|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
144.133.192.184
In Reply to: I'm not smart... posted by Presto on April 18, 2007 at 10:14:20:
The *less* circuitry and wires that you have, the better the end result. Complicated design just adds more opportunity to introduce jitter.What I'm looking for (which I haven't found yet) is an external multichannel DAC that has the following features:
- MUST use a *current generation*, flagship DAC chip (ie. BB PCM1794, AKM4396 or CS4398) capable of exceeding 120dB
- dedicated and switchable 44.1 and 48 clocks (NO PLL)
- located as close to the DAC chips as possible (to minimise EMI sensitivity)
- no upsampling
- The digital signal comes in ASYNCHRONOUSLY (ie. "unclocked") and gets clocked by the DAC itself - this can be a simple memory buffer but care must be taken that the wait states are consistent and predictable - static RAM would probably be betterUnfortunately, that rules out just about every DAC unit out there on the market today. Can't use Firewire - too much jitter. The only interfaces I can think of that supports asynchronous (other than proprietary ones) are USB and Ethernet. USB async mode is buggy in Windows XP, and Ethernet introduces too much latency.
So far, the only device that come anywhere close to the ideal (but still not perfect) is the E-MU 1820M, which uses a proprietary shielded (STP) cable to communicate asynchronously between the 1010 PCI card and the external AudioDock. Unfortunately the external unit also draws power from the PC via the STP (but I suspect E-MU wanted to minimise ground loop issues - as far as I can tell from looking at the board layout these guys are pretty smart and know what they are doing)
The only USB async device I've found is the Edirol SD-90, which uses a proprietary asynchronous USB driver written by Edirol (unfortunately, the unit only supports 44.1 and 48kHz sampling rates, and requires a "reboot" to change sampling rates). ThomasPf says the Audigy NX also supports async, but not sure whether Creative has written a proprietary async driver. Neither of those contain flagship DAC chips.
I would also include the Slim Devices Transporter (Ethernet connection, proprietary protocol) but only 2ch.
I'm thinking it should be possible to produce something similar to the 1820M, but using USB2 async and a proprietary driver (or the standard driver in Vista), and preferably using BB PCM 1794 in differential mode (or AKM4396). And externally powered. Such a unit should not cost much more than an 1820M, in fact potentially should be cheaper than a 1820M (if it is a DAC only and doesn't include ADC).
If you know of something like this, let me know. I think the problem is very few people (apart from myself) would buy such a unit. The market is obviously selling overpriced and overengineered SPDIF DACs to well heeled audiophiles. :-)
If I have spare time on my hands, I would build it myself. Unfortunately, I don't, so I keep hoping someone else would do it for me.
Follow Ups:
Chrstine,the Edirol UA101 is using USB2 and their own driver (at least on Mac), I am using it (and sometimes a TCElectronic Konnekt as well) with my Macbook and Altmann DAC, and it is as good as it gets. I do not remember exactly on what forum, but on one of the professional forums I read it is using async USB2 communication.
Most of the pro audio interfaces, eg. Digidesign Pro Tools, Lynx, RME, uses a voltage controlled oscillator (VCO) as a clock, rather than fixed frequency crystal oscillators (XO).The reason for doing this is so that sampling rates can be easily changed "on the fly" - a pretty essential requirement for pro audio.
However, VCOs are inherently more susceptible to jitter than XOs. They are very sensitive to logic induced modulation, and "drift" over time unless controlled using a PLL.
A device that accepts word clock in is effectively using the clock input to drive a VCO, and using PLL to compare the output with the input. This is less effective than syncing a VCO to an XO (which is what the "internal" clock does).
But using XO driving a DAC directly delivers better results than VCO.
Another technique is to use a "clock divider". This is common with DVD players, which generate a 44.1 or 48 audio clock based on the 27MHz master video clock. This is also not as effective as using fixed frequency XO.
XOs are also unstable (can drift based on temperature) but less so than VCOs. The main disadvantage is that they are fixed frequency, which means to change between 44.1 to 48 for example requires switching to a different XO.
This is not as easy as it sounds, because the DAC effectively needs to be "rebooted" when the XO is switched. This is why most devices that uses switchable XO requires a power cycle to change sampling rates (eg. my Edirol SD-90).
The E-MU 1820M is one of very few devices out there that uses switchable XOs but allow sampling rates to be switched on the fly. E-MU does this through very clever design, but even so it's rather cumbersome (effectively, they do a reboot of the DSP - they also mute audio outputs during the process so that you don't get a "screech" as the DAC resets).
So I should probably clarify, I'm really looking for a DAC with switchable XO rather than VCO. Unfortunately, that eliminates most of the DACs on the market.
nt
| ||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: