I am especially glad to see your 'Eyesee_v1.0' IPS. I always appreciate the simplicity of such circuits!
I will do the "eyesee" to add variety.
First , is the final design I will do for a CFA based amp.
I had the CFA-x or Nad (gonad) topologies to choose from....
I chose the 3 stage IPS (NAD).
I did so (even as the diamond input has been denigrated ) ...
because -
1.- It can be servo'ed
2.- input V/I is separate from the current feedback.
3.- I know more now 😀 .
SOME did not like the 3 stage CFA .... because -
1.- was not simple enough (give me a break) !! 😕
2.- Diamond buffer was unstable / scary ??
3.- 2'nd stage (CFB) gets too hot. This is an issue with both
CFA's as the CF stage HAS to provide that "current on demand"
to have fast slew and to satisfy a saturated VAS during overload.
I mulled over BV's enhancements (TMC , including the input in the
compensation loop, and the common ground for DC blockers.
Thimios said the original CFA-X sounded better anyways.
What exactly are we looking for ??
A.- reliability (free from magic smoke)- cool running ?
B. - sound quality 🙂🙂 .
C. - SPEED !! the almighty "slew" ....
(attachment 1) is my answer to "C" (400V/us EASY) !!
"A" is why I am a "VFA guy" .... the Baxandall/Hawksford VAS
takes care of that. OVER load creates just 3mA in the CF stage
and does not affect the diamond at all. (attachment 2- top/bottom plots)
-Clipping is textbook.
"B" will surely follow. 🙂😀
The final circuit (attachment 3) addresses most of the bugs I have seen
in all the CFA builds , give 5ppm @ <10K and 10-12ppm THD20.
Most of the original nad is the same ....
Except -
1.- CCS's Re's are fixed , DC is handled by servo ... final VAS current
is adjusted by R19 (super-pair gain).
2.- R13/14 can be changed for more/less gain with no effect on other
parameters.
3.- Q7/12 should be low Vce (BCxxx) ... all else should run cool
enough to not even need heatsinks. 😎
BTW - Unity gain is 1.6mhz and this NAD will simulate with no comp. at all !!
.... surprising , considering the "unstable" diamond + a Baxandall.
ON- to make it.
OS
First , is the final design I will do for a CFA based amp.
I had the CFA-x or Nad (gonad) topologies to choose from....
I chose the 3 stage IPS (NAD).
I did so (even as the diamond input has been denigrated ) ...
because -
1.- It can be servo'ed
2.- input V/I is separate from the current feedback.
3.- I know more now 😀 .
SOME did not like the 3 stage CFA .... because -
1.- was not simple enough (give me a break) !! 😕
2.- Diamond buffer was unstable / scary ??
3.- 2'nd stage (CFB) gets too hot. This is an issue with both
CFA's as the CF stage HAS to provide that "current on demand"
to have fast slew and to satisfy a saturated VAS during overload.
I mulled over BV's enhancements (TMC , including the input in the
compensation loop, and the common ground for DC blockers.
Thimios said the original CFA-X sounded better anyways.
What exactly are we looking for ??
A.- reliability (free from magic smoke)- cool running ?
B. - sound quality 🙂🙂 .
C. - SPEED !! the almighty "slew" ....
(attachment 1) is my answer to "C" (400V/us EASY) !!
"A" is why I am a "VFA guy" .... the Baxandall/Hawksford VAS
takes care of that. OVER load creates just 3mA in the CF stage
and does not affect the diamond at all. (attachment 2- top/bottom plots)
-Clipping is textbook.
"B" will surely follow. 🙂😀
The final circuit (attachment 3) addresses most of the bugs I have seen
in all the CFA builds , give 5ppm @ <10K and 10-12ppm THD20.
Most of the original nad is the same ....
Except -
1.- CCS's Re's are fixed , DC is handled by servo ... final VAS current
is adjusted by R19 (super-pair gain).
2.- R13/14 can be changed for more/less gain with no effect on other
parameters.
3.- Q7/12 should be low Vce (BCxxx) ... all else should run cool
enough to not even need heatsinks. 😎
BTW - Unity gain is 1.6mhz and this NAD will simulate with no comp. at all !!
.... surprising , considering the "unstable" diamond + a Baxandall.
ON- to make it.
OS
Attachments
Hello
I would love to have a question
Can the KSA992 & KSC1845 be replaced by Toshiba 2SA970BL & 2240BL, I have some of those transistors and I would love to use them
I read on the forum some DIY-ers like the Toshiba better than the Fairchild.
Also I have some MPSA42 & MPSA92, I did not check the pin out if the same as the KSA.
Thank
Greetings gabor
I would love to have a question
Can the KSA992 & KSC1845 be replaced by Toshiba 2SA970BL & 2240BL, I have some of those transistors and I would love to use them
I read on the forum some DIY-ers like the Toshiba better than the Fairchild.
Also I have some MPSA42 & MPSA92, I did not check the pin out if the same as the KSA.
Thank
Greetings gabor
"complexity" is a somewhat subjective idea. If for instance it "has a lot of transistors", it's considered complex. Even if it doesn't have a lot of transistors, if the schematic is just badly organized and jumbled looking, someone will say it's complex.
Furthermore, if it has a ton of resistors and diodes and bootstraps but not many transistors, it may still be considered simple, depending on who is looking.
Appearances can be deceiving... Within reasonable limits I think judgments on complexity are a false comfort, especially when there are more direct ways of rating a circuit like measuring it's actual behavior. For all you know there may only be 4 active transistors, the other 16 or so are just for protection and never turn on...
Furthermore, if it has a ton of resistors and diodes and bootstraps but not many transistors, it may still be considered simple, depending on who is looking.
Appearances can be deceiving... Within reasonable limits I think judgments on complexity are a false comfort, especially when there are more direct ways of rating a circuit like measuring it's actual behavior. For all you know there may only be 4 active transistors, the other 16 or so are just for protection and never turn on...
For newbie, every amplifier is complex 😀
I have 55V CT 20A trafo (second/used). I want to make an amp with single PCB (IPS and OPS). I want to use one of OS designed with minor modification. But now he comeback with new idea. This make me confused. Now I will waiting and learn again.
They are similar ... sure !
I'm sure that last schema would look complex if I did not "preen" it.
They have built the "spooky" with 80+ parts and a servo ... this might be
considered simple.
Speaking of the spook ... this CFA is a contender with it as far as coolness , thd,and accuracy. Finally , a CFA with VFA precision. 🙂
PS - none of the devices in this design are working very hard - decades
of service.
OS
Bimo, this is a somewhat "dangerous" approach, considering the new ideas will never stop coming (and they will not) 😀
Don't wait 😉 Build the one you like, test it, measure it, tune it, listen to it, check your own ideas - this will be a lot of learning with surprises, challenges, "weightlifting", failures and successes and a lot of satisfaction, pleasure, and - experience! - in the end
Bimo, I feel your pain. A while back I built a pair of OPS with MJL4302/4281 outputs figuring I would replace my poor old Leach Superamp once I decided on the best IPS for the job. I can't make up my mind. They all sound stellar including Valery's hybrid. Now OS comes back with some more tantalizing prospects claiming they may be better than the others. What's a man to do?
Bimo, this is a somewhat "dangerous" approach, considering the new ideas will never stop coming (and they will not) 😀
Don't wait 😉 Build the one you like, test it, measure it, tune it, listen to it, check your own ideas - this will be a lot of learning with surprises, challenges, "weightlifting", failures and successes and a lot of satisfaction, pleasure, and - experience! - in the end
I know. I have ordered 3U heatsink. May be 2 or 1 week later, the heatsink will arrive. So, I have 1 week to learn the new IPS from OS.
I will sim it. 😀
I have CFA (VSSA variant and simple CFA IPS with 3EF) and blameless. I search another topology. It must VFA with high slew rate dan low distortion 😎
I have 2 option of IPS. GLA and SYMASUI. I will compare those with OS new idea (VAS super pair with/out Hawksford).
The one with "Super-Hawksford" - my recently developed "building block" for VAS - will be better 😉
Bimo, please share the results - very interesting.
Bimo, please share the results - very interesting.
Hi OS,
Some time ago I simulated different VAS types inside the same CFA with lateral MOSFET OPS, to see what could be gained, and somehow EF VAS showed it's strength, quite low distortion and very good PM and GM.
I prefer input with Baxandall super pair in the input instead the diamond, but otherwise is similar to NAD. http://www.diyaudio.com/forums/solid-state/243481-200w-mosfet-cfa-amp-43.html#post3753533
Damir
PS. "2.- input V/I is separate from the current feedback."
I don't think this is correct and don't see why it's better to be separate, could you elaborate?
Last edited:
Excellent investigations, D... post 426 , I tried all those.
The "enhanced" is the Hawksford that most of the "slew" IPS's use
(successfully).
The last one " dimitri enhanced" has the Baxandall as a enhanced cascode.
Your Q1 would still saturate ?? .... but the cascode itself would not.
As far as my statement the diamond is a V/I and the 2'nd stage is definitely
I/I (CF).
The VFA equivalent would be the "H" bridge amp- with 2 true V/I's.
The CFA- X (vssa ...) uses just the first stage to do the same thing.
More current , less control .... poor little "stressed out" input pair.
I DID try the CFA-x with this VAS .... the input pair was still "stressed" ,
at 8+ mA - But it did work.
My VAS is the reverse of yours , with the Baxandall replacing the
single active device in the standard hawksford. NO saturation .. NONE!
(Attachment 1)
The current stage sees a symmetrical beta enhanced VAS and
the saturation clamp is within the FB loop of the Hawksford.
I probed the VAS , it has a peak at 10-12mhz .... I was even able
to make it oscillate at this freq. - the 22R base-stoppers (vzaichenko)
tame this nicely 😀 .
(attachment 2) no VAS cascode stoppers , no compensation
whatsoever on the IPS. All I can get is 10+mhz "glitches" 😎 ...
with the cascode stoppers , all is gone.
I would still use 33-39pf (C5).
Oh ... .asc below.
This is WAY better than any former CFA I've designed ! (stable as hell).
I maxed out at 65V in .1uS (650V/us) ...just a little overshoot.
OS
Attachments
That would be a easy as porting the IPS/VAS to 18-24V rails.
(below 1) is the "spook" with a few resistors changed .. running an EF2
class A OP into a 250R load.
The slewmaster IPS's can easily be ported to preamp/headphone OR
a larger power class A stage. NO mods needed (beside value changes).
If someone would give me a few ideas for a push-pull class A op?? .... 😎
Parts per billion territory for your headphones 😀 .
OS
Attachments
THD is not as low as typical blameless IPS, but clipping behavior is good.
Waiting for OS to make SYMASUI with superpair VAS 😉
Waiting for OS to make SYMASUI with superpair VAS 😉
The "blameless" (wolverine w/TMC) would be a more interesting "victim".
Super pair might push it into sub PPM THD20 + better overload behavior.
I do suspect it would still be "sterile" (quasi-subjective humor 😀)
OS
Good sound and VERY low THD have been achieved.
I am just going for better thermal / electrical / precision , especially
in the CFA design.
The VFA's are pretty much SOTA (cool , overdesigned , decades of
use/abuse).
Hey , at least you don't have to replace the OPS 😀 .
Just one more original IPS design - the TL072 "eyesee" based amp.
OS