Why Nakamichi use this configuration of output transistor

[ostie01]

Hi, would like to know why Nakamichi PA7 use this configuration of output transistors. Why they tied all connectors together instead of all emitters . Is there any advantages. Thanks


http://img342.imageshack.us/my.php?image=nakamichipa7sk9.jpg" target="_blank"><img src="http://img342.imageshack.us/img342/2029/nakamichipa7sk9.th.jpg" border="0" alt="Free Image Hosting at www.ImageShack.us" /></a>


http://img342.imageshack.us/img342/2029/nakamichipa7sk9.jpg

 

[jaycee]

It's the CFP configuration aka the Sziklai Pair. Except in this case, it's more than a pair as there are multiple transistors.

 

[davidallancole]

Looks like a CFP(Complementary Feedback Pair).

 

[ostie01]

What is the main advantage to use the complementary feedback pair
of transistor.

 

[djk]

Nelson Pass did that so he didn't need to use global feedback around the output stage. There is local feedback between the outputs and the drivers, but the main feedback loop is from the pre-drivers.

I think a lot of the nasty sound from a global feedback amp comes from the over-shoot, ringing, and settling behavior in the fast front end to compensate for the slow output stages. You can loop the faster stages and let the outputs take care of themselves. Motorola (ON Semiconductor) shows that their outputs can have on the order of 0.05%THD running open loop.

WM Leach takes this feedback from the same point in the predrivers in all the newer versions of his amps from at least the mid 80s forward, and really reduces the open loop gain in the front end in his latest version.

I really like the Threshold amps from the mid 80s that Pass designed too.

It would be interesting to use the Pass style output stage with the Leach front end.

LC Audio uses a similar output stage to the Leach, but with no loop feedback at all, I'd like to hear it too.

http://www.diyaudio.com/forums/attachment.php?postid=38107&stamp=1024679649

http://users.ece.gatech.edu/~mleach/lowtim/graphics/ckt.pdf

http://www.lcaudio.com/millprin.gif

 

[ostie01]

Hi, thanks for the reply. I'm asking because I have beggining the drawing for a Nakamichi PA7 clone. I'm really in the early stage of my drawing, some lay out from the inside, I'm looking for some transformer but I'm pretty sure what they will be. I know it will cost me around $400.00 to $450.00 to build it. The cloning will be more from the inside of the amp. I have almost everything needed as for parts, but miss some output transistors. I was thinking using 2SA1302 and 2SC3281 , 200v 15amp , 150w for the output transistors,.

Let me know what you think, Thanks

 

[forr]

Not really a Sziklai or CFP complementary feedback pair because of the lack of a low value sensing resistor between the node of the emitter of the input transistor and the collectors of the output transistors and the output.

I think it works like this :
on small signals, only the emitter followers of the output stage conduct, preferably in class A but not mandatory. When they deliver a certain amount of current, the voltage drop in the resistor of their collector is sufficient to make the common emitter power transistors entering in conduction.

Input emitter followers are in class A and output common emitters are in class C. This is a kind of dumping and was used by DPA amplifiers too.

This scheme of current dumping has been used in a different configurations, for example using only emitter followers in Elektor Edwin and Graham Nalty's amplifiers. The first using it were Phase Linear and ESS very high power amplifiers, a long time ago.

 

[Jennice]

IIRC, the PA5 & PA7 were Stasi (spelling?) designs, designed and licenced by Mr. Pass.

I wonder if he has anything to say ewhicvh can enlighten us.

Obviously, getting his attention would be the first step.

Jennice


-----
EDIT: I just sent a mail to Mr. Pass, asking for his comments.
-----

 

[jacco vermeulen]

Stasi (spelling?)

Stasis.
For Stasi you need to visit the former GDR region.

 

[jacco vermeulen]

I think it works like this :
on small signals, only the emitter followers of the output stage conduct, preferably in class A but not mandatory. When they deliver a certain amount of current, the voltage drop in the resistor of their collector is sufficient to make the common emitter power transistors entering in conduction.

That is a description of the Stasis principle.
IIRC, the bias of the PA7 Stasis section is 40mA(40mV), the big hurd should kick in above 160mW output power.

I suppose NP got the Stasis idea from George Sziklai's compound stage.

 

[lumanauw]

If you look at the famous JLH classA power amp (singleton input), this singleton input cct is CFP with gain

 

[Nelson Pass]

I think it works like this :
on small signals, only the emitter followers of the output stage conduct, preferably in class A but not mandatory. When they deliver a certain amount of current, the voltage drop in the resistor of their collector is sufficient to make the common emitter power transistors entering in conduction.

Input emitter followers are in class A and output common emitters are in class C. This is a kind of dumping and was used by DPA amplifiers too.

The parallel common emitter output devices are run in heavily
biased Class AB.

 

[StephenR]

I know this is a 10 year old thread. I am studying the PA-7 schematic and reading threads here. I can only refer to the PA-7 schematic.

If I understand correctly, at optimal bias, R122 ( emitter resistor for Q113 of the CFP) has 40mV drop at normal bias at idle, The voltage drop across R121 (3ohm) is about 120mV only. All the common emitter big transistors Q115, 116, 117, 118 etc. shouldbe all off at idle as there's only 120mV across the B-E. Seems like FORR is correct.

Am I missing something?