What cap for the CS8412 loopfilter?

[Elso Kwak]

Re: Re: Lm317/337

and still a 317 based thing is noisy compared to a decent discrete reg

Right! And that's why I won't use the LM317 as a preregulator, even if it is in a Wenzel Finesse circuit. Once the noise gets in your sytem it is very difficult to get rid off. Same for the follower.
Werner Ogiers made some very interesting measurements:
http://www.tnt-audio.com/clinica/regulators_noise1_e.html

BTW, who was it claiming that LEDS are too noisy.....?
http://www.diyaudio.com/forums/showthread.php?postid=336199#post336199

See also Post# 14 in this thread
Even more confusing: I did not find a direct correlation between noise of the regulator for a clockcircuit and soundquality.... The same applies for a preamp.......

 

[borges]

*46.2R would be the ideal value, according to my back of the envelope calculation that
Capacitor value C = (1000/R)^2 x 47
hence
Resistor = 1000 / square root of (C/47)
Cap values in nF, BTW.

Kit, can you enlighten me with the source of this equation?

Greetings,
Børge

 

[till]

my datasheet tells me 1k Ohm and 47nF from FILT to AGND. WHY are different values better???

 

[kittaylor]

Borges

The convention is that every time the resistor value is halved the capacitor value is quadrupled.

The eqautions to match cap and resistor values follow this rule, with the datasheet value of 1000R and 47nF as a starting point. Just a back of an envelope thing, so it might be cobblers

till

The values change the jitter attenuation of the CS8412, which I'm told is almost zero with the dataheet values.

What I want to know is if more is always better. Does the cap value eventually become too high, or the resistor value too low, and cause more jitter?

I've ben fiddling about and the 220uF Rubycon ZA is better than a 4.7uF X7R 1206. It's not night and day, but
pirch definition and timing seems superior and the music more relaxed.

 

[Guido Tent]

Borges

The convention is that every time the resistor value is halved the capacitor value is quadrupled.

The eqautions to match cap and resistor values follow this rule, with the datasheet value of 1000R and 47nF as a starting point. Just a back of an envelope thing, so it might be cobblers

till

The values change the jitter attenuation of the CS8412, which I'm told is almost zero with the dataheet values.

What I want to know is if more is always better. Does the cap value eventually become too high, or the resistor value too low, and cause more jitter?

I've ben fiddling about and the 220uF Rubycon ZA is better than a 4.7uF X7R 1206. It's not night and day, but
pirch definition and timing seems superior and the music more relaxed.

Hi

Lower PLL crossover = better as it reduces incomming jitter from the drive.

At a certain moment you run into limitations of sub LF supply noise en LF jitter properties of the Crystal receiver clock: Time for a secondary PLL using a crystal...............

look here for a practical example:

http://www.tentlabs.com/Products/DIYDAC/DIYDAC.html

cheers

 

[kittaylor]

To my surprise, the DAC works and works well with a 330u 16v SEP cap. It does take a couple of seconds to lock on and you get a blast of white noise first.

The tone seems silkier, but the sound overall is chunky and dynamic, with an alarming amount of energy in the low end.

The lock on delay is longer than with the ZA 220u, and that cap didn't make a burst of noise just a click, so I asume the bigger cap is providing more filtering.

 

[Jocko Homo]

One thing that no one has mentioned......

Changing RC values on these PLLS may lower the corner frequency, but usually do at the expense of peaking in the response. They do not lower the actual jitter any, just shift the frequency repsonse around a little.

Jocko

 

[kittaylor]

There's probably some truth in that.

I've upped the resistor value to 330R, and the sound is drier, cleaner and less massive. Different, not better.

 

[Peter Daniel]

My preferred choice is 0.47 BG N and 330R Caddock. I also tried 4.7 BG N with appropriate resistor value but it sounded too "thick" and not "fast" enough.