8 Channels ADAT converter

This thread is the following of a thread that was started in the brewery " Trying to get in touch with wavefront semi"

JD said

For best jitter performance you do want to use your preamp/converter's clock as WC master.

Click to expand...

JD you're right of course. ;D in theory.
In practice my system is based around two RME fireface 800. If you people know anything about RME, their WC is one of the very good in the business, so in pratice, I feel that I would get less jitter if I used my RME as master WC and slave the preamps converter to it.
My reasoning is that building a clock with less than a few parts in million deviation with what the average DIYfer like me can get is tougher than slaving to the WC.

Also in my system I use a video blackburst master signal that I distribute to my racks. Each racks contain a pro-tool video slave driver. End of the story here, but NO 96Ksamples possible, 44 or 48 only.
(I do multitrack live recording as a sideline business)

I looked around and it seems that SM pro has discontinued the adat converter they had (A0-8 and PR8IIa),
Rats!!
I checked MARIAN ADCON . They do good stuff but I don't need the ADAT to analogue converters, I checked also APHEX and it seems to be the best solution right now.
But I realy like the idea of building the converters in the mic preamp box, saves rack space, power cable and audio clutter... It's simply a neat idea.

I checked again, first reading really, on the wavefront site the pdf document they have about the ADAT chip and they discuss somewhat the protocol, but funny enough, no more application notes there!!  :

So here we are, wanting a cookie from the jar, but daddy has put the jar on the top shelf.  :'(

Luc

JD is still correct.

It's not really about the source of the clock, it's the interfacing and cabling that can affect the clock jitter.  Small amounts of impedance mismatch can cause small reflections that can affect clocks negatively in more ways than just jitter.  Generally clocking internally is a better practice unless you absolutely need to clock multiple units.

The wavefront part's jitter/phase noise(from internal clock recovery circuits) is MUCH worse than any internal clock if you read the datasheets(that you can't find!).  I think they quote jitter in the millisecond range!

But do we care?  Nah, it works doesn't it?  Can you really tell that the jitter is killing the signal? not really.

I think the whole jitter "problem" in audio is overblown and is frequently a scapegoat for other types of problems.  Much like anything else, spending a lot of money to cure a problem tends to "fix" it even though it might not have actually fixed anything.

The ears and the wallet remain connected..

I remember one of the heavier cats around here who dropped the octapre adat card into his preamp box.

This could save years...

blue_luke said:

If you people know anything about RME, their WC is one of the very good in the business,

Click to expand...

...according to their own sales literature. The measurements I've done on Hammerfall-class hardware (Multiface, HDSP9652) show that while they can lock onto quite jittery incoming signals, their clock output is noisier than necessary.

blue_luke said:

My reasoning is that building a clock with less than a few parts in million deviation with what the average DIYfer like me can get is tougher than slaving to the WC.

Click to expand...

For audio the absolute precision of the clock is generally irrelevant. If you want a simple, non-lockable (=master only) clock you can easily buy one for not much money (this is one). Locking a low-jitter clock to an incoming WC without degrading it is another matter. As was pointed out in a recent thread on Bruno Putzeys' PSW forum, it's harder to design a good PLL than a good master clock.

Svart said:

I think the whole jitter "problem" in audio is overblown and is frequently a scapegoat for other types of problems.  Much like anything else, spending a lot of money to cure a problem tends to "fix" it even though it might not have actually fixed anything.

Click to expand...

While it is true that jitter is used often enough to part an audiophool from his money, fact remains that it is a measurable phenomenon and that studies show that its effects are audible (depending on frequency/intensity). Just as improved converter technology has made it necessary to put more effort into analog ADC frontends, it forces designers to pay more attention to clock jitter in order to get the best out of their converters.

JDB.
[then again, putting a $10k rubidium-stabilised tube-driven Master Clock into a CD transport? Suuuuuure...]

EDIT: typo fix

*cough* SRC - the cleaner of all jitter

While it is true that jitter is used often enough to part an audiophool from his money, fact remains that it is a measurable phenomenon and that studies show that its effects are audible (depending on frequency/intensity). Just as improved converter technology has made it necessary to put more effort into analog ADC frontends, it forces designers to pay more attention to clock jitter in order to get the best out of their converters.

Click to expand...

Oh yeah, I'm not saying it doesn't cause problems sometimes.  What I was saying is that it seems that everyone who intends to "upgrade" their digital equipment either by buying an external device or doing internal changes generally buys into the "always need spend $$$ on a better clock" propaganda when generally their clocks are good enough and the sum of the rest of the system won't benefit from being clocked from a "cleaner" source anyway. 

I recently listened to an audio unit whom a friend had sent to be modified by one of the well known upgrade companies that lurks around here and gearslutz.  He was more than happy with his purchase of opamps and a new super low jitter clock and whatever else they did to it.

I couldn't hear the difference at all.  I took it easy on him and told him it sounded fine, which it did, but it wasn't any different than it was before he sent it off.  In fact I think it was a little more shrill but otherwise had the same sound and clarity it did before.

I've also experienced the same thing with external clocks.  Another friend added a big ben to his apogee rack and swore it was the best thing he'd ever done to the studio.  I didn't hear the change at all.

Rochey said:

*cough* SRC - the cleaner of all jitter

Click to expand...

Right, assuming you don't have too much of it, because then an SRC will happily apply the jitter to your output signal.

(Assuming we're talking about Asynchronous SRCs here, since a plain synchronous Sample Rate Converter won't do anything about your jitter at all).

Any real-world ASRC has to make estimates about its input/output frequency ratio. It has no way of knowing this ratio for sure (no a priori data), nor does it have any way to look into the future and predict ratio changes.

Any real-world ASRC has a finite input buffer. You don't want this buffer to be too large in any case, as it sets a lower limit to the ASRC latency.

Given these two facts, for each real-world ASRC there exists a jitter/wander level above which the ASRC will need to modify its ratio estimate to keep its input buffer from overflowing/underflowing. These changes affect the output data in a manner which is very similar to that of converter jitter.

Now, modern ASRCs are pretty good at keeping this from happening for normal input conditions. Still, the real world can be harsh, and mediocre engineering even more so. From some back-of-the-envelope calculations I did a while ago it looks like a simple 4046 PLL plus built-in VCO as used in the Beh ADA8000 will kick a SRC4392 from slow to fast mode often enough to be annoying.

The fix? Don't Do That, Then. Properly applied an ASRC can help quite a bit in a converter:

- Take a low noise crystal oscillator running at the highest frequency your ADC supports (say 13.5MHz for the PCM422x). Feed this clock directly to your ADC. Jitter at this level can be ~3ps

- Connect the data out of your ADC to the input of an ASRC. Have the output of the ASRC clocked by a CS2000 or any other clock cleaner; the output data of the ASRC goes to your ADAT encoder. Jitter at this level is 50-70ps

- Feed Word Clock or recovered ADAT clock to the clock cleaner. Jitter at this level can be several ns.

It helps a lot in avoiding ASRC artefacts if the ASRC input and output clocks are not too close together.

This procedure is used in a few hi-end commercial converters. It isn't totally immune to anything and everything; very jittery satellite/broadcast signals could still throw it off, as will Vari-Speed. The latter makes sense, because there is no way a ratio estimator can see the difference between deliberate continuous changes in sample rate and phase noise.

Bruno Putzeys (again) has some interesting things to say on the matter in his AES Masterclass (starting around Slide 90, but the intervening ones are worth reading as well).

JDB.
[but then you knew I was going to say this, no?]

Rochey said:

*cough* SRC - the cleaner of all jitter

Click to expand...

Rochey you cough an awfull lot! Poor you, do you live in Montreal??   

Anyway, SRC?  What is SRC? I can not claim I know every acronyms!

I can think of Société Radio-Canada
Saskatchewan Research Council
Security Response Center
And a few others....
Sure that's not what you have in mind!


Luc

Got IT!!
Sample Rate Converter....

I think I will have to play a bit of catch up before commiting any silicon to a design!  :-\

Luc

blue_luke said:

Anyway, SRC?  What is SRC? I can not claim I know every acronyms!

Click to expand...

The initialism "SRC" means, in this case, "Sample Rate Converter."

-a

In the course of finding more information for my new design I found this very helpful site, with a very good article by our own Rochey

I am very happy of this find!  ;D

Luc

anyway, did you check out the behringer ADA8000?  It's a decent unit for the price and already has preamps, converters and adat I/O for less than the cost that you could build one..

Svart said:

anyway, did you check out the behringer ADA8000?  It's a decent unit for the price and already has preamps, converters and adat I/O for less than the cost that you could build one..

Click to expand...

But do you really wanna use the ADA8000? I use the SM pro, pretty much the same thing and it's not THAT good. It's cheap and it's working, but the dynamic range in this preamp are bad. It works ok with shure beta 58/57 and rock music, but if you want to record an acoustic instrument the noise/signal level aint good.
I like the idè of making an AD converter inside a preamp and with excellent quality! 

babyhead said:

I remember one of the heavier cats around here who dropped the octapre adat card into his preamp box.

This could save years...

Click to expand...

Any idea who the heavy cat is that used an Octopre ADAT card?  Anyone have service info on that box to infer the pinout, power and signal levels on the two ribbon cables that feed this thing?



Sure would be nice thing to have on the back panel of the 8-channel 1290 project I have on my radar!

-Bob

But do you really wanna use the ADA8000?

Click to expand...

Are you sure that it isn't that good?  You use something that is "pretty much the same", not the actual unit.  So how close IS "pretty much the same"?  I'd like part-for-part documentation of the similarities and differences in the circuit design asap.

THEN I would like for you to search for the thread which contains the ADA8000 dissection and subsequent improvement that involves very little change to the design.

And then there is the whole jitter problem from the wavefront ICs.  Read the datasheets and gasp at the horror that is the recovered wordclock.  So it kinda nullifies the "excellent" quality portion of your requirements.

You'll get "decent", as I stated before, and you'll like it.

;D

stickjam said:

babyhead said:

I remember one of the heavier cats around here who dropped the octapre adat card into his preamp box.

This could save years...

Click to expand...

Any idea who the heavy cat is that used an Octopre ADAT card?  Anyone have service info on that box to infer the pinout, power and signal levels on the two ribbon cables that feed this thing?

Sure would be nice thing to have on the back panel of the 8-channel 1290 project I have on my radar!

-Bob

Click to expand...

Check out the box that Joe Malone did.

http://www.jlmaudio.com/phpbb3/viewtopic.php?f=2&t=22

At this stage, I have pretty much elected to design a very high quality converter board that can be used with pretty much any type of pre-amps.
By high quality I mean 24 bits TI converters and asynchronus clock converters, very good audio front end etc....
I intend to add a few goodies such as S-Mux for 8 channels in differential inputs etc....
All sketchy right now, but my line of thought is getting in line. I'll keep you posted.
It's not that difficult, I think, but VERY unforgiving.
It's like a loto ticket, you winn only if all the numbers are in the proper order. Not easy, but deasible!
Luc

Have you designed mixed signal before?  Drawing a schematic is the easy part.  The devil is in the details as they say. Also beware of layout issues.

I'm sure we will all help you in your quest but it's not as easy as you think if you haven't done it before.

Looks like Joe used a demo PCB or a board out of some other box..

Svart said:

Have you designed mixed signal before?  Drawing a schematic is the easy part.  The devil is in the details as they say. Also beware of layout issues.

I'm sure we will all help you in your quest but it's not as easy as you think if you haven't done it before.

Click to expand...

Yes I know about designing mixed signal design.
I can not claim I am the best there is but I know a trick or two!

Now the help offer is truly appreciated... and will be needed! Rochey, JD and some other have already helped by pointing me in the wind....

Stay tuned....
This project will be delayed some as my second grandson is due anyday now  

Luc

Joe used the SM Pro Audio ADAT Lightpipe Output for PR8MKII card

http://www.zzounds.com/item--STOADAT


How hard would it be, to do an 8 channels converter,  using this board?