Either I don't understand what you mean or your noise calculation is incorrect. For normal analogue circuit noise (as opposed to downconverted out-of-band quantization noise), the signal to noise ratio must be best straight at the DAC output with the filter board disconnected, as the filter doesn't have a negative noise figure (give or take 3.01 dB because some look at the differential and others at the single-ended signal).
Besides, 66 mV RMS at the 8.2 nF is too small when the filter board is removed.
Marcel,
I checked again and I made an error, this 66mV was indeed with the filter still attached.
Without filter the voltage is 424mVrms.
That makes the S/N 121.9dB.
Hans
Last edited:
Marcel,
Looking at available metal foil resistors and cost, it may be that the nominally 3k resistors you specified for the dac may more costly than some other resistance values. Do you have any comments and or recommendations for a range of resistance values that might still perform adequately well?
Thanks,
Mark
Looking at available metal foil resistors and cost, it may be that the nominally 3k resistors you specified for the dac may more costly than some other resistance values. Do you have any comments and or recommendations for a range of resistance values that might still perform adequately well?
Thanks,
Mark
The measurements done so far were as you know with the Amanero, not giving the best possible results.
I would really need a JLsounds USB converter that internally isolates from the computer.
Nautiboy offered to make a complete set available once he had new set of boards assembled.
Measurements can then be done within a week.
Hans
Comment: I hear what my friend above hears, and he hears what I hear. That has been confirmed. So what's going on? IMHO it has to do with preference. Restored to its original state as it is now, the dac has space, soundstage width and depth. But the sound is grungy and grating. Like sandpaper on my ears. I don't like it.
OTOH, the way I had the dac before reversion to original was very precise and well on its way to being very clean and dynamic sounding. But the illusion of space was much diminished as compared to its present state. I like that clean sound better. So do some other people I know who have heard it. To my friend in the quote above, the sound I like is inferior. The space has collapsed and is no longer good sounding.
Ideally we could find a way to satisfy both of us, but its not yet clear to me if that's possible. Part of the answer has to depend on the original recording and how it was digitized. If there were problems there then there may be no way to make reproduction sound a way that is satisfying to everyone.
Could be some kind of psychoacoustic effects processing could create an illusion that recordings are all around better than they really are, don't know. Do know we are trying to satisfy human perception for entertainment purposes though.
The above thoughts are things I have been mulling over for awhile but have not talked about. Still trying to see if I can learn more about what's actually going in terms of what is on a recording and how to best reproduce it for human consumption. Still a lot to learn, seems to me anyway.
3 kohm +/- 0.1 % is within the 3.01 kohm +/- 0.5 % range I specified, so that should be no problem at all.
With the original filter, 3.3 kohm or 2.7 kohm will result in a level change of just below 1 dB and a frequency response deviation of about 0.3 dB somewhere far in the ultrasonic region (80 kHz or so), that also doesn't seem much of a problem to me.
With Mark's filter, 3.3 kohm or 2.7 kohm will shift the corner frequency by about 10 %.
Last edited:
To continue Hans' noise calculations the thermal noise (Johnson noise) of 8 parallel 3k resistors is about 350nV from 20-20kHz. Low noise thin films (and foils) typically have a noise index of -40dB (10nV/V per decade). So the current noise of those resistors over the same BW is about 90nV. The total resistor noise is then about 360nV which is almost all from Johnson noise. So technically expensive ultra-low-noise foil resistors cannot be justified.
Last edited:
Acko,Hans, I have a JLSounds VIII, unused. Happy to send this to you if interested and yours to keep. Just let me know
Very kind of you, really appreciated.
The point is however that I need a whole set for a week, because I also don’t have the Dac + Filter boards, with still no plans to build them for myself.
Jesper kindly offered me to send his set for testing.
But he uses an external SOTA Andrea clock for his JLSounds board.
As perfect as his set will be, it can’t play DSD512 because of that clock, a rate that also has to be tested, so to my regret I had to pass up his offer.
Hans
Good that you finally admit that you don't posess any out-worldly given listening ability. If you take full insight into this new revelation of yours, I think it would be prudent to "pipe down" a bit going forward and don't be so self-confident and "uber" about everything you write.
//
No problems, understood
Nautibuoy has given us a very good reference with his listening experience and ideally we should measure a similar setup. Hopefully, you will get all that is needed for your tests and many thanks for your efforts!
Mark,
I still had another thought.
When using Marcel’s balanced output, the interlink is driven by 100R (2x49,9), properly matching the cable impedance.
However in your Transformer solution the 8.2nF is a short circuit at HF and you transformer is a open connection at HF.
With the high frequencies involved here, an ideal situation for reflections going up and down, possibly affecting the perceived sound reproduction.
To prevent that from happening a 100R termination in series with a proper cap value over the transformer would terminate the interlink with 100R at HF.
Hans
I still had another thought.
When using Marcel’s balanced output, the interlink is driven by 100R (2x49,9), properly matching the cable impedance.
However in your Transformer solution the 8.2nF is a short circuit at HF and you transformer is a open connection at HF.
With the high frequencies involved here, an ideal situation for reflections going up and down, possibly affecting the perceived sound reproduction.
To prevent that from happening a 100R termination in series with a proper cap value over the transformer would terminate the interlink with 100R at HF.
Hans
Hi Hans,
I2SoverUSB uses its own NDK SDA 45/49MHz clocks. If used with external clocks, they should be 45/49MHz as well.
Also, BCLK for NRZ DSD512 is 22/24MHz, depending. Should be not be a problem playing DSD512.
If there is a problem it would be that playing DSD512 requires the use of Native DSD and therefore ASIO drivers. It may be that JL Sounds has a problem with that.
Mark
I have repeatedly denied being a 'golden ear.' Its never been about anything like that.
A quick search for posts by me containing the term 'golden ear' turned up at least 9 times I have tried to make that clear. I have also tried to explain there is such a thing listening for discrimination instead of preference.
Seems that whatever I write, someone will find a way to misinterpret it.
A quick search for posts by you in this thread alone returns tens of post with claims of better sound.