AKM4399 dac

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Has anyone done this project or not
In Thailand, there are groups that are going to this project.
Do you think it would be worth a try. Give me some advise.
I have not been listening.
esoteric k03
Thank you very much. :D:D
 

Attachments

  • akm4399.JPG
    akm4399.JPG
    13.6 KB · Views: 1,267
  • akm43991.JPG
    akm43991.JPG
    83.5 KB · Views: 1,234
Given that the AK4399 produces a 2VRMS differential line-level output, I should think that perhaps the most critical stage will be your output buffer design and implementation. Particularly, should you want to produce a single-ended output from the AK4399 DAC's differential drive. An op-amp is probably the least risky solution to perform this function.

Should you want to stay away from op-amps then a transformer might immediately come to mind as the most elegant solution. However, while the AK4399 voltage outputs are rated to drive a 1k load, I believe that they also are only rated to drive up to 100pF of load capacitance. One meter of interconnect cable typically has more distributed capacitance than that, let alone including the input capacitance of the following component and the output transformer's parasitic winding capacitance. Which means that some form of active output buffer will most likely be necessary, even using an transformer.
 
Last edited:
One suggestion - try to rotate 90 deg the capacitors C3, C5, C7 and C12. Those PCB loops that you have now have higher inductance than straight(er) lines.

Hi SoNic_real_one,

It seems C3, C5, C7, and C12 that show in layout are the electrolytic capacitor. As we know, the purpose of these low value caps are used for decoupling/bypassing the high frequency noise. It should be small size such as the X7R or C0G dielectric or should be PSS or PEN dielectric (bigger a little bit). The low value (0.1u or 1u) conventional electrolytic cap may be used but I do believe the performance at higher frequency, either C0G or X7R better.

Roughly review, I see some abnormal thing in the evaluation board datasheet, the digital power supply pin (VDD pin1) and analog power supply pin (AVDD pin41) shared the same power supply source. But in chip datasheet suggest to seperate the digital power supply from the analog power supply, this is the right way to do to eliminate the digital noise that will be coupled to analog pin. The route of grounding must be considered also. Probrably, the guy who wrote evaluation board datasheet is not the guy who wrote the chip datasheet.

The problem is the route of ground return for both digital and analog also.
If we consider only the digital path of this 2 layers PCB. I do not see any ground via that close to ground pin so I assume that the designer will not fill any ground plane underneath the chip both top and bottom layer and you can see on the left hand side of AKM chip (digital section) BICK, SDATA, LRCK, and MCLK pin doesn't any ground plane underneath (on bottom layer) of these trace line also. This is the real loop antenna for EMI. You can imagine where is the trace will be for ground return? It's very important if DIY claim there DAC is pretty clear and their DAC can provide a lot of details but I always see their PCB layout is not good (almost is really bad), how can they here sound at the LSB. For myunderstanding is may be the sound is pretty warm due to the noise but the detail at LSB is gone.
To maintain the wide dynamic range with low noise in hositle digital environments circuit design techniques and also grounding schemes, particularly. Although, using the solid ground plane common to both analog and digital circuit it just for low resolution DAC. For higher resolution, more than 12 bit it extreamly required higher noise isolation, split plane connection?

For example with this DAC 5V reference level the LSB in 24 bit is equvalent to 0.3uV, approximately. :eek:
For 32bit you can imagine how much of the LSB will be equivalent to? Do they have 32bit digital reciever released to market yet? Where I can buy the 32bit music source?

Hope these are the fact but keep in my that the definition of sound quality in term of electrical performance (engineering) and audiophile listener/diyer is different. For example, I always see the people like the low noise circuits and components but they love the carbon/composite resistor. Some people like low distortion but they love tube. :)
 
No, the C3, C5, C7, and C12 looks like are decopuling ceramics for the electrolytics C4, C6, C11, C13.
And yes at least an inductor (followed by decoupling capacitors) should be placed between the digital and analog power pins if they come from the same regulator. Ideally would be separate regulators.

It is hard for me to see exactly, the pictures don't have the best resolution.
And yes, no DAC on market goes past 20-21 bit actual analog resolution (based on the THD+N figures, not just the noise). Some of them don't even go past 18 bit. 32bit is just pure marketing.

For reference:
20bit = 120dB = 0.0001% THD+N
21bit = 126dB = 0.00005% THD+N
 
Last edited:
No, the C3, C5, C7, and C12 looks like are decopuling ceramics for the electrolytics C4, C6, C11, C13.

And yes at least an inductor (followed by decoupling capacitors) should be placed between the digital and analog power pins if they come from the same regulator. Ideally would be separate regulators.

It is hard for me to see exactly, the pictures don't have the best resolution.
And yes, no DAC on market goes past 20-21 bit actual analog resolution (based on the THD+N figures, not just the noise). Some of them don't even go past 18 bit. 32bit is just pure marketing.

For reference:
20bit = 120dB = 0.0001% THD+N
21bit = 126dB = 0.00005% THD+N

It hard for me to see the exactly, too.

Yeb you right I'm worng. C3, C5, C7, and C12 top overlay look like are tantalum, it show the mark of polarity.

Overrall I agree with you. :devily:
 
A friend of mine
Thanks to all suggestions
Sorry for the language English it my Poor
I've tried many To be used Google translation To communicate
give understand in communication from me to you and from you to me
I believe that expertise from diyaudio member all to helps share knowledge
And a better approach For this project AK4399 It is as I expected
Now it's almost finished.
My team tried very hard.
So that it brings out the best
I still want suggestions from a friend everyone at diyaudio this
And thanks to everyone.
for read and think about this project together I would do it give certainly succeeds
Thank you to everyone again
Will be updated the next time.
 
Nice to see people diy-ing in tland
sawadee pbee mai kup
สวัสดีครับ คิดว่าคุณคงเป็นคนไทย สวัสดีปีใหม่เช่นกันครับ ขอบคุณที่เข้ามาทักทายกันนะครับ
I think you would be Thai. Thanks to greet them.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.