When a device is susceptible to RFI the enclosure and/or PCB design is not optimal.

Things can be done about this but takes some experimenting.
You might need some ferrite cores and small value capacitors on strategic places and this should do the trick.
The design itself already has a few anti RFI matters in it.
Are you sure the enclosure is connected to the signal ground ?

Sure am!  The chassis is indeed connected to the signal ground.  I have verified this several times.  The builder is JDS and I emailed him about the buzzing and he's like "Yep, that's pretty normal.  4G radios are high-power."

I have hopes that the newly forming Desktop Amp at DIYaudio which is derived from the O2 (by agdr) will fix some of that. It looks good, so the tinkerer in me might just build one. One thing that I think will bug nwavindividual is that my NuForce Icon HD is superb, solidly insensitive to mobile phone RFI, and almost perfectly channel-balanced on the volume knob as far as my ears can tell.

Things are different out here in the real world.

Did NwAv resurface at all yet?  Shenanigans.

I've realised I don't like the sound of ES9023 in any DAC, I hope the new ES9018-2M lives up to it's namesake.


I'm basically taking a long break from serious audio, but I'll prove the sameness-illusion one day.  It's intuitive fact to me, I think that's where most research starts nonethelesser.

As far as I know, the ES9012 chip has eight DAC's inside of it as well, in essence eight D/A circuits in parallel, however AFAIK there is no picture of the insides of the ES9012 / ES9018 chip to actually verify this, I can't find a direct statement from ESS themselves at this moment either.

If we take a look at the Qualia Indigo http://qualia-highend.com/index.php/products/usb-dac/ it says it's using four ES9012's, so in essence it's then using thirty-two D/A circuits total, sixteen for L and sixteen for R.

This raised a question in me "could the ES9012 / ES9018 chip sound akin to eight ES9023 chips in parallel?"

If we assume the ES9023 and ES9018 both share the exact same hyperstreaming concept, and the same D/A circuit architecture, a notable difference would be the I/V attachment to the ES9023.  As ESS puts it "integrated 2Vrms driver", so pretty much... it's using an extra attachment which could very likely have inherent colour, especially if you're asking an audio enthusiast like myself which thinks even AD797 is inherently coloured.


Returning to parallel DAC design, this do have an important place in audio design which I'll cover now - non-oversampling.

Modern companies which like to focus exclusively on non-oversampling design such as MHDT http://dhost.info/mhdtlab/ are using DAC chips with weaker specs than virtually all the modern oversampling chips.

A way around this dilemma is to parallel the NOS chips, for instance in the TeraDak Chameleon parallels sixteen http://www.teradak.com/products/57.html

However, adherents to the PCM1704 will happily say the Chameleon using 16 x TDA1543 can not rival a well designed 2 x PCM1704, even if the measured specs of the Chameleon appear superior.

In the same fashion, we could hypothesize a 16 x Sabre ESS can not rival a well designed 2 x AK4399, 2 x PCM1794, or similar.

The advantage 16 x has is what I tend to label as "high resolution", it's akin to something like moving from 720P video to 1080P video.

I'm not certain this is the correct label but the presentation is nonetheless vastly improved.

If a complete newcomer to audio systems with a neutral, clear mind is presented with different DAC selections in a speaker system, can they hear them?  If yes, can this person communicate what they're hearing, or only feel / perceive it?

Likewise, if someone looked at a professional computer monitor for the first time with life-like colours, high contrast, high resolution, deep black levels, zero ghosting, zero input latency, fast grey-to-grey, fast hertz rate, in a dark room, they may say... "That looks great"

"What looks great?"

"You know, it looks really nice"

"Nice?"

"Umm, it just looks really clear"

"What about the deeper blacks and vivid blues?"

"Oh!  The blacks are deep!"

"What about the faster G2G and faster hertz rates?"

"Huh?  Where do I see that!"



Now, if we return to sanity for a second, all of this is audio talk is actually complete overkill.  Humans are attracted to higher specs with lots of zeros after them, and get endorphin kicks paying for them with lots of zeros after them.

No one ever locates the spec they are apparently listening to with evidence via controlled blind listening.

In essence, the entire re alm of audio design falls to pieces as soon as you apply the "clearest window achievement" concept.

To quote an esteemed EE on this concept...

"Ultimately, the ES9023 properly implemented, is an audibly transparent DAC. The window glass is completely clean. There are no perceptible flaws. It's not possible to get the glass perceptibly cleaner. While the ES9018 or ES9012 might even better measurements, I honestly don't believe they can improve on the sound quality when listening blind."

Now, since even Sabre ESS with all their advanced technology and deep listening sessions can't even show a single blind listening attempt, sanity highly favours the clearest window achievement concept, CWAC.

If someone really likes the endorphin kicks they get from purchasing and listening to a $50,000 Qualia Indigo, that's their right.  After all, they're in an exclusive position to listen to it, and they may even all avoid shockingly easy blind listening simply to keep the elements of faith and suspense.


That took a while to write.

/kiteki