind of design. Am I wrong?
 
Well, never passed in my mind to change PNP with NPN. That will not cause a disaster. Only a BUMMM!! :D

I've forgot one thing, The output stage is all with NPN and the drivers are designed to work with that.

The question is: what are the differences in the signal? Is it better, or not?

Tanks 1 more time.

Pedro Martins
 
I am not the person to answer this question but just to clarify. I think the question is:

Does the use of NPN transistors only in a push-pull OPS (quasi-complimentary?) result in a better performance than using the conventional complementary NPN and PNP OPS?

I suspect the question is based on the PSSAUDIO (they use is I think) statements in another thread?

A MOSFET version of this concept often discussed the N-Channel?:
http://www.aussieamplifiers.com/downloads/

/UrSv
 
If the NPN devices are driven by identical, but exactly 180degrees outta phase signals, the output stage will in theory be more linear since the devices are more alike. NPN / NPN pairs have much closer input/output characteristics than NPN / PNP pairs.

As for different impedances as the signal swings positive and negative is wrong. The load sees transistors, transformer, wiring, etc, in series no matter which order they're connected. An amp with NPNs at the negative rail, PNP at the positive and load connected to collectors will have same output impedance as one where the transistors are emitter followers,(with all devices being equal).

Regards Nick
 
So, I conclued that I must follow the PNP/NPN complementary type of OPS.
I can achieve full simmetrical sinal swing from input to output. Is that right? I must never forget to insert a biasing transistor between the pre-drivers. Is an obligation that this transistor have 'thermal tracking' with tor output transistors?
I'm new in designing amplifiers.
I've always constructed from designs in books and webpages.

When I full design my own amplifier, I'll put it in these forums for discussion.

Many thanks to all of you who reply.

Best Regards,

Pedro Martins
 
sapito said:
I like the designs of PSS audio and I've made one of it few years ago and it still work.

Well,

Does the use of NPN transistors only in a push-pull OPS (quasi-complimentary?) result in a better performance than using the conventional complementary NPN and PNP OPS?

Yes, that's what I need to know.


Pedro Martins

Pedro,

Two different issues here:

- The quasi-complementary OPS is using an NPN output device and PNP driver to *simulate* a hi-power PNP. That is not very good, generally.

- The circlotron topology uses two NPN (or FETs) in a symmetrical topology that is much better (like two tubes in a PP arrangement). Again, check the treads, all you ask is there!

Jan Didden
 
Am I wrong?

Yes I am afraid so peranders........

"You get emitter out (low output impedance) for positive signals and collector out (high output impedance) for negative"

Ummm.... I don't think so.

"I would say that there are no high performance amp with this kind of design."

http://www.passdiy.com/pdf/citation.pdf

http://www.creekaudio.co.uk/products/5350.asp

The 5350 uses a high current, N channel only, MOS-FET power output stage, driven by a P channel MOS-FET phase splitter and double differential voltage amplifier with symmetrical class A driver, to achieve ultra-low distortion and high speed with minimal negative feedback.
 
Re: Am I wrong?

HarryHaller said:
Yes I am afraid so peranders........

"You get emitter out (low output impedance) for positive signals and collector out (high output impedance) for negative"

Ummm.... I don't think so.


I don't want to start a sandbox debate (sandlådedebatt in swedish) but isn't there some truth in the unsymmetry thing? When we talk large signal properties, we have very different output impedance, or? The transistors aren't very "bidirectional" sort of speaking.

What does our guru Nelson say in this matter?

Anyway, the reason to use only NPN devices must be (as I see it) that you can't get (or won't) any PNP devices. I can understand if we talk MOSFET's because there aren't that many P-channel as there are N-channels.
 
Hi Peranders

The output transistors can be regarded as paralleled from an AC point of view so I assume it is not quite correct to say that for one of the half cycles the impedance is very low and for the other one very high.

I also once read an article by Self on that subject (quasi complementary) a long time ago and can only remember that he came to the conclusion that it was inferior to a complementary, but I can't remember why.

But it is a fact that there is better availability for N-channel MOSFETS and NPN BJTs than their opposite-polarity counterparts.


Regards

Charles
 
Also Naim use only NPN output trannies.
It's always important in these matters to think like electrons when considering topologies rather than looking for visual symmetry in the schemtic diagram. PNP transistors are not electrically opposite to NPNs - there are no "positrons" to carry the current in a PNP device. Instead you end up with what are called "holes". The upshot is that it is very hard, if not impossible in practice, to find a power NPN and a power PNP that are identical electrically. Add to this the greater availability of NPNs, especially those with high speed characteristics, and the reason for quasi-complementary outputs becomes clearer.

I would argue that a quasi-complementary output, implemented with care, will be better than an NPN/PNP. Of course, you then have to match the driver transistors instead since they will be NPN/PNP but I would argue this is much easier because of better availability of devices and the dominance of the output transistors characteristics on the sound. In theory, if the drivers and surrounding components are perfectly matched then a quasi-pnp will be electrically identical to its all-npn darlington counterpart. That is to say the electrons don't look at the schematic diagram.

My amps use quasi BJT outputs. I have tried using complementary but found the choice of appropriate BJT transistors limited. The MJ15022/23 seem to work quite well together, but they are slowish. You can get a polite sound from these. I have found FETs even harder to match up. When trying to find FET complements don't just look at devices with complementary part numbers - you'll see the P-channels usually have significantly different gms or capacitance values or both.
 
The NPN that works on the negative side is driven by a PNP transistor, this setup is called compound pair and has some advantages compared to the darlington setup on the positive range.
It virtually works like a PNP but has better input linearity amongst other things.
 
Traderbam,

please note that in a quasi-complementary output stage the
output transistors will not operate in the same way. One will
work in CE configuration and the other in CC configuration and
this will most likely make them appear much more mismatched
than the NPN/PNP pair in ordinary push-pull configurations.
 
I'm talking about the linearity of the input impedance, further the compound pair seems to have some kind of local feedback.
It doesn't matter that the NPN is connected as common emitter, the driver is the one that counts because it controls the current through both transistors and compensates for the lower bandwidth of the power transistor in CE mode.
And of cause you can get closer to the rails.
What I want to say is that not the compound pair is the problem, but the darlington with it's extremly varying input impedance.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.