WM8805 upgrade board (cs8414 pins) - dissapointed

[jean-paul]

That is clear now, at first I got the impression that you wanted extra functionality.

[AudioLapDance]

The naked bypass caps were included because they were "audiophile" but they are a terrible design choice for digital bypass. They are too physically large and too far away to do any serious high freq noise shunting.

As far as I know, they aren't good for coupling the digital signals either so eventually you should remove them all from critical locations.

I have this board too and its gonna be hard to fix it.

For power to pin 1, you should move the OScons (can anyone verify that they're real?) to where the naked bypasses are (drill a new hole for the gnd led and scrape off the paint to solder to gnd plane) and then put small sm caps as close as possible to the 8805 (scrape off paint on pwr trace so sm cap is touching power pin and scrape gnd side to solder to gnd plane)

Good luck and keep taking pics!

Cheers,
Jeff

[AudioLapDance]

Ooops, too late to edit:

That should be:

(drill a new hole for the gnd leg and scrape off the paint to solder to gnd plane)

[KlipschKid]

Hi,

Thanks - that's something I overlooked. Pin 1 is dvdd so that needs local decoupling with 0.1uf x7r, easy enough, and I think pin 11 is the supply for the pll - pvdd - so that needs the same, but isn't so convenient.

I have no idea if the os-cons are genuine, so I just tested one that I removed before. My DMM and it says it is 29uF. The sheath says 22uF but my DMM is very unreliable with large cap values so it could be correct. However, I have some 10uF 35V Rubycon ZA in my tray so I'll try them instead (0.33 ohm ESR @ 100Khz, and 350mA ripple).

I haven't been able to find an upgrade for the quad line receiver - most have very similar specs it seems.

Thanks,

Tom

[KlipschKid]

A bit too much wine = a bit too much solder !



They have followed the datasheet (page 64) and put a 1uF cap on pin 11 but it isn't that close; C* just out of the photo to the lower right.

[AudioLapDance]

Hey KlipschKid,

Now your speaking my language!
One of my mottos is the bastardized version of "measure once, drink twice, cut three times!"

[KlipschKid]

haha... yeah... I'll test the connections again tomorrow ;-)

[AudioLapDance]

Depending on what type of inputs you're using, you could just bypass the MUX. The 8805 only needs very simple digital input circuits (check the datasheet) and the traces are pretty easy to tap.

[KlipschKid]

Hi,

I'd like to keep it, but find a better part.... but no chance it seems. Hey ho.

Here's a photo of the top side. I scraped off some of the screen on the edges of the gnd so I could solder them together and create a more contiguous gnd plane. Hopefully, it will help with noise levels a little. It looks a mess and needs another clean but the connections are all good and it should work better than before. You might say it is an ugly master chef ?!

It's also been suggested I try to float/suspend the xo to avoid vibration. I could suspend the whole board when this gets boxed up...but it needs chassis mounting... hmmm... need to think about how to do that.... maybe just some foam under the xo and replace the solid wire with flexible stuff.

Before :

After :

Effectively :

[linuxfan]

Originally Posted by KlipschKid

How about this for 22.5792Mhz xo ?
R = 98.304 / 11.2896 (prescale =1 so 22.5792 becomes 11.2896 for f1) = 8.7074829932
PLL_N = 8
PLL_K = 4194304 x 0.7075 = 2D4766 (hex)

I confess I'm confused about how the WM8805 divides the master clock, and in particular whether any ONE clock speed can be optimal for different samplerate-families.
Please consider the scenario of a WM8805, under software control, sharing a common external clock with an ES9023 DAC -
can someone please tell me:

i) what would be the optimal XO for 44.1/88.2 material?
ii) what would be the optimal XO for 48/96 material?

As I understand it, the WM8805 in software mode will accept external clock freq up to 27MHz,
and the ES9023 will accept external clock freq up to 50MHz.

Thanks.