diyAudio (
-   Solid State (
-   -   Slewmaster - CFA vs. VFA "Rumble" (

ostripper 26th December 2013 12:38 AM

Slewmaster - CFA vs. VFA "Rumble"
7 Attachment(s)
It might be time to put the controversy to "rest'.

The only real way is to actually make some amplifiers.

The best way to start (I might be wrong) , is to assume
that the simulator lies (slightly).
RNmarsh commented that many of his simulated (CFA) amps performed
much better when actually built and I have had VFA's that did not perform
as simulated.

To give everybody some "holiday meat" to chew on ,
the top contenders - that could change .... are posted below.

DIYA allows edit of first post ... ALL current work is posted here ! :cool:

OPS - Amp is split into a output stage/ capacitance multiplier (simple regulator) PCB - 2/3/5 pair versions
are posted - Rated at 70/150/250 watts into 8R. With premium output devices this could be nearly doubled.
Triple output stage topology is used , as well as a 2 device compound thermal compensation circuit.
Similar to the Harmon /kardon 680 -990 series integrated amplifiers.

IPS - Separate small PCB's can plug in to the OPS. All IPS's will run off any supply the
OPS provides. CFA and VFA IPS's can be interchanged ... as the operating voltages
and currents are standardized.

Below is the "lineup" .....

1-3. SLEWMASTER OPS's (V1.2 -3 of them) the "baby" , "master" and monster".
......3 power levels of EF3 goodness. POST(s) #500 , 502 , 504

4. NAD-S-v1.2 IPS (CFA) - Based on the excellent "NX" with servo DC and the
Hawksford Cascode. Member NAF says "it sings" gracefully. :D Post #518

5. Wolverine-V1.2 IPS (VFA) - "blameless" based with PPM ultra low distortion.
Next logical refinement after the "Badger". POST #599

6. CFA-X and "XH" (CFA , of course) - Similar to the "VSSA/peeceebee" , a very simple amp usable
with the "slew" OPS's. POST #701 -perfected (684 is an error - do NOT use ! )
Edit ... post #2739 is the official schema for the "new" version for CFA-X "H" (hawksford powered).

7. Spookyamp (VFA) Based on the 2500$ Harmon kardon 990 receiver ... uses the classic
Georgia Tech "Leach" topology. Post #803 V1.1 ...

8. Symasui (VFA) - the DIYA "symasym" on steroids post #1628 , almost forgot this one .... Post #1628

9. Kypton-C (CFA) - better CFA-X with servo and super-pair. Post # 4729.

10. Kypton -V (VFA) "classic sansui" 3 stage IPS. Post # 5384.

11. EYESEE (CFA) - IC based amp - alexander amp #4944

12. Kypton ND - (CFA) - best CFA to date ... post # 6782 test - post 6863

More to come ...


ostripper 26th December 2013 12:55 AM

1 Attachment(s)
A proposed "blank" for the output stage.

I most likely will move the the main decoupling caps -
Rail inputs /ground to board center like AAK's leach/symasym.

Very short current paths - separation of
"clean" from "dirty" (G1-G2).


ostripper 26th December 2013 01:10 AM

Just a reference - "blameless"
1 Attachment(s)
Below is the THD (10k -1/2 power)
of the much abused "blameless" topology.
TMC was used as in the "badger" - not bad
for a simple circuit !
IF .... we go with a modular setup , I will
adapt this to the triple.

It might do better than with the standard

All simulations are done with 80mA bias on
the NJWxxxx OP devices.
PS- I will soon adapt the "VSSA" to the hawksford/EF3 ...
it could also be a real contender !

ostripper 26th December 2013 04:45 AM

OPS placement.. decisions - decisions
1 Attachment(s)

Having the PS in the middle would be worst of 2 "evils".

Having 2 sensitive Vbe's crossing high current returns would
be worse than long traces coming from the fuses (rails) to
supply the regulators (1.) .. in attachment below.

Regulators will have separate decoupling , so this will not matter.

Vbe 1 will attach to the driver heatsink , Vbe 2 will be under the PCB
close to an output.
This is very similar to my HK680 .... even a little better.

As far as grounding ... 1 wire :o (G2) will run from the input stage star
to the main star. "Star within a star "

Any small signal device will decouple to G2. G1... (the big 'dirty' star)
will only decouple OP devices ,the main rail caps , and the zoble.

I checked the 680 , it also follows these rules.
Also , I am now at 3 pairs for the main OPS. :)

Bigun 26th December 2013 08:00 PM

Looking good :D

ostripper 26th December 2013 08:53 PM

1 Attachment(s)

Originally Posted by Bigun (
Looking good :D

Thanx ...

While not logged on , I noticed 100's of views/downloads of the "material".
Yours is the first post.

I really want advice , have not built a good CFA yet.
:confused: The NFB path can not (should ;) not) "snake" around the
sensitive predriver/Vbe area. Shown are the wires (in pink/red - below).
On a double sided PCB one could bring it down along a rail - later date.
Is there a better way ?
Current feedback will "dump" +/- 50ma or more into the clean ground ,
simulation shows to NOT connect to the "lifted" ground directly .... as
you will modulate the 0v reference. With a thick wire between the 2
grounds (G1 and G2) , the reference should be good , even with CF.
This IS NOT an issue with VFA - need a few pointers ....

This is a 1 sided creation for prototypes.
I will keep the big drivers , NJWxxxx is only $.50-75 more than the 15032/
15033. Semitech high current or MJL21193/4 will be options for output devices.


jkuetemann 26th December 2013 10:23 PM

My only CFA experience has been with the VSSA, and I must say I like it very much.

Looking forward to following your take on a CFA design, OS. So far things are looking good to my eyes.

nattawa 26th December 2013 10:42 PM


Originally Posted by ostripper (
.........The NFB path can not (should ;) not) "snake" around the sensitive predriver/Vbe area.....

I'd put a daughter card to the bottom side of the main PCB, and a string of multiple resistors on it, including the shunt resistor. the sample current can then return right back to the speaker ground terminal. I saw a few suggested it and I tried it in Dadod's 200w CFA SMT layout. Many wont favor this arrangement for obvious reasons, but it seems to be a very quick and clean solution to me.

Bigun 27th December 2013 03:13 AM


Originally Posted by ostripper (
Current feedback will "dump" +/- 50ma or more into the clean ground ,
simulation shows to NOT connect to the "lifted" ground directly .... as
you will modulate the 0v reference.

I could wrong (often am) - ultimately I have to think in terms of voltages. The 'error' amplifier of a BJT front-end will receive a control voltage across its base-emitter (this is true whether 'CFA' or 'VFA'). Hence, this BJT 'sees' only a differential voltage and will modulate it's collector current based on the difference in voltage at the base with respect to the voltage at the emitter. If they have different ground references the input error amplifier will not differentiate between the intended signal (feedback & input) and the parasitic signal between the ground references.

And also, although the lifted ground gets modulated by the current, it's linear - since it's lifted with a resistor.

Bigun 27th December 2013 03:44 AM

I design all my CFA amplifiers as if they were VFAs :D

All times are GMT. The time now is 04:19 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio