8 Channels ADAT converter

[Svart]

Read the review on that page.  The single ADAT lightpipe is an output only.  Only good for an A/D.

 

[stickjam]

Read the review on that page.  The single ADAT lightpipe is an output only.  Only good for an A/D.

Still if the objective is to include lightpipe outputs on a DIY 8 channel mic preamp (like the 8-way 1290 that I'm planning to do) it'd be perfect, presuming the converters can do reasonable justice to the 1290s of course.    The SM option looks simple enough, judging from that picture--it looks like 8 channel audio in on the header on the back edge and power input on the little header on the side.  (10 pins?  8 channels + ground + ??  hmmm--maybe the tenth pin is strapped to the ground pin to "tell" the main board that the A/D option is installed--just a WAG.  Might be worth picking up one to mess with)

Although a bit pricier, I like the Focusrite A/D option for the Octopre better, making the assumption that the converters are at least as good as SM's--it's the Platinum series after all.  I like it because it has SMUX for high rate use.  I'm also guessing it provides the interface to the front-panel bit rate indicators on the Octopre. 

It'd obviously be nice to have schematics for either.


OTOH, the SM option looks attractive because it already has a back panel bracket--mounting looks less problematic than the F.R. one.

 

[blue_luke]

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 using this board?

Not much really.... I haven't seen it but it would probably take about 5 min to figure it out.
My personal problem with it is that this stupid thing is about 100$ in the US and about 300$ Canadian, and the importer is CANADIAN!!!  And most US large chain stores like ZZ music, guitar center, musician friends etc... won't sell to Québec! Yup, anywhere in the rest of Canada but not in Québec!!  ??? ??? ??? ???

Looking at this, and after some discussion here and with others on another audio site (French) I figured that "with a little help from my friends" I can build something that I've always wanted.
I figure that making a good one is not more difficult than making a bad one!
Now before the pros cringes at this last statement, I say that because it is awfully complex and finicky to make these things!
And like someone said earlier, making the schematic is the easy part. Getting it to work at all is a big hurdle, then next step is to get it to work better, another big step, and the last, and hardest, step, is to get it to work perfectly!
My concern right now is, can I design and troubleshoot such a project with only the test instruments I have? (100Mhz 2 traces scope, THD analyzer, AC voltmeter, a good audio generator and a 48 ch logic analyzer). Would be nice to get my hand on a Audio Precision test set!!

This project is probably above my head and somehow the final result is secondary... I will get a great experience though!!  Which is one of the goal of DIY ehh!!

Luc

 

[Svart]

How about a nice spectrum analyzer too?

I guess the biggest question for quality would be, are you clocking it internally, externally or externally through the ADAT protocol?

 

[stickjam]

I will soon have one of these cards and an Octopre in hand.  I plan to give it a "DIY Group greeting" with a screwdriver and digital camera.  ;D

-Bob

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

This could save years...

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

 

[walter]

I have been thinking about this kind of project, but using the ADAT expansion card for the ddx3216. This card has two ins and two outs so 16 A/D D/A. Clocked via lightpipe, or internally so there must be a line from the mixer-that could be used from any clock source. I'll have to dig out that box of cards and have a look at what's going on.

 

[Skylar]

I have an OctoPre with ADAT card.

I just took the cover off...

The preamp circuit is divided into two boards.
Even channels on lower board.
Odd channels on upper board.

The ADAT card connects to the main board with two 20-pin ribbon cables.
2 rows, 10 pins on each row = 20 pins.

The small ribbon cable [J2] connects to the upper preamp board (odd channels).
The long ribbon cable [J3] connects to the lower preamp board (even channels).

The converters are AKM5383.
http://www.asahi-kasei.co.jp/akm/en/product/ak5383/ak5383_f03e.pdf

These are pin-compatible with the 5393...a much better converter. This chip is in the RME ADI-8DS and other stuff I'm sure.
The 5393 is superior in specs to the 5392, which is what the Apogee AD8000 and maybe Rosetta use.
http://www.asahi-kasei.co.jp/akm/en/product/ak5393/ak5393_f04e.pdf
You could upgrade to this chip with some careful SMD rework.

Of course, the converter is just part of the story. Dropping a 5393 in here is not necessarily going to make it better than an Apogee!
The 5383, though, is still a very respectable chip. 110dB dynamic range...very much pro level.


The ADAT board uses a TI DSP chip and some wavefront stuff.

Here are pictures:










Jumpers J2 & J3, where the ADAT board connects to the main boards:



NOW THE FUN PART:

I've tested the pin-outs of these jumpers. In short, looks like we have a suitable add-on AD converter for any 8-ch pre.
You'll need to supply +16.5V, -16.5V, +5V, +3.3V.

Here's what I've figured out tonight about the pin-outs for J2 and J3:



Note: I'm obviously missing the grounds and where the dither & sample rate control signals are placed.


That's it for me tonight fellas...more later.

 

[hobiesound]

as we say in amsterdam, KANKER VET!!!  ;D ;D ;D

greetings,

Thomas

 

[jdbakker]

Great stuff, Skylar; thanks for the detailed breakdown.

These are pin-compatible with the 5393...a much better converter. This chip is in the RME ADI-8DS and other stuff I'm sure.
The 5393 is superior in specs to the 5392, which is what the Apogee AD8000 and maybe Rosetta use.
http://www.asahi-kasei.co.jp/akm/en/product/ak5393/ak5393_f04e.pdf
You could upgrade to this chip with some careful SMD rework.

Trouble is that AKM chips are notoriously hard to get in small quantities. None of the major distributors carry them, and AKM itself is only interested in talking quantity.

JDB.

 

[stickjam]

I have an OctoPre with ADAT card.

I just took the cover off...

That's it for me tonight fellas...more later.

You rock, dude! 

 

[Skylar]

Yes, I've heard that AKM converters are hard to find, but I've never looked myself.

I was able to find one distributor though.

https://webapps.nuhorizons.com/storefront

Do a search for AK5393VSP


They've got 200+ 5393s in stock in America and 300+ in stock in Europe.
Apparently they let you purchase in small quantities too.
They're $21.26 each. You'd need four 5393s to handle 8 channels.

 

[gang of elk]

another option.. 

http://www.smproaudio.com/produkte/adda-konverter/a08-analog-adat-wandler.html

looks like the same card discussed earlier in this thread but now in a box w 8 analog inputs.  at $50 more than the raw card i wonder what's going on, if anything, before the A/D conversion.. 

 

[Skylar]

another option..  

http://www.smproaudio.com/produkte/adda-konverter/a08-analog-adat-wandler.html

looks like the same card discussed earlier in this thread but now in a box w 8 analog inputs.  at $50 more than the raw card i wonder what's going on, if anything, before the A/D conversion.. 

This particular box does not support sample rates above 48kHz (no 96kHz MUX).
It uses the Cirrus Logic CS5333-KZ.
With a max dynamic range of 98dB, I would consider this a consumer-level converter.
http://www.datasheetcatalog.org/datasheets2/38/384956_1.pdf

A SoundBlaster Live! has similar A-weighted specs (see page 4 of Cirrus datasheet).

Live! RightMark results

 

[Skylar]

Okay, I've done some further testing on the Octopre.

Where I was reporting 3.3V before I was really measuring 3.7V, but I approximated thinking that 3.3V was needed.
In actuality, we're feeding 5V to the LEDs and returning from the cathode (to the pins I erroneously marked "3.3V") which is why we see the voltage dropped to 3.7V–3.8V.

I reverse engineered the circuit as best I could...I still could have made some errors.

It looks like there are some switch de-bouncing capacitors in there.
They are unmarked SMD.
My multimeter only goes up to 400nF; when I tested them, they overloaded the DMM.
So we know that the caps are at least 0.5uF
They are non-polar...probably in the range of 1uF–10uF if they are indeed de-bounce caps.

But then again, maybe they're not switch de-bouncers because I don't really see an RC network in there, which is what most hardware de-bouncers consist of.


Anyway, here's what I've got now.
I'm going to breadboard and test it:

         

 

[Skylar]

Good News!


I just tested this out...IT WORKS!

And the phase is even correct!


So, if you want an 8-channel, pro quality A/D converter,
you should be able to buy an Octopre ADAT card & use the most recent pin-out/schematic I posted above.

That being said, I haven't tested J3 yet, which is where power is applied to the ADAT expansion board.
Probably going to build a power supply for this, make some boards, throw it all in a chassis and then sell the Octopre (minus the ADAT card of course)!

[OFF TOPIC]
The Octopre pres are decent but not that great. Nothing special...boring sounding if you will.
I had been using them for drums for a long time before I built a pair of Seventh Circle N72s.
The first time I fired them up, my jaw practically hit the floor..."so THAT'S what a real preamp sounds like!"
[/OFF TOPIC]

I'm sure the whole thing would cost less than $400 to build from scratch, and you'll have something that's maybe one step below an Apogee.
Then if you want to swap the converters to the 5393, you'll be that much closer.

 

[Svart]

These are pin-compatible with the 5393...a much better converter.

NO, they SAY they are a much better converter but in practice I found no difference.

Loopback testing using an old MX2424 analog card deck showed no bit differences between the 5383(original parts) and the 5393(parts I put in).  Pin for pin compatible.  Data output comparisons on a logic analyzer also showed no bit differences..  If you feed them the same signal you get the same digital out. How can one be better?

I'm just trying to save someone time and money here.  I spent my money for the purpose of testing you don't need to spend yours now.

There used to be an IC vendor that I bought handfuls of the AKM converters from.  I can't remember what the name was now and I don't see it on their website anymore.  If I run across one of the old receipts I will find out where I got them from.


After re-reading some of the previous posts:

The 5393 is superior in specs to the 5392

Again, they SAY they are.  I'll bet MONEY that you can't find an appreciable difference when clocked from a single clean clock source using the same signal.  Pretty much all commodity converters are similar these days.  Most people screw up the clocking and analog portions which gives converters used in cheap boxes a bad name.

The Alphalinks use AKM CODECs that are 3$ each.  Does that mean they suck? 

With a max dynamic range of 98dB, I would consider this a consumer-level converter.

That's pretty darn good when you mix all of the chassis noise, PSU noise, flicker noise, cosmic radiation and whatever else you have beating your nice clean signal to death on a product that is built for consumer OR pro usage.  You always lose bits to noise floor which equates to less dynamic range.  People have become easy prey when it comes to spec'ing a device.  Most companies will simply give you the dynamic range of the converter(which is also very optimistic most of the time), while some might actually give you test results.  I belive that the creative card is giving you *real* numbers..

SSL seems to have fallen prey to that as well.  My book that came with the Alphalink states 105db but their website now states 128db(which matches the codecs they used).  Go figure.

Then if you want to swap the converters to the 5393, you'll be that much closer.

Again,  NO.  It's the clocking and analog that makes the difference.  Focus on those because those will either make you or break you.

As for the Wavefront ADAT chips, go read the specs.  I'm going to keep harping on this until someone does...

JITTER in the hundreds of nanosecond range...  That's all I have to know to stay away from Wavefront clock recovery.  If you have to, design your nice clock(s) and do all clocking internally.  That should be enough to keep you busy for a long, long time.

I'm sure JDB has some stories about clock design he can share.

I'm glad there are people who are proactive and really getting into this but please know what you are getting into before leading folks down a certain path.  If you don't fix the fundamental problems of these boxes it really doesn't matter what preamp you stick in there, you are getting the same box that "sounds better" because your wallet is a LOT lighter..


EDIT: Wow I sound like PRR now..  ;D

 

[mitsos]

There are way to many confusing specs, numbers and crap when you get to digital, that I admit I don't understand much of anything. But it's still pretty f'n cool to know you can build an 8 pre box and give it a digital out for less than $200... 

I hear ya Svart, and your post begs the question... What to do about the analog/clock of this thing? Has anyone messed around with this?  Is it possible to "fix" this part of the card and make a killer A/D or are we talking major modification?

 

[jdbakker]

Is it possible to "fix" this part of the card and make a killer A/D or are we talking major modification?

Very hard to say without measuring. A closer look at the card (/the card's Gerbers + schematic) could give a tentative verdict, but it'll be as fuzzy as 'yes this might work' or 'no that's definitely FUBAR'.

JDB.
[it's a great pity that none of the currently commercially available converter chips have differential clock inputs]

 

[Svart]

First thing to do is to actually get a card and see what the clock looks like by itself.  if it's passable then we should look to see what it looks like with noisy motherboards/hdds/processors/etc around it.  Lastly if it appears decent we can see where the clock is derived from and work from there. 

Temp stability is of great concern here since the area the card is in has a high amount of heat and air moving about which can make the XO wander.  We could look to see if the source crystal or oscillator can be replaced with better parts.  We can also see how the clock is distributed/divided/etc.  Certainly they have to change the frequency around a bit to get different sample rates. 

There are lots of things to look out for and as JDB mentions, even layout can be critical here.  The circuit could be great but the layout terrible and that would undermine any amount of "fixing" we could attempt.

 

[Skylar]

Svart,

The tests I've seen between the 5383 vs 5393 show different results than what you have seen.

A little background...I worked for E-MU systems for several years as a test engineer.
The main products I worked with were the 1820m/1820/1616m/1616/1212m/0404 PCI audio interfaces and pretty much every product since then until 2008.

I have run 1000s upon 1000s of audio tests at the lab where I worked with RMAA and an AP-one.
Of course, the engineers who designed these products had much more modern/sophisticated tools at their labs in CA.

These E-MU cards were designed by a world-class audio team headed by an utterly brilliant guy who also designed the Apogee AD8000, C768 low jitter clock, AMBus, and other Apogee stuff during the time period he worked there.
I wouldn't be surprised if later Apogee products used variations of his earlier designs.
He now is the VP of engineering at Univeral Audio. I'm not sure if I should say his name here or not, but you can find him on LinkedIn.
So, in testing various revisions and prototypes of E-MU gear, I've seen what difference these two converters can make given an otherwise identical circuit.
In my experience and observation, the 5393 is a measurably better converter than the 5383.
I've seen 3-4dB better performance [dynamic range / SNR] from the 5393 compared to the 5383.


As you've said, though, the clock source and analog path are other elements of good A/D/D/A converter design that, if not done well, can foul up the whole thing.
So, my statements in earlier posts are really over-generalized and don't tell the whole story.
I'm sorry about that—I admit they were "sensationalist."

And as I mentioned before, dropping a 5393 in an Octopre ADAT board does not give you an Apogee.
But technically, I think it does get you one step closer because you are now using the same (or at least the same class) converter chip used in some Apogee units.
However, you won't necessarily have an Apogee-quality clock or analog path. But it is Focusrite, be it Platinum.
Because the Octopre ADAT card is capable of external sync, you could still use a Big Ben or comparable master clock if you are worried about Focusrite's ability to deliver a good internal clock.


About the Cirrus Logic converters/SMPro ADAT unit:
98dB dynamic range is bad when compared to so many other options that are better.
There are consumer converter/audio interface products (and cheap pro converter/audio interface products) that blow those specs away.
I know this from experience, and I'm talking about real-world measured results (as well as listening tests)...not datasheets.
Most low-midrange priced pro audio interfaces yield measured dynamic range results of 97–109dB.
Signal-to-Noise ratios on these products are usually similar numbers -90 to -109dB.

I'm talking about stuff like the M-Audio Delta series on the low end to the Digi 002 on the high end of the numbers above.

For what it's worth, the E-MU m-level PCI cards deliver real-world results of 117–118dB dynamic range in RMAA tests, 120–121dB when tested on an Audio Precision.
The only card I've tested that came close to this is the RME hamerfall (Multiface), which had a dynamic range of 115–116dB if I remember correctly.
I've also ran RMAA on a Rosetta 200...118dB dynamic range, again, if I recall correctly.

Most stuff that I have tested measures below 105dB dynamic range.
So, when I see 110dB, as in the Octopre specs, I think..."Okay, this is good quality."

BUT what about the clock?
Who said the Octopre ADAT card's clock was bad?
Focusrite says the clock on this unit has a jitter of 200 pico seconds.

Here are some RMAA results with an 1820M synced via ADAT to the Octopre ADAT board's internal clock.
[1820M line output >> Octopre ADAT board line input, Octopre ADAT output >> 1820M ADAT input, 48kHz]


Dynamic range, dB (A)  111.6


Noise level, dB (A)  -112.0


The measured results are actually BETTER than the specs that Focusrite publishes!
http://www.focusrite.com/answerbase/en/article.php?id=041
And this is even with my ghetto circuit hack & using "******" Hosa cables!
     


Kudos to Focusrite for not fudging their specs.



So, my opinion is that the internal clock on the OctoPre ADAT board is just fine.
If you're concerned, just run external sync via wordclock.