In case it helps someone else in a purchasing decision I have enclosed measurements of an affordable AK4396 DAC. (It was only $16 on AliExpress.)
The measurements were done with REW with the optical output of a USB SB0490 driving the AK4396 routed back to the Line 1/2 inputs of the SB0490. ASIO drivers were used (ASIO4ALL at 24 bits 48 kHz).
The distortion is at the limit of my SB0490 to measure. It is still worth measuring because it could help detect a major layout or power supply issue. I have measured other DAC boards and found quite bad performance due to layout, design, power supply, etc.
Two sets of measurements are enclosed. One set with the stock onboard power supply (LM7812/LM7912) for the analog section. The second set uses an external LM317/LM337 supply as a pre-regulator to feed the LM7812/LM7912. This eliminates the 60 Hz (and harmonics) seen with the stock configuration. The pictured LM317/LM337 kit only costs $5 from AliExpress (you solder it together) and it greatly reduces 60 Hz. (And no modifications to the DAC board.)
I think it is a very nice board especially since it costs only $16. I still might add 10 uF ceramics under the film supply bypass capacitors around the AK4396 and the output op amp. I am using an NJM4580D as the output op amp but I did not see very much difference in measurements with the stock NE5532P.
FYI: The spike near 2 kHz is something in my system that I can't seem to track down. That is not the DAC under measurement.
The measurements were done with REW with the optical output of a USB SB0490 driving the AK4396 routed back to the Line 1/2 inputs of the SB0490. ASIO drivers were used (ASIO4ALL at 24 bits 48 kHz).
The distortion is at the limit of my SB0490 to measure. It is still worth measuring because it could help detect a major layout or power supply issue. I have measured other DAC boards and found quite bad performance due to layout, design, power supply, etc.
Two sets of measurements are enclosed. One set with the stock onboard power supply (LM7812/LM7912) for the analog section. The second set uses an external LM317/LM337 supply as a pre-regulator to feed the LM7812/LM7912. This eliminates the 60 Hz (and harmonics) seen with the stock configuration. The pictured LM317/LM337 kit only costs $5 from AliExpress (you solder it together) and it greatly reduces 60 Hz. (And no modifications to the DAC board.)
I think it is a very nice board especially since it costs only $16. I still might add 10 uF ceramics under the film supply bypass capacitors around the AK4396 and the output op amp. I am using an NJM4580D as the output op amp but I did not see very much difference in measurements with the stock NE5532P.
FYI: The spike near 2 kHz is something in my system that I can't seem to track down. That is not the DAC under measurement.
Attachments
-
AK4396 Picture.jpg -
AK4396 1 kHz Spectrum SB0490 SPDIF Optical to AK4396 to Line In 12 24 Bits 48 kHz.png -
AK4396 512k 8 Sweeps SB0490 SPDIF Optical to AK4396 to Line In 12 24 Bits 48 kHz.png -
AK4396 1 kHz Spectrum SB0490 SPDIF Optical to AK4396 to Line In 12 24 Bits 48 kHz LM317 LM337 An.png -
AK4396 512k 8 Sweeps SB0490 SPDIF Optical to AK4396 to Line In 12 24 Bits 48 kHz LM317 LM337 Ana.png
Last edited:
Look here:
https://www.diyaudio.com/forums/digital-line-level/186245-dac-2496-ak4393-dac-kit-
cs8416-ak4393-5532-a.html
https://www.diyaudio.com/forums/digital-line-level/186245-dac-2496-ak4393-dac-kit-
cs8416-ak4393-5532-a.html
Thank you. Very helpful. The price has gone down on this board over the years. It will take a long time to wade through the full thread and try to distill the key points. So far I see 22,000 uF caps (I might use pre-regulated DC instead), solid polymer capacitors, LM4562 and polystyrene in the filters as the key points in what I have read. What is the executive summary of the most important changes that you see in that thread?
I plan to take a slightly different track since I would like to include measurements before and after my changes. And I would like to try a few different approaches such as pre-regulators and perhaps shunt regulators. Right now my idea is to not bring AC power onto the board but instead use pre-regulators in an aluminum project case and then twisted pair and aviation connectors between the power supply and the DAC board. Currently I am using just LM317/LM337 with a huge improvement in AC line related harmonics (as show in the first post spectrums). I think I might end up with a much simpler and cleaner looking modified board that way. Perhaps the external pre-regulated power supply and then 10uF ceramic bypasses under the board plus a few low ESR caps under the board.
Right now I am working on my measurement setup to see if I can remove the measurement setup glitches. (Trying the stepped sine measurements in REW.)
I only place Digikey or Mouser (authorized distributor) orders once (maybe twice) per year and I plan on including a few op amps this year. I could perhaps also include some solid polymer capacitors also.
Try to find Bill of Materials (BOM) for AK4396, where each element is precisely talked about.
BOM) AK4396 - Google Search
BOM) AK4396 - Google Search
Last edited:
I have also got another BOM, I think this is somewhat newer. I don't know who the author is and if I can share it here, but I'll try.
Those BOM are very helpful and will help organize and plan.
Since I am already using LM317/LM337 pre-regulators for the on-board LM7812/LM7912 analog supplies I thought the next logical step would be to look at the bypassing of the op-amp (first), the AK4396 (analog then digital) and then finally the CS8416.
I would like to include good quality measurements in this project, at least as far as my equipment allows. I can use REW to measure spectrums and also sweeps of distortion.
However I was wondering if the REW RTA can be used to make more directly applicable spectral measurements of supply noise before and after I add improved bypassing. For example, perhaps I use a 10uF film capacitor as a DC blocking capacitor and then use REW and the audio interface to measure the noise spectrum across each supply in question by soldering the blocking cap and a shielded cable under the existing bypass capacitors (from under the board.)
Then there would be a recorded quantitative measurement/spectrum of supply noise that could be compared before and after each modification.
Does this sound interesting/workable? Is this a good idea? Bad idea? I would appreciate any expert advice/suggestions.
Since I am already using LM317/LM337 pre-regulators for the on-board LM7812/LM7912 analog supplies I thought the next logical step would be to look at the bypassing of the op-amp (first), the AK4396 (analog then digital) and then finally the CS8416.
I would like to include good quality measurements in this project, at least as far as my equipment allows. I can use REW to measure spectrums and also sweeps of distortion.
However I was wondering if the REW RTA can be used to make more directly applicable spectral measurements of supply noise before and after I add improved bypassing. For example, perhaps I use a 10uF film capacitor as a DC blocking capacitor and then use REW and the audio interface to measure the noise spectrum across each supply in question by soldering the blocking cap and a shielded cable under the existing bypass capacitors (from under the board.)
Then there would be a recorded quantitative measurement/spectrum of supply noise that could be compared before and after each modification.
Does this sound interesting/workable? Is this a good idea? Bad idea? I would appreciate any expert advice/suggestions.
I think you will go quite a ways under the noise floor of your measurement setup. The trick is to add an Low Noise Amplifier (LNA) of about 60dB so you can drive the noise level up so it is measurable. Trileru is pretty knowledgeable about this and is using this type setup in the Denoiser thread by Elvee. He actually sent me the Gerbers for a a nice LNA by Doug Self. I would be happy to send you a board, but be forewarned, the board does not have a silk screen for parts placement and a couple of the cap pads are backwards. It was a real challenge to assemble, LOL. I'm waiting on a case so I can use it. At this level you pickup all the stray EMI/RFI/ lines noise from everything in your house, the case and proper grounding are essential to success.
The 2k spike may be USB induced, it interacts with the 1k fundamental some how. Try using a fundamental just under 1k, around 900, and see if the artifact goes away.
Hi! I wanted to buy this DAC, many thanks for the measurements - I'm convinced to buy it now. I have one question only - did you use two separate transformes for 9v and 2x15v, or the one that is suggested by seller from the Aliexpress (single transformer with 5 secondaries)?