View Full Version : Slow construction of ESS9018 dac
Some time ago a colleague from my work gave me a fried populated ESS9018 DIYINHK pcb. Well, he has already removed the dac itself.
I was quite intimidated by the prospect of soldering an expensive chip with so tiny legs. I considered buying another board with ESS9018K2M.
I mentioned the project to my father, who suggested he could ask the people from his former lab whether they would solder the chip for me.
I ordered the chip and it got soldered in. The quartz has been replaced as well.
Unfortunately, the dac would not work. I pretty quickly purchased the Bus Pirate and was pretty relieved that the chip was not totally dead. It actually talked to the bus pirate. I asked at diyaudio. People suggested I should check the contacts between the pins and vias. Again easier said than done.
After more session with the bus pirate I concluded that the sampling frequency was 0. People at diyaudio suggested that there was no incoming signal. At that stage I started poking the legs of the chip with my meter. I actually attached some sewing pins to my meter's probes. Using the magnifying glass I could check for contacts between the pins and via. Indeed two of I2S pins showed no contact. With hammering heart I resoldered them (I cannot really say I have much smd soldering experience).
The blessed thing started playing music!
I also got (on a loan) the usb->i2s board and psu's from the guy who gave me the fried dac board.
The PSU boards were from DIYINHK. The most important voltage for the this dac is 3.3V AVCC (analog part of the chip). The DIYINHK claims 0.8uV noise in their PSUs. I was not convinced that such a low noise figure is soooo important here. I had to figure it out myself. I was aware that LM317 will not do. So my first attempt was ADP151 preregulated with LM317. ADP151 seems pretty decent (on paper) and certainly better than AMS117 widely used as a 3.3v reulator. I cobled a PSU with two LM317. One provides 5V. This will be used from the Arduino controller. The other provides some voltage which is further egulated down to 3.3v.
I compared this PSU with the DIYINHK. The sound of my PSU was pretty flat and uninvolving.
So I started looking around.
sq225917
14-01-2019, 23:41
Nice work on saving the board. I'd be looking at splitting the rails for the clock, avcc and the opamps. The opamp look like they have good local decoupling already.
Hi sq! I will continue, hopefully tonight. As for the PSUs your contribution will be mentioned ;-)
Nice work on saving the board. I'd be looking at splitting the rails for the clock, avcc and the opamps. The opamp look like they have good local decoupling already.
Any of the setups I have tested used separate +/-12V for the opamps and a separate 3.3V for the digital part of the chip and another 3.3V for the analogue part. The pcb provides a possibility to split the later (AVCC) into the left and right channel supplies. This apparently improves the sound even more.
Coming back to my experiments. As mentioned above I was not happy with the sound of adp151 regulated 3.3v AVCC. I attacked the problem in two ways. I decided to daisy chain opamp regulator (with LM49710) after the 3.3v regulator (APD151 provide V ref for the LM49710). Opamp regulation seems to be recommended by ESS themselves. Even if it does not provide LT3045-like 0.8uV noise benchmark. The other line of attack was to make sure that I have my own LT3045 supply. Sq225917 kindly offered me on diyaudio some left over pcbs from stammheim. These are dual boards which can take up to 3 regulator chips per channel be making them good up to 2x1.5A. I populated them for now with one LT per channel, but I figured out I will better off with these PCBs because I will be able to add more regulators if they prove to be good sounding. In the attached pics you can see stammheim's boards revealing my humble smd soldering skills. As for the opamp regulator, I piggy backed it on the board with two adp151. ADPs are soldered on the adapters (underneath, what you see on top are in and out caps). 3.3V DVCC is meant to be taken from one ADP, 3.3V AVCC after the opamp. I know I am open to critique about my grounding here...
A better shot of my humble opamp regulator.
The is really a slow thread. I got distracted by puting RPi with Allo Boss into a (reasonably) nice box. I am also pretty happy with the sound and functionality of the Pi so Es9018 has been put aside.
Still, I guess it would be nice to finish it, especially that I get tempted by some DACs in the Private Exhibitions here! So finishing this project might save me some money as well!
I have two possible donors for a chassis. A Sony Cd player (with TDA1543) and a Samsung TV box. Samsung is a bit smaller which suits me very well. The only issue is my 20x4 LCD display does not fit there.
I will have to look for something smaller maybe an OLED display.
Anyway, it will be nice to put the DAC to service. Even without extra functionality provided by the Arduino with the LCD display. This is what I managed to do today.
27001
27002
27003
27004
Even more. The white PCB with tall capacitors is the dual 3.3V PS (one for digital circuits; one for analogue) the for the 9018. The PCB in the last photo is the +-15V opamps PS.
I got the dac operational today. The pictures will follow.
At the moment it works in the hardware mode, ie it jumps into some default state. I cannot set up filters etc. It is configured to accept usb.
When arduino comes, or rather I have a display that fits in my enclosure I should be able to have volume control, set up filters. I will try to add an spdif input. But I am running ahead of myself. Maybe I will stop here as there is so much music available through usb...
The nice thing would be to feed i2s signal from the RPi, but this apparently does not work with different sampling rates and bit depths.
Btw I had a wonderful evening (still more to come, but now with headphones using Pelerano's Zen head amp) listening to NPR Tiny desk concerts with Hillary Hahn and Yo Yo Ma and many others...
The promised pictures.
As you can see I installed my humble home mage PS for the DAC chip. It provides 2x5v using LM317 preregulators, one 5V is unused for now. The other one is regulated to 2x3.3V using adp151 and one 3.3V is regulated with LM49710 as recommended by ESS.
I still can't decide whether I prefer this 3.3V supply or a dual LT3045.
The chassis is predrilled to accept both now.
Yet more.
I see that I miss a X-mas tree picture with all LED on (there are more LEDs on when everything is hooked up and running!)
I listened yesterday to more tracks.
The DAC sounds really good.
I have noticed that the opamps regulators are pretty hot.
Will increase the output voltage (now +-13V). I will go to 15-16V. I should probably order some small heatsinks. Unfortunately I can accommodate something like 20-25K/W. Maybe I should screw them to the bottom plate and solder from underneath.
Ordered 1.3" OLD display. Shipping from London so should be able to run experiments soon. If I can master OLED display (as opposed to 20x4 LCD used in HiFiDuino) I will probably order a bigger screen
The other exciting piece of news is that the project got attention of Burson Audio!
I will be getting two V6 Classic to try them out in the buffer duties (OPA134 at the moment).
So stay tuned! ( I feel almost like a blogger ;-) )
SS-Audio
27-11-2019, 22:56
Hi Greg.
I'm confident the V6 will add a nice touch to your project.
Cheers!
The OLED display has not arrived yet.
So I've cut small heat sinks today.
This is mainly for the op amps power supply as they heat up very much. I have also increased the output voltage of the op amps PS. I do not see any reason to keep it as low as 12V as indicated on the PCB.
I used to regulate down to +-13V, now went up to almost +-16V. This will also keep the regulators cooler.
The 3.3V PSU got the same heat sinks, why not?
1.3'' oled has not arrived yet, but I still ordered 2.42''.
Programming these cannot be rocket science!
Still no 1.3" OLED in my mailbox. Apparently London where I ordered it from is not the one I live in. Probably somewhere in a big Asian country which name starts with Ch.
Lawrence001
07-12-2019, 09:19
Still no 1.3" OLED in my mailbox. Apparently London where I ordered it from is not the one I live in. Probably somewhere in a big Asian country which name starts with Ch.I ordered my daughter some fruit shaped rubbers nearly 2 weeks ago from Manchester, they haven't arrived yet. Strangely when I got the PayPal receipt, the seller's details were in a Chinese script. Make of that what you will...
walpurgis
07-12-2019, 09:54
I ordered my daughter some fruit shaped rubbers nearly 2 weeks ago from Manchester, they haven't arrived yet. Strangely when I got the PayPal receipt, the seller's details were in a Chinese script. Make of that what you will...
Many Chinese sellers have stock held in the UK or EU. You deal online with them back in China, then the UK part of the business often run by friends or family send out the goods.
Lawrence001
07-12-2019, 16:12
Many Chinese sellers have stock held in the UK or EU. You deal online with them back in China, then the UK part of the business often run by friends or family send out the goods.Yes but then you'd expect to get them within a week...
I have been playing today with the power supply for the ESS chip.
In post #5 you can see a picture of a dual LT3045 power supply (PCB courtesy Sq225917 here). Top regulator chip nowadays 0.8uV/sqrt Hz. The PCB also from a respectable source.
So this this one vs my humble made LM317 -> ADP151 ->AD49710 (mind that ESS advises to use op amp regulator in their documentation).
I think I here more details using LT3045. But it sounds more HiFi than when using my own multi stage regulator. Mine is somewhat warmer, more analogue if you like. Johas Kaufmann and Anna Netrebko in O soave fanciulla are so dramatic! LT3045 is very good as well but I miss this passion!
The only drawback in my PSU is more pronounced sibilance I believe. Which sounds strange as I said I hear more details using 3045 which is also brighter. So I should be hearing more sss using LT3045 not the other one.
Strange. The only thing which come to my mind is that my PS oscillates somewhere. Not sure whether I have enough know-how to tackle this. So any advise is more than welcome.
But for now I think mine stays in the chassis not the 3045.
Also I discovered a wiring mistake in the connections of the +/-16V PSU. The overheating problem solved. I am too embarrassed to give you more details...
Meanwhile the smaller OLED screen has arrived from Far East London. I already got the money back as it was due a long time ago.
When it arrived I felt a bit bad about it showing up and started thinking how to make it up to the seller.
Until I hooked it up to arduino. No response. So I hooked it up to my Bus Pirate to read the i2c address. No luck.
The I found an Arduino i2c scanner.
Shall I ask the seller to reimburse the wasted time?
Shall I mention that I tested another screen with the scanner and it nicely reports its address.
Meanwhile another (bigger: 2.42'') OLED screen has arrived. This is the one I would like to use with my DAC.
I hooked it up to Arduino to see it working. I just tried an example program from the web to see how it looks like.
Not too bad. I will post pics when I have somewhat more customised stuff to show.
Anyway looks good. It will be a challenge to cut a nice display window in my enclosure.
I will also need to change available on the web programs to control ess9018 to accommodate my screen. So still a lot of work.
Ah! One more thing!
The Santa from Burson Audio has provided me with the tracking number for the V6 op amps!
I guess I should get some better op amps than OPA134 which I use for buffers to make the comparison more fair! At the moment only the buffers will be replaced with Burson Audio V6.
Btw I use LME49710 as IV converters.
Happy New Year!
A number of things happened in the last few days.
First of all the Burson Amps have arrived.
Unfortunately/fortunately they are duals and I need singles for this project.
So unfortunately I will have to wait a bit more to try (singles) in the DAC.
Fortunately I can use duals in my Whammy headphone amplifier (N.Pass/W.Colburn). I am very impressed with the Whammy now. I will write up my impressions soon somewhere else.
So for now just some teaser pictures.
I am also very curious what Burson does to an old TDA1541 CD player (already upgraded the opamps from NE5532 to LM4562 and teh output caps to Elna Silmic II).
I anticipation of Burson Audio amps I decided to try some other op amp buffers instead of OPA134. I picked OPA37.
I was very impressed. The difference is hard to describe as the DAC is already pretty good with OPA134. It is just more real, more natural.
Given the experience with Burson in Whammy my expectations are very high now!
Another thing is trying to understand the OLED screen.
I replaced Arduino UNO (which is pretty big) with Nano (small and cheap).
As I mentioned before I managed to run some test programs (see pics).
The problem I encountered is that the program does not run when the arduino is powered from an usb charger ie not conncted to the computer (it requires a manual reset to start properly).
That is a touch one. An audio forum is probably not the best place to discuss this (unless anybody is interested). The issue seems to be the initialisation sequence required by the display and arduino. There are some hints/solutions on the net. I haven't found a nice one. One (not very elegant: a capacitor between RST and GND which increases the time RST is low ie gives more time for things to initialise properly) will work for me (it requires disconnecting the cap when you want to upload a different program for the arduino).
A bit off topic here. I have written up experience with Burson Audio V6 Classic in my Whammy headphone amplifier. You can find it here:
https://www.diyaudio.com/forums/pass-labs/317803-whammy-pass-diy-headphone-amp-guide-191.html
Post #1908
Meanwhile I started adapting the real hifiduino script to work with my setup.
It turns out that Arduino Nano is too small to support graphics libs for my screen and a largish script.
For now I switched to ascii libs. You can still use different font size so I will be all right. For now at least.
Looks like I have mastered the Arduino and the OLED screen well enough to fit it in the chassis. For now I will also try to use the controls available in the chassis instead of a rotary switch. Saves me a lot of mechanical work!
Lawrence001
29-01-2020, 21:38
I've just bought this dac and the design (at least the output stage with 6 op amps) looks quite similar to yours.
https://www.ebay.co.uk/itm/274226725245
Interesting! Looks good!
Certainly cheaper than mine ;-) I feel like an idiot spending all that time - but it is part of the fun I guess.
As far as I can see it really looks good. Good power supply (probably the most important part in any dac - there are so many examples of diy dacs using humble dac chips but where a lot of attention has been given to the power supply).
Op amps are mounted on sockets so you can upgrade them (probably will be worth doing).
Finally the box is much nicer than mine!
In terms of my own project, I got distracted recently. Mainly because of the Burson V6 playing now in my Whammy head amp.
This led to many hours of headphone listening using the dac described here but also to my vintage Philips TDA1541 CD player with DIY Pass D1 IV converter - very analogue sound! I guess when the dac is finished I will have to build a balanced D1 IV stage for it.
Being distracted by the headphone listening I ordered PCBs for yet another head phone amp... Pioneer Super Linear by EUVL from diyaudio.
Lawrence001
30-01-2020, 22:29
In terms of my own project, I got distracted recently. Mainly because of the Burson V6 playing now in my Whammy head amp.
This led to many hours of headphone listening using the dac described here but also to my vintage Philips TDA1541 CD player with DIY Pass D1 IV converter - very analogue sound! I guess when the dac is finished I will have to build a balanced D1 IV stage for it.
Being distracted by the headphone listening I ordered PCBs for yet another head phone amp... Pioneer Super Linear by EUVL from diyaudio.I thought you got the boards free and used an old case? It's nice to build something yourself anyway I wish I could!
I think I got a good deal. It sounded a bit bass light with the USB direct so I used my JKenny USB-S/PDIF convertor and it sounds really good now :)
I like the 2 filter option (fast or slow) but I'm probably going to use the fast roll off one.
Fortunately I have 6x OPA234s from another dac I already upgraded that cost me £150 so no extra expense there :)
I thought you got the boards free and used an old case?
Fortunately I have 6x OPA234s from another dac I already upgraded that cost me £150 so no extra expense there :)
Yes, the board was free but without the dac chip. That alone is £55 for a private customer.
And then all the bits and pieces. I also bought the BusPirate (£30) to debug this particular board.
As for the op amps: The annoying thing is that you need 6 singles while 3 doubles would be almost half the price.
Having singles is probably marginally better in terms of separation.
I would prefer to spend the same amount of money on 3 doubles than 6 singles. I am pretty sure the result would be better.
Yesterday I have done some dreaded mechanical work related to the screen. It looks actually better in real life than on those pictures. Maybe some flakes of polyester need to be sanded off.
The lines are straight.
The actual screen is sandwiched between the front plate and the old pcb and fixed with 3 distance screws. 2 screws are glued to the front plate, 1 to the pcb.
Got distracted by tinkering with my Whammy and Zen headphone amplifiers. So not much progress here (except the the present DAC is used daily in combination with the aforementioned amps).
Ah! Settled for Lt3045 preregulated by LM317 for the AVCC. I guess the diff compared to my previous experiments with similar (and more potent) LT3045 is that a small PCB with lt3042 is mounted directly on the DAC AVCC input. I guess distance matters!
Still working on the window for the Led screen. Already on the third version. This one is passable. I did not take pics of the failed approaches. If I am happy with the present one I will make some pics tomorrow.
Otherwise listening to the dac pretty often.
As promised some pics of the screen window.
The first attempt was just a simple rectangular window. I could not figure out how to glue it nicely.
So the second one was a rectangular piece of transparent plastic but, I routed a small lip around it. The idea was that I will glue it from the outside and the lip will rest on the front panel.
I used a speaker surround glue which dries transparent. Still it did not dry nicely and I did not like the result.
Notice my great gluing weight!
Not liking the previous result I went for a simple solution. An aluminium frame which covers the joint between the front panel and the window.
The transparent plastic has been glued to the alu frame from the inside.
One pic shows the result before cleaning the excess of the glue.
You can see a blob/stain on the screen - this is the protective foil which I haven't removed yet.
Finally some progress! The other projects got stalled as a bag pnp transistor is at home, the bag of npn transistors arrived at my work. But I work from home now given circumstances and it is difficult to match pnp and npn which are in different locations.
So here's the control unit for the display and the DAC chip.
Almost ready! Still misses the IR diode socket. All the other connectors are there: the power supply, push buttons (can support plenty of them), display and the dac.
The code is based on HiFiDuino but adapted to my display.
I have decided not to use the rotary encoder at this stage but hijack the buttons available in my chassic.
It saves me some mechanical work and maybe will look better.
Should pressing buttons be too cumbersome I can always install the rotary encoder.
On the other hand I do not expect to use the push buttons much.
IR controller should be more practical
It is taking shape!
I got the arduino program working reasonably well.
The push buttons which were in the chassis are not working great (I believe they are too noisy). I tried different ways of debouncing but could not achieve nice feel to it.
So maybe a rotary encoder after all.
Some pics attached. The protective plastic on the screen not removed yet.
After some tinkering with the arduino code I achieved satisfactory results.
The volume can be controlled using the rotary encoder (no knob yet!)
Works like a charm! Very happy with it.
The encoder allows to navigate the menu and choose different digital filters and change other properties of the on-board DSP of the DAC.
I think I am almost there!
The remaining work:
1. Still some more work is needed on the code to get a bit nicer layout of the info on the screen.
2. Remote control (should be easy peasy)
3. Add SPDIF input(s) (maybe Toslink as well) - that will be a steep learning curve
4. Buy an audiophile approved knob for the encoder.
5. Put a lid on the box!
The DAC has been in normal service for some time now.
The volume control (hardware ie not loosing bits of the actual digital material) allows me to skip my preamp (Pass B1 clone) which is actually audible in the end result.
But the news is that finally, amid Covid-19 situation, I received Burson Audio V6 Classics for the DAC.
Today I installed 2 op amps in the buffer positions and started initial listening tests.
Yummy!
I need to listen to them more. I will try them out in the IV converters positions as well and report.
I promise pictures in the next post as well!
Finally I have spent a week or so with Burson V6 Classic in the buffer positions.
In this DAC I used OPA132 as buffers (I was not really happy with them) then OP37 (I wanted to try OP27 but could not get it) which I actually liked a lot.
To be honest the I kind of knew already what to expect given my experiments with dual channel Burson V6 in my Whammy headphone amp (where it replaced AD823).
It is not about deeper bass or more treble or more (or less) of whatever. Burson gives natural, organic, fluid, relaxed presentation of music. Maybe the thing worth mentioning is that I still had some sibilance with OP37 (less than with OPA134). This was gone with Burson. Very appreciated!
You know by now that I have received Bursons for free from Burson Audio in return for some pictures of them in my creations and my honest opinion.
So mind I am not not obliged to say they are good. But I am completely sold.
Now I have a chance to listen to OPA627 in my Whammy headphone amp (and quite a few other good opamps). OPA627 stood out from the crowd. I have found it very good just a bit less than Burson. So I thought I have found a holy grail of opamps. Well until I checked the price of OPA627. RS Components had them for some £40 a pop. Well dual Burson V6 is around £60. I think IT IS better. £40 is already silly money for an opamp, so if I considered spending it I would probably spend £20 more just not to think about it anymore. But actually this £20 extra pays back in the sound as well.
As a bonus:
Today the full monty.
Extra 4 Bursons V6 replaced LME49710 in I/V positions.
The diff is small, but everything gets yet more relaxed, the soundstage gets somewhat deeper. The timber of instruments gets yet more natural.
The saxophone in Hanna Paulsberg's Catalan boy is very nice. Instruments in the background are better placed in the stage. The best way to describe this is that they are not part of the wall of sound.
Katie Melua's Fields of gold are absolutely seductive...
I have a pretty damn good DAC!
What remains is to redesign the whole I/V conversion based on opamps and use some discrete I/V converter. To be honest I am not in a hurry.
What will certainly happen soon is that the opamps will get a better power supply.
Powered by vBulletin® Version 4.2.3 Copyright © 2026 vBulletin Solutions, Inc. All rights reserved.