ESS Sabre Reference DAC (8-channel)

Recently in Japan, some diy people has been trying a DSD input for ES9018 based DACs, for example, Buffalo II, Fidelix CAPRICE. Their DSD sources are USB Audio interface board developed by ElectArt or diy-tapped commercial SACD players. ...
I'd like to report a tip for playing DSD on ES9018.

A Japanese Buffalo II user played DIFF files downloaded from "1 bit consortium" web site. (Their original file format was "WSD" and the format was converted into "DIFF" using a free software "AudioGate" developed and distributed by Korg.) He played the files using "USB Audio Interface board" developed and distributed by ElectrArt. AUDIO? FPGA?? (In Japanese)
For some files, his plain Buffalo II was unable to play them. Bunpei also reproduced the problem by himself.
Bunpei tweaked some register settings of ES9018 and found that;
the setting of "Automatically detect SPDIF input" causes the problem and
selecting "Must manually select SPDIF input" yields a normal play of those files.
 
there is also some nice software by weiss that will convert these files, but also the new HDMI board for the ackodac will play them natively, i'm looking forward to this release, as I have always enjoyed SACD and DVDA, but never got around to getting a modded player together. good tip on the korg software, I think I even have that somewhere as I planned on getting one of their portable dsd players at one point for my portable rig
 
I'd like to report a tip for playing DSD on ES9018.

A Japanese Buffalo II user played DIFF files downloaded from "1 bit consortium" web site. (Their original file format was "WSD" and the format was converted into "DIFF" using a free software "AudioGate" developed and distributed by Korg.) He played the files using "USB Audio Interface board" developed and distributed by ElectrArt. AUDIO? FPGA?? (In Japanese)
For some files, his plain Buffalo II was unable to play them. Bunpei also reproduced the problem by himself.
Bunpei tweaked some register settings of ES9018 and found that;
the setting of "Automatically detect SPDIF input" causes the problem and
selecting "Must manually select SPDIF input" yields a normal play of those files.

Good to know. Thanks. I am implementing an auto detection mode by manually switching between spdif and i2s/dsd and detecting if there is signal lock.

In auto spdif it probably thinks it is spdif. DSD is 2.8224 MHz which is the same as spdif for 44.1K
 
Last edited:
Hi RayCtech
I also esperienced the same glitches in turning on lamps, etc. The Buffalo II unlucks for a fraction of a second, stopping the sound temporarely. I never pointed where the problem originates. I have a Mux, BuffaloII, IVYIII and Placid power supply. They are all TP products as you can see.
I am folowing your experiences with great interest since I find the situation anoying.

I investigated this more closely as I previously only assumed the mains noise/spikes trigged the LOCK / UNLOCK issue.

I have previously also believed that this problem and the DPLL settings issue was related in some way as both issues have been improved by improving the power supplies etc..

I have now confirmed that the mains noise/spikes triggers the UNLOCK /LOCK.
The next step I will do are to check if there are any correlation between the different DPLL bandwidth settings and the triggering of the UNLOCK / LOCK by the mains noise/spikes.

There should/could be many out there that experiences these UNLOCK / LOCK by the mains noise/spikes??
For me this issue was also the most annoying....
 
I investigated this more closely as I previously only assumed the mains noise/spikes trigged the LOCK / UNLOCK issue.

I have previously also believed that this problem and the DPLL settings issue was related in some way as both issues have been improved by improving the power supplies etc..

I have now confirmed that the mains noise/spikes triggers the UNLOCK /LOCK.
The next step I will do are to check if there are any correlation between the different DPLL bandwidth settings and the triggering of the UNLOCK / LOCK by the mains noise/spikes.

There should/could be many out there that experiences these UNLOCK / LOCK by the mains noise/spikes??
For me this issue was also the most annoying....

Thank you RayCtech for your comments. I tought the Placids could solve everything, but apparently not. I suspect that some off the problems could come from ground connections with the different equipments (ground bounce or alike). There are a lot of power supplies there. Unfortunately, I have no facility to check this theory.
Anyway, I read TPA is about to release a new power supply for the Buffalo, I do not know exactly what they are about, but it sould be a big improvement, as I understand. I hope they have considered solving the unlock problem if cure is coming from there.
I know of one person who has the same problems with Buffalo, unlocking.
I will be happy to read your developments on this subject.
 
Thank you RayCtech for your comments. I tought the Placids could solve everything, but apparently not. I suspect that some off the problems could come from ground connections with the different equipments (ground bounce or alike). There are a lot of power supplies there. Unfortunately, I have no facility to check this theory.
Anyway, I read TPA is about to release a new power supply for the Buffalo, I do not know exactly what they are about, but it sould be a big improvement, as I understand. I hope they have considered solving the unlock problem if cure is coming from there.
I know of one person who has the same problems with Buffalo, unlocking.
I will be happy to read your developments on this subject.

Actually Ray is using his own DAC.

Our experience is that in the cases where this has occurred for a few people the problem was resolved with better EMI shielding for their project and/or wire routing.

Of course it could be that increasing the DPLL bandwidth could indeed help. I really wouldn't know because I have not been experiencing any issues in this regard.

I am working on a new power supply, but the motivation to do so has nothing to do with this. :)

Cheers!
Russ
 
Hi Peter,

The bit lenght for the coefficients is 32, the inter MAC unit is 64 bits. Each of the 8 channels can have its own unique coefficients. ThIs is ideal for making an active Xover that will not have an phase problems. (well other than the speakers phase responce itself). We are almost done with a new GUI that can be used if you have a demo board that actaully has a graphical filter builder built into it. Soon..



Dustin

Hi Dustin,

Is this available for people that want to program the filters into the DAC?
Thanks...
 
Is this available for people that want to program the filters into the DAC?
I suppose the GUI-based filter design software might be "Filter Designer" included in "Sabre32 GUI" software. When you buy ESS ES9018 Evaluation Board, you can obtain the benefit of those programs. Moreover, if you tweak the board and add external I2C extensions, you can connect your own ES9018 chips. The "Filter Designer" seems very powerful and convenient.
 
In slow rolloff mode you have 63taps with completely custom coefficients. You can implement any coefficients in this mode. (asymetrical, phase warped - to eliminate "pre ringing"....) Downside is there is only 64 taps. This is a 4x oversmapling filter. The first coefficient MUST be 0, the rest are the 63 programmable ones.

In fast roll-off mode you have a two stage interpolated FIR filter.

There was supposed to be an app note coming on how to use them, but I finally just broke down and figured it out for myself. No simple task, but very doable. :)

The datasheet example actually gives almost all the info you need.

In any case, after a several attempts and emails to Dustin and others I determined that the filter was never intended for lowish audio frequencies (say less than 5Khz). It really won't make for a great active crossover. That is unfortunate as it was something I was really looking forward to doing.

Still it is quite possible to do all sorts of cool low pass filters. That is really the intent. For filters with fs/2 corners you will have superb results.

I hope that helps answer some of the question.

Cheers!
Russ
 
Hi,

I'm looking to build high-end cost effective 8-ch DAC/DSP combo, and thought of ESS as a possible alternative to four separate DACs (probably PCM1798) at DSP's output.
Do the ESS have remarkable performance on it's outputs in 8-channel configuration? Or it is good in 2-ch mode only (it's dual-differential thing, by using 4 channels to create one channel output).

Thanks.
 
ES9018 in 8ch mode

The ES9018 is a true 8 channel chip. It "may" be used in the 2 ch mode with better DNR performance. Here is a guideline of we have achieved in the lab.

8CH mode
THD+N = -120dB across entire audio band.
DNR = -129dB

2CH mode
THD+N = -120dB across entire audio band.
DNR = -135dB


Dustin
 
Hi,

I'm looking to build high-end cost effective 8-ch DAC/DSP combo, and thought of ESS as a possible alternative to four separate DACs (probably PCM1798) at DSP's output.
Do the ESS have remarkable performance on it's outputs in 8-channel configuration? Or it is good in 2-ch mode only (it's dual-differential thing, by using 4 channels to create one channel output).

Thanks.

The ESS option may not seem quite so cost effective when you consider that you need the I/V stage as well. Depends on your budget I suppose, but I've recently bought 3x Opus for my 6 channel DSP based system.

I'm curious, what DSP are you using?
 
The DSP is going to be TAS3108. It's pretty easy in programming, as all you need is to burn the EEPROM with a firmware generated by PurePath studio.
In addition, you can change biquads using external MCU via I2C.
3x Opus is... too much :) And pretty overkill... Specially for the subwoofer/woofer channels :)
The I/V stage is pricey, each opamp costs around 8-12$, and you need at least 8 of them for the I/V, and a bit cheaper opamps for filter.
Still, it's well over 50$ i'm looking for as a generic solutuion :) Maybe the WM8740's, they are cheap and nice, but i prefer the current output DACs.

In addition, the ESS chip isn't that DIY/HiFi friendly when used in 8-ch configuration - you get looong signal paths from the IC to I/V converters. On the other side the power supply for the DAC could be nice and clean... Many pros and cons to review :)
 
Member
Joined 2007
Paid Member
The ESS option may not seem quite so cost effective when you consider that you need the I/V stage as well. Depends on your budget I suppose, but I've recently bought 3x Opus for my 6 channel DSP based system.

I think implementing DAC filters in the ESS chip per se don't really justify the cost of the ESS-based solutions in a multi-channel rig. There are other solutions.

For several years I used 4X Opus in a multichannel rig with great success. However, the hobby causes one to itch... So I can also attest to the fact that a stereo Buffalo II board sounds nicer when subbed-in for one Opus to provide the 'main L-R' mids/highs. The Legato is a delight. It sounds great as is but also invites and benefits from select capacitor tweaks (that will cost more than the $60 orig. kit). Adding in just one BuffaloII/Legato took my system to another level partly by increasing midrange dynamics over what Opus outputs. Naturally, YMMV...