Build thread for Diyinhk ES9018 DAC on Ebay

Status
Not open for further replies.
asyn 100M clock with CM and XM is tested:D it will have lock issue.
lower than 40M clock is also tested:D it will have serious sound distortion problem, I don't know if it's call quantization noise.

Sync works faultlessly from the CM and I have no issues with the sound quality like this. :D I'm using OPA627 in the I/V and LME49990 in the buffer, playing 44.1Khz.

Unfortunately, when I play 88.2 Khz or higher, the sound is seriously distorted. Is this a setting problem for the CM asio driver / Windows 7 ? I'm using Foobar and the SOX resampler.

Thanks,

Tom
 
Last edited:
Sync works faultlessly from the CM and I have no issues with the sound quality like this. :D I'm using OPA627 in the I/V and LME49990 in the buffer, playing 44.1Khz.

Unfortunately, when I play 88.2 Khz or higher, the sound is seriously distorted. Is this a setting problem for the CM asio driver / Windows 7 ? I'm using Foobar and the SOX resampler.

Thanks,

Tom

The sync clock is connect from X49/GND and X45/GND (not from MCLK) to the es9018 X1/GND, remember to disable the local clock. And only 6631"A" version will work, 6631 will not work. If the old 6631 driver is installed, it must be uninstall and reboot, and install the latest 6631A driver
 
Last edited:
Hi,

The problem was the ASIO driver. For over 48Khz, it only works at 24 and 16 bit. It can only do 32 bit for 44.1khz and 48Khz.

DIYINHK - if you have tried many clock frequencies with this DAC - which one do you think sounds best ? From the very limited tests I've done, I think an ultra-low jitter 90.3168Mhz would do well, with a logic gate to divide that to 45.1584Mhz and feed that back into the CM6631. Just an idea, probably not a good one, and I have no idea where I can source such a clock.

Anyway, it's working perfectly now and has been for hours. No issues at all and sound quality is better than I expected. I hooked it up to my hifi and found it has a very wide soundstage, with a warm signature. So thank you very much for making a good low-cost DAC like this !

Next for me to try is a WM8805 receiver, and then next week I'll start fitting better resistors etc.

I've put a lot of information on the first page of this thread too, aimed at beginners. Please have a look at it and PM me if any of it is obviously factually wrong.

cheers

Tom

EDIT - I have a question. The two boards have two ground connections - the I2S ground and the XO ground. Is this a potential ground loop ?
 
Last edited:
Hi,

I thought I'd post an update now before I get busy with work again and this project starts to gather dust.

I've been trying different I2S sources with different clocks, and comparing that to a 384Khz dual AD1955.

So far, the only way I have managed to get as good or slightly better is to use a Crystek CCHD-957 and the WM8805 in async. I can say with some certainty that faster clocks sound better.

What is exciting about this - I'm still using the thick film resistors and TL1963 reg @3.3V. So it seems very likely that this DAC, with a CCHD957 clock, decent resistors and the Sigma11 reg @3.8V, will be something very special.

Does anyone know if a 90.3168 Mhz low jitter oscillator exists ? Maybe Guido Tent would be willing to develop one ? It could be divided into lower frequencies with a 74VHC163, and then used to synch with the I2S source.....

cheers
 
Last edited:
Hi,

...
So far, the only way I have managed to get as good or slightly better is to use a Crystek CCHD-957 and the WM8805 in async. I can say with some certainty that faster clocks sound better.

...
cheers

Thanks for the reports, KlipschKid,

What is the speed of your 957? 100 MHz CCHD-950 is also a low jitter clock...
 
Found this overpriced part :

FVXO-HC73BR-90.3168 Fox Electronics | FVXO-HC73BR-90.3168-ND | DigiKey

I have two 957 - the 24 and 22Mhz. They don't work well with the usb but they're great with the wm8805. Thanks for heads up about the 950 !

I've tried several of the Fox Xpresso and definitely faster is better, but lower phase noise is better still, so that 950 looks like the best choice ? Has anyone actually tried all of Crystek's to see if higher freq or lower phase noise is better ? i.e. compared the CCHD957 45m with the cchd950 100m ?

How's your build coming ?
 
Last edited:
Some numbers

I changed the power supplies so the Sigma11 is only feeding the AVCC analogue section @3.76V, and the lt1963 has been bypassed.

I tapped the rectified but unregulated voltage on the Sigma11 and fitted an RC filter - 12R 5W and 1500uF - followed by an 7805A reg and a 820uF Nichicon PLG after that, and fed this to the wm8805 and the 3.3V adp151 which powers the DACs digital sections.

With the 125Mhz clock, the 12R drops 3.2V from the unregulated voltage of 10.7V, so that means the wm8805 + DAC's digital use 266mA with a 125Mhz clock.

At the other extreme, the 12R drops 1.51V from 11.8V = 126mA with a 22.5792Mhz clock.

I was really surprised by the difference so I double checked those voltages - they are correct. I know the wm8805 board uses 70mA so we have a 56mA to 196mA range just by changing the clock. This is at 3.76V AVCC, which is independent . 196mA is close to the ADP151 limit so the 125Mhz clock is risky.

The graphs on hifiduino show a linear relationship between clock freq and power, so it should be easy enough to work out what current demand each clock freq will require.

The sound quality is excellent with the Crystek and Sigma11. I'm very pleased and I'll be very interested to see what difference the resistors make, and I think I'll upgrade the 7805 to something like an adp7104.
 
Last edited:
...
With the 125Mhz clock, the 12R drops 3.2V from the unregulated voltage of 10.7V, so that means the wm8805 + DAC's digital use 266mA with a 125Mhz clock.

At the other extreme, the 12R drops 1.51V from 11.8V = 126mA with a 22.5792Mhz clock.

I was really surprised by the difference so I double checked those voltages - they are correct. I know the wm8805 board uses 70mA so we have a 56mA to 196mA range just by changing the clock. This is at 3.76V AVCC, which is independent . 196mA is close to the ADP151 limit so the 125Mhz clock is risky.

...

That is too close to max current of the ADP regulator. With a 100 Mhz clock, the current would be around 160 mA. Perhaps it is a good idea to have two regulators (right and left) feeding the 1.2v core supply

There is also the increased temperature in the DAC chip

Which 8805 module are you using?
 
Last edited:
Yeah, it seems I should not go over 50Mhz for the adp's sake, or use an ADP7104 (15uV noise) that can handle 500mA.

The WM8805 is some "crap" I got on Taobao that I spent a week "fixing" so it works really well, and damn, it sounds great with this DAC and a Crystek.

WM8805 ¶à¹¦ÄܽâÂë½ÓÊÕÇл»°å ´ø8ÖÖÒôƵÊý¾ÝÊä³ö¸ñʽ-ÌÔ±¦Íø

I want to feed the spdif from the CM6631A into the Wolson so I can isolate it. The CM6631A output is CMOS or TTL so I'm using a potential divider to drop the levels so the wm8805 can get a lock. I've done this with a CM6631, no problems. The CM6631A is not the same it seems because its volume is 6dB higher. So get distortion unless I turn the volume down -6dB in Foobar.

Any idea how I can solve this issue ? It can't be bit-depth issues because 16 to 24 is 8 bits which is far more than 6dB. I think it must be either the firmware or the driver ?
 
Status
Not open for further replies.