Interface : USB or NOT

[renaudagnes]

Imagine you are designing the ultimate DAC in the box.

Everything is here: finest converters, all the connectors you need (SPDIF, glass, plastic XLR...), clocks from Guido and all and all.

Yeah ! But there is a question you're unable to answer: do you need (or not) an USB in ?

You're thinking USB is a not required because :
- USB is by definition poor quality : who need the ultimate DAC with such a poor quality source ?
- you need to add something in you design that means less quality package, a sort of compromise to add the interface you're not a great fan

But in other hand you know many people use their computers like a drive with thousands records on their HDD.

So what to do ? USBing or not ?

[analog_sa]

Originally posted by renaudagnes
USB is by definition poor quality

Are you considering a commercial development? If i was in the market for a ready made dac i wouldn't consider anything without USB support. Maybe because i have no hi-rez sources and all i'm interesed in is 44/16.

The out-of-the-box USB solutions are really not that great and i doubt the addition of a single 270x chip would do USB much justice, unless you reclock.

Apparently async USB is a completely different animal and not "by definition poor quality"

[renaudagnes]

Thanks analog for your response.

I am considering the commercial development of a little product.

I want explain to the technical team WE NEED USB ! But the technical team considers computer is devil and USB are the daemons

I use my computer since many years to store a lot of records, and the more I look, the more I see people dematerializing their records.

Today, I've seen B&W and Peter Gabriel are starting a new service with the ability to download records (err, one per month) in looseless format (APE I think) here :

http://www.bowers-wilkins.com/display.aspx?infid=777

And so many radios from the web, etc...

I would appreciate positive returns for USB, it will serve the cause!

[falis]

I have a Glow Audio One (tube) amp with a USB input and an internal DAC. It sounds great, and I really appreciate having that feature in the amp. I know, French engineers are perfectionists - I've been working with them for 25 years. Your guys probably call themselves scientists instead of engineers like ours always have. Don't rely on them for marketing decisions. ;-)

The result of the USB input may not be the best, but a lot of people will appreciate its convenience. "Don't let the best be the enemy of the good"!

Best wishes in your venture!

[Eric Juaneda]

Originally posted by analog_sa
The out-of-the-box USB solutions are really not that great and i doubt the addition of a single 270x chip would do USB much justice, unless you reclock.

Apparently async USB is a completely different animal and not "by definition poor quality"

There is a discussion on implementing PCM2704 here : USB to SPDIF converter

Here is some specification from datasheets
SLES081E-JUNE 2003-REVISED NOVEMBER 2007

The PCM2704/5/6/7 accept 16 bits data audio at 32, 44.1 et 48 kHz sampling frequency.

USB interface
All data to/from the PCM2704/5/6/7 are transferred at full speed.

The PCM2704/5/6/7 employs SpAct™ architecture, TI's unique system that recovers the audio clock from USB packet data. On-chip analog PLLs with SpAct enable playback with low clock jitter.

Page 26
When receiving the audio data, the PCM2704/5/6/7 stores the first audio packet, which contains 1 ms of audio data, into the internal storage buffer. The PCM2704/5/6/7 starts playing the audio data after detecting the next subsequent start-of-frame (SOF) packet.

The PC send data to PCM2704 so PLL is needed. There is no ASRC (Asynchronous Sample Rate Converter) in the device. Keeping 3 ms of audio data permit to keep jitter lower than continuous tracking. PCM170X is not sync device nor real async device!

Eric

[analog_sa]

Originally posted by Eric Juaneda

The PC send data to PCM2704 so PLL is needed. There is no ASRC (Asynchronous Sample Rate Converter) in the device. Keeping 3 ms of audio data permit to keep jitter lower than continuous tracking. PCM170X is not sync device nor real async device!

Eric

I didn't quite understand your post. The 270x works in isochronous mode with a PLL. It's intolerably fussy towards everything and invites $300 usb cables to work well. I only like it if followed by an ASRC, otherwise the jitter is unacceptable.

Sadly, due to diy difficulties and laziness i have no direct experience with chips working in async mode. Hopefully, positive results are more easily achievable with those.

[renaudagnes]

Beware guys, techs are arriving

Eric, maybe you can explain us a bit more (a bit more than the specs) ?

Thanks again for your valuable posts !

[Eric Juaneda]

Hi analog_sa,

I never implement PCM2704 too, I decrypts datasheets and confuse post previously linked.

I link picture from page 26 PCM270X datasheets. It explain data transfer between PC and IC. PC don't send continuous audio data stream to IC. It send 1ms audio packet at max USB 1.0 speed.
Synchronizing PLL on continuous data stream could generate high and low frequency jitter. With 1 ms buffer you can smooth (or cancel) high frequency jitter. [Is dealing only with low frequency jitter inside audio band is better than high frequency jitter out the audio band?]
Jitter level can be 'dilute' by buffer size: big buffer large dilution, low buffer small (or no) dilution.

I agree with you, device quality is SpAct PLL quality design.

Eric

[jmar]

USB is poor quality? Rather vague statement.

Why?

Is there a tradeoff between the WAV format and USB or something?

What do you mean?

[Eric Juaneda]

Hi jmar,

Originally Posted by jmar

USB is poor quality? Rather vague statement. Why?...

Originally Posted by Eric Juaneda

...PC don't send continuous audio data stream to IC. It send 1ms audio packet at max USB 1.0 speed.
Synchronizing PLL on data stream could generate high and low frequency jitter...