Ok, I admit i don't make any sense out of this

[analog_sa]

PD was recently kind to post a fragment of the ML38 schematic.

For the last couple of hours i've been staring at the PS arrangement and feeling very dumb. Power to the collectors of the matched transistors seems to be supplied on both sides from CCS feeding a unity gain opamp. What's the point?

 

[jan.didden]

Frank,

Very strange indeed. Could there be a drawing error in the interconnection of the terminals of those matched pairs? That doesn't make sense.

Jan Didden

 

[Peter Daniel]

I'm not sure if you read it right, but the dual transistor is in 8-DIP package, so it is represented by the frame outline on a drawing. The power is supplied to one of the collectors, the other collector is connected to the emiter (of the other transistor).

 

[Eva]

This appears to be a funny and exotic way of cascoding

All transistors appear to work at a fixed Vce of about one diode drop

 

[jan.didden]

But that transistor that has the collector to the emitter of the other, that places its collector-base diode in conduction, isn't it? That would influence the operation of the input, no?

Edit: that collector doesn't appear to get any supply, as it is connected (with the other emitter) directly to the output via 10 ohms.


Jan Didden

 

[Eva]

Left side transistors work with Vc = Vb so its Vce is one diode drop
Right side transistors work with Vc = Vb + diode drop so its Vce is two diode drops

Note that left side transistors have its bases tied together while right side transistors have its emitters tied together

Note also that small signal transistors may work with Vce as low as 0.1 V [ie: collector potential between base and emitter potential

Redrawing the schematic may help to understand it

 

[Peter Daniel]

Here's the redrawn version. It looks much clearer now.

 

[jan.didden]

Thanks Peter, that's MUCH better.
Eva, does that mean that the right side transistor works in what looks like bootstrapped common-base?

BTW, who designed this?

Jan Didden

 

[andy_c]

Is this better?

Here's yet another interpretation. It's a diamond buffer with the input collectors bootstrapped to the output emitters, and the output collectors bootstrapped to one diode drop away from their respective bases.

 

[Peter Daniel]

BTW, who designed this?

This circuit is used in Mark Levinson ML380 preamp, as well in some other ML units.

Is there a need for dual NPN/PNP transistors here?

 

[jan.didden]

Re: Is this better?

Here's yet another interpretation. [snip]

Neat!

This is a buffer, right? Looks like a lot of work to replace a piece of wire..

Jan Didden

 

[jan.didden]

This circuit is used in Mark Levinson ML380 preamp, as well in some other ML units.

Is there a need for dual NPN/PNP transistors here?

I think for best results they should all be matched. Difficult without duals or better, quads. The THAT series quad matched transistor come to mind.

Jan Didden

 

[andy_c]

Re: Re: Is this better?

Looks like a lot of work to replace a piece of wire..

It's a curved wire without gain

 

[analog_sa]

It's a diamond buffer with the input collectors bootstrapped to the output

After some redrawing i figured that out but what is the purpose of the opamp+current source?

 

[Eva]

The current sources force the input transistors to work at a nearly constant Ic

The op-amps work as 'cascodes' and force the output transistors to work at nearly constant Vce

Look how andy_c has redrawn the circuit in a much clear way

 

[Steven]

Thanks Andy_c, only after your drawing it was clear to me what the actual design was.
Funny, how one needs to draw a schematic in a certain way to recognize things. People are severely restricted in this compared to Spice.

Steven

 

[ChocoHolic]

In fact a cool advanced diamond!

 

[analog_sa]

The op-amps work as 'cascodes' and force the output transistors to work at nearly constant Vce

Many thanks for the explanation Eva.

 

[andy_c]

Thanks Andy_c, only after your drawing it was clear to me what the actual design was.
Funny, how one needs to draw a schematic in a certain way to recognize things. People are severely restricted in this compared to Spice.

Yes, our circuit pattern recognition seems to want to see things in a certain way. I've had circuits I've designed myself put into a CAD system by a draftsman who redrew them in such a way that I could not understand what the circuit was! Curious how the mind works.

 

[Destroyer OS]

If anyone stills looks at this thread, is it by chance the Borbely style? I am slightly lost, but I have been show this,

http://www.diyaudio.com/forums/attachment.php?s=&postid=805978&stamp=1136397897