ESS Sabre Reference DAC (8-channel)

because there isnt any gain to get to that level, it is simply converting current to voltage and by producing less than unity gain you have indeed lowered the snr and dnr. the sabre produces about ~3vrms with no gain, to get less you would need to add some sort of attenuation, adding noise in the process.
 
Last edited:
Hello diyers, I have a question for you regarding the output stage of this ess dac and of any dac in general.

I see that most dacs output 2Vrms unbalanced and double the voltage for balanced operation. Having my amplifier an output sensivity of 4dbu=1.227Vrms I would like to have a dac with a balanced output matching this value at 0dbfs, or less.

I recently had a talk with a designer who told me that decreasing the gain of the output stage of the dac would compromise the SNR of the signal. His advice would be not going lower than 1Vrms unbalanced/2Vrms balanced.

I don't get why this should be the case since in my opinion the gain of the output stage should increase the both signal AND SNR, and outputting a lower signal should be always better than outputting a higher one and then attenuating it (in any way).

Hope you can enlight me on this?

Thank you

One thing you could try is attenuating the signal using a digital volume control like Twisted Pear's Volumite. You'd then be able to connect the DAC directly to your power amp(s). I saw a comment by Russ that said the trade-off is worth it in his opinion.
 
because there isnt any gain to get to that level, it is simply converting current to voltage and by producing less than unity gain you have indeed lowered the snr and dnr. the sabre produces about ~3vrms with no gain, to get less you would need to add some sort of attenuation, adding noise in the process.

One thing you could try is attenuating the signal using a digital volume control like Twisted Pear's Volumite. You'd then be able to connect the DAC directly to your power amp(s). I saw a comment by Russ that said the trade-off is worth it in his opinion.

Thank you for your replies. What I wanted to avoid is indeed a high degree of attenuation, being it digital or analog. Having an amplifier input sensivity of +4dbu or 1.227vrms results in a sensivity mismatch when inputting both a 2vrms or worse a 4vrms signal. I'd want maximun level of my source to be <=1.227vrms.

So qusp what are you saying is that the sabre outputs 3vrms with no gain o the output stage..are you sure? I thought the sabre had both current and voltage outputs. Also with 3vrms are you referring to unbalanced or balanced mode?

Thank you
 
yes i'm sure. i'm not sure you quite understand the concept of an iv stage. it doesnt have current and voltage outputs, it only has one set of outputs. the higher the impedance of the following stage, the more voltage and less current it puts out. zero ipedance and it will be all current, high impedance and it will be 100% voltage; the equivalent voltage of the output is 3vrms p2p regardless. its balanced as its a balanced dac, there is no 'unbalanced mode'. so feeding a zero impedance stage and all that current then converted to voltage with no gain, or feeding a higher impedance stage directly, the dac will produce about 3vrms.

AndriyOL: no there isnt and you wont find one either, as even though the nda requirement has been relaxed somewhat (well you can just ask for the datasheet, but you cannot distribute it), there would still be a lot of difficulty running a group buy.
 
AndriyOL: no there isnt and you wont find one either, as even though the nda requirement has been relaxed somewhat (well you can just ask for the datasheet, but you cannot distribute it), there would still be a lot of difficulty running a group buy.

I asked for datasheet of ES9018 and get it without any obligations. I didn't asked why, however I can ask for permission before disclosure.
I also received pricing for bulk quantities of ES9018 as follows:
1 - $65.6
50 - $61.29
100 - $52.44
What do you think guys, shall we organize a Group Buy for ES9018? The price for 100pcs is far attractive!
I guess it should be a person in UK, as the distributor company located in UK.

Can I find at this forum a schematic for 8ch ES9018 with one chip?
Regards.
 
Last edited:
yes, weve been through all of that some time ago, the nda has been relaxed for several months, but you have to ask for the datasheet and be cleared, thats my point, you cannot distribute the datasheet or a schematic that discloses information found in the datasheet, which makes it difficult to run a group buy. people must get it through ess and bulk buys afaik must still be for furnishing a finished product, not individual sales of the chip. they really need to publish the ds to put an end to all this imo
 
Last edited:
no you dont, but the wording on the ds has not changed since it was required, it still states that it is confidential and not to be distributed; they have not posted it for public consumption anywhere, so its still a bit grey imo and i wish they would make it clear. if they wanted it made public they have the mechanism to do that, yet they have chosen not to
 
Last edited:
ESS recommendations?

RayCTech commented in exaU2I thread;
http://www.diyaudio.com/forums/digi...hronous-usb-i2s-interface-20.html#post2512675
If you read the datasheet - ESS recommends 352.8k -> 1536k samplerates to be used with OSF disabled.
If you manages to make a work-around like using a higher clock frequency and thus be able to use the OSF it is up to the designer to solve any issues.

Likewise ESS recommends the "Best Default" setting for the DPLL.
If a product use other DPLL settings it is up to the designer to solve any issues.

I've heard such ESS recommendations for the first time in his post.
Though I've read the datasheet carefully, I failed to find any direct statements corresponding to these.
Are these points well-known to people in this community?

My understanding is ES9018 can accept I2S input of sampling frequency up to 500kHz using the internal OSF.
As for DPLL Bandwidth Parameter, "use the best DPLL Bandwidth settings" is a hardware default. However, I think they do not say explicitly "It's our recommendation.".
 
24bit/32bit switching

Real 32bit I2S output capable transports such as SDTrans and exaU2I become available for DIY users now.
Is the "24/20/16 Bit for Serial Data Modes" setting as "32Bit" essential for playing real 32bit effective I2S signals? When "24Bit" is set, does ES9018 ignore the residual 8bit?

I remember someone said "Even if 24Bit is selected, ES9018 processes I2S data as full-32Bit-effective".

Does anyone compare those settings of "24Bit" and "32Bit" for real 32bit effective music sources?
 
you said it already, if there is a better performing and better sounding chip than the lme49990 i dont know what it is. some others to look into are the lt1007/lt1037, but i doubt they measure up to the national chip, which also sounds superb in the right design. my favorite used to be ad797 followed by opa827. if i were you i would also look into the fully differential chips like ths4131
 
IIR BANDWIDTH setting for PCM

The datasheet tells us,

IIR BANDWIDTH
Normal (for least in-band ripple for PCM data set to Normal)

Does anyone kindly tell me the meaning of "Normal"?

Whenever we play PCM sources, is it better for us to set the parameter to "Normal" though it's hardware default is not "Normal"?
 
Implementing dual mono mode with all 8 channels in parallel

Is such a kit or commercial product available that implements dual mono mode of ES9018 DAC chip with all 8 channels in parallel?

What I want to mean in the term "all 8 channels in parallel" is;
all DAC1, DAC2, DAC3, DAC4, DAC5, DAC6, DAC7, DAC8 pins are tied together for one differential positive output and
all DAC1B, DAC2B, DAC3B, DAC4B, DAC5B, DAC6B, DAC7B, DAC8B pins are tied together for one differential negative output.