Review: MOTU 1248 & Ultralite AVB

FAMILY PROTOCOL

The entire family has two protocols in common — AVB and USB 2.0. They’re divided into two main camps from there: interfaces with Thunderbolt (1 and 2-compatible), and those without. In the ‘In’ camp are the 1248, 8M, 16A and 112D. On the outer are the Ultralite AVB, Monitor 8, 24Ai and 24Ao. The thinking is that you’d always go for one of the former interfaces first, and they’re more likely to be closer to your computer than an expansion unit. The exception being the portability of the Ultralite AVB, where USB 2.0 makes most sense.

The upshot is that you can run 128 inputs and 128 outputs to and from your computer via Thunderbolt, up to 96k (half that at higher sample rates), and MOTU’s USB driver can run 64 I/O at 44.1k/48k, which is… frankly, incredible compared to most USB 2 interfaces.

There’s no restrictions on what you can route as an input. Oftentimes with audio interfaces, if you plug a mic in, you lose one of your analogue line inputs. It’s almost the opposite on the 1248, MOTU has so much bandwidth to spare, if you can plug it in, you can route and record it.

Routing is fairly easy to accomplish on MOTU’s web app. It’s a matrix setup with the ability to click and drag your cursor over multiple blocks to assign I/O in a hurry. The web app is locally served from the devices themselves. Once you download MOTU’s AVB manager, a little icon appears on your dock with a menu of any available devices. Hitting a device opens the web app, where you can adjust routing, set device preferences, as well as adjust the gain of your preamps or trims of your line inputs and outputs, as well as control phase, phantom and pads. Every element is well set out, at first the gain dial seemed a little difficult to set accurately, but when you touch it a box pops up allowing you to type in an exact value; which can be up to 56dB on the Ultralite AVB, and 63dB on the 1248. There’s also a little gain meter above the dial, where the single colour scheme strikes again. No red for overs. MOTU’s looking into that.

If you have just two devices, you can forego the AVB switch, and hook them up directly using a Cat5e or Cat6 ethernet cable. If you only have one device, you can use the ethernet port to connect to a standard Wi-Fi router for wireless control. I had the luxury of reviewing the AVB switch as well, so I had both devices plumbed into it, with an ethernet output plugged straight into my Wi-Fi router. Over the last few years I’ve used this system as a ‘worst-case’ scenario for Wi-Fi ‘ease of use’. I’ve got an Ethernet over Power adaptor that works as a two-port router, mini Wi-Fi hotspot, and extends my internet connection. Most audio-related devices, even though they’re only using the network for control purposes, do not play nicely with this arrangement. The MOTU gear had no issues whatsoever.

WIRELESS CONTROL

This remote control-ability is a central advantage of the whole AVB idea, or any networked studio gear. Being able to sit at the drums and adjust mic preamp gain from your iPad or phone is really handy. Of course, it depends on whether you’re using external preamps or not, but even then, you could still hook them up to your MOTU converter and see the input levels coming into that channel on your device.

What’s more, the analogue line trims are only supposed to give you gain of between +2 to +22dB. But in the web app, the trim on each channel actually ranges from -96dB to +22dB, and seems to respond as such. Of course, there’s still a maximum input level of +24dBu on the inputs. But it does give a degree of control I’m not used to seeing on line level inputs.

The web app is clearly set out, and updates in a snap. There are four main tabs: Device — where you set clocking, control inputs and set your number of streams; Routing — a matrix connecting any input to any output; Mixing — a console for any input with insertable processing; and Aux Mixing — where you can send inputs to create ‘zero latency’ monitor mixes, or group channels to record.

For some reason, I was expecting a lag between changing, say, the number of AVB input streams on the Device page, and seeing it update in the Routing tab. But it’s instantaneous.

What’s more, it’s nice that the Routing matrix doesn’t display every possible input and output by default. You can enable and disable banks of I/O, the number of AVB streams (which are in banks of eight), the number of mixer inputs, and the number of to/from computer channels, to condense or expand the matrix at will.

ROUTING SAMPLE SET

The routing worked very simply. I’d route I/O over the AVB network from the Ultralite AVB into the 1248, and pump it into my computer via Thunderbolt. The clocking locks in without a hassle, achieving sample accurate phase lock over the ethernet connection. As well as routing direct inputs to outputs, you can also send I/O in and out of MOTU’s internal DSP mixer. Which gives you HP filter, gate, four-band parametric EQ, and a compressor on each channel, as well as a global reverb, and EQ and limiting on the group output, master bus, and reverb return. There’s also three stereo groups, and seven stereo auxes you can use to mix headphone sends, submix tracks live, or use in a stage environment.

Pro Tools recognised the CoreAudio I/O… for the most part. I did run across one snag. When I was trying to send audio over the AVB network to the line outputs of the Ultralite AVB at the end of the chain, rather than coming up as ‘AVB Stream 1 1’, etc, Core Audio was labelling those outputs as ‘Out (Multiple Destinations)’. Resetting the 1248 to factory settings sorted that little niggle out. I’d put it down to the bad treatment I often give interfaces during a review — re-routing with sessions active, unplugging interface cables while still operational, hard quitting sessions, and turning the units off and on hundreds of times. The MOTU devices performed pretty admirably despite all these ‘no-nos’.

DSP RETURNS

One thing worth noting is that there are limitations to what the DSP can handle. When the interface is set to 48k, you can run the full 48-channel internal mixer with every last effect, EQ and dynamic control instantiated. Turning up the dial to 96k restricts the channel count to 32, and severely limits the leftover DSP for effect processing, leaving enough power for just 13 full channel strips and no reverb. Once I added the reverb, I was down to a parsimonious two and a half channel strips. There’s a meter for DSP usage, and once you get to the upper limits, a little dialogue box essentially tells you, ‘no more, sorry’ any time you try to add another module.

After nixing a bunch of processing to make way for the reverb, it was a little bit of a letdown. It had that ‘pebble bouncing over concrete’ digital grain. The controls are flexible, and you can create your own presets with varying amounts of spread, pre-delay, reverb time, etc. But the underlying algorithm isn’t doing the effect any favours. I tried freeing up even more DSP to see if that helped, but it remained the same. The EQ and dynamics were quite good though, and switching the compressor to RMS detection made it particularly useful as a quick level control without fussing too much. The limiter is an LA-2A emulation, and a handy bus processor.

You can create channel presets, and route mixer inputs from the channel, which is much easier than jumping to the Routing matrix. You can also drag across on/off buttons to instantiate a whole row of compressors, for instance. Another time-saving use of the click and drag motion.

LATENCY GAP

The latency specs on MOTU’S AVB network are extremely good. The standard AVB protocol specifies a fixed latency of 2ms, but MOTU has managed to claw back some digits for a fixed network latency of 0.625ms, which is staggeringly low. And that’s fixed over an entire network of up to 512 channels, with up to seven switches and hundreds of metres of cable. You have to add a skerrick of time for internal processing, but you can expect to shoot things around the AVB network in under 1ms when you factor in internal DSP, etc.

Measuring the actual roundtrip latency of the 1248 interface over Thunderbolt returned a result of 70 samples, which equates to about 1.4ms at 48k, plus double whatever buffer you’re using. It’s very good, and if you’re operating at 96k, with a low 32 sample buffer, your total round trip latency will be within that 1.4ms mark — super snappy. This puts it at level pegging with the RME Fireface 802 according to the latency tests Andrew Bencina recorded in Issue 105. The latency over USB is not far behind, and if you have a Mac running Yosemite you can also connect directly to the computer via an ethernet cable, but the performance is more computer dependent.

My main issue at the moment is the lack of delay compensation between your host interface (connected via Thunderbolt or USB) and your other interfaces feeding signal to it via AVB. It’s a common issue when expanding your I/O count via ADAT — there’s a discrepancy between when sounds arrive from your host interface and those connected via ADAT.

When I split a signal to the analogue line inputs of the 1248 and Ultralite AVB, and recorded the result, the Ultralite AVB was returning the signal 25 samples later than the 1248, even though its signal was passing through the 1248.

I would have thought, given AVB provides more information about the device than a ‘dumb’ ADAT connection, MOTU could have delayed the host interface to time align it with other interfaces on the AVB network, but it doesn’t seem to be the case. This is one area Avid has licked with its HD I/O interfaces, and Focusrite Rednet routes multiple interfaces to a computer via a switch to eliminate the host bias. 25 samples isn’t very much — about 0.5ms at 48k — and it’s a rare occasion this would create an issue; phase and alignment issues if you’re splitting stereo inputs over different interfaces, for instance. Something to keep in mind.

MOTU has implemented a workaround, which involves routing everything through the DSP mixer to align the latency. But it limits the typical implementation of the mixer as a Post-FX monitor mixer. Understandably, MOTU is responding to its user’s request for the lowest latency possible; an admirable quest. But I don’t see the conflict between letting a solo host interface operate at its fastest throughput, then matching latencies when an AVB expansion is connected.

RATTLING THE SABRE

The ESS Sabre 32 output DAC is the same top-of-the-line chip used by the Apogee Symphony. The key point here is the 32-bit word length is lowering the THD+N and increasing dynamic range. In the case of the Symphony, Apogee quotes a dynamic range of 129dBA, which is the chip’s spec, while MOTU pegs its dynamic range as 123dBA, which is a more reliable measured spec. Noise is essentially undetectable. The other advantage is being able to use the digital main output control without losing resolution; a common criticism of 24-bit DACs, leading many to rely on external monitor controls. I would still probably tend that way with the 1248, just because I prefer having a big knob-type of passive controller in front of me. But if the 1248 is in arm’s reach, it’s not a problem to use the Main Output pot, and pressing down on it mutes the output.

The DAC is really quiet. I couldn’t really detect any noticeable difference in the sound stage between either the Avid HD I/O or the MOTU 1248. There was no noticeable noise on the MOTU, which is an improvement over what I remember of the 828 mkII.

On the line input side, the Avid HD I/O out-specs the MOTU for dynamic range, 122dBA to 117dBA. But again, whether quoted or measured, I find these numbers somewhat negligible in normal recording — they’re both very good. I was evaluating a Shure mic kit against an Audix drum mic kit as part of my tests, and after going through all modes of evaluation, it only reinforced to me that the impact of using different equipment goes down exponentially the further you get down the chain. A change at the source was huge, likewise, different mics netted completely different results, but the general picture of the drums was still there. Then, using the MOTU preamps versus some Neves, the difference was noticeable, with the Neves smoothing out the response and adding some more body to the drums. Still the MOTU preamps performed really well and make great all-round preamps. Getting down to comparing converters, it’s extremely difficult to detect any differences when you get to this level. After comparing recorded results, mixes, and just general listening, I was completely satisfied with the reproduction of the 1248 conversion. The soundstage appeared the same as the Avid HD I/O, and likewise, I found no difference in the response and extension at both ends.

MOTU UPCONVERTS

While MOTU has often been a great buy for the price conscious, this time it has leapfrogged up a converter chip generation to the best of what’s currently available, while still keeping things affordable. I’m sure Black Lion will come out with a mod for those really intent on eking out the best from the units. Yet, with the specs already on offer, it’s hard to see the justification. On the hardware side, after running the 1248 all day the unit exhibited only the slightest change in temperature. I’ve tested other interfaces packed with I/O that have almost melted into my desk. Even though I didn’t get to test the bigger units like the 16- or 24-channel converters, it bodes well. Combining the software and hardware upgrades with MOTU’s dedication to supporting its units over the extra-long term, I’d have no hesitation jumping onto its AVB network and discovering a whole new, flexible way of recording.