Diamond buffer, class A, CFP

[00940]

I've been reading audio forums rather than law courses again today.

On diyaudio and headfi, the diamond buffers seem trendy. But there's a variation on diyaudio that didn't seem to raise much interest here. It consists in replacing the output emitter followers by a pair of transistors in CFP configuration. According to the proponents of the thing it could cut distortion down quite a lot (tenfold, optimistically). In class AB, it has stability problems and isn't supposed to sound that great. In class A however (with heatsink on BD139/140), it is supposed to shine. So it's out for portable amps but could be a fun thing to try as main amp.

Basic concept :

The whole thing is heavily biased, a bit too much perhaps. All 2sc1815/sa1015 have 15ma going through and the output transistors have 100ma of bias, so they wouldn't drop out of class A, leading to oscillations. The LM317 ccs will have to sink roughly 12ma.

 

[00940]

It doesn't seem to raise much interest. Here's the thread who raised mine : http://www.diyaudio.com/forums/showt...&pagenumber=10

Anyway, I played a bit more with the simulator. A bit with eagle too, to get an idea of how compact it could be. New schematic and layout follow. It's a rather simple design, easy to put onto a single sided pcb. The output devices have a quiescent dissipation of 1.5W (100ma bias). A 3 channel structure and an improved isolation of the opamp gain section would help I guess. Using a two channel opamp to provide for its own virtual ground seemed amusing, perhaps not optimal. If 3 channels are used it would draw quietly 500mA @ 30V.

 

[steinchen]

sounds interesting and the parts are rather cheap

 

[00940]

Basically, it's a scaled down version of this : http://www.pha.inecnet.cz/macura/next.htm

It should be more stable using 2.2 resistors rather than 0.1 at the output.

Parts are rather cheap indeed. Furthermore, which serious diyer hasn't a few 2SA1015/2SC1815 laying around, a 30V PS and a pair of OPA2134 ? The costs of testing the thing would be minimal. A few heatsinks, a few bd140/bd139, some resistors.

Sadly, I won't be able to play with it in real life anytime soon. I'll only be home by two months and no soldering iron allowed in the dorms here.

 

[blip]

This design looks rather interesting... I like the idea of an amp designed from the first step to use diamond buffers and run in class A.

Keep us up-to-date on this... You may have just created the 940-Amp. (Other name idea: DiMETA)

 

[Glassman]

you have T4 wrong in the schematics.. anyway, CFB is one nice config no doubt about that, have you saved your spice results?

 

[00940]

Right on T4, I need to change that. I've not run extensive spice tests, just basic stuff to get the values roughly ok. On the other hand, this guy on diyaudio did run some serious tests here : http://www.diyaudio.com/forums/showt...threadid=26681

edit: schematic updated

 

[00940]

Oh, it's worth checking the last post of this thread : http://www6.head-fi.org/forums/showt...highlight=sddb