Obviously you havent listened to a krell, accuphase Myryad, Arye and even low cost Cyrus models. Nothing wrong with the diamond. Your ears are lying to you, fortuneatly others know better.
That was selling advertisement, don't worry we know that diamond buffer(or Baxandal pairs) sound excellent. Manso you promised to send me an email and I still expect it to arrive.
BR Damir
How about telling us what evil
Diamond i/ps buys for us amp makers? I've already mentioned one. Got any more?"Looks like Krell bla bla" dun kunt.
But if you'd like to summarize how Krell ... Cyrus bla bla get around da Diamond i/p problems, I'm sure we'll all be grateful.
Here's what 'simple CFA' buys us. If you so wish, you can have 1nV/rtHz noise or better. Of course not everyone needs this.
I don't support topologies for religious reasons .. only for stuff (which I want) they provide more conveniently or better than other topologies.
Last edited:
How about telling us what evil
"Looks like Krell bla bla" dun kunt.
But if you'd like to summarize how Krell ... Cyrus bla bla get around da Diamond i/p problems, I'm sure we'll all be grateful.
Here's what 'simple CFA' buys us. If you so wish, you can have 1nV/rtHz noise or better. Of course not everyone needs this.
I don't support topologies for religious reasons .. only for stuff (which I want) they provide more conveniently or better than other topologies.
Thats a important one.
Complexity, Simple Diamonds can be had using as little as 22 ohms degeneration, must I really post some of these manufactureres schematics. How about one for yourself as proof of concept in exhange for one of the virgins
. While trying to free yourself from 2 capacitors you create more complexity than if you would just use the diamond norm.
Other reasons for use of the two extra trannies with the diamond which can be given, are not relevant to audio amplifiers but do become so with IC opamps which are used for other purposes. Maybe soon youll see a application where it becomes relevant, I still do owe Dadod a email.
Yes one loses on noise but creating a special amp for just 0.00001 % of the audiophile world population does not make sense, its ridiculous.
Diamond ips is not a problem, there are enough examples of commercial schematics around. A little matching helps, maybe something wrong with the servos ....
Why Krell dun kunt ....?? I repaired one a while back, there were more diamond buffers in there then my little daughter could kunt.
Da larger emitter degen resistors reduce i/p Gm and also introduce noise.
I'm claiming this as a win for simple CFA like VSSA & what I posted in #822 etc
You need a servo or twiddle pot for both evil Diamond & simple.
The disadvantage of course is needing big evil electrolytics.
But I have a supply of electrolytics handcarved by virgins from solid Unobtainium which have vanishing levels of THD, supa musicality bla bla.
Send me $450 in used bank notes and I'll send you a sample. You won't be disappointed.
.. but seriously folks .. I'm looking at one of Guru David Zan's applications which needs 1nV/rtHz noise at the i/p and zillion W o/p so that precludes evil Diamond i/p.
As I pointed out earlier, simple CFA is a Classic Very Low Noise topology.
______________
I actually have an active ribbon mike application which requires matched Vbe. Can't figure out how to do even a simple trim without introducing noise. Things are difficult below 1nV/rtHz
I cannot see how you would need 1 nV per root Hz on a practical power amp Richard. Sorry, it just does not make sense.
The diamond is great . . . I would certainly not go of something that requires 'lyrics
I chose to increase the degeneration resistors to overcome the Vbe issues. However keep in mind:-
1. If both genders of the input stage are true complements and both the buffer and level shifter run under the same Vce and Ic, the problem is much reduced.
2. If you want to keep the degen resistors low value, then alternatively just run the buffer emitter currents higher than the level shifter current. eg 3mA on the buffer and 1 mA on the level shifter
3. If you feel strongly about it, you can always Vbe match - after all, the complementary JFET input designers do this as a matter of course.
So, there are plenty of options here, all of them providing good solutions.
1. If both genders of the input stage are true complements and both the buffer and level shifter run under the same Vce and Ic, the problem is much reduced.
2. If you want to keep the degen resistors low value, then alternatively just run the buffer emitter currents higher than the level shifter current. eg 3mA on the buffer and 1 mA on the level shifter
3. If you feel strongly about it, you can always Vbe match - after all, the complementary JFET input designers do this as a matter of course.
So, there are plenty of options here, all of them providing good solutions.
The Vbe differences create current hogging which then does not allow offsets to be correctly dialed out. It's nothing to do with intrinsically solving offset with increased degeneration voltage - there are two separate mechanisms.
He forgot to tell you, the virgins are 85 years old. It takes that long for them to gain that skillI ll settle for the virgins only thanks, now how much are those ?? 450$
Dont claim a win so fast. Its easy to get around this problem. Members here just havent found the correct solution. Look at Nad, Myryad, Cyrus, Accuphase and others, do you see them having this problem ?? No
Do they rely on SMD closely matched pairs ?? No.
This is for John Curl too. I wonder if he knows that the recently released products from Ayre which is getting high accolades in the audio press are CFB designs, and all feature the dreaded Diamond buffer. I mention John as I believe he respects Charles work. See what Charles Hansen says about the diamond (Hifi News Oct 2013). Dont if you dont like the diamond
Even own some of those cr.., they're fine enough to circulate among adds. As long as they're out of the chain you're on a safe side.Obviously you havent listened to a krell, accuphase Myryad, Arye and even low cost Cyrus models. Nothing wrong with the diamond. Your ears are lying to you, fortuneatly others know better.
Er.rrh! You still haven't told us what advantages Diamond i/p gives over Simple CFA.Thats a important one.
Complexity, Simple Diamonds can be had using as little as 22 ohms degeneration, must I really post some of these manufactureres schematics. How about one for yourself as proof of concept in exhange for one of the virgins
While trying to free yourself from 2 capacitors you create more complexity than if you would just use the diamond norm.
Other reasons for use of the two extra trannies with the diamond which can be given, are not relevant to audio amplifiers but do become so with IC opamps which are used for other purposes. Maybe soon youll see a application where it becomes relevant, I still do owe Dadod a email.
Yes one loses on noise but creating a special amp for just 0.00001 % of the audiophile world population does not make sense, its ridiculous.
Diamond ips is not a problem, there are enough examples of commercial schematics around. A little matching helps, maybe something wrong with the servos ....
Why Krell dun kunt ....?? I repaired one a while back, there were more diamond buffers in there then my little daughter could kunt.
I've already mentioned the caps. You may like to look back on some of my previous posts with examples of Simple CFA for circuits and compare complexity.
And in your list of Krell ... Cyrus, there are several which are unstable into 'real life' speakers .. though I'm not sure CFA has anything to do with this.
And I DID say, 1nV/rtHz isn't for everyone but you may like to work out what you really need for your zillion bit D/As
______________
Andrew (Bonsai), I'm really just making sure we keep Simple CFA on the table (and yanking your chain a little).
I was disappointed to hear of your troubles with nx-amp
but if I build a CFA, it will likely be one of your excellent designs. 'Tried & tested' beats theoretical flights of fancy any day. I'm re-thinking my 'favourite' amp topology because of shortcomings in areas which are nothing to do with 1ppzillion THD or zillion V/us slew. To really move forward, it's important to question all previously held beliefs & prejudices and concentrate on very basic & crude parameters which have been checked with Double Blind Listening Tests.
The chosen topology/design should address these issues .. not the other way round.
Last edited:
Lazy Cat said:
May-i suggest to both of you to bring positive and documented (based on experience) inputs instead of what is souding like autosatisfaction for one and designer jealousy for the other, while both does not help ? This is unworthy of both of you.Your ears are lying to you, fortuneatly others know better.
Sorry for this 'old man' reaction against two guys i appreciate.
Diamond input can bring obvious advantages to the VSSA, by the removal of the big electrolytic capacitance in the feedback path. And the associated distortion at low frequencies. I believe L.C. when he found the diamond version he tried was sounding worse than the simple one. And had experimented myself such issues (something less transparent and sparkling, what L.C. describe as "nervous").
I'm sure the solution is to dig further in the problem to find the solution (resistance in serial between emitter of the first transistor and the base of the second one ? Various current combinations ?
"Andrew (Bonsai), I'm really just making sure we keep Simple CFA on the table (and yanking your chain a little).
I was disappointed to hear of your troubles with nx-amp but if I build a CFA, it will likely be one of your excellent designs. 'Tried & tested' beats theoretical flights of fancy any day.
I'm re-thinking my 'favourite' amp topology because of shortcomings in areas which are nothing to do with 1ppzillion THD or zillion V/us slew. To really move forward, it's important to question all previously held beliefs & prejudices and concentrate on very basic & crude parameters which have been checked with Double Blind Listening Tests.
The chosen topology/design should address these issues .. not the other way round. "
Build a set. As they say, 'Listen and you'll Hear'
You are right - making claims that are not tested properly, or do not stand up to scrutiny are simply a waste of time and nonsense.
That said, no amp is perfect, and they all have tradeoffs - you have to pick your poison.
I was disappointed to hear of your troubles with nx-amp but if I build a CFA, it will likely be one of your excellent designs. 'Tried & tested' beats theoretical flights of fancy any day.
I'm re-thinking my 'favourite' amp topology because of shortcomings in areas which are nothing to do with 1ppzillion THD or zillion V/us slew. To really move forward, it's important to question all previously held beliefs & prejudices and concentrate on very basic & crude parameters which have been checked with Double Blind Listening Tests.
The chosen topology/design should address these issues .. not the other way round. "
Build a set. As they say, 'Listen and you'll Hear'
You are right - making claims that are not tested properly, or do not stand up to scrutiny are simply a waste of time and nonsense.
That said, no amp is perfect, and they all have tradeoffs - you have to pick your poison.
I have been off-line for a while due to the failure of my internet provider. Nice to find someone still looks out for my concerns
.I did use the time to work on a circuit and have a pretty decent contender.
Close to 1nV/rtHz, distortion at full power is only a few ppm, several hundred V/uSec slew rate.
I haven't been more specific because at those levels it's already better than any rational requirement, whichever parameter people want to obsess about. And details of implementation will probably dominate the simmed results anyway.
The amp is unconditionally stable, better than 90 PM.
It clips nicely, not that I care, because the compression driver efficiency that make the noise an issue conversely makes overload a non-issue, your ears will bleed before the amp clips at >130 dB.
14 transistors in the front end (IPS + "VAS") and an EF2 OPS, no error correction, so not too complicated.
And it avoids some of the problems of a typical "CFA" that create the need for ancillary complexity.
Nicely symmetrical with low offsets, so no servo should be needed either and decent PSRR so simpl(ish) power supply.
But I am still interested in "CFA", Some of the ideas for my contender were clarified here.
And I found that the noise cost of the diamond buffer can be fairly small, certainly adequate for the vast majority of requirements, as Andrew says.
So it still seems a close call to me whether the elimination of the theoretically undesirable capacitors in the simple CFA is worth a little extra noise from a diamond buffer.
If competently done I expect the difference would be totally inaudible in a properly conducted (blind of course) test.
Best wishes
David
Last edited:
Here's my 
2 cents.
At least in simulation , a "tweaked out" VFA "leach" can slam any of the
CFA's.
And not have offset/PSRR issues.
The IC makers have absolute matching by process ... so CFA is easy for them.
With our crude though- hole DIY construction , this is more difficult.
I'm new to symmetrical topology's , but the VFA approach can exceed
in all 3 categories (psrr/slew/offset correction).
250V/us 100db+PSRR/1mv offset is easy. I struggle to get the last 2
attributes with a CFA ... I might have to use 14 devices for IPS ... but it pays
off.
OS
2 cents. At least in simulation , a "tweaked out" VFA "leach" can slam any of the
CFA's.
And not have offset/PSRR issues.
The IC makers have absolute matching by process ... so CFA is easy for them.
With our crude though- hole DIY construction , this is more difficult.
I'm new to symmetrical topology's , but the VFA approach can exceed
in all 3 categories (psrr/slew/offset correction).
250V/us 100db+PSRR/1mv offset is easy. I struggle to get the last 2
attributes with a CFA ... I might have to use 14 devices for IPS ... but it pays
off.
OS
Hello Bonsai
There are no offset issues in a correctly designed VFA - the correct cures for this in discrete designs are now clear also , one approach is to use a current source . Transistor input VFA opamps have very low offsets using this approach.
Regards
Arthur
O stripper, Now exceed is bold statement.
To my experience the data we're so bravely using here to determine performance has so little to do with exactly that. We debate hearing thresholds on distortions. BUT I find the real performance leaps in how the amplifier we create deals with back EMF and funny reactive loads, how well the grounding system is made and how good we are at boxing the full shebang. off course the minute differences in PPM has academic value, but to draw conclusions on the amplifiers real sonic performance from that is like building an F-1 car based on computer analysis alone. NOT possible if you want the front row. Simulation is good to prevent smoke and explore circuits, but to really look at performance you need to ride the beast.
To my experience the data we're so bravely using here to determine performance has so little to do with exactly that. We debate hearing thresholds on distortions. BUT I find the real performance leaps in how the amplifier we create deals with back EMF and funny reactive loads, how well the grounding system is made and how good we are at boxing the full shebang. off course the minute differences in PPM has academic value, but to draw conclusions on the amplifiers real sonic performance from that is like building an F-1 car based on computer analysis alone. NOT possible if you want the front row. Simulation is good to prevent smoke and explore circuits, but to really look at performance you need to ride the beast.
MiiB, totally agree, measured parameters not always reflect real performance, they are indicative and example with F1 is very good. Most of F1 cars are built on the same basic principles, but actual performance vary a lot due to a lot of nuances. However, looking at performance parameters table, you most likely can determine the group of favourites from the group of outsiders. So indication is still useful. And this indication is what engineers use when improving their designs.
In a nut shell - measured parameters don't give you the full "picture", but worth to assess.
In a nut shell - measured parameters don't give you the full "picture", but worth to assess.