[jefemeister]

I have not been able to find any good information comparing the Analog Devices AD1896 to the TI SRC4192. Anyone have any insight or know of a good comparison done somewhere? Here's my take on it currently:

Both these chips are pin compatible ASRCs. The are very similar in functionality although I don't know how the TI part is implemented. The 1896's theory of operation is featured in the datasheet.

The major differences are that the TI has a 1:16 I/O range where the AD is only 1:8. The TI part can handle rates up to 212kHz, while the AD part is limited to 192kHz. Despite this, almost any application falls within the means of the less capable AD part except I would think the extra couple kHz of bandwidth in the TI part could be useful if upsampling to 192.

The specs on the front page of each datasheet are also very similar. TI claims 144 dB of Dynamic Range while AD claims 142. TI claims -140 dB THD+N, AD says "up to" -133 dB. The 1896 says "up to" for a reason.

The real differences between these parts are seen later in the documents under the "Typical Characteritics" sections. These show a bunch of plots (around 50 plots each) of harmonic distortions, intermodulation distortions, THD+N vs frequency and amplitude and RateIn:RateOut, DNR, Freq response, Linearity, etc. The 4192 absolutely destroys the 1896 in every way. The AD part has some good numbers in some very specific instances, while the TI part posts better and virtually constant numbers for all amplitudes, frequencies, and conversion rates. The AD part is only better in one regard: it has roughly half the pass band ripple at .01 dBFS. This isn't very significant though.

Since the devices are pin compatible, I am very interested in designing/building a DAC and switching between the two. We all know that datasheets aren't everything, but it's also hard to see why people would use the AD1896 when the documentation shows that the TI is clearly the "better" part.

Edit: The datasheet pdfs:

Texas Instruments SRC4192
Analog Devices AD1896

 

[Wodgy]

Despite the note on the first page of the SRC4192 datasheet, the SRC4192 is said to have very poor jitter rejection in practice, while the AD1896 is known to have extremely good jitter rejection.

Other than that, the specs of the SRC4192 are superior. There is also a new part from Cirrus/Crystal that you should be considering as well if you're doing a new design.

BTW, if you plan to swap out the SRC4192 and AD1896, buying the eval board for the 4192 is a pretty good deal, since it includes a rather expensive ZIF socket that makes swapping easy. For the price of the eval board, you're almost getting the socket for free.

 

[00940]

Quote:

Originally Posted by jefemeister We all know that datasheets aren't everything, but it's also hard to see why people would use the AD1896 when the documentation shows that the TI is clearly the "better" part.

the post 99 of that thread is one explanation. It has been discussed here and on diyaudio.


edit : wodgy was faster

 

[aos]

There is a new part from Crystal? Which one?

Unless it also shares the compatible pinout though or with "FREE jitter clock offer! Limited time only!" - the other two have now the advantage as you can swap them (TI will surely come out with a fixed part, and AD won't sit idle either).

 

[maczrool]

Quote:

Originally Posted by aos There is a new part from Crystal? Which one? Unless it also shares the compatible pinout though or with "FREE jitter clock offer! Limited time only!" - the other two have now the advantage as you can swap them (TI will surely come out with a fixed part, and AD won't sit idle either).

It's Crystal's CS8421 with up to 32-bit 192 kHz ouput!

Stu

 

[00940]

This post could interest you : http://www.diyaudio.com/forums/showt...=&pagenumber=1

 

[jefemeister]

Thanks for the info guys. I just came across this website that has some really good info on it. Doesn't talk about ASRC, it's about jitter, how to measure it, what it sounds like, etc. I have our librarian trying to locate the "CD Differences" white paper mentioned towards the bottom.

http://www.nanophon.com/audio/

 

[Wodgy]

Quote:

Originally Posted by aos There is a new part from Crystal? Which one? Unless it also shares the compatible pinout though or with "FREE jitter clock offer! Limited time only!" - the other two have now the advantage as you can swap them (TI will surely come out with a fixed part, and AD won't sit idle either).

As Maczrool pointed out, it's the CS8421:
http://www.cirrus.com/en/products/pro/detail/P1082.html
In general, even better specs than the SRC4192, including 175dB dynamic range!

The compatible pinout of the AD1896 and SRC4192 is not really all that handy in practice, as far as I can tell. The sockets that enable you to swap them are just too expensive (at least I haven't found any for less than $120 each), and these devices are too high performance to solder to some kind of jerry-rigged browndog-style DIP adapter. I'd also imagine the optimal configurations for each device in terms of jitter rejection are slightly different.