ESS Sabre Reference DAC (8-channel)

But in "less-critical" locations their bang/buck is unbeatable since they need no I/V. I'm not qualified to debate, but I can repeat what I've read from those who are that 'quality of implementation is critical'... Cheers!

If you need only I2S input (max 192k/24bit) the ES9022 have nearly the audio performance of the ES9018 and I/V included.
The output voltage swings around GND / 0 volt at it also have an internal charge pump for a -3.3 volt - the ES9022 needs only one +3.3 volt..
It will even drive low ohm headphones without any problems..

EDIT: ESS have changed the name to ES9023?? - but the link on the ESS web does not work...
 
Last edited:
UNLOCK issues

I have investigated the DPLL UNLOCK issue further and I found some interesting informations:

When playing 44.1k/16bit using the SDTrans192 SD card player I can normally use the lowest DPLL setting..

But since I have had some few UNLOCKS and I investigated this and found that all albums that played OK I have ripped by myself as bitperfect and plays without UNLOCK problems.

Some albums that I have received as FLAC (I have converted to wav with MAX) or as burned copies gives UNLOCKS.
Some UNLOCKS are repeating at the same position and some are spurious.

The status are then that the albums I have ripped bitperfect to wav in 44.1k/16bit plays with the lowest DPLL setting without any UNLOCKS....

And that albums I have received in FLAC and converted to wav or albums I have received as burned copies may / can give UNLOCKS...

I will conduct one additional test on some CDs I have experienced problems with in some CD players. Some CDs have some kind of copy protection that triggered similar dropouts at consistent positions on my Nakamichi MB1000 CD player. At the time I believed the CDs to be defective and the shop replaced the CDs, but it was the same dropouts in the same positions..
On my Accuphase DP75 the same CDs played without dropouts..

I will rip some of these CDs and test in the SDTrans192.

With SPDIF there are error correction.
With I2S there may not be any working error correction and I have confirmed that "bad" data gives UNLOCKS!

The Nakamichi MD1000 CD player (1991) was possibly to "old" to understand the copy protection coding and it resulted in audio dropouts.
I expect this could be the issue when SDTrans192 or other similar products are used with I2S as they might not have any working error correction mechanism...
 
If you need only I2S input (max 192k/24bit) the ES9022 have nearly the audio performance of the ES9018 and I/V included.
The output voltage swings around GND / 0 volt at it also have an internal charge pump for a -3.3 volt - the ES9022 needs only one +3.3 volt..
It will even drive low ohm headphones without any problems..


EDIT: ESS have changed the name to ES9023?? - but the link on the ESS web does not work...

indeed its a nice chip, much nicer than PCM1798 IMO, , it will drive lowZ headphones, but the lack of volume control is a bit of a deal breaker
 
indeed its a nice chip, much nicer than PCM1798 IMO, , it will drive lowZ headphones, but the lack of volume control is a bit of a deal breaker

You can use a series resistor and a optocoupler as a shunt to ground.
The optocoupler can be controlled with a potentiometer or a Arduino PWM output (256 levels).

I used two ES9022 (one inverted) to make balanced outputs and then the shunt optocoupler are connected between the series resistors and not to ground.

I tested the ES9022 with up to 80MHz clocks, but 50MHz was the optimum and with higher frequencies it get noisier...

A properly setup with two ES9022 chips for balanced performance with optocoupler shunt volume outperformed a Stock BuffaloII in respect of audio quality :eek:
And due to this I used the ES9022 DAC for several months before I started to work with the Sabre32 prototypes.
The performance of the Sabre32 was clearly better, but the ES9022 may have the best performance vs price...
 
yeah of course, you can use whatever Volume control you like, but I find all analogue controls have issues and so you dont see them anywhere in my system. still sounds like a nice little dac you built. sounds similar to one I have seen EUVL build.

no arguments with price vs performance though, its amazing for the money you can build a 9022 dac for considering so much internally. normally I dislike voltage out dacs, I find tghe internal opamps usually ruin it, but not this one.

i'm actually interested in the 9016 for a workshop dac. i'm choosing between 9022 and 9016 for the bass of my 3 ways and may just go with 9022 because it could fairly easily be built into the same enclosure making active 'satellite' digital input subs. 9016 would be easier to interface with the master MCU I have planned for the XO (to get 4th order)
 
Member
Joined 2007
Paid Member
Help on I2S quality?

To feed the ESS 9018 I lift I2S from a rack-mount computer sound board. This worked really well for Wolfson DAC chips but is proving trickier with the 9018. Even with the largest available sample latency, I still sometimes hear a click or pop in the output - only on a few tracks, not on most. However, I have no problem with 'lock' - that is continuous. The I2S is transferred by fine copper wires about 8 cm in length soldered into vias of the sound board. I'm hoping for suggestions that might improve the I2S signal. For example, might it help to disconnect the input pins of the original DAC chips still on the board so as to lessen the signal load? Which would be superior for lifting the signal, stranded or solid wire and might silver be beneficial? Should these wires be paired and twisted, and what would be the ideal positioning of I2S input wires relative to 5V. DC power lines? [A 90 degree orientation, I assume?] Any and all suggestions will be much appreciated. My ultimate goal is to be able to decrease the sample latency in the computer.

Thanks in advance,

Frank
 
The ES9018 has it best performance really only in stereo or mono mode. Though it is still certainly quite good in 8 channel mode.

I would recommend reading up on the chip and getting the datasheet. That should answer your questions.

Cheers!
Russ

If ES9018 is operated in 2 channel mode, does that makes it comparable to ES9012 performance wise?
 
To feed the ESS 9018 I lift I2S from a rack-mount computer sound board. This worked really well for Wolfson DAC chips but is proving trickier with the 9018. Even with the largest available sample latency, I still sometimes hear a click or pop in the output - only on a few tracks, not on most. However, I have no problem with 'lock' - that is continuous. The I2S is transferred by fine copper wires about 8 cm in length soldered into vias of the sound board. I'm hoping for suggestions that might improve the I2S signal. For example, might it help to disconnect the input pins of the original DAC chips still on the board so as to lessen the signal load? Which would be superior for lifting the signal, stranded or solid wire and might silver be beneficial? Should these wires be paired and twisted, and what would be the ideal positioning of I2S input wires relative to 5V. DC power lines? [A 90 degree orientation, I assume?] Any and all suggestions will be much appreciated. My ultimate goal is to be able to decrease the sample latency in the computer.

Thanks in advance,

Frank

A couple of thoughts. a) make sure you have small value resistors before the I2S wires. This will reduce slew rate and thus radiated RF. b) use miniature coax cables to also reduce radiated and received RF.
 
Member
Joined 2007
Paid Member
Thanks Spartacus! ...will see what I can do - the vias are somewhat spread-out. For the shielding of the coax, any ideas on the best source of ground? [This board is outside the computer chassis and I have large clip-on RF filters on the source cables...]

Frank
 
Last edited:
My understanding is that I2S is limited to distribution to very short distances, and the quality of the signal is not as much affected with the quality of the wires but with the length. Is it possible to use transmitters to send this kind of signal on the longer distance, like from one case to another case without loss?
 
The 9018 will have very slightly better performance potential due to limitations of the internal routing of the outputs.

Thank you Brian. That sounds great. I am thinking on starting a multichannel project for ultimate active stereo 4 way system, most likely after some DSP. Still searching what that will be software / hardware wise. Any recomenndations are welcome, but not miniDSP quality level. With that in mind, than I could go with ESS 9018 all the way and maybe go for mid - high in two channel configuration and for low - sub use 4 channel configuration. If I am going in this I would like to keep the highest quality, but without making it irrational price wise. This is great help if 9018 is on par with 9012, makes life much easier as well as makes it easy for possible future changes.

What is happening with clocks in the set up like this. Does that means that each DAC would need to have it's own local clock set to be slave and one clock at DSP that will be master for the whole system. I am not sure about this since all that I know is that clock has to be close to DAC and that matters the most?

Thanks!
 
Last edited:
Member
Joined 2007
Paid Member
Do you not already have a ground connection?

I'm running into a Buffalo II. On second thought, the particular vias for this channel are close together, so it's not such a problem to transfer the ground in the shields D1, D2, and DCK and then re-join them into the ground input. The other channels, w/ different DACs, don't suffer the same problem...

Thanks
 
Last edited:
I'm running into a Buffalo II. On second thought, the particular vias for this channel are close together, so it's not such a problem to transfer the ground in the shields D1, D2, and DCK and then re-join them into the ground input. The other channels, w/ different DACs, don't suffer the same problem...

Thanks

You're welcome. At the least, try connecting a ground wire between the Buffalo and the soundcard, keeping it close to the I2S lines. That itself might be enough to solve your problem. If the ground is floating between the two, it's a wonder it's working at all.
 
If the ground is floating between the two, it's a wonder it's working at all.
I did this by accident once - connected I2S without a ground wire & surprisingly, it worked & seemed to not suffer any sonic degradation. I wondered how this could be possible & figured that one or some of the I2S signal wires were also acting as a ground. I aboandoned it as bad engineering but wondered none the less how it worked!
 
Member
Joined 2007
Paid Member
That is what we call "luck"

...of which I need all I can get! ...but the ground IS connected - by a single wire rather than distributed as a shield.

I'm willing to accept that the ES9018 simply needs more time to complete it's work than does the WM8741. But for whatever reason, running with the lowest possible latency always seems to sound better. The BII needs 1024, the Opus is fine with 256 (from the exact same I2S wires).

Frank
 
Last edited: