SD Card Memory FLAC/WAV 192/24 player - ESS crescendo II - AK4495 - DV20A

Hi all,

Haven't seen this one around on the forum so I thought to give it a shot.
Been on the forum for a couple of years but haven't contributed yet and thought now is a good time :)
I've been playing around with a couple of SD card FLAC players for a while now and ran into this one on eBay:

Finished Digital Turntable Lossless Music Player Ape WAV MP3 AK4495 DAC | eBay

Decided to buy it and see how it can be modded to sound any decent.

[IMGDEAD]http://highendaudio.ca/images/DV20A/DV20A.jpg[/IMGDEAD]

After opening up, the player shows a modular approach. The SD card section is based on the ESS Crescendo II soundbar chip. I spoke with my contacts in ESS to get confirmed that this chip is actually real and capable of doing FLAC decoding on the fly.

[IMGDEAD]http://highendaudio.ca/images/DV20A/DV20A-ESS-Crescendo.jpg[/IMGDEAD]

The ESS chip reads SD cards and USB sticks, clocks the data through it's built in DSP blocks, which at this point should be all 32bit wide audio path.
Apparently this is also where it decodes FLAC up to 24bit 192ksps and does digital volume control.
Via i2s and spdif the resulting digital audio is available.
Seeing the small smd crystal oscillator that clocks the whole thing, I'm thinking there may be something to optimize here.

The second pcb is the i2s AK4495 and power supply board.

[IMGDEAD]http://highendaudio.ca/images/DV20A/DV20A-AK4495.jpg[/IMGDEAD]

First thing I had to do was a checkup of the infamous dipswitch that controls the filtersettings of the AK4495 and of course the dipswitch design flaw is present in this board. Removed and replace by 2 jumpers.
Second thing, check the level of pin 5, the SSLOW signal. Luckily this is connected to GND already, so disabled.
The SSLOW signal makes the filters extend their roll off for use at lower than 44.1 kHz sampling frequencies, if I'm interpreting the datasheet correctly, page 28:

http://www.akm.com/akm/en/file/datasheet/AK4495EQ.pdf

The next thing was checking the output circuit. They used a double opamp by JRC, the MUSES8920
I'm not familiar with that chip at all, I'm definitely going to experiment changing it for something else.

The output is connected via 22ohm resistors, thank goodness no crapacitor in the output... BUT because they apparently used 5% resistors or something there was 45mV DC offset in one channel and 16mV in the other. Not so good.
My mod shows a 47K5 fixed resistor / 50K trimpot pair over each DAC output resistor to level away the DC offset. See the image above.

I have to say my first impression, after the mods, is quite positive. I am going to do a lot of listening now and prepare a list of further mods to improve the way the thing sounds.

By the looks of it, there is room for improvement in the decoupling, power supply, clocking and output filter.

Never a dull moment :)

I'm curious to know if there's anyone out there that already has done something with a unit like this.

Cheers, Hans.
 
The player has now been playing for almost 2 days and I've been doing some first impressions listening to it.

First immediate thing I noticed using the headphone is that the channels are reversed on the 3.5mm headphone output! Annoying... The line outputs are connected OK however.

Studying the decoupling of the DAC chip itself revealed an interesting mod:
The VREFH and VREFL for both channels are decoupled with 100n ceramic parallel to a 22uF elco. According to an article on hifiduino - https://hifiduino.wordpress.com/2014/12/07/akm-verita-4490-dac/ - there's something to be gained by the value of the decoupling capacitors there.

vrefcap.png

(image courtesy of hifiduino)

I paralleled 2200uF/35V to it and that has made some good changes to the lower midrange and bass.

[IMGDEAD]http://highendaudio.ca/images/DV20A/DV20A-vrefmod.jpg[/IMGDEAD]

Next mod will be focusing on the output opamps, the one currently in there is the MUSES8920 which I'm still not familiar enough with the have a feel how it sounds like compared to other opamps out there.

I have ordered some other dual opamps, such as the OPA2604 and 2134 and might build a couple double SO8 SMD->single DIP8 contraptions to test other single opamps

Since the AK4495 is voltage out compared to current out, it could be that other than my usual goto workhorses such as the OPA627 might shine here...

Any recommendations from the field?
I'm going to shoot an order in with Digikey this Monday

Cheers, Hans.
 
I'd not trouble yourself swapping out the opamp, that MUSES has decent enough specs on paper. You could though increase the impedances of the networks surrounding it (Rs and Cs) as when I did this on my WM8741 player it improved the low frequency dynamics a lot. If you can identify which of the recommended output circuits in the AKM datasheet is being used, I can calculate some new R and C values to try.
 
Can you also plug in a USB hard disk ? (Probably a really dumb question but what the heck)

Not a dumb question at all...
I was more focused on playing from SD card but intrigued by this option I connected a 1TB external USB disk and it does show the file structure no problem, however, when I try to play a FLAC file from it the player freezes.
I will experiment a bit more to try get this going because that would turn it into a neat low cost music server...

Cheers, Hans.
 

Jogi

Member
2008-01-14 12:30 am
Hans,
any news on this project?
It looks quite interesting, especially regarding the possibility of adding another dac, for example the Soekris. Should be a nice one box solution, if the Scard Player is up to task soundwise.
Playing via computer and USB is surely very convenient, but I guess SD sounds better.
 
Bought a DV20A as well. Slightly different than Hans his version, powersupply for the SD/USB has been moved to the i/o board itself. Now i'm trying to identify the oscillator on the i/o board. It's a 4 "legged" SMD verion, brand : KSS and maybe it's a 27 MHz version ?? I will add a picture, a sharper one once i'm back home again.
 

Attachments

  • IMAG3238b.jpg
    IMAG3238b.jpg
    228.1 KB · Views: 643
Okay , first thing that i wanted to do was to split up de powersupply for the opamp.
The double voltage 15/0/-15 was rectified using one bridge.
The 0 wire was connected inside the transformator with the other secundairy windings.
Using just 3 wires gave a loud hum coming from the transformer.
I had to cut the trafo open to make 4 wires for the double rectifier.

I then removed the two regulators, LM317 & LM 337 and surrounding caps + resistors.
Changed them for 2 NEWclassD regulators i had laying around.
Removed the 2x "Elna for audio" 4700uF caps for 2x Nichicon KZ 1000uF
I also made a new rectifier with 8 Schottky diodes, the existing diodes were also Schottky's.
 

Attachments

  • IMAG3252.jpg
    IMAG3252.jpg
    498.5 KB · Views: 544
  • IMAG3255.jpg
    IMAG3255.jpg
    437.4 KB · Views: 515
Last edited:
With this newer model DAC is the powersupply for the USB/Sd card /display etc. moved to it's own board. The DAC board itself remained unchanged, only without parts. So I cut the board up to make room for 2 trafo's to make 4 powersupply's.
The last picture is of a later date, there i allready changed all the elco's surrounding the AK4495 for real? Silmic 2 and Panasonic FM.
I also moved the main switch away from the rectifier, too close, it became unsafe.
Funny thing about that switch, it was placed between the 2 primairy 115V coils. That gave some problems when installing my own trannies :)
 

Attachments

  • IMAG3315[1].jpg
    IMAG3315[1].jpg
    462.7 KB · Views: 182
  • IMAG3260[1].jpg
    IMAG3260[1].jpg
    350.5 KB · Views: 194
  • IMAG3237[1].jpg
    IMAG3237[1].jpg
    493.9 KB · Views: 445
Last edited:
The stock transformer delivers 8V/1A to the I/O board and 2x 15V ?A to the DAC itself. There only the negative side is used to feed the MUSE8820. The positive side is also used for the opamp AND to feed 2 LM2941 regulators.

The LM2941 one the right powers VDD/L/R and VREF/L/R, the other regulator powers DVDD and AVDD. I wanted to split that up, the original supply is now only used for the opamp.
With my 2 new trannies i can make 4 power supplies. One NewclassD reg for VDD/L/R at 6.9V, a TPS7A4701 based reg for VREF/L/R also at 6.9V.
In the pictures you can see where i cut the lines.

TPS7A4701 Low Noise RF LDO Regulator Power Supply Module 1.5V-34V 3.3V 5V 12V 1A | eBay
 

Attachments

  • IMAG3271[1].jpg
    IMAG3271[1].jpg
    850.7 KB · Views: 177
  • IMAG3273[1].jpg
    IMAG3273[1].jpg
    318.1 KB · Views: 130
  • IMAG3332[1].jpg
    IMAG3332[1].jpg
    504.1 KB · Views: 147
Last edited:
Another TPS7A4701 reg at 5.25V was placed in the small space left. That one powers a Twisted Pear AVCC reg with 2x3.6V output. Those outputs feeds the DVDD and AVDD of the AK4495 chip. Luckily the sound improves with every change, better image, cleaner sound, but maybe a little too bright.

-As mentioned before on other threads, you can easily cut the single powersupply for analog and digital up by removing the 2 ferrite coils on the board.
- I tried the 2 big capacitors on VREF/L/R, the sound got gloomy and slow. I currently use only 2 Wima 6.8uF filmcaps.
 

Attachments

  • IMAG3334[1].jpg
    IMAG3334[1].jpg
    440.9 KB · Views: 166
Last edited: