DAC ESS9023 - decent standard measurements, very poor IMD - does that make sense?

[truscellino]

Hi everyone,
I am a keen music listener, interested but not a specialist of audio electronics and I was delighted to discover this forum. Very informative and enlightening. I am learning a lot. Thank you all.

Recently I put together a DIY raspberry pi based audio player with an ESS9023 I2S DAC. I know it is a bit dated now but I am interested in the process of assessing the design. I was curious to see audio measurements and came across this post on another web site:
DAC comparison WM8741 vs ESS9023

Basically, the ESS9023 looks better on standard measurements (THD+N etc.) but IMD is terrible, and according to the author, this shows in listening tests. The author makes conclusions about DAC design criteria and measurements. Does that make sense to anyone?

 

[solderdude]

It may well be caused by an incorrect analog filter behind the ESS.
See: IMD hump.

Some claim the IMD hump is audible others say it doesn't. In any case it appears it becomes an issue when post filtering isn't done correctly.
So possibly the PCB with the 9023 on it is poorly designed.

Often it is NOT the DAC chip but the implementation and components/circuit around it as well as the used digital filter settings in some cases.

does not apply... see post below.

 

[Bamboszek]

ESS9023 doesn't really use analog output filters based at opamps as other DACs does.
It was cheapest chip from Sabre family with build-in charge pump and 2Vrms output stage from single 3.3V supply.
Original ODAC was based at this chip, later switched to PCM5102A with Rev B.

 

[solderdude]

Ah indeed.
Only a 4.7nF to ground.

 

[ZolaIII]

Put an graphite shielding (2~3 mm) over the DAC IC. First you need to apply some electric isolation.

 

[Veri]

Put an graphite shielding (2~3 mm) over the DAC IC. First you need to apply some electric isolation.

Why?

 

[ZolaIII]

Why?

Because it fixes all the things mentioned.

 

[JohnYang1997]

Because it fixes all the things mentioned.

Measurements?

 

[truscellino]

Interesting, thanks all. The tests described in that post are for cheap DIY-type DAC boards. Probably poorly designed (I don't have skills to judge here), not tested and using components of variable quality...
In terms of "how to rate DAC performance", this still raises a question. Maybe the ESS9023 tested would have been not catastrophic on a SINAD chart, but still a very deceptive product. The author raises the point "It seems very contradictory that the ESS9023, which performs flawlessly on sine waves, falls over with a test signal like this. I’m afraid it is perfectly understandable that if an engineer is asked to design to specifications such as THD, SNR and cost, those are the only places that will receive attention and optimisation, often with some very clever tricks to shine under those measurement conditions."
Amir mentions IMD in the description of measurements method but still the overall DAC ranking doesn't account for IMD flaws... or I might be missing something?

 

[ZolaIII]

Measurements?

If you read this little saga the "standard" measurements where never a problem. I don't have equipment to do it & neither you do. You either can try it & judge afterwards as really it's simple to try out or just call me crazy, it suits me anyway.

 

[JohnYang1997]

If you read this little saga the "standard" measurements where never a problem. I don't have equipment to do it & neither you do. You either can try it & judge afterwards as really it's simple to try out or just call me crazy, it suits me anyway.

I have access to APx555B. so. I only talk measurements.

 

[ZolaIII]

I have access to APx555B. so. I only talk measurements.

@solderdude mentioned actual EMI lab for conducting measurements. I think all modern ESS DAC IC's are prone in picking up what ever they are. You probably have much better equipment & some easy accessible ESS designs laying around to put this to simple test.

Best regards.