What cap for the CS8412 loopfilter?

[kittaylor]

The recommended values seem to be 470R/220nF. I'm having good results with 75R/10uF, which seems to give a fuller and more impactful sound on my clone of Peter Daniels 3D dac.

I've tried caps on pin 20 from 10nf to 4.7uf. THey all make the sound lusher, but there doesn't seem to be anything between them.

I'm using a 1206 SMD resistor, assuming that shorter is better, and 10uf 35v Rubycon ZA. Is this an ideal cap to use, in terms of properties rather than capacitance, and do you think the dac would still work if the filter moved it up to 50R/20uF?

TIA

[Chris Daly]

Kittaylor
Have you a schematic for CS8412 ? Crystal may have a
application sheet download. You also need to define
from a schematic what you are doing

Resistors and Capacitors have many functions
when combining together such as frequency rolloff
frequency emphasis. You need to understand
the manufacturers intent. Indeed 50r and 20uf might
be better but why, how are you influencing the circuit.

Electronics is benefitted by experimentation but
only when the terms or boundarys are known.

Is this making sense ?

Cheers / Chris

[Guido Tent]

Hi

When using 75 ohm, I'd suspect you need about 8 uF, so 10 uF will do

I prefer a film cap like the Wima over any lityc. The 10uF Wima MKS2 is available in 5mm pitch, 16V working voltage

http://www.reichelt.de/inhalt.html?S...OVID=0;TITEL=0

I hope the link will still work

Even better is a cacscade of a second PLL containing a low jitter VCXO, following the Crystal

succes

[Guido Tent]

Originally posted by Chris Daly
Kittaylor
Have you a schematic for CS8412 ? Crystal may have a
application sheet download. You also need to define
from a schematic what you are doing

Resistors and Capacitors have many functions
when combining together such as frequency rolloff
frequency emphasis. You need to understand
the manufacturers intent. Indeed 50r and 20uf might
be better but why, how are you influencing the circuit.

Electronics is benefitted by experimentation but
only when the terms or boundarys are known.

Is this making sense ?

Cheers / Chris

Chris

Its seems kit knows what he is doing, given the way he scales the parts, he only asked for parts reccomendation

Ofcourse lowering the PLL corner is a good thing, since the jitter attenuation increases.

The manufacturers intent is to sell as much as possible silicon, they do not care for good sound - ever considered why they put the filt pin (20) next to the MCK output (20) pin ?

I guess they don't know how the crosstalk affects the PLL jitter, would they ?

That is dealt with in a great eal by adding some 10nF directly to ground. Ofcourse this advise is not in the datasheet.................

[kittaylor]

Some findings that might be of interest.

I've built a new dac with 10nF NPO 1206 on across pin 19 & 20.

The first loopfilter was 100R 1206 and 4.7uf Ruby ZA. Swapping the cap for a 22uF Oscon SP gave more of the same, more forward, bigger and tonally thicker and more rounded. I think it takes a split second longer to lock on, I'm not sure.

Swapping the resistor for 43R* made quite a difference. Fatter, cleaner, more liquid and greater clarity in terms of timing and separation.

Downsides? It sounds a bit too well organised, I tend to go for more of a thunderous wall of sound effect, and bit harder in the upper mid.

*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.

[fastcat95]

Guido:

Can you elaborate on the use of the 10 nF capacitor as
to where and how it is used with the CS8412 chip?

Also, what is the current thinking about the filter
component values and configuration for best
performance with the CS8412 reciever? I also
seem to remember that you once said that
seperating out the analog and digital grounds
for this chip was an improvement, right?

Fastcat

[Guido Tent]

Originally posted by fastcat95
Guido:

Can you elaborate on the use of the 10 nF capacitor as
to where and how it is used with the CS8412 chip?

Also, what is the current thinking about the filter
component values and configuration for best
performance with the CS8412 reciever? I also
seem to remember that you once said that
seperating out the analog and digital grounds
for this chip was an improvement, right?

Fastcat

Hi

The cap reduces crosstalk from pin 19 to pin 20. It should be from pin 20 to pin 21 (Agnd)

See also

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

and

http://www.tentlabs.com/Info/Article...decoupling.pdf

on layout issues

NEVER sperate groundplanes (regardless of what others or industry say)

What happens when you seperate them and why would it improve ?

Pay attention to quality of analog supply (LOW noise)

cheers

[kittaylor]

Surprisingly the dac still works with 43R followed by a 220uF Rubycon ZA, it just takes a second to lock on after the transport engages.

An improvement I feel, maaaaking sure that the rest of the system is setup and tweaked to match. Quite a subtle change, there's not really any more of anything but the presentation feels more relaxed.

[Guido Tent]

Originally posted by kittaylor
Surprisingly the dac still works with 43R followed by a 220uF Rubycon ZA, it just takes a second to lock on after the transport engages.

An improvement I feel, maaaaking sure that the rest of the system is setup and tweaked to match. Quite a subtle change, there's not really any more of anything but the presentation feels more relaxed.

Hi Kit

You lowered the PLL lowpass frequency. Next step could be to lower the sub LF noise on the analog supply pin of the 8412

enjoy

[kittaylor]

That's what I'd hoped to do Guido. Does there come a point where the cap value is too high for the CS8412 to function at all, or point where it causes more jitter than it cures?

As for the CS8412 analog supply I'm getting good results with an LM317AT with 4.7uF X7R 1206 on the ADJ pin, two 56uH Epcos mini chokes in series and another 4.7uF across pins 22 & 21.

Any improvement suggestions that aren't too radical?

TIA