diyAB Amp - The "Honey Badger"

[dadod]

Originally Posted by ostripper

That is like the pioneer. It is real nice , BUT .... more sensitive to compensation (it tends to oscillate) if you don't get it EXACTLY right. Likes to (is more conducive) run a push-pull VAS. The pioneer only used the cascode as it did not have the reduced supply points like your German amp.

I would love to have suggestions on WHAT FET to use ?? , so I could physically
implement it (layout wise). Some are D-S-G vs. a BJT E-C-B , I need to know this to allow for BJT/FET interchangeability.
OS

LSK389B or C from Linear Systems.
dado

[Tekko]

Heres another jfet input circuit from another amp i used to have some years ago: http://i.imgur.com/MhM8f.png

Though the 2SK270 is made of unobtanium. The 2SK332 is also obsolete and only found on one single place on the entire net, on ebay sold by syracuse semiconductor/components.

[ostripper]

Originally Posted by dadod

LSK389B or C from Linear Systems.
dado

I'm on it ... your talking about this "dualie" , correct - http://www.linearsystems.com/datasheets/LSK389.pdf

(below 1)- would be how it would lay out as compared to a BJT (the option). To use the same pads/layout the BJT will be "back to back" ... no biggie (a dab of thermal compound and shrink tubing - AOK) . Correct me if i'm wrong , I do not believe this will impact performance.

Input section is the one where you "sweat the details".


OS

[ostripper]

A tiny adapter for SOT23-323 bjt's/fets. Double sided , would solder right in place of the through- hole devices. Could be supplied separately , a very simple option for the input pair. just a 10 X 10mm pcb.

OS

[farjon]

Originally Posted by ostripper

I actually just redesigned the whole thing (below 1) , wanted it to "look" SOTA. Is the choice of resistor OK ? (below 2) 10mm lead- spaced noble emitter resistors - non-inductive ?

Ok, but... can you leave some space around in order to vertically mount more conventional resistors, if desired?

[Conrad Hoffman]

IMO, this is a really super design and project. One comment. If the performance is as good as I suspect, you'll likely see a difference between output inductors wound on a resistor, vs wound on a plastic form separate from the resistor. The type of resistor and leads probably determine how much effect it has. At least that's what I found in a particular fussy design I once did.

[ostripper]

Originally Posted by Sound_Buster

Ok, but... can you leave some space around in order to vertically mount more conventional resistors, if desired?

If that all you had , I see nothing in the "keep out " zones that would not allow for those 5W monsters to be mounted vertically.

By C. Hoffman - IMO, this is a really super design and project. One comment. If the performance is as good as I suspect, you'll likely see a difference between output inductors wound on a resistor, vs wound on a plastic form separate from the resistor. The type of resistor and leads probably determine how much effect it has. At least that's what I found in a particular fussy design I once did.

With BJT's , I know we are at least at the silicon chip level (.0006% THD20), the FET option is slightly less "super" . This can be offset with smaller current mirror degeneration , 10-22R on the Jfet's .... all this to get the loop gain in the same league as the BJT option. All this will be documented. I have only thoroughly tested the simple and cascoded BJT combo's (ss9014/ksc1845 - cascoded and just the ksc1845 as input pair).

The OP output coil is separate from the 4.7R resistor , my screenprint is generic and shows the resistor inside (wrong-resistor is right next to coil). As with any PPM amp , even the OP stage, voltage stage/input, and supply cap traces "star" from a single rail point right at the fuse. Not having an exact center NFB takoff would add hundreds of ppm, as well - wait till you see the input stage/final labeling.

OS

[Samuel Jayaraj]

2SK389 is nearly unobtanium and LSK variation is no better. Although 2SK170 matched pairs can be used, this is also not easily available. A better option would be to use something like the BF862, easily available, SMD and cheap; but very good performance. You ought to either keep its dissipation low or use parallel devices. This way, the amp will be practically realisable for many across many countries.

[ostripper]

My reason for post #34 , allowing for E-C-B bipolars or a plug-in 10mm sq. smd fet PCB. This is why most of my amps are "hardware store" variety BJT semi's - sourcing !!!

OS

[Bonsai]

Very nice OS! I like the concept and it means anyone coming fairly new into amp construction can get a first class result. Looking forward to hearing from all those that are going to build this!