Just got this unit and early testing looks good.
 
Connects via USB2
LOW round-trip latency (smallest buffer setting yields 4.9ms total round-trip latency at a 64-sample ASIO buffer size/44.1k)
Under decent load (Sonar demo projects), CPU use was identical to a Lynx AES-16e (same 64-sample ASIO buffer size/441k)
Average noise-floor is ~-112dB.  That's darn good for a $650 audio interface.
 

• Two decent onboard mic preamps via XLR
• Two Hi-Z 1/4" guitar inputs
• Six Balanced 1/4" TRS inputs
• Two "Main" 1/4" TRS outputs (connect to studio monitor speakers)
• Six additional 1/4" TRS outputs
• Lightpipe I/O (switchable to optical S/PDIF)
• MIDI I/O

 
Want to run more extensive tests before giving a final opinion, but thus far this looks like a great fit for many users.
With outboard preamps, you've got enough I/O to record a full band.
 
Haven't mentioned the onboard DSP and Mixing features... as (IMO) those aren't the primary reason to choose an audio interface... but they are a nice addition.
 
More to come...

Thanks Jim, good to have quality choices that are not over a grand. 
Hey, any chance of a loop back test? 
Lately I'm obsessing about this.. 

Hi Jim,
this unit also connects via Ethernet, doesn't it?  I assume you've not tried that connection yet?
 
I have an ultralite MKIII hybrid (USB & Firewire) and I've loved it.  but this unit looks like an upgrade from that for an extra $100.

Currently, the Ethernet connection (as audio interface) only works on a recent make Mac.
Connection via LAN (via any protocol I've seen thus far) typically results in ~10ms total round-trip latency.
 
Sub 5ms total round-trip latency (via USB2) is great performance...
Since we're kind of "in-between" with Thunderbolt on the PC, it's nice to see USB2 units (other than RME) performing well.
 
The average noise-floor for most sub $1000 audio interfaces is typically between -103dB and -107dB.

Cactus Music
Hey, any chance of a loop back test? 
Lately I'm obsessing about this.. 

I haven't done a loop-back test... but I have measured the round-trip latency in several different audio applications.
Also, you can check to see if audio is offset vs. the original (re-record a single existing track and compare).
If it's spot on, then the reported round-trip latency should be accurate.

That's interesting.
I'm seeing 5.6msec with the RME BabyFace at 64 samples.

At 48 samples I see 4.9msec

She "runs" at 48, don't know how complex a project yet.
This is on my Office Rig; Win7x86 i3 4gb ram.
Is the difference a "system" thing or are the MOTU drivers better than the RME?
Tom
 

Hi Tom,
 
The "X factor" with audio drivers is each one's hidden-safety-buffer.
All audio interface drivers have a hidden-safety-buffer.
 
A larger safety-buffer can help avoid drop-outs/glitches under less than ideal circumstances (ie: high DPC Latency)
But it comes at the cost of higher latency
 
Most audio interfaces don't allow you to adjust the safety-buffer (thus why I call it "hidden").
Some audio interfaces (Focusrite, Behringer X32 series, and the new MOTU ABV units to name several) allow you to adjust the safety-buffer's size.  This allows the end-user complete flexibility to decide how to use the safety-buffer.
In the case of the MOTU Ultralite AVB, that 4.9ms is the result of using the smallest safety-buffer size... and the smallest ASIO buffer size (which is 64-samples).
 
Don't throw the Babyface out with the bathwater!   
The ASIO I/O buffers *AND* the safety-buffer (combined) yield the unit's total latency.
Babyface:  

• Hidden (fixed) safety-buffer is a littler larger than the MOTU's (at the smallest size), but it can use a smaller ASIO I/O buffer size (48-samples)

MOTU Ultralite AVB:

• Safety-buffer can be adjusted to a little smaller than the Babyface, but the smallest ASIO buffer size is larger (64-samples)

Total round-trip latency of the Babyface and Ultralite AVB is identical (4.9ms at 44.1k). 
Both allow you to run substantial loads (glitch-free) at 4.9ms total round-trip latency.
Note: The computer has to be able to keep up with the load...
 
Both are great performers.

I was just running the BabyFace on the "Cat".
She's running purrfectly at 48 samples.
28 audio tracks, lots of ProChannel:
EQ and Console Emulation on all tracks & busses, Concrete Limiters, Compressors, Gate scattered about.
LXP on 3 busses, some FabFilter comps and De-Esser.
Finally the FG-X on the Master Buss.
Even loaded 1 instance of Amp3 (my problem child at the moment, or not?).
I failed to "bring her to her knees"
But that's a "good" thing!

Off to install Platinum!
T
 

Well, I took the plunge and ordered the Ultralite aVB.
It will take a few days to get here, and I'll give it a workout - and report back.
 
I'm just going to set it up as a basic interface (skip the mixer and FX etc) and will only use the analog i/o. As long as I can get low latency and the CPU load isn't too bad (this is my first USB interface) then I'll be happy.
 
To be continued...

I have the older Ultralite MkIII (firewire only) and am still pretty happy with it. Anyone feel I would get great gains if I upgrade to this one?  I do mostly virtual instruments (medium projects), but also some vocals and occasional tracking of "real" instruments on an individual or duo basis. Been happy with MOTU thru the years, and the only other I've considered as an upgraded would be the RME UCX.
 
Thanks.