Enough excursion to my layout.
Let's get back to circuits.
I want to refer to Richard's schematics.
Would it not be better to use a voltage reference or LED's (to replace e.g. R5) in combination with a CCS diode like J511 (to replace R1) to bias the MOSFETs ?
I expect at least better PSRR ??
Patrick
Let's get back to circuits.
I want to refer to Richard's schematics.
Would it not be better to use a voltage reference or LED's (to replace e.g. R5) in combination with a CCS diode like J511 (to replace R1) to bias the MOSFETs ?
I expect at least better PSRR ??
Patrick
Hi Bogdan,
John Curl, who designed the preamp, is selling it commercially. He has stated he has no intentions of releasing the schematic. I'm sure there are unit to unit variations to compensate for component variables. I fully support John's decision to not release the diagrams for those reasons.
If you read the entire thread, there are many clues. This is more of a gift as you can apply these lessions to your own low level designs.
Read and learn.
-Chris
John Curl, who designed the preamp, is selling it commercially. He has stated he has no intentions of releasing the schematic. I'm sure there are unit to unit variations to compensate for component variables. I fully support John's decision to not release the diagrams for those reasons.
If you read the entire thread, there are many clues. This is more of a gift as you can apply these lessions to your own low level designs.
Read and learn.
-Chris
back to the psu ...
I’m trying to put the ideas gathered here to work (serie reg + CCS + shut reg).
I still use the basic Curl follower running under switchercad. I think this circuit is both less sensitive to the psu and “harder” to the psu than the symmetric folded cascode we are talking about here. Sorry If I’m lagging ideas by Richard and allied…
How can I improve the line regulation provided by the simple shunt reg ? I imagine that higher gm transistors would help (which ones?). At about 100ma the gm of the irf640 and irf9540 are just a bit above 1S. Regulation is quite similar to the one provided by a .85ohm internal resistance voltage source.
I can obtain a slightly more linear regulation in high frequency with a small (about 470nf) in // with the reg. It also helps to eliminate the overshot one can see with square waves (I think that Richard published them a while back). Funny, John says “no cap” and he must have good reasons.
A “massive” approach to passive filtering, 3 cells composed by a 10ohm resistor and a 47000µf 1nH ESL and 3mohm ESR cap (I hope these values are realistic), provides a much higher attenuation in most of the audio frequencies and above, but this attenuation is changing with frequency and there is a severe bump at very high frequency (due to very low esr ?). I know that only the last cell is useful in term of regulation.
The active reg shines in terms of line rejection. I probably obtain one more time meaningless results because real live noise masks the simulated behaviour that lies at impossibly low levels. Anyhow, I think that the curves are interesting. For instance, you have a less acute peak in HF using 3 different caps in the passive way and you can see the interest of multiple cells.
Any help and comment gratefully welcomed.
Philippe (who as difficulties to explain himself in English and whose mail is broken :-/).
I’m trying to put the ideas gathered here to work (serie reg + CCS + shut reg).
I still use the basic Curl follower running under switchercad. I think this circuit is both less sensitive to the psu and “harder” to the psu than the symmetric folded cascode we are talking about here. Sorry If I’m lagging ideas by Richard and allied…
How can I improve the line regulation provided by the simple shunt reg ? I imagine that higher gm transistors would help (which ones?). At about 100ma the gm of the irf640 and irf9540 are just a bit above 1S. Regulation is quite similar to the one provided by a .85ohm internal resistance voltage source.
I can obtain a slightly more linear regulation in high frequency with a small (about 470nf) in // with the reg. It also helps to eliminate the overshot one can see with square waves (I think that Richard published them a while back). Funny, John says “no cap” and he must have good reasons.
A “massive” approach to passive filtering, 3 cells composed by a 10ohm resistor and a 47000µf 1nH ESL and 3mohm ESR cap (I hope these values are realistic), provides a much higher attenuation in most of the audio frequencies and above, but this attenuation is changing with frequency and there is a severe bump at very high frequency (due to very low esr ?). I know that only the last cell is useful in term of regulation.
The active reg shines in terms of line rejection. I probably obtain one more time meaningless results because real live noise masks the simulated behaviour that lies at impossibly low levels. Anyhow, I think that the curves are interesting. For instance, you have a less acute peak in HF using 3 different caps in the passive way and you can see the interest of multiple cells.
Any help and comment gratefully welcomed.
Philippe (who as difficulties to explain himself in English and whose mail is broken :-/).
EUVL said:
As I said in an earlier post, each time I have tried a current source I came back to a resistor, I really don’t know why, but of course replacing a CCS by a resistor implies generally a “perfect” power supply. If you replace R5 with a few leds you need also a resistor in series with the Mosfet gate to avoid peaking at high frequencies, moreover sound get worse, but i can't totally explain this too, maybe someone here?
In resume:
1) bad PSU, CCS is better.
2) good PSU, resistor is better
2 is better than 1
Patrick did you tried something like this in your output stage:
An externally hosted image should be here but it was not working when we last tested it.
anybody read through the latest stereophile yet?
the ayre k5xe preamp is discussed a bit with more tech tidbits than i've usually seen elsewhere on the ayre circuitry.
sounds a bit like what we're talking here.
no good pix, though. :-(
maybe charles might be encouraged to say a bit more ... ?
mlloyd1
(who is REALLY digging this discussion - my favorite thread in quite some time! seriously tempted to send my wife on an out of town shopping trip so i can have time to play)
the ayre k5xe preamp is discussed a bit with more tech tidbits than i've usually seen elsewhere on the ayre circuitry.
sounds a bit like what we're talking here.
no good pix, though. :-(
maybe charles might be encouraged to say a bit more ... ?
mlloyd1
(who is REALLY digging this discussion - my favorite thread in quite some time! seriously tempted to send my wife on an out of town shopping trip so i can have time to play)
> Patrick did you tried something like this in your output stage.
Richard,
Have not tried (still trying to get Toshiba MOSFETs), but this is actually very similar to the original output stage of Borbely, which is based on Hitachi MOSFETs. (Since it is his circuit, I am not going to post it here. Please check his website.)
This is the easy and obvious solution. But there is good reason why I would want to bias the MOSFET independently from Stage 2. (Check you private email.) And I think an independent unity-gain MOSFET follower circuit can be put to good use elsewhere. Just see how many people use BUF634 for audio.
So I hope someone can still come up with a better solution than my post #779, which I do not find elegant enough.
Patrick
Richard,
Have not tried (still trying to get Toshiba MOSFETs), but this is actually very similar to the original output stage of Borbely, which is based on Hitachi MOSFETs. (Since it is his circuit, I am not going to post it here. Please check his website.)
This is the easy and obvious solution. But there is good reason why I would want to bias the MOSFET independently from Stage 2. (Check you private email.) And I think an independent unity-gain MOSFET follower circuit can be put to good use elsewhere. Just see how many people use BUF634 for audio.
So I hope someone can still come up with a better solution than my post #779, which I do not find elegant enough.
Patrick
Well - I have read the article and it has not surprised me. There was a thread here called ""What's your reasoning?" and not "What's your belief?". " by mikeks, see
http://www.diyaudio.com/forums/showthread.php?threadid=41770&perpage=10&pagenumber=1
The article is about belief and not the reasoning. Different we are and different believes we may have ...
http://www.diyaudio.com/forums/showthread.php?threadid=41770&perpage=10&pagenumber=1
The article is about belief and not the reasoning. Different we are and different believes we may have ...
A factor that many technical people ignore is the 'reality' of making a good sounding circuit. For the most part, just about any IC or discrete circuit should sound the same. There really should not be any significant difference between them.
I might make a slight exception with tubes, since they usually have a significant amount of distortion (usually lower order) that is difficult to remove completely. Just measure a Dyna tube amp some time, if you don't believe me. However, even a Dyna, properly maintained, and maybe 'improved' with a few quality parts, can really sound good, even on a $5,000 pair of loudspeakers. I have tried it, myself.
What audio designers, over the decades, have tried to do, is to understand WHY equipment that measures almost exactly the same in frequency response, distortion level, and has the same absolute polarity, can still sound different, except when compared in a double blind ABX listening test.
Many of the 'skeptics' just think we are imagining things, but then why do we hear differences, when we don't want to hear differences? Why doesn't our design ALWAYS sound better? I think there is something wrong with the ABX test, instead. Your opinion may vary!
I might make a slight exception with tubes, since they usually have a significant amount of distortion (usually lower order) that is difficult to remove completely. Just measure a Dyna tube amp some time, if you don't believe me. However, even a Dyna, properly maintained, and maybe 'improved' with a few quality parts, can really sound good, even on a $5,000 pair of loudspeakers. I have tried it, myself.
What audio designers, over the decades, have tried to do, is to understand WHY equipment that measures almost exactly the same in frequency response, distortion level, and has the same absolute polarity, can still sound different, except when compared in a double blind ABX listening test.
Many of the 'skeptics' just think we are imagining things, but then why do we hear differences, when we don't want to hear differences? Why doesn't our design ALWAYS sound better? I think there is something wrong with the ABX test, instead. Your opinion may vary!
I don't think we measure the right things, we measure the deformation of a sine wave in steady state and call that distortion, but not what happens during transients or in multi tones. I belive that our sound sensors (ears) are extremly sensitive. I know for certain that 2.5 KHz reproduced by a midrange (that extends to more the 10 KHz) sounds totally differnt than the same tone produced by my Raidho ribbon tweeter. Point is its not worth spending too much time wondering why things sound differnt, accept this as a fact and then concentrate on listening and tweaking the results. There's no such thing as simulating to better results as you tend to focus on distortion, By listening before measurement you rid yourself of any expectations and thus free yourself from the low distortion hunt. Or in my case for speaker liniarity hunt.
Michael
Michael
There is nothing like 2 power amplifiers, 2 preamplifiers or 2 opamps that do measure same. In case you make a set of measurements (characteristics and responses, not numbers) like freqchars, stepresponses, frequency analysis, IM characteristics, multitones, noises etc., they ALWAYS do measure different. The case is that we are unable to make right correlation and blame it on "same numbers do not mean same sound". There is nothing like "same numbers".
I do use multitones, frequency modulation and other different non-traditional methods. Do I know more? Not much. But I am very sure that high distortion (like some non-feedback designs) matters. It matters sonically, making nonsense mismatch when listening philharmonic orchestra. Just listening, not measuring. That's why I do not believe the fairy tales like Hansen's in Stereophile 5/2006, p. 19.
I do use multitones, frequency modulation and other different non-traditional methods. Do I know more? Not much. But I am very sure that high distortion (like some non-feedback designs) matters. It matters sonically, making nonsense mismatch when listening philharmonic orchestra. Just listening, not measuring. That's why I do not believe the fairy tales like Hansen's in Stereophile 5/2006, p. 19.
The ear is very sensitive, I often compare music to fractals, if you miss resolution (low level resolution) you don’t hear all, moreover, have you noticed that with very good audio material the sound seems to slow down, why? Because the ear and the brain take more time to analyse the music...
- Status
- Not open for further replies.