diyAudio (
-   Digital Line Level (
-   -   DAC: CS4398 with CS8416+CM102s (

SoNic_real_one 22nd April 2012 05:56 PM

DAC: CS4398 with CS8416+CM102s
4 Attachment(s)
As title says I did buy this eBay DAC board and associated enclosure from along1986090. Used my transformers.
CS4398 with CS8416+CM102s, provides coaxial SPDIF and USB inputs.
First impressions:
Enclosure definitelly well done, with (almost) all necessary hardware, holes in exact locations to match the board. Kind of long (10", plus 1/4" front aluminium plate), I guess it can be used to add a dedicated headphones amp.
I did buy the pre-solderd board. Inspected with the magnifier glass showed an OK grade of soldering. PCB is in filled ground technique and parts are like shown in the picture.

Initial tests had a sligtly distorted signal and analized with my E-MU1820m showed a really bad signal. I did replace the 5532 that was on board with LM4562 that I hade around and everything started to look better. I suspect that the 5532 was either a fake or a damaged exemplar. There is no capacitor or resitors on OpAmp outputs, so it is easy to be damaged.
At -2dB looks decent. The output stage overloads my ADC input at higher than -2dB (all the other DAC's that I tested didn't manage to do that).
The output stage is not done per recomended schematics in the datasheet.

USB input:
CM102 locks only onto 16bit/48kHz signals. 16bit/44.1kHz is not supported in Windows7 mixer (WASAPI).
If looses the connection with the PC, static noise is outputted till the next power off/on cycle. The automatic switching could be done better I guess...
Altough in Win 7 there is no need for drivers, here is a package that provides a little more that basic USB. This driver looks like it does the resampling to 48kHz automatically.

Locks on 44.1-96kHz, 16 or 24 bit. I don't have a SPDIF source capable of 192kHz to fully test the CS8416.

Using Foobar with resampling to 48kHz would allow USB connection to work at any rate, but I do prefer the coax.

All in all, sounds decent, especially for the price. THD+N with LM4562 is at -94dB. OpAmp upgrade is strongly recomended and looks like some of the AC grid noise could be reduced by better filtering.
Direct coupling of the outputs leaves some higher levels of 10Hz noise creep into the output chain as seen in the coax graph (at USB I forgot to extend the range below 20Hz).

The seller give the schematics (at my request).

wwenze 23rd April 2012 04:57 AM

Hmm...? I'll try 176.4 and 192kHz on this when I get home.

How's the stereo crosstalk on yours?

SoNic_real_one 23rd April 2012 10:54 AM

I don't have measurements saved, but it was OK with the LM4562. For some reason, the original 5532 sucked bad. I still have it in a bin :) I will do some measurements latter.
I don't know why, CS8416 in my board (in hardware mode) doesn't go over 96kHz. Maybe is the same as for the WM8804 receiver.

wwenze 23rd April 2012 12:55 PM

1 Attachment(s)
192kHz from my PC's onboard Realtek sound works.

Realtek doesn't support 176.4kHz, but I tried Musiland Monitor 01 USD's 16/176.4 and that works too.

SoNic_real_one 23rd April 2012 10:17 PM

Your graph is weird, looks like noise... I will try again, with another source.
Be carefull, if you use a player that uses DirectSound to send the audio data, DS will resample the datarate to match what your device "reports" that is capable of or what is set in Windows mixer. I am using Foobar because I can choose exactly what path to use for data.

TheGimp 25th April 2012 01:19 PM

1 Attachment(s)

I have one of these due any day now (reached NY yesterday).

One thing that stood out to me was the original graphs shown by the seller had the frequency response going past 40KHz at -3dB with the sample rates at 44.1KHz and 48KHz.

I thought that was not possible according to Niquest-Shannon sampling theorem.

Certainly the LP filter could allow it and the plots for 96Khz and 192KHz would be accurate in that region.

But would the 44.1 and 48KHz plots?

I bought just the assembled DAC board.

SoNic_real_one 25th April 2012 10:34 PM

I think those graphs are wrong. Plus they are useless to me, I am more interested in distortion level.
That's why I did my own measurements.

TheGimp 26th April 2012 03:19 AM

I figure if one is off that much, the rest are garbage.

wwenze 27th April 2012 05:06 AM

Well the 4 curves do at least look different so this might not be impossible...

TheGimp 27th April 2012 12:54 PM

Near 100 years of sampling theorem says it is impossible.

You have to sample twice the highest frequency of interest to reconstruct it without massive distortion.

At 44.1KHz, the highest frequency that can be passed undistorted is 22.05KHz (minimal distortion actually because one can not guarantee synchronization between the frequency of interest and the sample clock). In reality 22.05KHz will be distorted as well so normal procedure is to filter below this.

There is a good explanation here:

Nyquist?Shannon sampling theorem - Wikipedia, the free encyclopedia

All times are GMT. The time now is 01:11 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio