Universal Audio have combined their considerable audio expertise and renowned DSP platform to create their first audio interface: the Apollo.
Universal Audio's first computer interface, the Apollo, has been one of the most keenly awaited new products of 2012. Sound On Sound had an early sneak preview before NAMM in January, when the software was still in a beta stage, but the Apollo is now shipping for Mac users (the PC software is due in the summer) and I am very pleased to review one of the first Apollo Duo units in the UK.
The Apollo is essentially a UAD2 Satellite DSP unit combined with a 18x24 audio interface, but the whole is far greater than the sum of the parts, because it combines the unique capability of being able to run UAD plug-ins in the monitoring and recording paths with very low latency. A bespoke 'Console' application allows UAD plug-ins to process signals on the way into the DAW, and/or for the artists' headphone or monitoring mixes, all in real time. The Apollo can also be used in exactly the same way as a UAD Satellite, powering plug-ins inserted in DAW channels in the usual way, and it serves as a normal I/O interface too, with dedicated monitoring outputs.
Latency is a major concern in this kind of application, naturally, but I measured mic to monitor and headphone latencies of just 2.3ms at a 44.1kHz sample rate, even with plug-ins in the monitor path. Universal Audio's marketing claims latency of 1.1ms, but the small print reveals that this (like most interface manufacturers' specs) is measured at 96kHz, when the A-D and D-A filter latencies will inherently be halved. I don't think 2.3ms of latency is going to worry anyone in practice, though.
The review unit was equipped with two Firewire 800 ports, but a blank panel covers an accessory slot for an optional Thunderbolt interface module (slated for release this summer). The Apollo is available in two versions, the Apollo Duo and the Apollo Quad, fitted with either two or four SHARC processors, respectively. As you would expect, the DSP performance matches that of the Duo and Quad Satellite and PCIe versions.
Before going any further, I should mention that the Apollo won't play nicely with all Apple hardware, and requires OS 10.6.8 (Snow Leopard) or higher. I discovered this the hard way when the 15-inch Macbook Pro (model 3,1) I wanted to use turned out to be the only completely incompatible Intel-based Macbook Pro! The Apollo also won't work with any PowerPC, nor will it work alongside any UAD1 devices. A number of MacBook Pro models with FW800 ports will only work at FW400 speeds, too, apparently — full information can be found on UA's web site. Fortunately, a friend generously allowed me to use his 12-core Intel Mac Pro, which worked flawlessly.
The current Apollo software (the review unit shipped with v6.1.1 but v6.20 is available on the web site and was used for this review) only allows one Apollo per system, but UA promise multi-Apollo support with a future update. However, a UAD2 Satellite can be daisy-chained on the Apollo's Firewire bus for additional DSP power, if required.
The Apollo ships with a basic 'Analog Classics Plug-in Bundle' of UAD 'physical modelling' effects, comprising the 1176LN Limiting Amplifier, the LA2A Leveling Amplifier, and the Pultec EQP1A Program Equalizer. A $100 voucher is also included to help fund additional user-choice plug-ins, and the entire UAD suite can be auditioned using 14-day demos. We've covered the UAD2 plug-ins in SOS many times before so I won't cover the same ground again; suffice to say that I'm a big fan, running a PCIe Quad card myself.
The Apollo hardware is a 1U, rackmounting box extending 306mm behind the rack ears, with a black steel case liberally punched with ventilation holes on all sides (it gets warm around the mic input area). The rear panel is busy but clearly labelled, starting on the left-hand side with a four-pin XLR power inlet which connects with a supplied 'power brick'. This is a standard third-party unit (from Skynet Electronics) that accepts 100-240V AC mains via an IEC inlet at one end, and provides six amps of 12V DC through a fixed cable terminating in a four-pin XLR plug at the other.
As regular SOS readers will know, I much prefer hardware to have integrated power supplies, but that's not always possible, for reasons of space or to keep mains-related noise away from sensitive preamps. Fortunately, Universal Audio have employed a high-quality outboard PSU with a universal IEC mains inlet and a secure, latching DC connector, which is a reliable, professional solution.
Continuing the back-panel tour, a removable plate covers the optional Thunderbolt (Light Peak) adaptor socket. Thunderbolt is nominally 12 times faster than FW800 and, while the latter is more than adequate for a single Apollo, the additional bandwidth makes multi-Apollo rigs, or combined Apollo-Satellite rigs, more practical. With two Firewire 800 sockets available, other FW devices can be daisy-chained through the Apollo if necessary, and if a Thunderbolt interface is installed, both Firewire ports can be used for external drives or UAD Satellites.
Two pairs of ADAT Toslink ports provide S/MUX capability, and the second ADAT output mirrors the first at base sample rates, while the combination supports eight S/MUX'd channels at double sample rates and four at quad rates. Moving right, S/PDIF in and out is via a pair of RCA-phono sockets supporting all sample rates up to 192kHz, with sample-rate conversion available for non-synchronous and alternative rate inputs.
A pair of BNC connectors handle word clock in and out, with a push button to select 75Ω input termination. Much of the rest of the rear panel is taken up with two rows of balanced (TRS) quarter-inch sockets for the eight line inputs (impedance: 11kΩ) and eight line outputs. These can be configured using the Console application for either +4dBu or -10dBV nominal levels, and they will accept balanced or unbalanced connections. Two more TRS sockets provide a dedicated stereo monitor output that can also be duplicated on the S/PDIF output if preferred, perhaps to drive digital monitor speakers directly. Four female XLRs accept microphone connections, which can be selected individually in place of the first four line inputs.
There are no MIDI facilities at all, largely, I suspect, because there isn't any panel space left for the sockets! This may be deemed a negative point by some, but in practice I doubt many will be put off, as many keyboards have MIDI-over-USB connections, and serious MIDI rigs need much more flexible facilities than simple In/Out/Thru connections anyway.
There are only so many ways you can arrange the front panel for a product like this, but the Apollo does look very similar to the Prism Sound Orpheus. To the left of the smart, brushed-aluminium face plate is a pair of high-impedance (2.2MΩ) J-FET instrument input sockets (unbalanced TS types), which have priority over the first two channels' mic or line inputs. A rotary encoder with a green illuminated ring around the periphery adjusts the gains of channels 1 to 4, and a press-switch action allocates the encoder and associated buttons to each channel in turn (the selected one being indicated above the bar-graph meters). The gain range is 55dB and the minimum mic signal level necessary to achieve 0dBFS at maximum gain is -50dBu.
Six silver push-buttons configure the first four channels, switching between mic and line inputs; introduce a 20dB pad, phantom power (mic mode only), and a 12dB/octave 75Hz high-pass filter; invert the channel polarity; and link adjacent odd-even channels for stereo working. LEDs in the meter section confirm status, while 10-segment bar-graphs show the input level for the eight analogue channels, spanning -27dBFS to clipping in 3dB increments. A back-lit Universal Audio logo serves as the 'power-on' indicator, and three more LEDs show the current clock source (internal or external) and whether the Apollo is communicating with a host computer (UAD Link). There is no indication of the current sample rate.
Two further bar-graphs show the monitor level prior to the rotary encoder, which controls an analogue attenuator after the D-A to maintain the 'resolution' of the output signal. Pressing this encoder mutes the outputs and turns the coloured ring red instead of green. A pair of smaller knobs adjusts the two independent headphone outlets, while a large power on-off toggle switch completes the front-panel facilities.
The Apollo provides a total of 18 physical inputs — eight analogue (the first four accepting mics, and the first two also instruments), stereo S/PDIF, and eight ADAT — and 24 outputs (eight analogue lines, stereo monitor, stereo S/PDIF, eight ADAT, and two stereo headphone outputs.
The analogue I/O has very good technical specifications (see box) and performs extremely well. My only criticism concerns the headphone outputs, which seem unusually sensitive to headphone impedance. Using 32Ω headphones, I found there was plenty of level available, even when tracking alongside other performers in the studio. However, with 600Ω models the maximum level was noticeably lower, and there were times when I would have liked more 'oomph'. If you work, as I do, with 10dB or more headroom when tracking and mixing, the headphone outputs may seem a little feeble. On a more positive note, the headphone quality is actually very good; not quite in Benchmark or Grace Design territory for sheer power, weight and clarity, but very detailed and neutral, and significantly better than most console and recorder headphone amps.
The instrument inputs have an extremely low noise floor and a phenomenal gain range that can easily cope with any passive or active instrument. The four mic amps also performed pleasingly, with a clean and reasonably quiet noise floor and a 55dB gain range, which is adequate for close-miked sources and high output mics. They might struggle with low-output ribbon mics or distant mic placements, but I didn't have any problems during the review. These are not clinically clean preamps, but impart a subtle musical character which I found pleasant and very acceptable. The harmonic distortion is very low (less than 0.03 percent at maximum gain) with, if anything, a slight tendency towards second-harmonic emphasis which hints at a rich mid-range character.
The Apollo software installs the standard UAD meter and control panel alongside a new 'Console' application and a console plug-in. The Console app looks like a vintage mixing desk with delightful visual details and controls the Apollo's input and monitoring configuration. The Mac Pro's owner queried whether it was a case of 'form over function', contrasting strongly with RME's Germanic 'function over form' approach, but I liked it and found it very easy to use.
The layout places the analogue inputs on the left, followed by the S/PDIF and ADAT inputs, and the two internal stereo Aux returns. The monitor, headphone and aux bus controls are on the right. The first four channel input 'strips' show all the preamp controls, with a green LED indicating to which channel the Apollo's own controls are currently allocated. Virtual buttons and an encoder knob control the mic/line, HPF, pad, phantom, polarity and stereo link functions, and a numerical display shows the channel gain. Frustratingly, it's not currently possible to click in the gain display box and type a desired value, as I am used to doing with my current RME interface. The four line-only channels just have switches for +4dBu or -10dBV sensitivity. All input channels have four insert 'slots', with on-off and plug-in selection buttons to install any UAD plug-ins available on the system. Of course, the Console application includes automatic delay compensation to maintain accurate alignment between processed and non-processed channels.
Below the insert slots is a pair of Aux Send level controls and associated pan pots, followed by a pair of headphone-bus level controls with pan pots and on-off switches. Finally, each strip has a main monitor-bus pan control, fader, solo and mute buttons, and a bar-graph level meter alongside the fader. The Aux buses feed two dedicated Aux returns, which have their own four insert points and headphone bus level controls. These aux returns can be used to create effects loops, or to put effects in the headphones, but they can also be routed to any selected hardware output.
The monitor section features a pair of analogue-style meters sensibly marked with a scale changing from green to yellow at -9dBFS. DSP usage meters are shown below, and a rocker switch moves any effects plug-ins between the monitor and record paths. The monitor output level control is supplemented with buttons for mono, mute and solo-clear, plus buttons to audition each of the headphone outputs. Oddly, there is no facility to audition the aux sends. The headphone outputs can be switched independently to mono, and sourced from the monitor mix or their own independent buses. Entire Console configurations can be saved and recalled.
The Console Recall plug-in I mentioned earlier is available in VST, RTAS and AU formats, and is designed to be inserted in the DAW to make it easier to control the Apollo's monitoring facilities and to integrate console store/recall functions with the DAW project. Essentially, it avoids you having to run the full Console app when mixing. The plug-in comprises the monitor section, with metering, monitor volume and source-selection controls, headphone selectors, configuration recall and synchronise facilities, and the switch to move effects between the record and monitor paths.
Setting up monitoring paths for the Apollo with the Console application is very straightforward, and I was pleased to find everything is remembered when the unit is powered down. Subsequently, if the Apollo is powered up but not connected to a Mac, the same routing and mixing stays in place and the unit can be used as a stand-alone mixer. Plug-ins are removed, because the necessary DSP code resides on the computer, not in the Apollo hardware.
Installing the software and hooking the Apollo up to the Mac Pro proved trouble-free and, after registering the unit with my UA account and downloading the appropriate authorisations, it all worked immediately. I was using Nuendo, but Logic, Pro Tools, Ableton Live, Sonar, Cubase and most other DAWs are also supported, and the software is compatible with both 64- and 32-bit DAWs (the plug-ins are currently 32-bit but will be upgraded to 64-bit in the near future). Pro Tools has to use the RTAS plug-in format, although Universal Audio are working on adding AAX compatibility as a free upgrade soon.
I quickly established an optimal buffer setting of 192 for this specific Mac Pro system, with a reported latency of 11ms in and 7ms out. These figures were roughly double those achieved with the RME MADI interface I normally use, but presented no issues at all when playing virtual instruments or live tracking (thanks to the Apollo's low-latency monitoring). Smaller buffer sizes reduced the latency further, of course, but occasional clicks appeared during periods of frenetic activity!
Nuendo's output routing could 'see' the monitor and headphone buses, as well as the analogue line, ADAT and S/PDIF outputs. Sensibly, the DAW's list of Apollo outputs starts with the monitor feeds in positions 1-2, which is where most DAWs expect to route the main output signals by default anyway. Similarly, all the physical inputs and buses could be selected to Nuendo's record inputs. The processed version could then be recorded straight off by switching the effects to the record path. It is not strictly necessary to run the Console application once the line I/O and monitoring routing has been configured, because input gains and monitoring levels can be adjusted directly from the Apollo front panel.
One interesting and surprisingly workable application of the Apollo is in live performances, combining backing tracks from a computer with live mic and instrument inputs, and applying UAD processing and effects in real time with negligible latency. I can see this being a major selling point to a whole raft of potential users, as it's something no other interface can offer to the same degree.
The Apollo won't suit everyone, of course, and there are other interfaces around with more I/O and lower system latencies for similar money. However, for ease of use, sound quality, routing flexibility, the DSP engine, and the ability to track with UAD plug-ins in real time and with negligible latency, the Apollo is a very attractive proposition with no direct competition. It was well worth the wait, and I hope the PC drivers will be released soon! .
The RME UFX is a fairly natural alternative for consideration, as is the Prism Sound Orpheus — although the latter is a far more expensive option. The UFX doesn't quite match up to the Apollo's technical specifications, nor does it have the vast array of plug-ins or the dedicated monitor outputs, but it does have more I/O and integrated MIDI ports, as well as a built-in power supply and RME's legendary low-latency drivers.
Universal Audio claim that the Apollo performs as well as or better than interfaces costing far more, so naturally I ran a full series of bench tests using an Audio Precision system. My own bench tests matched the published figures very closely. For example, the AES17 dynamic range and signal-noise ratio figures for the line inputs (to a digital output) are 117dB with A-weighting (107dB without). The THD+N figure is 107dB and the line outputs (from a digital input) performed similarly with a signal-noise ratio of 118dB (A-wtd) and THD+N figure of 106dB, which is well below 0.001 percent. The monitor output performance was a little below that of the line outputs, but only by two or three decibels on the dynamic range, S/N ratio and THD+N figures — a difference I attribute to the analogue gain-control circuitry employed.
The first four preamps provide a 55dB gain range, giving maximum and minimum microphone input levels (for 0dBFS) of +6dBu (+16dBu with the pad) and -50dBu. The line inputs have maximum and minimum levels of +20dBu and -36dBu, while the corresponding instrument sensitivities are +10dBu and -38dBu.
Harmonic distortion on the outputs is very low indeed (below 110dB), with evenly distributed even and odd components. The inputs have a slight tendency towards odd-harmonic distortion when the levels are pushed towards 0dBFS, but at around -20dBFS the balance is very even and all distortion components are below 110dB.
The input noise floor spectrum is very smooth, with a distinct reduction in level above 3kHz. The strongest mains-related harmonic is at 100Hz and measured about -105dBFS, with the bulk of the noise floor sitting around -125dBFS. The frequency response for all inputs is ruler-flat, with the low-frequency 3dB roll-off at about 3Hz, rising to 75Hz with the HP filter switched in.
I measured the Effective Input Noise (EIN) figure for the mic preamp to be -128.4dB, which is pretty good, if not class-leading, and the input impedance measured 2.7k.
The Apollo's internal electronics are constructed to a very high standard, with surface-mount components on three main circuit boards and a couple of daughterboards on the front and rear panels. Texas Instrument PGA2500 digitally controlled microphone amplifiers are used for the first four inputs, with NE5532 op-amps handling more or less everything else. The A-D converters comprise a pair of four-channel AKM AK5388s, while four Cirrus Logic CS4398s provide the D-A conversion. The headphone amplifiers are based on Texas Instrument TPA6120A2 chips, optimised for use with low-impedance headphones.
The main digital board carries either two or four Analogue Devices' ADSP21469 SHARC processors and their supporting components, with an ARM STM32F1 processor and a Xilink Spartan-6 FPGA (field programmable gate array) to control everything. The Firewire interface is handled by a Texas Instruments TSB81BA3 transceiver.
One of the key selling points of the Apollo is that it makes the UAD2's low-latency mode available to laptop users for the first time. In tests, this worked fine for me with most plug-ins, although, as UA point out on their web site, some of their mastering plug-ins introduce more latency than some people will be comfortable with in this situation. Still, I can see this being very useful for anyone wanting, for example, to track direct to 'tape' using UA's Studer tape emulation plug-in, for example. Indeed, as the processing is all done in the Apollo itself, you can operate it as a stand-alone processor running with imperceptible latency. Brilliant!
Accompanying the launch of the Apollo was a new plug-in bundle, the 1176 Classic Limiter Collection, which provides brand-new emulations of their 1176. Actually, make that '1176s' , because there have been several revisions over the years. Unlike the Waves CLA76 (one of the better competitors), which offers a choice of two separate models in a single plug-in, UA have created three separate plug-ins in this bundle, which costs $299. Each plug-in mimics a different hardware model: a Rev A 'blue-stripe', silver-face model, based on the original FET compressor created by UA's founder Bill Putnam; a 1970s 1176LN revision E (LN stands for 'low noise'), which brought one of the biggest changes to the design; and UA's hot-rodded 40th anniversary version, the 1176AE.
I've used a number of different revisions of the 1176 and, while all share the same basic FET compression character, they leave rather different sonic imprints — but noise aside, which one is 'best' is largely a matter of personal taste. That said, all of the new plug-ins sound excellent. I used them on a range of sources for tracking and mixing, A/B'ing them against other manufacturers' plug-ins along the way. The Rev A sound is harder and brasher than the LN, almost a little more deliberately 'trashy'. It offers a distinct retro character that could prove particularly useful on vocals in the right track. It wouldn't be my first choice on a drum bus, but thankfully that's one role in which the 1176LN Rev E excels. By way of comparison, I found the Waves CLA76 set on Blackface mode to sound rather harder and punchier than UA's 1176LN Rev E, and while I'm sure that the Waves version will suit some material — hard-hitting pop mixes, for instance — as an out-and-out analogue emulation I much preferred the UA version: it still sounded punchy, but in the rather warmer, fuller, more rounded way that I'd expect from the real thing. More of a rock and less of a pop vibe, if you like.
Sonically, the Anniversary Edition lies somewhere between the other two, but it also includes slightly different features. There's no 'all buttons' mode, for example, and no 12:1 ratio, but there's a gentler 2:1 than you'll find on the others, which should make it useful in situations (tracking, bus-compression) where you might not want to squish things so hard.
While I love the sound of all three plug-ins, I missed a couple of easy-to-implement enhanced features such as a wet/dry mix control and a side-chain high-pass filter (as you find on, for example, Softube's FET Compressor), which would really add to these plug-ins' versatility.
It's also worth mentioning that the driver software you must install to run the 1176 bundle is the first that doesn't support UAD1 cards: you'll need a UAD2 or an Apollo. Perhaps this is for the best, as running all three versions of the 1176 concurrently used 12 percent of my UAD2 Quad's available DSP power, compared with four percent for three instances of the old 1176LN.
These minor quibbles aside, then, what we have here are yet more wonderful-sounding and faithful software recreations of classic analogue outboard from Universal Audio. I doubt I'll give up on hardware for a while, but that day is getting nearer, and these will probably become my go-to software choices for 1176-style compression.