john curl said:
You are entirely correct, and indeed, HPS3.x has an equivalent of 0.25-0.3nV/rtHz with sorted (for noise) JFETs and 0.3-0.35nV/rtHz with regular, unsorted devices. Which is, as you said, exactly about 5 and respectively 7ohm equivalent noise. Thanks for confirming this, I was not aware of the SCP-2 transconductance.
john curl said:
Absolutely agreed on this one as well, no need to go back the Yishay et. al. references, and that's exactly what I said myself earlier:
http://www.diyaudio.com/forums/showthread.php?postid=1720582#post1720582
I'm also looking into various ways to use bipolars; the main thing is to find a reliable way to cancel the input current. And no, the Leach and other similar common base input stages are not the right answer for this. But that's already another project...
MRupp said:
No pictures of THD, oh well. I think it's one of the great hoaxes of audio. People talk about line-contact stilii but I can't find a single photo of one improving the (relatively bad) distortion on an LP.
Years ago I bought a Monster Alpha 1, at the time very highly rated and had a square cut stylus with a very narrow groove contact. I also had a full set of RCA STR test records (this set BTW can do all the alignment and measurements you would ever want). No difference from my Grado Signature 8.
You can do the numbers for some very well liked contemporary carts, they simply can't track the grooves even if they were there.
Yes Scott, the test records are flawed. However, the phono cartridges all have a unique sound, and the expensive ones tend to sound the best. I wish it were not true, but I have found it to be so, up to this day. I have never been able to measure much difference between phono cartridges that should make the distinction between them so obvious. But then, there is little measured difference between my Porsche 944 and my Acura, either, but they are rather different vehicles on the road.
Syn08, I am glad that we finally agree on this. The applied math is there for all to see, and it is what engineers do best.
Now, as to performance of real world fets:
You may have lucked out, with your batch of 200. However, I have measured 1000's of Toshiba fets for noise and I have not been as lucky. I doubt that Charles Hansen, who bought 500,000 2SJ74's recently, could select out 10 sets of components that are virtually perfect.
Now, as to performance of real world fets:
You may have lucked out, with your batch of 200. However, I have measured 1000's of Toshiba fets for noise and I have not been as lucky. I doubt that Charles Hansen, who bought 500,000 2SJ74's recently, could select out 10 sets of components that are virtually perfect.
You hire someone to do it. They can work at a reduced wage. They could test perhaps 1000 per day and it might take 2 years. It is that difficult a task to make 10 of what Syn08 is predicting, in my opinion. The variable is mostly in the construction of the jfets. It can be shown that once in a while, a fet with very very low 1/f noise can be produced. Usually, below a few hundred Hz, there is added significant 1/f noise that makes meeting an ideal spec virtually impossible.
For PMA
I cooked up this little demo of the crossover distortion removal (797) applied to a fairly ordinary amplifier architecture, the only difference being the differential ”chasing its own tail” VAS. BTW with some work this would make a nice little discrete op-amp even without the error correction.
Because this is a two stage amp, the trick involves a unity gain buffer in addition to the cap. The simulation results show three plots. The amplifier is in a gain of 100 with +-100mA at 10kHz being shoved into the output. The orange plot is with G = 0 on the buffer (little triangle in schematic) and shows the glitch due to the crossover (PMA, sort of looks like your “other guys” plot?). The white plot is with G = -1 showing virtually perfect cancellation. The green plot is with G = +1, horrors positive feedback? No because it’s not feedback, the error is just doubled. In fact if you break the loop at “B” you will see that there is no loop gain (in fact huge attenuation). The little triangle is indeed just a gain of -1 buffer and this is just subtractive neutralization. In all three sims the displacement current in the comp cap (C12) contains the same distortion signal, but at G = -1 it is cancelled exactly by subtraction.
I cooked up this little demo of the crossover distortion removal (797) applied to a fairly ordinary amplifier architecture, the only difference being the differential ”chasing its own tail” VAS. BTW with some work this would make a nice little discrete op-amp even without the error correction.
Because this is a two stage amp, the trick involves a unity gain buffer in addition to the cap. The simulation results show three plots. The amplifier is in a gain of 100 with +-100mA at 10kHz being shoved into the output. The orange plot is with G = 0 on the buffer (little triangle in schematic) and shows the glitch due to the crossover (PMA, sort of looks like your “other guys” plot?). The white plot is with G = -1 showing virtually perfect cancellation. The green plot is with G = +1, horrors positive feedback? No because it’s not feedback, the error is just doubled. In fact if you break the loop at “B” you will see that there is no loop gain (in fact huge attenuation). The little triangle is indeed just a gain of -1 buffer and this is just subtractive neutralization. In all three sims the displacement current in the comp cap (C12) contains the same distortion signal, but at G = -1 it is cancelled exactly by subtraction.
Attachments
I challenge anyone to find an analog master tape that has less the 0.3% distortion at operating level or O Vu.
To clarify things further, many of the original test records were made in the stone ages of the 50's, and early '60's, with inferior disc cutting systems. You can verify this by comparing different phono cartridges with the same test record. Too much similarity. The cartridges sound different, however.
To clarify things further, many of the original test records were made in the stone ages of the 50's, and early '60's, with inferior disc cutting systems. You can verify this by comparing different phono cartridges with the same test record. Too much similarity. The cartridges sound different, however.
john curl said:
scott wurcer said:
http://www.diyaudio.com/forums/showthread.php?postid=1722023#post1722023
scott wurcer said:
Probably a safe bet based on the dimensions. Even a "line contact" stylus has a radius that is much larger than the curvature of the of the groove at high frequencies. You will get a distorted reproduction at that ratio. The line contact stuff was created to get more of the subcarrier on the CD4 process. And the RIAA curve helps reduce the THD with its 6 dB per octave rolloff giving better numbers.
Reference level on a tape can be pretty low distortion if you pick the right reference level, eq curve (IEC) and very high output tape. Quantegy Platinum is rated at .06% 3rd at reference level (not clear if thats 250nW or higher).
Demian, I must agree that the spec sheets show improvements in analog magnetic tape of more than I was aware of. It doesn't make sense to me, but I applaud progress.
However the distortion on tape is always fairly high, and the same is true with phono cartridges.
The only redeeming virtue of either system is that the distortion falls in a reasonably predictable way with lower levels, so that LOW LEVEL information is reproduced with great fidelity. I can't say the same for digital.
However the distortion on tape is always fairly high, and the same is true with phono cartridges.
The only redeeming virtue of either system is that the distortion falls in a reasonably predictable way with lower levels, so that LOW LEVEL information is reproduced with great fidelity. I can't say the same for digital.
- Status
- Not open for further replies.