Brother Val has the "clean" one - I have the "dirty one".
Already built and tested this 5 years ago - the ol' "BX" (below).
Bootstrapped VAS / 7 device simplicity - as "low/dirty" as you can go ....
(omit ALL CCS's and the servo)
I'll make this layout a virtual "playground" for P3A and "DX" lovers.
2 options so far.
OS
OS, I design VAS with bootstrap too. Here.
It seem, some people on my country love it. May be it is simple, easy to understanding, and have enough performance.
They are nice and simple .... but have a"thump" at turn on. Servo cancels that.
The LED CCS drops THD 20 .... .01 down to 15-20ppm. Their choice -
How "dirty" they want it.
BTW - nice design , like my original -5 years ago
.OS
I have also tried dirty and simple amplifier "Douglas self" low distortion but I don't like the Sound
Merry Christmas and happy New year to all��
Symmetrical CFA is a way to go
a lot of options in this thread. 
Marry Christmas and Happy New Year to OS & guys. 
Symmetrical CFA is a way to go
Don't discount a fast symmetrical VFA - The spooky leach VFA indeed is nice.
At low volumes it "sings better" than one of my OEM's and matches my HK680.
Next (to hear) is my take on your VSSA (kypton-C) , direct comparison to the spook.
Then , to see if it is just "gobs" of slew-rate that makes for magic - the infidel.
I suspect a CFA's low Z feedback is just like the bootstrap - "dirty" feedback
right from speaker = believable reproduction.
PS - I've yet to do a 75V rail full power test - this where an amp "comes out
of the closet" - shows it's "wild side".
Merry Xmas /happy new year to all
.OS
Dirty VFA - harmonics.
Perhaps the correlation of 2nd harmonic domination to perceived listen-ability IS
no B$ !!
(below 1) = higher distortion with NO CCS's (.01%) , tail runs off the -12V zener
direct. But , harmonics are "right" (perfect).
(below 2) = add the LTP CCS , looks more like the "badger/wolverine". H3
is dominant , but total THD is 1/5th the first plot (<20ppm).
Now , actually KNOWING- (seeing) this .... would this alone subjectively
push one towards the H2 dominant circuit ? Or , is there something more ?
PS - even as simple as it is .... this amp accounts for over 1/3 of DIYA's
projects (from the beginning) - it MUST be explored.
OS
Perhaps the correlation of 2nd harmonic domination to perceived listen-ability IS
no B$ !!
(below 1) = higher distortion with NO CCS's (.01%) , tail runs off the -12V zener
direct. But , harmonics are "right" (perfect).
(below 2) = add the LTP CCS , looks more like the "badger/wolverine". H3
is dominant , but total THD is 1/5th the first plot (<20ppm).
Now , actually KNOWING- (seeing) this
.... would this alone subjectivelypush one towards the H2 dominant circuit ? Or , is there something more ?
PS - even as simple as it is .... this amp accounts for over 1/3 of DIYA's
projects (from the beginning)
- it MUST be explored.OS
Attachments
Oh, Ive tried it ...
5+ ma for the 1381/3503 gives the lowest H3 (on all the ips's). 7 or more is
"diminishing returns" (very little change). 7+ heats up the VAS , as well.
OPS bias (on my real amp) is the "biggie" ... 60+ma , you can really hear the
improvement. Up to 100ma (diminishing returns , again).
Never heard any IPS's on lateral FET's , BJT's have such solid bass. Just
one pair 0281/0302 - geddy lee's (rush) bass riffs are so distinct. These (slew EF3)
are as good as the HK680 in this regard. 5 pair OPS - the bass definition
must be "insane"(100's of watts) ..... ( I WON'T push my 1 pair ... don't have any extra pairs).
OS
5+ ma for the 1381/3503 gives the lowest H3 (on all the ips's). 7 or more is
"diminishing returns" (very little change). 7+ heats up the VAS , as well.
OPS bias (on my real amp) is the "biggie" ... 60+ma , you can really hear the
improvement. Up to 100ma (diminishing returns , again).
Never heard any IPS's on lateral FET's , BJT's have such solid bass. Just
one pair 0281/0302 - geddy lee's (rush) bass riffs are so distinct. These (slew EF3)
are as good as the HK680 in this regard. 5 pair OPS - the bass definition
must be "insane"(100's of watts) ..... ( I WON'T push my 1 pair ... don't have any extra pairs).
OS
Guys always plays around output's bias but for me the most important is Voltage Amplifying Stage bias since changing it within acceptable perimeter nicely correlates H2 or H3 dominance, at least in diamond transistor CFA, try it in simm
I will say that the most important stage is input stage in a amplifier, which is also documented by Douglas self " . Inputstage" transconductance" and ability to charge VÁS stage "cdom"
Please prove me wrong I'm open to learn
I will say that the most important stage is input stage in a amplifier, which is also documented by Douglas self " . Inputstage" transconductance" and ability to charge VÁS stage "cdom"
Please prove me wrong I'm open to learn
Typically - you are right.
But with the Hawksford and super-pair VAS , combined with less Cdom - it's
the VAS design that steals the show.
Example = "Spooky leach" with hawksford runs 1/2 cdom and has 1/5th the thd
as the professor's original.
OS
the feedback injection is also crucial... This is why I have great expectation for the double diamond (infedel) I still see this as a CFA, as the injection into the amplifier is pure current. The second diamond is a fantastic V/I converter. this is why the slew can be so tremendous and why it retains the expansive current on demand from the CFA.
Quite sure there's plenty of work to be done in the OPS. there must be a way to create a sliding bias or a way to prevent the output transistors from ever switching. or a way to make the turn of and on so soft and gradual that there's no spikes to be handled through the FB network. There has been a few attempts Nelson posted some of his "old" patent papers in the Blowtorch thread, and I remember some KRILL OPS claiming non switching.
Quite sure there's plenty of work to be done in the OPS. there must be a way to create a sliding bias or a way to prevent the output transistors from ever switching. or a way to make the turn of and on so soft and gradual that there's no spikes to be handled through the FB network. There has been a few attempts Nelson posted some of his "old" patent papers in the Blowtorch thread, and I remember some KRILL OPS claiming non switching.
Last edited:
I agree that the exploration of a non-switcher is warranted. It would be interesting
to just tell people that the OPS is (when it really isn't), and see if they can tell
the difference from a finely tuned EF3 ??
By the sound I hear now , "diminishing returns" is the buzzword that comes to mind.
PS - running true class A might be different (2 pair MT-100 = 50-75W)
).Edit - concerning Nelson's old patent's - look at the old devices involved , I'm sure
our 35mhz new modern OP's switch much nicer.
OS
Last edited:
What about tripple diamond amplifier, Infidel plus buffered diamond output stage (tripple output stage) from pic below.
Integrated error correction (differential diamond stage) and minimal output bias current defined with second pair of CCS.
Attachments
Last edited:
No poetry, neither 'dirt' in low Z feedback. Just necessary to get no phase turn at HF with the parasitic capacitance in the emitter of the input transistor. it means an accurate feedback up to high frequencies. That makes all the difference with VFA where the feedback signal has to across a transistor (with the added poles) before to be subtracted.
What about tripple diamond amplifier, Infidel plus buffered diamond output stage (tripple output stage) from pic below.
Integrated error correction (differential diamond stage) and minimal output bias current defined with second pair of CCS.
If the Infidel is the one to be pushed as the SOTA one, I wonder why OS not using thermaltrack transistors from the start. No other options. Is it still in production?
So that's why you sacrificed the input stage a bit
What input stage ? Infidel ? I could of gone full cascoded / CCS ... but then the
perfect thermal cancellation of those exact P/N gender devices would be thrown out.
As it is now , resistive current sourced , already 14 devices versus 20+.
Sanken has an equivalent integrated diode package ( with 2 diodes) , SAPxxx
- darlington with 5 diodes , ON has the D series with 1 diode.
There is no standard here. design something with thermaltrak's , anyone who can't
source them is left out in the cold.
HK has the right idea for the 990 , they just use 2 thermaltraks for the output Vbe
(other 4 pair are regular OP's).
while using the same separate driver Vbe as the HK680/slewmaster.
It would be easy to make the slew accept one pair thermaltrak (with the option of a standard Vbe).
As it is now , the 680/slewmaster vbe multiplier is the BEST I've seen (OEM or
typical project).
OS