Build thread DIYINHK ES9018 DAC balanced setup

[Suprted1]

Finally its alive

First pictures on the scope just arrived. And it is working...

The board has a lot of messy modifications. But thats the reason i got this board.

mod 1 - Separate ground for analog / digital signal
mod 2 - Balanced output (following ESSTech´s design ??)
mod 3 - Power supplys

Components used is at the time a mix of some good ****, and some that I had available. Esstech rated AD797 on top. So of course I had to get a few of these expensive op-amps.

Tomorrow I hope I will manage to do a simple test, that its actually working with amps instead of scope. After I have fixed one of the outputs that didnt behave as supposed to. The negative output on one of the channels.

USB --> I2S by Musiland 01US 2012

Some trouble with Musiland if turning on/off other equipment. It shuts down immediately :-( Hopefully Ill find a way around this later, or replace it with another USB -> I2S.

Good night folks ;-)

 

[diyinhk]

PCM2707 also has the same interference shuts down problem, using an adum4160 USB isolator can solve it, but it's usb1.1 only.
CM6631 do not have the shuts down problem even without isolator, and the windows driver is stable, it's much better.
I also want to try balanced setup, it should have another excellent sound experience

 

[themystical]

Hello diy mekkern
That looks very impressive! Have you got "sound" out of it yet?
I don't have the board but am considering getting one to use in a balanced mode with an external IV arrangement.
Previously I had assumed that to get the differential outputs from the Sabre, you could connect to pads as shown in this picture.
http://hifiduino.files.wordpress.com/2013/01/1-dsc03262-006.jpg
I notice from your pictures that you have a few connections directly to the chip pins. What are these for?
Can you please explain in a little more detail the mods you have had to do get a balanced output and seperate the Digital/Analogue grounds?

 

[Suprted1]

Well, after a short discussion, my thought about separating the Agnd and Dgnd is not even close to a good idea. but the chip did actually work, so I luckily didnt kill it because of different potentials....

Balanced output :
Replaced 680 ohm with 2720 as in ESS´s papers, which probably is because the single ended output is removed, otherwise i dont know.
See pic´s.

 

[Suprted1]

Hello diy mekkern
That looks very impressive! Have you got "sound" out of it yet?
I don't have the board but am considering getting one to use in a balanced mode with an external IV arrangement.

I notice from your pictures that you have a few connections directly to the chip pins. What are these for?

Direct connections is Dig gnd, which will be removed....
Red and yellow direct connection is digital 1,2 and 3,3 Volt supply. Delivered from another PSU then the analog parts.
Crystal has its own ADP151 3,3V regulator, and crystal is one of the few 100MHz I found on RS online. Crystek seems to be non-available on ebay / ELFA / Rs-Online / Farnell.

Thanks , but still no sound.
Other obligations is killing my time ;-)

 

[themystical]

Thanks for sharing this information. Much appreciated.

 

[qusp]

Well, after a short discussion, my thought about separating the Agnd and Dgnd is not even close to a good idea. but the chip did actually work, so I luckily didnt kill it because of different potentials....

Balanced output :
Replaced 680 ohm with 2720 as in ESS´s papers, which probably is because the single ended output is removed, otherwise i dont know.
See pic´s.

put the 680R back, you are following the 8 channel IV values... 2720/4 = ?? with 2 channel you have 4x the current, so need 4x lower resistor to get the same voltage. with stereo and 2720 the output would be erm...slightly hot.

the ESS App notes that are available sometimes use one or the other and either dont mention it, or its a bit vague. to be sure though 680R is correct, not 2720

 

[Suprted1]

PCM2707 also has the same interference shuts down problem, using an adum4160 USB isolator can solve it, but it's usb1.1 only.
CM6631 do not have the shuts down problem even without isolator, and the windows driver is stable, it's much better.

But I like the possibillities with the musiland and 32Bit / 384Khz....
Nice to have them available in such a setup, even if there isnt any audio files available yet.... Not as I´ve seen anyway....

But a nice tip, because if i cant get rid of this shut-down issue. Then it will be replaced.

 

[qusp]

actually you better watch out with that resistor value, I would replace it before switching the dac on, you may kill the dac with output voltage compliance

 

[Suprted1]

actually you better watch out with that resistor value, I would replace it before switching the dac on, you may kill the dac with output voltage compliance

It has already been on, but not connected to anything yet, except scope.

I´ll replace them before any further testing. And do som calculation afterwards, to se what settings that will match my ASP1000 modules the best.

Thanks

 

[qusp]

cool, yeah short periods will probably be OK, but calculate for just under 16mA for a stereo channel. so if you are running the balanced out and dont have the second voltage divider on the second stage BAL-SE convertor opamp to bring the voltage down, then youve still got a pretty damn hot output with 680Ω

 

[Suprted1]

Schematic of dual ES9018 balanced setup.
Did I forget anything ?

Is it okey to use the same crystal for both ?
Is it okey to just connect all data inputs in parallel ?

edit : Added offset circuit which i had forgot.

 

[AAK]

Hi diy mekkern,

For 8 DACs in parallel the 680 on the I/V is to high. 8DACs x 3.9ma/channel = 31.2ma x 680ohms = 21.2Vpeak. In order to get down to a standard 2.0Vrms/2.82peak you need to use 90ohms but it's not practical using your I/V set up, the op-amps will overheat. As far as the clock I would use a clock distribution chip. I would not connect the inputs in parallel without a series resistor followed by a resistor to ground on each input as illustrated in ESS Tech's ES9018 Eval board schematic. Hope the info helps.

Best regards,

Al

 

[Suprted1]

Hi diy mekkern,

For 8 DACs in parallel the 680 on the I/V is to high. 8DACs x 3.9ma/channel = 31.2ma x 680ohms = 21.2Vpeak. In order to get down to a standard 2.0Vrms/2.82peak you need to use 90ohms

Best regards,

Al

The reason it still says 680?? is because i havent calculated anything yet. But also I didnt know they output 3,9mA per channel. Thanks a lot.

Then I might be better off, just to remove the whole output stage, when running all dacs in paralell. And use two resistors to match the voltage to my amplifier.
My power amp has max limits off +/- 12,8V at input.
And a circuit like added picture.
Also this will be a shortwired 1 unit setup with DAC / power amp (and analog inputs in the future).

31,2mA max @ 12,8V equals approx 300ohm. What voltage is the ES9018 able to output ?

No im not going to push the limits. Just find them and figure out a reasonable place to try out afterwards.

Thanks a lot

 

[Suprted1]

Forgot the picture of the power amp input....

 

[Suprted1]

Finally had some time, to undo some mods, and connect anything needed.

Musiland US01 2012 :
I disconnected ground and power from the computer cable, and added a switch to connect ground if needed. This because when gnd is connected, the computer senses the audio card. But afterwards it works fine without it. (usb data dont need gnd reference to transfer data)
Just hold the button a few seconds at startup, and everything is fine.
That also solves some of the lock-down issues with the Musiland DAC. Need further testing to know if it works flawlessly now. But it seems much better on the bench, due to no lock-down when iI turn on/off other equipment. Which it did before.

All ground connections has been connected as supposed to on the DIYINHK board. Only difference is the Balanced output, and that I put some 1117 3,3V regulators directly on the board.

Output signal is measured at +/- 11V at almost max volume on computer. I dont use any pre-amp yet. And therefore dont modify the I/V converter any more yet. (to decrease the output voltage).

AD797 and es9018 easily temp and functionality checked on the bench, and nothing gets hot. AD797 is a littlebit hotter then cold. ES9018 is, I guess about 45-50 degrees.

See pictures for visual details.

I also made a new board in eagle for dual ES9018 chip´s. But its pretty messy.

Time to play, Ill be back.

 

[themystical]

Finally had some time, to undo some mods, and connect anything needed.

Musiland US01 2012 :
I disconnected ground and power from the computer cable, and added a switch to connect ground if needed. This because when gnd is connected, the computer senses the audio card. But afterwards it works fine without it. (usb data dont need gnd reference to transfer data)
Just hold the button a few seconds at startup, and everything is fine.
That also solves some of the lock-down issues with the Musiland DAC. Need further testing to know if it works flawlessly now. But it seems much better on the bench, due to no lock-down when iI turn on/off other equipment. Which it did before.

I have connected a 500 Ohm resistor between incoming USB +5V and 0V, which lets the host computer "see" a quiescent current of around 10mA. These are both galvanically isolated from the "DAC". I do have an "Earth Management" switch which allows the 0V to be connected through via a inductor/resistor arrangement but the set-up has worked flawlessly with galvanic isolation on all the USB DAC's that I have. No need to mess around holding buttons at start-up.
PS Good work on your DIY PCB! I am just waiting for my two diyinhk boards to turn up in the post before I tread in your footsteps to create a dual mono balanced set-up.

 

[Suprted1]

These are both galvanically isolated from the "DAC". I do have an "Earth Management" switch which allows the 0V to be connected through via a inductor/resistor arrangement but the set-up has worked flawlessly with galvanic isolation on all the USB DAC's that I have. No need to mess around holding buttons at start-up.

The computer usually recognizes the board without connecting gnd, but not always... Dont ask me why.

Got any schematic of your galvanic isolation ?
And which components used ?

 

[Suprted1]

Finally the DAC has been running for 30 hours, and it works like a charm.

Enjoyed the whole Yello - Baby album, and it was just about a perfect hour.
Still have a little noise and some glitches. But nothing loud.
Any suggestions to reduce noise/glitches will be greatly appreciated.

I guess this will be better when I replace all capacitors with proper ones, instead of the ones I had around at the moment. Also it works better when connected to my laptop, instead of to my desktop. Which I also experienced before. There is a lot more noise on the USB / gnd on my desktop.....

I´ve finally reached a quality level that makes me pretty satisfied with my stereo.

Complete setup as follows :

PSB Bronze speakers bought used for 350USD.
Musiland 01US 2012 80USD.
DIYINHK empty board, 15USD.
ES9018 chip from ISMOsys, 70USD.
Components from around, 50USD.
PSU from the shelf, 50USD ???
B&O ICE Power ASP1000*2, 600 USD.

A sweet total of 1215USD, and that I believe is hard to beat. For a kit including speakers / DAC / and amps.

All the hours makes a completely different total, but thats the nice thing about hobbies... You dont count that.

If I am in need of explaining the sound, my word would be crispy. And i love it.

It outperforms any setup Ive had before. Like Rotel RSP1068 / Yamaha RX-V 3900 / Denon something amp /Denon DVD 3910 / Rotel RB1092 / and things like that.

I would like to have a run of this setup on my old speakers, Infinity Kappa 90. But they are temporarily stored at my parents place due to WAF :-(.

 

[themystical]

I haven't got a schematic drawn up of the galvanic isolation but the principles are shown in these pictures.
ImageShack® - Online Photo and Video Hosting
I use a switch to switch between the two options and use an inductor instead of the 47 Ohm resistor.