DIY ES9018 Hi-end USB DAC

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I mistook a calculation...
But at other than BPZ, the noise will reside in analog output amount of |N-32|/32*Vn at N level output of 64 pk-pk level even in infinite CMRR.
Therefore the performance will degrade proportional to RMS swing level.
e.g. at full scale sine wave have the level of 32/sqrt2 = 22.8 in pk-pk 64 level,
then (32-22.8)/32*Vn=0.288*Vn will reside in analog output.
Thus it have more than ten times smaller impact than other opamps...
I will consider finite CMRR case.
 
In the ESS application note, a simple circuit is shown that uses an opamp to clean up the AVCC supply (Fig4.) The way the circuit works it that noise on DVCC is filtered by R4 and C1, while noise on the 12V supply is reduced by the PSRR of the opamp.

It should be easy to implement such a circuit on a perfboard, using an AD797. Using this circuit would leverage the decent 15V supply to produce a clean AVCC. Main problem in actual use is grounding.

This would be a more worthwhile improvement to this system than trying to implement I2S (I think.) It is also in the spirit of DIY audio, no?

no, the AVCC power supply and ground is the return for the IV stage, it needs to be as direct, unimpeded, low noise and low impedance as possible. perfboard wont cut it and I wouldnt trust AD797 running unity gain on perf anyway.
 
shinja, yes thats what I meant, but also for Clock. remember this board is not set up for any AVCC bias, so any output is not swinging around AVCC once it its the output stage. also remember that although the output stage opamps have higher CMRR, you are calculating with perfect cancellation in the output stage, with ideally matched resistors yes? that will certainly not be the case here.

as well, just help/clarification with your English so I make sure I understand. the definition of finite is 'clearly defined', as in not infinite; it does not mean small.

its really not that much trouble to do from the beginning, but i'm not sure it will be so easy here. you could use cheap opamps and a reasonably cheap reference with some passive filtering and get it done for less than $5 in parts for each regulator I reckon
 
Last edited:
ODAC and es9018 review

I got the opportunity to borrow a brand new ODAC and i have spent the evening listening to different setups. I thought i might give you a short review.

The setup is the following:

Sennhieser HD540 reference gold (much better after replacing the faulty cable)

Amps:
Lovely cubewith maybe fake opa627 (Lehmann black cube liner clone)
Topping D1 MKii portable USB DAC and AMP
Bravo V2 modified 'toy' hybrid tube amp

DAC:s
ES9018 with maybe fake opa627
ODAC
Topping D1 MKii portable USB DAC and AMP

Source:
Spotify over USB:eek:

Test music:
Herbie Hancock Maiden voyage
System of a down- Attack
Regina Spector Sailor song

Setup 1:
es9018 and Lovely cube

Good, fast accurate but kind of unmusical. The kind of sound that is hard to criticize because you can´t find a specific where it stands out, good or bad. Tonal balance- very good, nice controlled bass, mid range kind of laid back, almost flat, hi end is detailed without hash. The imaging is ok, kind of narrow.

Bass-4
Mids -4
Highs-4
Imaging -3
Musicality -3
Over all 4

Summary: very good, but lacks musicality.

es9018 and the bravo V2
I haven´t used the Bravo since i got the Lovely cube and repaired the cable of the HD540. What a mistake! This combination shines! The amount of details is not far of the Lovely cube, but with fantastic musicality. The bass is much fuller than the cube, perhaps not as controlled, but very good. The mids are rich, full but still quite true to the recording, fantastic vocals. The highs are surprisingly detailed. The imaging is fantastic, you can pinpoint every instrument with ease. The biggest difference is from the cube is that you get not only the direction but also the distance, a kind of 3d effect.

Summary: My new primary setup. The cube clearly need some tweaking because it should outperform a 30 usd toy with switched power supply!

Bass-3
Mids -4
Highs-3.5
Imaging -4.5
Musicality -4
Over all 4.5

ODAC and Lovely cube
Now it gets interesting, for the first time i can compare and evaluate the dac. Well, now i realize that the 9018 is good, really good. I could not find a single pint there the ODAC outshine the 9018. The bass is a little more in front, but not as controlled. The mids are more or less the same, maybe, maybe a little less detailed. The highs are the big difference The 9018 has a huge amount of details and a kind of smoothness i like. The odac is kind of hash and over bright.

Summary: Surprisingly (really) i prefer the 9018. It is the highs that set it apart.

Bass-3
Mids -3.5
Highs-2.5
Imaging -3.5
Musicality -3
Over all 3.5

ODAC and Bravo V2
As I suspected the 'Tubeness' of the bravo evened the score a bit, but the kind of strange highs are there and the musicality never reached the same level as with the 9018. The same goes for the imaging, a little less 3D effect.

Summary: Surprisingly (really) i prefer the 9018. It is the highs and the imaging that set it apart.

Bass-3
Mids -4
Highs-2.5
Imaging -3.9
Musicality -3,5
Over all 4

The topping d1 MKii
It is a good portable dac, but it is out classed in every aspect, not worth reviewing in this competition.

Aver all:

I do not know how good the ODAC is but i assume that it is a good dac in its price range, about the same money as the es9018 with case and USB card.

I learned that the es9018 is a good dac for the money and my Lovely cube sucks!:mad:
 
I got the opportunity to borrow a brand new ODAC and i have spent the evening listening to different setups. I thought i might give you a short review.

The setup is the following:

Sennhieser HD540 reference gold (much better after replacing the faulty cable)

Amps:
Lovely cubewith maybe fake opa627 (Lehmann black cube liner clone)
Topping D1 MKii portable USB DAC and AMP
Bravo V2 modified 'toy' hybrid tube amp

DAC:s
ES9018 with maybe fake opa627
ODAC
Topping D1 MKii portable USB DAC and AMP

Source:
Spotify over USB:eek:

Test music:
Herbie Hancock Maiden voyage
System of a down- Attack
Regina Spector Sailor song

Setup 1:
es9018 and Lovely cube

Good, fast accurate but kind of unmusical. The kind of sound that is hard to criticize because you can´t find a specific where it stands out, good or bad. Tonal balance- very good, nice controlled bass, mid range kind of laid back, almost flat, hi end is detailed without hash. The imaging is ok, kind of narrow.

Bass-4
Mids -4
Highs-4
Imaging -3
Musicality -3
Over all 4

Summary: very good, but lacks musicality.

es9018 and the bravo V2
I haven´t used the Bravo since i got the Lovely cube and repaired the cable of the HD540. What a mistake! This combination shines! The amount of details is not far of the Lovely cube, but with fantastic musicality. The bass is much fuller than the cube, perhaps not as controlled, but very good. The mids are rich, full but still quite true to the recording, fantastic vocals. The highs are surprisingly detailed. The imaging is fantastic, you can pinpoint every instrument with ease. The biggest difference is from the cube is that you get not only the direction but also the distance, a kind of 3d effect.

Summary: My new primary setup. The cube clearly need some tweaking because it should outperform a 30 usd toy with switched power supply!

Bass-3
Mids -4
Highs-3.5
Imaging -4.5
Musicality -4
Over all 4.5

ODAC and Lovely cube
Now it gets interesting, for the first time i can compare and evaluate the dac. Well, now i realize that the 9018 is good, really good. I could not find a single pint there the ODAC outshine the 9018. The bass is a little more in front, but not as controlled. The mids are more or less the same, maybe, maybe a little less detailed. The highs are the big difference The 9018 has a huge amount of details and a kind of smoothness i like. The odac is kind of hash and over bright.

Summary: Surprisingly (really) i prefer the 9018. It is the highs that set it apart.

Bass-3
Mids -3.5
Highs-2.5
Imaging -3.5
Musicality -3
Over all 3.5

ODAC and Bravo V2
As I suspected the 'Tubeness' of the bravo evened the score a bit, but the kind of strange highs are there and the musicality never reached the same level as with the 9018. The same goes for the imaging, a little less 3D effect.

Summary: Surprisingly (really) i prefer the 9018. It is the highs and the imaging that set it apart.

Bass-3
Mids -4
Highs-2.5
Imaging -3.9
Musicality -3,5
Over all 4

The topping d1 MKii
It is a good portable dac, but it is out classed in every aspect, not worth reviewing in this competition.

Aver all:

I do not know how good the ODAC is but i assume that it is a good dac in its price range, about the same money as the es9018 with case and USB card.

I learned that the es9018 is a good dac for the money and my Lovely cube sucks!:mad:

Thanks for taking the time to write a review of this ES9018. This latest DAC has superior performance in term of technical specs compared to all the DACs you have in your possession. I would expect it(ES9018) to outperform your other DACs. You have concluded that.

I am sure this ES9018 design can be improved on and should consequently sounds better i.e. "will probably improve on its musicality". For what it is worth $160(including all parts, opamps and labour), I think you are getting good value for money if you include say a conservative profit margin of 15%. For an est. cost price of $136 for this item I challenge anyone here to come up with a better solution(implementation of ES9018) hence better sound and still costs less than $136. If you can make one I WILL be the first one to place an order for sure. Otherwise it is all talk.
 
@ttan98
The dac chip is only one of many critical components in a dac design. A high-end chip in crappy design is pointless and probably worse than a crappy dac chip in a high end design. I know that the design and component choice of our 9018 is not what i call high-end. You should not compare this dac with other ES9018 dac, because they will be much more expensive and well designed or cheap designed holding back the dac chip big time. qusp argues the same and he is quite right about this. I expect our es9018 implementation to loose against all or most of the other 9018 designed out there.

The design and execution is more important than the mark and model of the dac chip. The impression of the ODAC is that a lot of thought has gone in to the design and it uses a high quality chip. Therefore i was surprised to find the ODAC not outperforming our dac.

@qusp
If you compare our dac against a good 9018 implementation it will loose and is probably a waste of a good chip, but if you compare it with other dacs the same price category it is excellent value. May be there is some cheap easy tweaks with a big return we could implement to increase the SQ and the value for the money. And have some fun on the way.

It is good enough for 135$ and i do not think that you should throw money at this platform and expect it to shine, but I suspect that this design is hastily put together and there are some relatively cheap simple tweaks to improve the SQ.

One example is my Bravo v2 amp. A cheap crappy amp that transformed to a descent one by replacing the output transistors, cutting and rerouting a via. Modification 5$ and 30 mins. Insane value for the money! There are several other tweaks that can be done, but i stopped there. If i want to spend more time and money i better do that on another better designed platform.
 
shinja, yes thats what I meant, but also for Clock. remember this board is not set up for any AVCC bias, so any output is not swinging around AVCC once it its the output stage. also remember that although the output stage opamps have higher CMRR, you are calculating with perfect cancellation in the output stage, with ideally matched resistors yes? that will certainly not be the case here.

as well, just help/clarification with your English so I make sure I understand. the definition of finite is 'clearly defined', as in not infinite; it does not mean small.

its really not that much trouble to do from the beginning, but i'm not sure it will be so easy here. you could use cheap opamps and a reasonably cheap reference with some passive filtering and get it done for less than $5 in parts for each regulator I reckon
Thank you, qusp!
AVCC bias means the 1/2AVCC bias at noninverting input of IV opamp? Ah, I forgot it.

yes yes,I know the CMRR and the cancellation will be limited by resister matching, but ignored to make my first attempt simpler.
I intended the "finite" to be such as 40dB of CMRR : 1% of gain mismatch due to 1% tolerance resister.
 
The board is set for AVCC bias, as shown in this schematic of the I/V stages.
 

Attachments

  • BiasIV0a.jpg
    BiasIV0a.jpg
    24.9 KB · Views: 955
Positive and negative output pin of one differential output have noise from AVCC
each N/64*Vn and (64-N)/64*Vn
where N is the output level of positive pin and 0≦N≦64, Vn is AVCC's noise voltage.

At differential to single summation ,when their in-phase component is put as N/64*Vn ,thus differential component is |(64-N)/64*Vn-N/64*Vn|.
The in-phase component will canceled by CMRR, therefore output noise is
N/64*Vn/CMRR+|(64-N)/64*Vn-N/64*Vn|
arranging this,
{N/64/CMRR+|1-N/32|}*Vn

for example
Full scale range sine wave have the level of 64/2/sqrt2 = 22.8 in pk-pk 64 level,
Adding 32level offset, this signal is equal to 54.8 rms level.
(revising previous post)
When CMRR is put as 40dB,
54.6/64/100*Vn+(54.6/32-1)*Vn = 0.715*Vn
 
Whilst I can't comment on a lot of what has been discussed here, I also have one of these recent 9018 DAC boards from Weiliang. As a deviation from standard, mine has the TXCO oscillator and funky looking 9710/9720 transistor op-amps for which I paid an additional $78.

I am using it with a standalone CM6631A USB -> s/p-dif converter with onboard 220v transformer also from the same supplier. I opted for this over the onboard USB module as it doesn't use the host USB power. With the aid of these two boards and an old gold tipped video rca cable I can play 192kHz files from a Raspberry Pi straight into my hifi.

The sound of this DAC really is very good indeed. It was an obvious and noticeable upgrade from my Cambridge Audio DACMagic. Depth, clarity, separation, warmth, punch, detail, excitement - everything improved in a very obvious way although this took a little while to get used to - it almost sounds analogue in comparison and I was mistaking the extra content for distortion at first. I don't have experience of a large number of DACs, but did hear a Cyrus DAC XP recently which comes in at nearly 20 times the price of this board, and would struggle to compete let alone justify it's price tag.

The one problem that I am having is that of audible artefacts. I don't know if this is caused by the DAC, the USB board or even the cable but playback isn't completely smooth. It almost sounds likes dust on a piece of vinyl. I'm a little disappointed as it's quite noticeable, especially with classical music.
 
Whilst I can't comment on a lot of what has been discussed here, I also have one of these recent 9018 DAC boards from Weiliang. As a deviation from standard, mine has the TXCO oscillator and funky looking 9710/9720 transistor op-amps for which I paid an additional $78.

I am using it with a standalone CM6631A USB -> s/p-dif converter with onboard 220v transformer also from the same supplier. I opted for this over the onboard USB module as it doesn't use the host USB power. With the aid of these two boards and an old gold tipped video rca cable I can play 192kHz files from a Raspberry Pi straight into my hifi.

The sound of this DAC really is very good indeed. It was an obvious and noticeable upgrade from my Cambridge Audio DACMagic. Depth, clarity, separation, warmth, punch, detail, excitement - everything improved in a very obvious way although this took a little while to get used to - it almost sounds analogue in comparison and I was mistaking the extra content for distortion at first. I don't have experience of a large number of DACs, but did hear a Cyrus DAC XP recently which comes in at nearly 20 times the price of this board, and would struggle to compete let alone justify it's price tag.

The one problem that I am having is that of audible artefacts. I don't know if this is caused by the DAC, the USB board or even the cable but playback isn't completely smooth. It almost sounds likes dust on a piece of vinyl. I'm a little disappointed as it's quite noticeable, especially with classical music.
 
ok i made a first attempt to get i2s yesterday with no success. Connections were towards diyinhk xmos board. The connections from the dac pins to xmos were the following:
pin 59 DATA1 = LRCK
pin 58 DATA2= DATA
pin 60 DATA_CLK=BCK
pin 51 gnd= gnd
Also to note that the dac by default has data3-data5 shorted. I didnt fiddle with that and left as it was.
it was a real pain to even make the connections so a bit disappointed not to get anything out of it, but happy that i didnt bricked the dac in the process :)
The xmos now is not recognised by the pc which is something experienced by this guy as well until he modded the xmos grounds: XMOS 384Khz Asynchronous USB to I2S | myl8test
i guess that if i disconnect it from the es9018 it will get recognised normally as before.

any help would be much appreciated, as i am just learning how these things work

Looking at the PCB traces on my version -
DATA_CLK - No connection.
DATA1->SPDIF.
DATA2 to DATA8 connected together by PCB tracks.

Looks just about possible to cut the tracks (But don't blame me if you trash it :eek:).
 
Does this mean it might be possible to get an I2S signal into this board? I have a WaveIO now and am still using spdif as I emailed Weiliang and he said that I2S wasn't supported.

Thats fairly understandable, as the USB card is an upgrade and this is probably primarily designed to be a traditional standalone dac using optical/spdif inputs.

Also as an update to my post above, the noise I was experiencing was nothing to do with the DAC, but related to disk buffer settings on the streamer. This thing now sounds very smooth and detailed with no imperfections to distract from the music.
 
Last edited:
Does this mean it might be possible to get an I2S signal into this board? I have a WaveIO now and am still using spdif as I emailed Weiliang and he said that I2S wasn't supported.

Thats fairly understandable, as the USB card is an upgrade and this is probably primarily designed to be a traditional standalone dac using optical/spdif inputs.

Theoretically, yes.
But it will involve some very fine surgery on the PCB tracks.
Do you have a good magnifier to see the tracks I am talking about ?

I also appears that I2C control would be easy to implement as
SDLA and SCL have resistors connected with nice size pads.
 
Yes I can see them, but considering the board with upgraded opamps etc cost me over £150 I'm not sure I want to risk doing this as my first modding project.

However if you can get it working and also post come clear enough photos of the mods I will give it a try.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.