Slewmaster - CFA vs. VFA "Rumble"

"Project spooky" (leach)

The reason for "Spooky" is this thread ...
http://www.diyaudio.com/forums/solid-state/245032-work-progress-leach-based-amplifier.html

Looking at the "eye candy" put me on to actually designing one for
this series. The original leach article was one of my first reads from the
internet back in the 20'th century 😀 .
Leach Amp Plans - Part 1

I have "digested both the original and the "double barreled" versions. The good
Professor was very generous with such a design. 🙂

In my version , I will attempt to preserve the "sound".

1. Leach used feedback from the predriver stage for global HF feedback. I
will also allow the option of a feedback path from a slewmaster pre-driver
stage to the NFB node. The main OPS will feed the servo as is conventional.

Of course , a unified feedback path will also be available.

2. I used the hawksford .... but it just "smokes" leach's VAS , about 1/6
the full amplitude THD with this "upgrade".

3. CCS's - leach used basic zener/resistive loading for his LTP(s).
The led CCS + buffer 1K resistor I use should also have the effect of
less thermal drift / better PSSR (confirmed)

4. Faster devices , unavoidable ... 30 mhz ft drivers/OP's are standard today.
One could use the MJL21193/4 to return to the "classic speed" 🙄 .

Below I begin , having studied Leach's original layout .... I'm on the right track.

PS - Leach's split NFB scheme - 200ppm 120v p-p ... unified NFB = 41ppm - same output level 20K ...
OS
 

Attachments

  • original leach.jpg
    original leach.jpg
    96 KB · Views: 790
  • spookypcb1.JPG
    spookypcb1.JPG
    75.4 KB · Views: 703
Last edited:
Now we just need someone to iron that trace down and build it to make sure all is well. I will be waiting for sure to hear the end result.

Buzzford I can't wait till you test that circuit and let us no the BUZZ! 😀

Member NAF's NAD IPS was a little "squirrely" at first(simulation).
But it ran flawlessly in real life (squirrely models ?).

This one is FAR more stable/simple , can run without compensation caps.
The NAD is faster , as it has a capacitor less DC feedback path and
that diamond buffer.

CFA-X CAN'T fail .. it's too simple. Almost like hooking a 1K resistor + led
to a 9V battery (correct polarity) .. it HAS to work. 😀

OS
 
Please forgive me.
In the board POST # 666 and POST # 684 in my opinion are inaccurate.
R14 and R17 are not attached to power rails as in Scheme

No , I COMMEND you ... 😱😱😱😱 That would be a serious error.
I have now corrected this /.... do NOT build CFA-x post 666/684.

the new corrected one is below.....

Thank you , friend !! 😎:Olympic:🙂

OS
 

Attachments

Last edited:
Os,
It is only a request, but a bridged version is what I would like to build. If it's not your things, it's cool, but my system is balanced and I would like to keep it that way. If there is a jumper board that I can use to connect two SE versions, I can get started on them, but you seem to be a stickler on layout.
 
Keantoken,
When you say a model of a particular transistor could be improved what do you mean by that? As someone who has just downloaded and installed LTSpice I have no real experience with it yet, so I am wondering what makes a good model and what is missing in the models that are being traded around this site?

A Newbie to spice,
Steven
 
There needs to be enough accurate parameters in a model that they behave reasonably close to the transistor. There are several parameters that need to be specified correctly or any detailed simulator work will be futile.

There are any number of things that could be wrong with a model. Vbe could be wrong, Hfe could be just a straight line (or zero across the board), Cbe could 2pF (common with bogus japanese models), or there could be some bogus parameter that makes the simulator refuse to work. Basically anything you can think of could go wrong if you use just any model out there, because the computerized solvers will give parameters outrageous values when they fail to get the curve they want.

If you want to know what's wrong with the models, all you need to do is set up a simulation to reproduce the datasheet curves. As many of them as you can. That will work better than if I try to explain it all.
 
Os,
It is only a request, but a bridged version is what I would like to build. If it's not your things, it's cool, but my system is balanced and I would like to keep it that way. If there is a jumper board that I can use to connect two SE versions, I can get started on them, but you seem to be a stickler on layout.

With a simple OP-amp buffer , balanced to unbalanced could be done.
That is one way (below 1).

The better way is just to power 2 amps with a buffered balanced input.
They will be bridged at this point.

Example - 2 baby slews with MJL21193/4 , running at 40V rails bridged
would be 200+ w /8R. You would need a 40 CM long heat sink.
I would not layout a bridged setup unless someone tried and tested it first.
(with separate amps)...
Never built one 😱 .

OS
 

Attachments

  • BtoUBconv.jpg
    BtoUBconv.jpg
    20 KB · Views: 706