[wushuliu]

Surprised this has not been mentioned already. Fascinating presentation at RMAF by an ESS engineer on the basic hows and whys of their DACs and the history of sigma delta dacs. Sure is nice to hear a man of his background admit it was hard for him at first to believe people could hear differences. And justification for those of us who never quite embraced sigmadelta dacs.

Although he fulfills his promise of avoiding any equations (kind of funny watching him struggle to do so), things do get pretty technical by the end.
 

http://www.youtube.com/watch?feature=player_embedded&v=1CkyrDIGzOE

[wushuliu]

Wow, for once here is info straight from the/a horse's mouth with millions of dollars in resources, decades of experience, measurements out the wazoo, back history and all - and no comments?

Interesting.

[JohnR]

I can't sit still and watch online videos, I feel like I'm wasting time when I can scan a web page or an article in a minute and get the key points, and print it for later reading at the coffee shop if it needs more.

Can you summarize the "justification for those of us who never quite embraced sigmadelta dacs"?

[richidoo]

Thanks for posting this! I missed RMAF this year but I wanted to see this lecture.
I have a DAC using ESS9018 and I prefer it over other DACs I have compared it to.

Hopefully I can find an hour to watch it soon.

[wushuliu]

I can't sit still and watch online videos, I feel like I'm wasting time when I can scan a web page or an article in a minute and get the key points, and print it for later reading at the coffee shop if it needs more.

Can you summarize the "justification for those of us who never quite embraced sigmadelta dacs"?

Ack, but then whatever I write will make me the official spokesperson and the opposing rhetoric will focus on my words and not the Mallinson's talk. Grrr.

Synopsis: Mallinson has worked on SD  modulators since the 70s, now an engineer for ESS. He outlines the history of SD, going back to 11th century. Then the history of ESS, which has been around since the 80s but only recently entering the hi-fidelity market.  He then goes into how SD works in DACs. He notes that since they've been making SD dacs (not just the new chips but for a long time now in consumer electronics) they would hear some people note the sound quality concerns, but ignored them because the measurements were beyond audibility. Eventually ESS began doing their own experiments and tests and found that indeed there are people who can discern SD DACs every time. He then goes into why they believe that's the case, including the possibility we can hear far past 20khz in some fashion, as well as discerning 120db noise floor, amongst other reasons. Despite his and the other engineers' initial misgivings. He then explains how ESS sought to overcome these issues, which is frankly kind of over my head and I won't even bother to go into except to say they found a way to get even lower noise floor and reduced distortion. This has led to the current chips, of which the 9018 is the flagship. BTW, all this is as of 2005 (the hi-fi dacs). ESS uses audiophile feedback and testing to assist in developing the chips. They've even 'trained' engineers to listen for the same things audiophiles do.

It's a pretty dense presentation, albeit relatively easy enough to follow for a layman like myself. And it's graph-heavy so I really can't break it down much further with just text. But it is very, very interesting and Mallinson does a great job avoiding going into the math by using analogies. He is also genuinely excited by the yet unanswered questions of why some people can hear beyond what the measurements indicate and admits he's not sure why that's the case.

He also mentions why ESS keeps a low online profile (piracy), but gives an email for anyone wanting to get more information on what he's discussing and recommends direct contact w/ the company. You def. get the sense that the hi-fidelity products are just one part of a much larger company.

[wushuliu]

Thanks for posting this! I missed RMAF this year but I wanted to see this lecture.
I have a DAC using ESS9018 and I prefer it over other DACs I have compared it to.

Hopefully I can find an hour to watch it soon.

Please watch it. He's very good and I learned a lot. And again it's nice to hear from 'behind the curtain' instead of endless speculation and debates.

[DaveC113]

Thanks for posting. I haven't watched it yet but I will when I have time.

I think it's great an engineer can actually believe people can hear things that aren't (yet) measured or understood. I'm a ME myself and I have to say many engineers don't have open minds and think their schooling taught them all they need to know about everything. Those that think that way are really limiting themselves via their belief systems.

[stevenkelby]

Thanks for the link. Probably the most talk on audio I've ever seen, can't believe some people, who are into audio, can't find an hour to watch it...

The info presented has impact on half the threads here regarding cables, digital vs analogue, active Vs passive, measurements vs ears etc. etc.

Very educational and worth watching, thanks again:)

[JohnR]

Thanks wushu for the summary. Looks like the sniping has started already in this thread, at least it's not at you!

[JohnR]

BTW looks like these are the slides:

http://www.esstech.com/pdf/noise-shaping-sigma-delta.pdf

[Rclark]

That was a fantastic lecture. A, now I must have a Sabre Dac! lol, and B) interesting stuff, coincides with what Bruno Putzeys was saying last year in the Ncore thread on DIYaudio, that there is not yet a complete measurement suite for audio and that we don't know everything. Anyway, looks like interesting times lie ahead for digital gear in general.

... I think I might have a Sigma Delta type system in my cdp. The lecture left me really wanting to try a Sabre dac.


edit: This is relevent too, I recall a paragraph in which Bruno said he was developing a dac and his goal is to exceed the ESS chips.

[*Scotty*]

They are all Sigma Delta DACs now days , with the exception of the non-oversampling DACs.
Scotty

[medium jim]

Very interesting and telling how much they relied on the human ear over measurements.

Jim

[*Scotty*]

However, I did run across an interesting DAC recently, the PCM 5100,5101,5102 series. Accepts up to 32bit 384kHz input and out of band noise up to 300kHz is suppressed better to better than -100dB. There FFT plots showing noise between 0Hz and 20Khz at -160dB down.
Here is a link to the PDF doc
http://www.ti.com/lit/gpn/pcm5102
Junilabs? is offering kits using this DAC.
http://www.junilabs.com/en/kits/kits.html
Scotty

[medium jim]

... I think I might have a Sigma Delta type system in my cdp.

Your CDP has a:
emotiva.com/erc2.shtmAdvanced Analog Devices AD1955 DAC with 384X oversampling filter.

Jim

[Rclark]

Very interesting and telling how much they relied on the human ear over measurements.

Jim

Of course, the lecture was about listening tests of equipment, not room measurements.

[Rclark]

Your CDP has a:
emotiva.com/erc2.shtmAdvanced Analog Devices AD1955 DAC with 384X oversampling filter.

Jim

Went to Emo and looked further, it has Delta-Sigma processing. As good as it sounds it's time to try another dac, maybe a used Eastern Electric or something.

So far I've only heard 3 variations of this same dac and nothing really different in my system.

[*Scotty*]

What was missing from the slide show is a graph of the performance of a PCM 1704K operating on 24bit 96Khz input. It might have looked closer to the ideal than the 16bit DAC. Also missing was an extension of the graph out to 300Khz.
Within the audio band the Sabre DAC looks cleaner than the PCM5102 by 50dB.
It might sound similar or better in some ways than a PCM 1704K based on the in-band noise levels. I would try it in a non-upsampling mode using the current out pin. Asynchronous up-sampling never sounds right to me.
Scotty

[medium jim]

What was missing from the slide show is a graph of the performance of a PCM 1704K operating on 24bit 96Khz input. It might have looked closer to the ideal than the 16bit DAC. Also missing was missing an extension of the graph out to 300Khz.
Within the audio band the Sabre DAC looks cleaner than the PCM5102 by 50dB.
It might sound similar or better in some ways than a PCM 1704K based on the in-band noise levels. I would try it in a non-upsampling mode using the current out pin. Asynchronous up-sampling never sounds right to me.
Scotty

Scotty:

Non-upsampling is clearly the best mode.

Jim

[stevenkelby]

Looks like the sniping has started already in this thread, at least it's not at you!

I assume you're referring to me? Apologies if so!