Improved Diamond Buffer Design

[TimS]

Hi All

I was wanting to build a discrete buffer for a preamp/headphone amp, so I started with a classic diamond buffer design and I found that I had to increase the bias to drive the low impedance loads at low distortion, but then the output devices would get hot because of the heavy bias.
I needed a design that can supply large current to the output devices when the output devices needed it.

So I came up with this design here.

Enjoy

Tim

[TimS]

The previous iteration of the design was very good, but its input impedance was about 7k which is fine if the circuit is to be used as a buffer to a opamp but not so good if it will be used by itself.
The changes have increased the input impedance to about 150k but reduced the voltage swing by +- 1V and have lowered the distortion in to 32R slightly from about 0.003% to 0.002%

Otherwise the rest of the specs are the same with 0.12R output impedance, theoretically 470MHz 3dB bandwidth and vanishingly low distortion into 100R

Cheers

Tim

[TzeYang]

no degenerative resistors for the drivers and the mirrors?

[TimS]

I was wondering when somebody would ask me that.
According to the unreal world that is Microcap 6, I don't need them because I can't get the distortion any better than it is and if I add them to my simulation the distortion gets worse.

But you are right, in a real world my parts won't be so well matched and I will need some degeneration.

I will be having a go at the prototype soon so we shall see.

Originally posted by TzeYang
no degenerative resistors for the drivers and the mirrors?

[GK]

Originally posted by TimS
But you are right, in a real world my parts won't be so well matched and I will need some degeneration.

Alternatively:

http://au.farnell.com/1354180/semico...n-that340p14-u

[Elvee]

Originally posted by TimS
I was wondering when somebody would ask me that.
According to the unreal world that is Microcap 6, I don't need them because I can't get the distortion any better than it is and if I add them to my simulation the distortion gets worse.

It isn't simply a distortion matter: without resistors, it will go into thermal runaway at an incredible speed.
I suggest you power your prototype with a good, current limited lab PSU. Otherwise the experiment may end in heath and smoke...

[TzeYang]

Circuit looks very cool to be honest (though with emitter resistors it'll be even better).

There's one thing I don't understand, which of the following is of higher Zi (input impedance):

1) Source sees two diodes.
2) Source sees the bases of the driver transistors (typical diamond buffer structure).

[peranders]

Originally posted by TimS
I was wondering when somebody would ask me that.
According to the unreal world that is Microcap 6, I don't need them ..

... but you have a real world also with not equal parts at different temperatures.

Your statement is interesting. Here is where the experience comes in

[TimS]

OK

This version I have included the degeneration resistors and the BC847B's are intended to be BC847BV dual transistor pairs to improved temperature tracking, possibly matching as well as reducing parts count.
Doubled the output devices to reduce the output impedance,

The THAT 300 devices look very nice, if they were not so expensive I would probably give them a try.

Cheers

Tim

[TimS]

Originally posted by peranders

... but you have a real world also with not equal parts at different temperatures.

Your statement is interesting. Here is where the experience comes in

Other peoples experience is what I am here for.