Gustard X16 Balanced MQA DAC Review

[Bob from Florida]

I don't care what Amir measure . this chip should follow it's datasheet , it means 32bit should have 194.4dB SNR, so this another marketing lie

I think you are confusing the capability of converting 32 bit PCM data with Equivalent Number Of Bits. If you had a 32 bit PCM file to convert to analog, this chip can do that task. However, that does not mean the output will result in 32 ENOB. That number is a function of SINAD which is 20 times Log of (signal divided by (noise + harmonic distortion)). Low signal reduces ENOB as does high noise or distortion. Assuming the DAC puts out sufficient signal levels - 2 volts unbalanced and 4 volts balanced - the biggest problem is noise and distortion. The real world has both.

 

[Geert]

Amir , This Gustard has ES9068A , this chip has 32bit , your measurement show 21 bit dynamic , and you say it is superb , I don't get it !! with 32 bit you should measure 194.4dB .

The '32 bits' refers to the sample width it supports as digital input, and used for internal processing. It doesn't mean the chip delivers 32 bit resolution end to end (at the analog output). This is what the chips data sheet specifies:
+130dBA SNR
+126dBA DNR
-120dB THD+N

Amir measured 0.000085 dB THD+N. That's -121.4 dB, so better than the data sheet specs.

 

[JohnYang1997]

I mean SNR , 24bit system should have noise floor 144dB , 32bit should have noise floor 194dB, with normalize SNR, and for your knowledge I have M.Sc. EE degree with 33 years of HW design in RADAR , 5G , work in past from Ericsson AB Sweden , RUAG Space , Emerson

It supports 32bit data input. That's it. This has always been the case in audio converters. Real world SNR, DNR, SINAD performance metrics are independently spec'd. Either you are trolling or you have been living under the rock for the past 20 years.

 

[Belker]

Disclaimer: Ericsson AB do not back user Vahid’s statements in this matter.

 

[JohnYang1997]

Yes, BUT with big BUT , when they say in their datasheet it take 32bit PCM , my question is now, how that A/D convert that from beginning the analog signal and convert it to 32bit, what was the requirement for noise floor in that A/D converter ?( what ever it is !) ,

The same is here. As long you can output 32bit PCM format. This is the convention in the audio converters. However how many bits are not very important. The quoted specs are often SNR and THD+N directly.

 

[Geert]

if they claim that this DAC take 32bit PCM , it means from studio the A/D should have noise floor -194dB , which IT IS VERY HARD TO GET , so they claim only lie .

Since when do data sheet specifications of a chip take a full system into account? And how would they know the specs of every application someone can come up with?

 

[JohnYang1997]

SYSTEM NOISE FLOOR deiced at end how many bit you can get of your DAC or A/D , ( ENOB in other term) if you have great 24bit DAC but noisy power supply , ( Jitter and other stuff ) you can say Goodbye to your 24bit . here I mean if they claim that this DAC take 32bit PCM , it means from studio the A/D should have noise floor -194dB , which IT IS VERY HARD TO GET , so they claim only lie .

Difference between
A: 32Bit DAC
B: 32Bit input DAC

 

[JohnYang1997]

Datasheet will not say anything except defining the requirement of noise floor , Again I mean here if they claim that if this chip take 32bit PCM , so it means some how there is A/D converter in studio that generate 32bit PCM, am I right ?!
NOW my big question : what is noise floor (SNR) for that A/D converter in STUDIO , I mean SNR not SINAD . We can exclude distortion for now.

It doesn't matter what noise floor is. There are ADCs can support 32Bit PCM output stream. Then this is all it needs.

 

[Geert]

Datasheet will not say anything except defining the requirement of noise floor , Again I mean here if they claim that if this chip take 32bit PCM , so it means some how there is A/D converter in studio that generate 32bit PCM, am I right ?!
NOW my big question : what is noise floor (SNR) for that A/D converter in STUDIO , I mean SNR not SINAD . We can exclude distortion for now.

Fact 1. The ES9068A accepts 32 bit PCM. This has been verified.
Fact 2. It has been proven there are ADC's which produce 32 bit PCM files
Fact 3. ESS provides clear analog performance specifications in the data sheets, which have been verified.

Conclusion: there's nothing wrong with specifying the ES9068A supports the 32 bit PCM format. The format exists, and its practical limitations have been specified.

 

[Veri]

Datasheet will not say anything except defining the requirement of noise floor , Again I mean here if they claim that if this chip take 32bit PCM , so it means some how there is A/D converter in studio that generate 32bit PCM, am I right ?!

32-bit is a digital format. Often an intermediary before dithering. Very common for digital processing.
How are you not getting this.

The actual signal/noise specifications can be found very clearly in the datasheet. Never is there a claim of 190dB SINAD. Not in this world

 

[JohnYang1997]

NOISE FLOOR is everything , how do you want achieve otherwise , if we make simple calculation for 32bit DAC , 2^32 = 4294967296 , with voltage refence say 2.5V such as REF6125 Texas Instrument ( which is still not good enough ) , you have LSB around 0.58nV !! . so your noise floor should be lower then this , 1nV is -180dB .
I can tell you with this level of system noise floor , you can never never get 32bit PCM. chip manufacture clime a lot of things but they don't tell you how to achieve it , only marketing claim .

You are still mixing up format and actual performance.

 

[JohnYang1997]

do you have fully complete datasheet , or you just flat-earth guy , if you don't have datasheet or verified measurement then this forum is wrong for you , measurement and verification , that's it , show me for example what kind voltage reference noise they have ?!

Voltage reference is provided externally. The noise performance is limited by the dithering in the modulator and the output impedance of the output. You can calculate the output impedance by the output current and output voltage swing. Dither value is specified in the register description. The real world performance fulfills the SNR/DNR/THD+N performance spec in the datasheet.

 

[Geert]

do you have fully complete datasheet , or you just flat-earth guy , if you don't have datasheet or verified measurement then this forum is wrong for you , measurement and verification , that's it , show me for example what kind voltage reference noise they have ?!

I already presented the specs I copied and pasted from the ES9068A data sheet, which is available online, more than an hour ago. These specs are in line with what Amir measured.

And I think you'll soon find out you're the one that doesn't belong here. Who do you think you are, busting in and offending other people?

 

[JohnYang1997]

then what is reason you buy high RES , now we get new definition , "format" !!! very irony , read RED BOOK first to know how CD format works!

I don't care about format. You shouldn't too. You are the one pointing the 32bit like no tomorrow. Read the actual spec.
Plus don't pull CD format in this. It's irrelevant.

 

[JohnYang1997]

which chip is reference voltage then , do you have datasheet for it ?! you should back your comment with datasheet , I want really know what kind of reference voltage they use to get 1nV noise floor !?

You design the reference voltage. The one used in reference design of 9068AS is opa1611/opa1612.

 

[JohnYang1997]

AGIN , you should back your comment with fully datasheet , the only datasheet which is avabaible on their home page is 5 page briefly version , Again they claim some thing without actual measurement , I include it here their datasheet . FACT is everything in engineering , claim is not FACT ,

Why would I share confidential information that can get myself into trouble? Nice bait won't bite.

 

[Geert]

AGIN , you should back your comment with fully datasheet , the only datasheet which is avabaible on their home page is 5 page briefly version , Again they claim some thing without actual measurement.

Than you back up your statement the chip doesn't perform as specified (refering to the analog output specs in the data sheet) and they are lying? Where are your measurements?

 

[AdamG]

Stop feeding the Trolls!!!

 

[Veri]

Agreed. Fun's over guys

in any case. Specs are correct, even exceeded. Sign of very good design, as any reasonable person would acknowledge.

 

[Raindog123]

The noise (and noise floor) of a device is caused by more than one factor and is a combination of those. Some factors are internal - thermal noise, quantum noise… while some are external - eg, noise in input signal, reference voltage stability… The bit depth of the input signal (24bit, 32bit) supports [potential] lowering of the input noise level. However, it does not affect/reduce other, internal noise sources that might become dominating and define the noise floor of a DAC.