Author Topic: nuendo audiolink 96 (2002)  (Read 735 times)

0 Members and 1 Guest are viewing this topic.

Online chrisNova777

  • Administrator
  • Posts: 6765
  • Gender: Male
  • "Vintage MIDI Sequencing + Audio Production"
    • | vintage audio production software + hardware info
nuendo audiolink 96 (2002)
« on: December 26, 2017, 03:43:12 PM »

Steinberg are keen to establish their Nuendo brand in the hardware arena, and to this end have formed an ongoing alliance with German soundcard manufacturer RME.

The Nuendo Audiolink 96 system on test here thus has an identical counterpart in RME's Hammerfall DSP range, while other items in the Nuendo range include the 9652 -- a rebadged RME Hammerfall DIGI 9652 -- and the DD8, also known as RME's ADI8 DD. Unlike the 9652 (first reviewed as an RME product by Martin Walker in SOS September 1999), however, the audiolink system is more than just a PCI card.

Currently, the items available are two interface cards -- one a PCI type and one a PCMCIA type -- and two breakout boxes, the Digiset and Multiset. For the review, I was supplied with both interface cards and the Multiset breakout box. Drivers and utilities are supplied for both Windows and Mac OS 9, and there are Mac OS X versions slated for imminent release; currently, the units are supported under Windows 98, 2000 and XP, and Mac OS 9.1 and above. RME, never content to rest on their laurels, plan further additions to the range as well as developments to the driver software to give new features.


Finished in Nuendo's now familiar silver and pale blue livery, the Digiset and Multiset are both half-width rack units (a matching full-width rack tray to house the two units side by side is available). The Digiset offers three ADAT optical digital inputs and outputs (making 24 channels in and out) plus stereo SPDIF digital I/O, a nine-pin ADAT Sync port, word clock I/O via BNC connectors, two pairs of MIDI I/O and a headphone output. The Multiset has more of an analogue focus, with eight analogue inputs and outputs via 16 TRS balanced quarter-inch jack sockets, a single eight-channel ADAT digital input and output and stereo SPDIF digital I/O, along with the ADAT Sync port and word clock I/O, one MIDI input and output, and a headphone output.

The front panels of both devices are very similar, with a pair of MIDI sockets on the left-hand side (the Digiset's second pair are on the back panel). To the right of these are LEDs that indicate MIDI activity, the presence of word clock, S/PDIF or ADAT Sync, a warning light that illuminates if breakdown in communication with the host card occurs, and finally a quarter-inch jack socket which can act as either an analogue output pair or a headphone jack. The breakout boxes interface with the host cards through a FireWire port, but cannot be connected directly to a computer's FireWire port as they use a different protocol.

The Digiset will provide a maximum of 26 inputs and outputs at 48kHz, which becomes 14 if you are working at 96kHz. This is because the ADAT protocol supports eight-channel audio transfer only at 48kHz or below. RME have applied a technique they call 'Sample Split' to achieve four channels of 96kHz audio per cable, so 12 ADAT channels and two S/PDIF channels can be obtained from this box at high sample rates. The Multiset offers a maximum of 18 channels of input and output at 44.1/48kHz, which again reduces to 14 when operating at 96kHz (eight analogue channels, four ADAT channels and two S/PDIF).

Connection Options

As you may have gathered by now, the Audiolink system is modular, but if you buy both cards and both breakout boxes you will not be able to connect both Digiset and Multiset to one of your computers, because they cannot be linked or chained like FireWire devices. Each box requires its own card, so two typical setups could be as follows:

-You have a desktop and a laptop computer, each one has an appropriate card, and you move your breakout box of choice between the two.

-You have a desktop computer, but need 32 ADAT channels. You need two PCI cards, and could run a Digiset from one and a Multiset from the other.

The Multiset box comes with an optical cable, an installation disc containing drivers and applications (for Windows and Macintosh) and an operation manual. The Audiolink 96 PCI kit contains, as well as the PCI card, another copy of the CD, a three-pin internal cable and a FireWire cable. The Audiolink 96 Mobile kit contains the PCMCIA card, a copy of the CD and a CardBus-to-FireWire cable. There are also three power-cabling options, since the CardBus interface does not supply power to the breakout boxes. There is a small-form 12V mains power adaptor with mains lead, a 12V car cigarette-lighter power connector similar to a mobile phone car charger, and a battery cable for a European battery size with which I am not familiar. The mains lead supplied with the review unit was also the European two-pin type, and Steinberg UK need to pay a little more attention to this packaging detail if they are to rebadge German hardware successfully for the British market.

PCI Installation

The small-form (12x6 cm) 32-bit PCI card slotted easily into a spare slot in the review machine (a dual-1GHz Power Mac) with connection to the Multiset just as easy as any FireWire connection. The installer CD contains quite a lot of software support and updates for various products, and the first thing I checked out was the readme file for the Audiolink 96.  A couple of points drew my attention, the first of which was that the driver software installed by the CD might require the firmware of the PCI card to be updated. Initially I was unable to find the updater, but it turns out be installed when the driver install utility is run, and can only therefore be run after the drivers have been installed(!) Running this firmware updater informed me that the card was up to date, so I needn't have worried.

The second item in the read-me that drew my attention was a list of features not to be found in the driver version installed by the CD. I immediately checked the Nuendo web site but was disappointed to find that the driver version was the same as supplied with the CD (2.01). Popping over to RME's web site revealed a newer version of the driver (2.10), but Steinberg technical support advised against its use, on the grounds that it was not yet tested with their software. The missing features (which are actually still missing with RME's driver version 2.10) include software-controlled varispeed of ADAT machines, and PCI Check, a display of the computer's PCI/CardBus efficiency. I am looking forward to seeing these in action.

The Audiolink installer places the Audiolink 96 ASIO driver and Settings control panel in the ASIO Drivers folder of the software, which you choose during installation. If you run more than one, you will need to run the installer again and choose the other(s) or simply copy these two items to each ASIO Drivers folder. An extension is placed in the Extensions folder, so a restart is required before proceeding.

The Audiolink hardware is made available to Nuendo by choosing Device Setup from the Devices menu; you then choose VST Multitrack in the list of devices and select Audiolink 96 ASIO in the ASIO driver selection panel. This window also allows you to specify the number and size of the disk buffers. Clicking Control Panel on this window opens the Audiolink 96 Settings utility.

In Logic, you select Audio Hardware and Drivers from the Audio menu, then select the Audiolink ASIO driver in the driver selection panel. Clicking Control Panel again opens the Settings panel, a welcome change from some other soundcards (such as the MOTU 828 and 896) that require Logic to be closed down before any changes can be made to the settings.

PCMCIA Installation

The CardBus card connects to the Multiset using the supplied cable, which is then inserted into the PCMCIA slot on the PowerBook. The power jack from the switching power supply is then inserted into the auxiliary power socket on the back of the Multiset -- the PCMCIA card is unable to deliver power to this unit, unlike the PCI card. After connecting the power supply to the mains, I booted the laptop and the card announced its presence on the desktop as 'Hammerfall DSP'. Mac users are required to drag this icon to the trash before ejecting the card. Software installation is the same as for the PCI card.

Setting Up

The digital I/O on the Multiset is made more flexible by the Audiolink Settings utility. This allows the optical output to become another S/PDIF output in parallel with the co-axial one. The user can also select the type of audio being sent to the S/PDIF output, with a choice of Professional, which doubles the output voltage level to 1.6V to permit interfacing with AES-EBU gear if an appropriate phono-to-XLR adaptor cable is made up, Emphasis (for when pre-emphasis has been used on the audio material), and Non-Audio, which allows the playback of AC3 signals.

Also in this utility under Clock Mode is an Autosync option, which allows the use of the internal master clock until a valid external sync is detected, in which case it will slave to any sample rate between 25kHz and 105kHz. You can force the device to run from the internal clock only by ticking off Master, or if you have more than one external connection providing valid sync source, you can choose which one to slave under Pref Sync Ref.

In Use

All the Multiset's inputs and outputs appeared in Nuendo and Logic as expected. The only bugbear was Logic's usual refusal to label different types of output -- to select ADAT output 1 from the Multiset, for instance, you have to remember that ADAT outputs are numbered 9-16 (on a mono channel), 1-8 being analogue and 17-18 being S/PDIF. In Nuendo this isn't a problem, everything being correctly labelled in the VST Outputs window.

Steinberg also provide a monitoring utility called TotalMix which has several useful features, despite its very basic graphic layout. With a total lack of trim controls (except for hardware jumpers inside the Multiset), or for that matter XLR inputs and mic preamps, you'd be right in thinking that the Multiset and Digiset are designed to be used with a hardware mixer when recording. However, the TotalMix utility can be used to monitor any combination of inputs (with zero latency) and playback channels via any outputs; you can save eight different setups per card installed and even set up a separate submix for the headphone output.

Both Multiset and Digiset provide MIDI in and out (16 channels on the Multiset, 32 on the Digiset) and will require the use of good old OMS if you are a Mac user. Make sure you have the most up-to–date drivers, though, as OMS's control panel does not function correctly with earlier versions.


RME's technical information promises data transfer rates of up to 130MB/s from both the PCI and PCMCIA interfaces. This should easily support very high track counts at 16-bit/44.1kHz, depending on the performance of your drive. Since the audiolink supports 24/96 operation I decided to run some track count tests using Nuendo and Logic on desktop/PCI Mac and Powerbook/PCMCIA combinations. I used the 24-bit/96kHz 'continuous' audio files which formed the basis of some of my tests in the FireWire Drives article in last month's issue, and the results are summarised in the table above.

While Logic has fairly limited optimisation options -- the Larger disk buffer option in Audio Hardware and Drivers was selected -- the number of samples per buffer, chosen in the Audiolink Settings utility, has a slight bearing on the above results. In most cases a setting of 256 samples was found to give best performance, especially with the desktop machine. My conjecture is that this relatively small buffer size allows the processor(s) to share more of the workload with the hard drive. The one figure that deviates slightly from the norm in the table is where the single track of recording reduces Nuendo playback from 40 to 37 tracks on the Powerbook. The impact of recording activity is obviously greater here, where the processor is less capable.

Although RME/Steinberg claim similar figures for data transfer rate for both the PCI and CardBus interfaces, and further that on-card processing minimises load on the host, track counts were down on the Powerbook. Although this had only one CPU operating at nearly half the speed, the limiting factor as shown by the performance indicators in both applications was the hard drive -- and since the hard drive and associated files were identical in both cases, one can only assume that the CardBus interface is slightly less efficient than its PCI equivalent.

I was also interested to see how the Audiolink interfaces compared, in terms of track count, with MOTU's 896 FireWire device. Logic played back 32 tracks of 24/96 audio from the same drive using the 896 on a G4 800MHz iMac, which is only one track more than the Audiolink/ Powerbook combination and 11 tracks fewer than the Audiolink/ Power Mac combination. Although like-for-like comparison cannot be made, the PCI/PCMCIA system appears to offer better data transfer rates than FireWire. What is also clear is that Nuendo and its hardware are optimised for exactly this kind of project, the Audiolink system enhancing Nuendo's reputation for high track counts.

Low Overheads?

Although RME/Steinberg claim that the system should use no host CPU overhead, even at the lowest latency figure, it was noted by Martin Walker in his review of the RME DIGI 9652 that the smaller the buffer size, the more IRQs (interrupt requests) are made on the PCI buss, which will show up as activity on the CPU performance meter of the DAW software. I decided to put this to a practical test by counting the number of plug-ins available at the two extremes of latency offered by the soundcard.

As in previous test, I added mono instances of PlatinumVerb as inserts in Logic Audio until the processor overload warning message was given. The maximum working number (ie. one below this) was the number I recorded as the test result. In Nuendo I added Nuendo Reverb plug-ins as inserts (four per channel) but found that a finite overload point was more difficult to pinpoint. As in Cubase, as the CPU performance meter level increased, operations in the software were dramatically slowed and in each case the program hung (not a 'soft' crash, either -- the machine had to be restarted each time) before the overload light lit up. The number of Nuendo Reverbs recorded was therefore the one below that which caused the hang.

What is curious here is that although the maximum number of Logic plug-ins reduces by three at the minimum sample buffer size (a performance decrease of 10.3 percent), the Nuendo plug-in count is reduced by 54.2 percent. As I noted in my Mac processor upgrades feature in SOS December 2001, the number of VST plug-ins available to Cubase (and now Nuendo) responds to available processing power in a more linear way than native plug-ins in Logic. PlatinumVerb, as Logic's most CPU-intensive native plug-in, demonstrates this more effectively than any other, suggesting a reliance on other hardware (and software) factors. The most likely of these is probably that the two applications employ the second processor on the dual-1GHz Mac in different ways.

Scope For Improvement

Using ChannelD's Mac The Scope, an integrated software signal generator and oscilloscope, I made some measurements that might help to quantify sound quality. Pink noise with no low-cut filter revealed a frequency response of combined analogue input and output of +0/-0.5 dB from 20Hz to 20kHz at a sample rate of 44.1kHz. This did, however, roll off to -3dB at 6Hz. I measured the Multiset's noise floor at -132dB against digital full scale, but probably the most impressive measurement was that for jitter, which was better than -123dB. This low level of noise due to clock jitter bears out Steinberg/RME's claims for the stability of their clocking, which gives rise to excellent clarity of spatial information.

Final Thoughts

While the Audiolink system does have a few limitations, such as the lack of mic preamps, it does offer the best of two worlds, laptop and desktop, for those not able or willing to take the FireWire route. Adding high audio quality, 24/96 capability (not necessarily the same thing!) and comprehensive sync options to either your desktop or laptop (or both) makes for a very flexible system indeed. While FireWire interfaces such as the MOTU 896 offer greater simplicity and elegance in terms of connection to the host computer, this system does offer slightly better performance, especially if you are a Nuendo user.

Users of other software should not be put off by the Steinberg badge (you could always seek out the RME original), however the niche for this product will probably be with those who are happy to use an external hardware mixer and wish to use the same hardware with a variety of computers, since it is equally at home on Windows or Mac, desktop or laptop.
« Last Edit: December 26, 2017, 04:34:29 PM by chrisNova777 »

Online chrisNova777

  • Administrator
  • Posts: 6765
  • Gender: Male
  • "Vintage MIDI Sequencing + Audio Production"
    • | vintage audio production software + hardware info
Re: nuendo audiolink 96 (2002)
« Reply #1 on: December 26, 2017, 04:49:10 PM »
links to other related harware:

Steinberg Nuendo 8 I/O (2002?) 96Khz A/D D/A Converter,4929