AK4479 or AK4499 DAC Kit????????

chinook9

Member
2013-08-15 4:00 pm
My home system is Parasound A21 driving Thiel 3.6. These match well but I do not like my Buffalo IIISE with them (no preamp). The Sabre chip just seems too harsh.

I have a DIYINHK AK4497 DAC that I built and it sounds good but it is not easy to get it to work with the DIYINHK XMOS USB to I2s converter and my Android tablet. I run TIDAL off the tablet.

I am looking for an AK4497 or AK4499 dac I would be happy with either of the following:

1. A complete kit that can be built without extensive audio electronics knowledge. I can read instructions and I can solder. I like kits where a build has been documented online with many pictures.

2. A good ebay or alibaba chinese board that has been proven as a quality product. The following may be one of those: Dual AK4497 AK4497EQ DAC DSD HiFi Decoder audio +Bluetooth 5.0+Amanero USB NEW | eBay or 2019 Dual AK4497EQ DAC AK4118 AK4497 DSD256 Native DAC 32Bit 384KHz USB AK4497 DAC,XMOS XU208 Amanero USB Bluetooth 5.0 APTX HD-in Amplifier from Consumer Electronics on AliExpress


Any help would be appreciated.
 

PJN

Member
2004-12-31 1:27 am
Pa, USA
Maybe the problem is with the XMOS board implementation. Perhaps a better implementation .might give you what you're looking for. JLSounds makes a isolated asynchronous xmos board that sounds very nice. I've used them in two builds and have been very satisfied, easy bettered other xmos boards that I've used.
 
No AK4499 kits as of yet, although evaluation boards are available (but need some know-how to implement).

IME, all Chinese boards are built with cheap parts often disguised to look like quality parts. They sound like cheap parts too. With a lot of work replacing parts they can be made to pretty good. It can take a pretty fair amount of time and more money spent on it. At least AK4497 boards tend to be simpler than Sabre boards.

Regarding your B-III, don't know that its the best sounding example of what can be done with Sabre parts. Never heard one myself, but did buy and try out some related TP boards that didn't end up being 1st choice for some of my projects.
 
Markw4 do you have have any opinion on the following board.

Sorry, no. Only AKM dac I have tried is AK4499. One thing I would say is that since AKM dacs get MCLK from the USB baord it is important to use a USB board with very low-jitter clocks. Best one I know of is I2SoverUSB: I2SoverUSB - I2S over USB Audio
The same company makes some well implemented AK4493 dacs. Haven't tried one myself, but do know that good implementation is just as important as the dac chip used (maybe more important than the dac chip in some cases).
Products - I2S over USB Audio
 
One thing I would say is that since AKM dacs get MCLK from the USB baord it is important to use a USB board with very low-jitter clocks.

Where did you get this from? Perhaps from some obsucre chinese crap board? The AKM AKD4499 reference board (I got one) gets the MCLK from the AK4118 DIR (or, selectable, from an external source), exactly as it should.
 
Where did you get this from? Perhaps from some obsucre chinese crap board? The AKM AKD4499 reference board (I got one) gets the MCLK from the AK4118 DIR (or, selectable, from an external source), exactly as it should.

AK4499D evaluation board (I got one too) can get MCLK from AK4118 if receiving SPDIF/TOSLINK, or optionally from an (unpopulated) local clock on the MB that can act as a reference for AK4137 (if it is used for ASRC following AK4118).
When using the external I2S input, say, a USB to I2S board then the USB board clocks are used. Most people around here are using USB as a source. That's where it comes from.
 
AK4499D evaluation board (I got one too) can get MCLK from AK4118 if receiving SPDIF/TOSLINK, or optionally from an (unpopulated) local clock on the MB that can act as a reference for AK4137 (if it is used for ASRC following AK4118).
When using the external I2S input, say, a USB to I2S board then the USB board clocks are used. Most people around here are using USB as a source. That's where it comes from.

I don't see in the AKD4499 board schematic any option to take MCLK from an USB board. The MCLK jumper has only the AK4118 or external MCLK sources options.

Now, if somebody is dumb enough to use USB to source the MCLK to the AK4499 board, as an external source, then he gets the performance he deserves. Those clocks are not designed for low jitter, good results are by pure luck.

This is not a board issue, but a user brain issue.
 
USB audio class receivers with asynchronous endpoints will have a local audio frequency oscillator. It’s going to need to go to the DAC unless you use the AK4137. That’s what Mark means. MCLK must be synchronous if you don’t use an ASRC.

An ideal design feeds this clock backward from the DAC board.

Of course. If I would have to use that clock I would not do it before properly PLLing it, or using the SRC (no idea how good the AK4137 is). I absolutely see no reason to use a $150 DAC chip with anything less.
 
I don't see in the AKD4499 board schematic any option to take MCLK from an USB board. The MCLK jumper has only the AK4118 or external MCLK sources options.

Now, if somebody is dumb enough to use USB to source the MCLK to the AK4499 board, as an external source, then he gets the performance he deserves. Those clocks are not designed for low jitter, good results are by pure luck.

This is not a board issue, but a user brain issue.

The above post should probably be preserved for posterity.
 
Last edited:
Things have changed some over the years since you may have dipped into this subject area, so it would seem anyway. No PLL used for audio can beat modern ultra-low jitter audio clocks such as NDK SDA series (I have done some audio work with ultra-low additive jitter PLLs, but they still add a little jitter to the crystal clocks). Those are the clocks used on the USB to I2S board I use and recommend. The main thing to be careful about is to keep jitter from deteriorating due to interconnections between the USB board and the dac board external MCLK input. In the case of AK4499 what I ended up doing was putting the USB board right above the external MCLK connector so that the interconnection wires are roughly in the range of 1/2" to 1" long. There is a little more that could be said, but that gets one pretty close to optimal for test purposes with an evaluation board. Orientation of the USB board has some effect, but I will leave that for another time. Integrating the clocks onto the dac board ground plane and putting them closer to the dac chip would be better, as Chris alluded to, but that would have to be reserved for a different dac board design.
 
Last edited:
Things have changed some over the years since you may have dipped into this subject area, so it would seem anyway. No PLL used for audio can beat modern ultra-low jitter audio clocks such as NDK SDA series

Excuse me for being dense today, but is there any way to recover the clock from a serial transmission using only an ultra-low jitter clock as the NDK SDA and without using a PLL/synthesizer? You think the AK4137 (or any other SRC for that matter) is, on the clock recovery side, anything much beyond an Integer-N PLL/synthesizer?

That would be indeed news to me; of course, an ultra low phase noise VCO is indeed desirable, since the PLL output is a combination of the phase noise of all components, see attached (for a synthesizer).

P.S. In a sense, you are correct: my old habits are telling me to avoid using a chip until I understand it's internals. This could be a non productive habit in the era of commodity chips that "just work", discouraging any optimization effort attempts.
 

Attachments

  • pll-phase-noise.JPG
    pll-phase-noise.JPG
    72.1 KB · Views: 519
Last edited:
There is just some confusion over semantics here. You don’t need a PLL with an asynchronous USB receiver. The data is clocked out to I2S by the clock you give it. So, it should be synchronous with the converter clock if you’re doing it correctly, which is feeding the converter clock back to the USB receiver IC. Asynchronous USB devices use a feedback endpoint to control data rate from the host (along with an internal buffer I’m sure).

If your USB receiver is on a separate board and has questionable clock quality then yes you could possibly benefit from a PLL + VCXO or ASRC.
 
We seem to be talking about two different things here. The first is serial audio such as SPDIF, which requires clock recovery. The other serial audio is also often called I2S to distinguish it from the first type above. I2S (sometimes referred to as I2S bus) has separate clock and data lines. Some USB boards can output both types of serial audio formats.

More specifically, the USB to I2S board I use can do both formats:
I2SoverUSB - I2S over USB Audio

For some reason they hid the manual for the current version. It can be found at:
http://jlsounds.com/uploads/I2SoverUSB v.III.pdf

It shows the galvanically isolated pinouts for I2S and for SPDIF. It can be jumper programmed to output 22/24MHz and or 45/49MHz I2S MCLK signals which are very low jitter.

In almost all cases it is preferable to use I2S rather than SPDIF inputs with dacs if possible. That is for a few reasons: (1) no clock recovery needed, (2) higher sample rates possible, (3) native DSD is possible (vs DoP).
 
We seem to be talking about two different things here. The first is serial audio such as SPDIF, which requires clock recovery. The other serial audio is also often called I2S to distinguish it from the first type above. I2S (sometimes referred to as I2S bus) has separate clock and data lines. Some USB boards can output both types of serial audio formats.

More specifically, the USB to I2S board I use can do both formats:
I2SoverUSB - I2S over USB Audio

For some reason they hid the manual for the current version. It can be found at:
http://jlsounds.com/uploads/I2SoverUSB v.III.pdf

It shows the galvanically isolated pinouts for I2S and for SPDIF. It can be jumper programmed to output 22/24MHz and or 45/49MHz I2S MCLK signals which are very low jitter.

In almost all cases it is preferable to use I2S rather than SPDIF inputs with dacs if possible. That is for a few reasons: (1) no clock recovery needed, (2) higher sample rates possible, (3) native DSD is possible (vs DoP).

I’m pretty sure syn08 knows the basics.

BTW, if your board has galvanically isolated MCLK then it’s going to be jittery. If only I2S is isolated then you don’t actually have galvanic isolation because you’re connecting MCLK around the gap.

All these boards putting the clock near the USB receiver are hacks.