Clocking Multiple MOTU FireWire Audio Interfaces: max tracks

3 replies [Last post]
Offline
Joined: 02/12/2009

If you are intending to use multiple MOTU FireWire audio interfaces, it works great... BUT ONLY IF you use the right clocking plan. If not, you'll be sorry. Here are my discoveries and solutions...

--------------
--------------

The safest bet is to use word clock via BNC cables. If you have just 2 interfaces, a word clock cable will do (or if you must, clock via FireWire but word is preferred).

If you have 3, definitely use word clock; you can get away with daisy chaining the word clock (in-out-in-out), although a word clock distrubution box is a good idea.

For more than 3, get a word clock distribution unit, and send clock via word clock cables directly to each MOTU FireWire interface.

MOTU's recent manuals (2005) say the following about synchronizing multiple FireWire interfaces:

Digital Performer 4:
"If you have three or more digital audio devices, you need to slave them all to a single master audio clock, such as a word clock distribution device or universal synchronizer."

...and their example scenario of just two interfaces (2408mkIII & 896) offers two choices: clocking via FireWire or via word clock.

Traveler:
"You can daisy-chain up to four MOTU FireWire interfaces on a single FireWire bus, with the following restrictions..."

NOTE they are not talking about CLOCKING four units together by FireWire, because in the next diagram, they show all clock carried by word clock:

"For two interfaces, slave one to the other." (word)

"You can probably get away with daisy-chaining three interfaces, although a word clock distribution device is recommended."

"[For four interfaces, daisy-chaining] is risky. A word clock distribution device is highly recommended."

----------------
----------------

I shopped around for a new clock master, and settled on the Lucid GENx6 96 Word/Super Clock Generator. It sells for under $600. Most of the other models which are BOTH a generator and a distributor were much more expensive. Be careful if you're shopping for one, because most of them in this price range are not clock generators, just distributors.

In the past, I would rent the Apogee Big Ben or the Aardvark Aardsync II when I felt I really needed it (actually, I needed it much more often than I wanted to admit). These two models have extra features not found on the GENx6 96. But the Big Ben sells for around $1350, and the Aardsync II doesn't seem to be available anymore. (The manufacturer Aardvark may be no more, possibly replaced by Antelope Audio?)

Other models out there which were at least as expensive as the Big Ben: Antelope's Isochrone OCX, the Drawmer D-Clock DSM-1 and the Rosendahl Nanoclocks/Nanosyncs.

I bought the Lucid GENx6 96 at zZounds:
https://www.zzounds.com/a--3778/item--LUCGENX696

Also, if you are using external mic preamps with ADAT optical outputs, such as the Focusrite OctoPre (with the Focusrite ADAT option card / ADAT/AES/SPDIF card), or the Focusrite OctoPre LE, or the Presonus Digimax, or the Presonus Digimax FS, an external clock generator is necessary to lock them to the same source clock as your MOTU system.

n/a
Offline
Joined: 02/12/2009

A follow-up report on the Lucid GENx6 96 clock:

It has made a big difference and I now STRONGLY recommend getting it or another external clock if you intend to use more than two MOTU FireWire interfaces. I recently did large-track-count test recordings with and without the Lucid clock.

I discovered that one recording, without the clock, had multiple dropouts across tracks. This recording was done by clocking the FireWire interfaces off of the first interface. It appears that periodically, all tracks being recorded on a particular FireWire interface would dropout. The first interface's tracks had very brief, fraction of a second dropouts. The later interfaces in the chain had longer dropouts, around a half-second. These dropouts were hard to notice until I looked pretty closely at the waveforms to spot patterns across multiple tracks.

[Edit, Feb 2006: Another possible cause may have been a FireWire bus bottleneck, see next post below.]

So, again, I think anyone who uses more than one or two interfaces should get an external clock or you might have serious troubles! These clocks do more than just improve the quality of A/D/A conversion (which is sometimes a debatable result), they are essential to the foundation of larger digital studios.

n/a
Offline
Joined: 02/12/2009

More developments to report:

First, I've discovered a way to increase my track count significantly. My current system, an iMac G5 20" 1.8GHz with 1.5GB RAM, only has one FireWire bus (unlike recent desktop G5s which have two). How can you tell how many busses your Mac has? Even if you have two FW jacks, you don't necessarily have two busses. It's not published in most specs, so call Apple.

The FireWire bus would max out after about 32 tracks of simultaneous recording. It's interesting to note that the processor was not overloading, it was the FireWire bus that was the bottleneck. It was hard to determine this because I have found no way to get a live readout of the FireWire bus bandwith. It is, however, possible to monitor your CPU with Menu Meters:

http://www.ragingmenace.com/software/menumeters/

But I digress. On to the new tip:

I removed my external FireWire hard drive(s) from the FireWire bus, and switched to a USB 2.0 connection for the drive(s). (These drives have both connections.) I have now successfully recorded 56 tracks at once! I ran this for an hour or so, and did get the DP "overload warning" once, but I dismissed it and the recording was unaffected. More tests to come.

In the past, on a different Mac with two FW busses, I placed the interfaces on one bus, and the drives on the second bus. This is a good call. But if you only have one FireWire bus, move your drive(s) to USB 2.0. (NOT USB 1, it'll never handle it!)

If you have FireWire 800 on a separate bus, try that too. But remember, if there is even one FW400 device on a FW800 chain, the whole bus is slowed down to 400.

n/a
Offline
Joined: 02/12/2009

One final (?) note:

If you use a Big Ben with MOTU interfaces, read this tip which explains why the MOTU devices appear to not be terminated correctly. It's just an oddity, apparently the termination is actually OK:

http://www.apogeedigital.com/pdf/appnotes/bigben_termination.pdf
(on page 3)

Quote:When a MOTU interface is connected without termination, Big Ben's
Termination LEDs light correctly when the MOTU interface is powered
off, but indicate under termination when the interface is powered on.
When a MOTU input is terminated with a 75 ohm load, Big Ben's red
(over-terminated) LED lights. Thus, do not terminate MOTU inputs; this
does not affect clock performance.

n/a
Share this