Paul Wiffen concludes his investigation into reliable means of audio interfacing on new‑style Macintoshes when the traditional PCI card route is not available to you. This month, he tests two more USB audio‑only interfaces, and reaches some more definitive conclusions about Tascam's US428 MIDI and audio interface. This is the last article in a four‑part series.
Last month, having said in the previous issue that I would road‑test various bits of USB hardware, I was only able to obtain two Mac‑compatible devices: Edirol's UA30 audio‑only interface, and Tascam's US428 control surface/MIDI and audio interface. My findings with the Tascam were then further compromised, as I was unable to get it to work correctly with Cubase VST v4.1 on the Mac (the then‑current version of the software). This month, I had rather more luck. Not only did I successfully test two new audio‑only interfaces, I also progressed with the US428 to the point where I feel I can now justifiably reach some reasonable conclusions about the Universal Serial bus, MIDI, and audio.
The first audio‑only interface, Media Assistance's 20‑bit USB One, came from Germany and is now being distributed in the UK by Et Cetera. Drawing power from USB, this device gives stereo analogue in and out, plus co‑axial and optical S/PDIF in, and worked first time without any drivers being installed when I plugged it in to a Mac OS 9.0.4‑equipped PowerBook. An audio CD was playing on the internal drive when I did this, and just a few seconds passed before the system recognised the device and switched over to playing via the USB One. The Sound Control Panel now offered USB Audio as one of the available inputs and this also worked immediately on the stereo analogue input when selected. It did however reveal the biggest problem with Sound Manager USB audio; a latency well in excess of 100 milliseconds. There also seemed to be no way to get the digital inputs to work under Sound Manager, presumably because of the lack of a System Control Panel specific to the USB One.
However, the solution to both these problems was provided by downloading the USB ASIO driver from the Media Assistance web site — and it was no great surprise to discover that this was a customised version of the Propagamma USB ASIO driver mentioned last month. Rebooting with the extension in place makes the USB One available to ASIO‑enabled applications. I used Cubase VST v4.1, and immediately found the driver in the ASIO driver list in the Audio System window. Selecting it halved the latency at a stroke from 30 milliseconds with the Sound Manager ASIO driver to 14mS with the Propagamma driver.
The USB One is well specified for its price (currently £139). It includes direct through‑monitoring capabilities to help avoid latency problems, and a dither option for reduction from 20‑ to 16‑bit without truncation errors. Recording with the USB One via ASIO was flawless, with the signal appearing in Cubase VST's Audio Edit window as normal. However, as so often with USB audio devices, I found that playback was very patchy with unexpected interruptions to the signal. I assume this is because of the extra burden that the lower latency of ASIO puts on the CPU (although I was using a 500MHz G3). I guess the good news is that you can make recordings in confidence, even if the playback is somewhat unreliable. I did find that by switching off the Propagamma extension, rebooting and using the Sound Manager ASIO driver, I was able to achieve proper continuous audio playback from Cubase through the USB One, and whilst this is not the smoothest of working practices, I guess it does make for a usable system, provided you do not need digital outputs.
For almost £100 more, the Ego Sys U2A from Korea offers several more features, including 24‑bit converters, a sample‑rate converter and S/PDIF co‑axial and optical connectors for output as well as inputs. I also liked the fact that there are LEDs to show when the unit is being powered. The Mac drivers, which I downloaded from Ego Sys's web site, come with a Mac Control Panel, so you can use all these features under Sound Manager.
The U2A also began receiving desktop audio within a few seconds of being plugged in and the Input became available in the Sound Control Panel. Once I had put the Control Panel in the System folder, I was able to access both the analogue and digital inputs simultaneously, as well as mix the audio from the computer with these without even rebooting. There is a switch to decide whether you are sending the analogue or digital input down the USB cable to the software on the computer, but the U2A can also mix the analogue and digital inputs it is receiving and add the audio signal coming back up the USB cable, then send different combinations to both its analogue and digital outputs. The direct signals therefore suffer from no latency (as they haven't made the journey through the CPU via USB) and can be used for direct monitoring. You could simultaneously use the analogue outputs to mix your direct analogue input with the output of the computer (say for the musician to listen to whilst recording) and the other as a 'digital through' to monitor what is coming in on from your optical co‑axial source (CD player, DAT or Minidisc) on another digital device. As this could be on the co‑axial digital output, the U2A effectively doubles as a co‑axial‑to‑optical converter (which would cost you £50 on its own). Furthermore, because there is a parameter to change the Digital Out from Consumer to Professional format, you could even use it to hook S/PDIF‑only gear to AES‑EBU equipment. Add in the fact that if you set the U2A to its internal clock, it automatically acts as a sample‑rate converter, and you have a damn useful little box.
An ASIO driver came as part of the Ego Sys download, and turned out once again to be the one from Propagamma (they seem to be the only game in town for third‑party USB ASIO drivers on the Mac at the moment). As before, installing the driver reduced the latency to 14mS. On opening the Propagamma Control Panel, I was a little disappointed to find that only 16‑bit resolution was available (all the others were greyed out). I imagine that 24‑bit recording is not yet supported on the U2A (despite the presence of 24‑bit converters). Mind you, it took Digigram eight months to get that working on the VXPocket under ASIO, and all the U2A control options available under Sound Manager can also all be used under ASIO.
Recording with the U2A gave the same glitch‑free results as the USB One, but playback was subject to the same occasional interruptions. Once again, I had to go back to the ASIO Sound Manager to get uninterrupted playback all the time. At least with the U2A you don't lose any functionality when running under Sound Manager. I would like to see USB devices get to the stage where you don't have to swap back to Sound Manager operation to achieve reliable playback for mixdown, but at least these devices do provide audio I/O for PowerBooks and iMacs.
As mentioned at the start of this month's article, I also managed to do some more testing with the Tascam US428 this month. The first major step forward was that I was able to get it to run as a control surface with MOTU's Digital Performer v2.7, thanks to the US428 Performer plug‑in, which I downloaded from the Tascam web site. When I dropped it into Performer's plug‑in folder and ran FreeMIDI in OMS‑emulation mode, the control surface of the US428 began operating Performer's on‑screen controls immediately. However, moving the EQ controls merely produced a string of error bleeps from the Mac. Maybe I didn't have everything assigned exactly right.
With Performer, you still need to use both the ASIO drivers for audio I/O and the OMS driver (albeit under FreeMIDI) to run the two external MIDI Ports on the US428, so I thought I would try this to see if the results were the same or better than I achieved with Cubase VST v4.1 last month. Once again, the audio was fine, but I didn't even get as far as a wandering MIDI click in Performer. Although I could get the OMS Studio Setup to acknowledge MIDI input on Ports 1 & 2 in the Test Studio mode, I could not get Performer to receive any MIDI input via the US428. However, I am not the most experienced Performer/FreeMIDI user, so I may have been missing something. I have heard of people both here in and in America who have apparently had the US428 working flawlessly with Performer, but I like to see (and hear) things working for myself before I recommend them wholeheartedly. In conclusion, I would say that if you want to use the US428 as an audio interface for Digital Perfomer v2.7 or Cubase v4.1, you should do fine with it, and the level of control surface response I got with Performer was certainly worth having. When I can get the EQ controls working with Performer, I will let you all know through the Apple Notes column.
While I was on the Tascam web site, I found a link to 'important notes on the use of Mac Cubase VST v4.1' with the 428. I eagerly read the lot, but was disappointed to find that it merely contained all the things I had tried last month to no avail. However, the site notes concluded with the observation that full support for the US428 would be implemented in Cubase v5.0 for the Mac — and at the time of writing, this was on the verge of being released. So the SOS production team moved heaven and earth to get hold of the final‑release v5 in time for this issue, and a copy dropped through my letterbox just hours before the magazine's deadline. I hastily installed the software and US428 ASIO drivers on an early iMac running OS 9.0.4.
Once again, I was able to record four tracks of 16‑bit audio simultaneously without any problems, proving it had not previously been a fluke on the G3 — and so many people had told me not to expect that level of ability from USB! Cubase v5 reported a reasonable latency of around 25 milliseconds.
Unfortunately, when it came to MIDI operation, I initially ran into similar installation problems as last time; OMS immediately found Port 3, the US428 Control Port, but not Ports 1 and 2. Tascam are now aware of this problem, however, and advise opening a completely new OMS studio setup when incorporating the US428 into your system (I had originally been trying to update the existing one in the hope of retaining all the existing connections). I followed Tascam's advice, and discovered to my joy that OMS found not only all three ports for the US428 but also all the other OMS devices previously connected as well. I strongly advise that you likewise start from scratch if you plan to install the US428 yourself.
This done, I sat down to some serious testing. Firstly, I tried setting the MIDI click to a percussive sound, and listened to both that and the audio click together. Those of you who read last month's instalment of this series will remember that I found the MIDI click with Cubase VST v4.1 to be wandering so much that it was unusable. Not so any more — and this was on the old iMac, a much less powerful computer than the PowerBook I was using last month. The MIDI click and the audio click (both of which you can hear through the US428 if you feed your external MIDI devices through the Tascam's audio inputs) were rock‑solid with v5, which gave me the courage I had previously been lacking to venture further. I then opened up some existing MIDI files with 16 channels of heavy MIDI controller usage, and these worked fine too. I had to go as far as filling up the second MIDI Port's 16 channels before I started to hear audible timing problems. In fairness to the US428, friends with other MIDI‑only USB devices report the same problems with full use of a second 16 channels, so I am inclined to blame USB's bandwidth for this, not the 428's implementation.
However, the big test was still to come, the one which I had been planning since first receiving the US428. Could it deal with the four channels of audio at the same time as the MIDI? I unmuted the audio tracks I had recorded earlier, to try playing them back simultaneously with 16 busy MIDI tracks. Now there was no audio at all, not even in the meters on the VST Channel Mixer. I switched the ASIO driver back to Sound Manager and the audio reappeared. But when I switched it back to the US428 16‑bit ASIO driver it vanished again. The same thing happened when I quit Cubase and relaunched it, and switching the US428 off and on didn't help either. Nothing short of restarting the Mac would bring the audio connection back. Now in fairness, I had been pushing MIDI to the point where it had clearly been butting up against the bandwidth limit of USB, so I assume that this side of things had monopolised the USB connection and squeezed the audio out.
However, once the audio connection was re‑established, I found I could indeed play back all the audio tracks I had previously recorded side‑by‑side with a full 16 MIDI tracks without problems. I was even able to simultaneously record four tracks with this much MIDI going on. I only managed to 'break' the audio connection between Cubase and the 428 once in several hours of testing, and again this was following heavy MIDI usage on all 32 channels. When trying out the 428 on an iBook which a chance visitor made available for a couple of hours, I wasn't able to break the audio connection at all, even when MIDI excess started to tell in the timing (so a 450MHz processor may help avoid this, in contrast to the 233MHz version in the iMac I was using).
Suitably heartened, I felt ready to tackle the final area of the Tascam's potential; to try to get the control surface working. In Cubase v5, there is now an entry in the list of Remote protocols to select the US428 itself (this was not present in v4.1). When I selected this, I found that a new window popped up in VST offering me the opportunity to select which bank of eight channels in Cubase VST's mixer would be controlled by the US428. I hurriedly opened the Channel Mixer in Cubase and waggled the 428's faders but to no avail. Pressing the 428's transport buttons was similarly futile. The LEDs above the buttons did not even light, let alone produce any response from Cubase. However, the LEDs did light when I operated the transport controls or switched banks from the Cubase end. Clearly some level of communication was going on from Cubase to the 428, if not vice versa. According to Tascam, one of the forthcoming Cubase updates will solve the remote problems with the US428. Until then, MOTU's Digital Performer leads the field in being able to drive the 428 as a control surface on the Mac.
So, after all this testing, what do I feel about the suitability of the Universal Serial bus as a replacement for other ways of getting MIDI and audio in and out of the new Macs? The answer to that still has to come in two parts, I'm afraid, one positive and one negative. For recording audio with an ASIO driver, USB can do the job. In stereo, it can clearly deliver better quality for recording than the Mac's built‑in hardware, and the Tascam US428 pushes this up to four channels with no apparent problems. So where there is no other choice for expanding the audio side of your new Mac, you will be able to make excellent recordings in the field, whether you use the Tascam or one of the stereo devices which works with the Propagamma USB ASIO driver. If you are running software synths on your computer, I would go for the Propagamma‑driven devices, as their ASIO driver gets the ASIO latency down to 14mS, as opposed to 25 on the US428 and 27 on Sound Manager. Monitoring at source on the Tascam gets you around latency during recording, but not when triggering software synths, unfortunately!
On the MIDI side, it does seem possible to run a certain number of MIDI channels reliably over USB. This means that for knocking out ideas with a portable PowerBook or iBook, the Tascam US428 should work as well as any other USB device via OMS. However, bandwidth limitations and the antique nature of OMS's core code may mean that you may start to notice timing problems as you approach 32 channels (or sooner, if you are heavy‑handed with continuous controller data or like to send SysEx dumps in the middle of a song). Of course, on a blue‑and‑white G3 or on a G4, you may already have MIDI interfacing solved by one of the methods I recommended in the earlier instalments of this series (eg. the Stealth port, or MIDI via PCI cards). In this case, I can wholeheartedly recommend the US428 to handle the audio side alone, as it is the most reliable and (with four simultaneous inputs) the most capable USB audio interface I have so far encountered. However, I cannot put my hand on my heart and recommend it as a complete solution for a fuller level of MIDI operation — at least, not for Macs. PC owners seem to have had far fewer problems with the US428, not least because the interface ships with a free version of Steinberg Cubasis for PC which drives the US428 directly. Apparently Tascam hoped to do something similar for Mac users, but developing a Mac equivalent to the US428‑specific Cubase is taking longer than they anticipated. I hear that their temporary solution is going to involve bundling a US428‑specific version of BIAS's Peak LE software, but even this is not yet available at the time of writing. Maybe when it turns up, you will be able to fully exploit the 32 channels the 428 offers without hearing timing inconsistencies or squeezing the audio out of operation. Be sure that if I get to this stage with the US428, I will let you know in Apple Notes.
So far, the US428 is the only shipping USB combination MIDI and audio device which is recommended for use with the Mac, but other similar USB units are due for release any week now from Event and Yamaha. In the meantime, please keep your feedback coming in to email@example.com, as it is in my regular Apple Notes column that I will be reporting back and commenting on your USB experiences from now on. In the meantime, I give a cautious thumbs‑up for ASIO‑driven USB audio, even if the OMS‑driven MIDI side still seems limited on the Mac.