"Those who cannot remember the past are condemned to repeat it." George Santayana said that, and I have tried hard throughout my life to develop at least a modicum of historical knowledge because I believe it. I've seen its truth illustrated over and over again in a number of different spheres of human life. Sadly, I still get bitten by it every now and then despite how hard I try to avoid its sting. This is one of those times.
The root of my foolishness lies in my choice of hardware for my home recording studio. As I've written previously, I had some nasty problems with the Mark of the Unicorn (MotU) 828 mkII device that I initially purchased. The MotU "technical support" was such a joke (and not a funny one) that I eventually ended up switching to a Presonus Firestation, which was a much better choice than the MotU unit for several reasons.
Or at least it seemed like a much better choice for nearly a month. My primary problem with the unit was that it would sometimes lose synchronization with the computer. When this happened, it was impossible to record or play back audio because nothing but clicks, pops, and nasty digital squeals would be transmitted from computer to Firestation and vice versa. I discovered relatively early in my use of the device that re-establishing sync could be achieved by switching the Firestation's sample rate using the mLAN control panel, and then re-applying my preferred mLAN configuration using the mLAN patch bay. That was, as my reader can no doubt imagine, a somewhat tedious two-step process, but it was a workable solution because the device was losing sync but a couple of times every few hours.
Unfortunately, and for no apparent reason, my Firestation simultaneously started losing sync far more often and stopped responding to my recovery technique just a few days ago. On Tuesday of this week I was surprised when it was losing sync after a period of roughly five minutes. I was still more surprised when I couldn't re-establish sync until after I had completely powered down both the Firestation and the computer. I was all the more surprised and far more displeased when I discovered that rebooting both devices was now the only way to restore sync, for using the mLAN control panel and patch bay no longer made any difference. That tedious two-step process was looking better all the time.
Since the problems had become intolerable, I called Presonus' technical support. Unfortunately, they were unable to do anything for me. I tried just about everything over the course of the next couple of days. I disabled Windows XP's system-restore feature, switched my computer from using ACPI to being configured as a "Standard PC", boosted the PCI latency of the firewire ports in use, disabled networking, used the stock Microsoft VGA video driver instead of the Catalyst driver for my ATI Radeon 9700 Pro, disabled lots of Windows XP services and other programs running in the background, ad infinitum, ad nauseam. None of these changes made any difference. I simply could not get the Firestation to retain sync.
And so it was time to head back to the store, speaking of which I should really put in a plug for Sam Ash Music. In almost all of my tales of technological terror I end up making multiple trips to my local Fry's Electronics. I give them as much of my business as possible for two reasons: (1) I prefer to support local merchants, and (2) they stand by the products they sell. Because so few hardware devices actually work anything like they're supposed to, or are radically incompatible with so many other devices, I would be wholly unable to do half the things I do with my computer were it not for local merchants like Fry's. Why? Because if I purchased things over the Internet I would have to pay multiple restocking fees, shipping and handling fees, and spend weeks waiting for shipping to sift through all the hardware dross to find the rare bit of gold that works.
As Fry's is to computers and general electronics so Sam Ash is to all things musical. I feel sorry for Claude, my contact at my local Sam Ash, because he's had to deal with my problems. But Claude takes it all in stride because the Sam Ash philosophy is to keep working with the customer until he's happy. I bought my original MotU 828 mkII from my local Sam Ash store. When it turned out to be defective they replaced it with a second MotU 828 mkII unit some weeks later. When I ultimately decided I couldn't make it work with my system I returned it in exchange for the Firestation. When the Firestation started acting up this week I called Claude to ask about my options, since I was nearing the limits of the Sam Ash 30-day return window. Thankfully, Claude told me that I could go beyond that window if needed because, in his own words, he "wanted me to be happy". You just can't put a price on that kind of commitment to customer satisfaction.
I used to do as much business as I could with mail-order companies and Internet stores, both because of their great convenience and because I could often avoid paying our state's ridiculous sales tax by doing so. Despite the inconvenience and ridiculous sales tax, however, I've switched over the years to doing my business with local merchants whenever possible. Sure, I pay a few extra dollars here and there for products that I could get more cheaply elsewhere. But how many mail-order or Internet merchants are going to work with me like Fry's or Sam Ash? How many are going to let me try and return two, three, or more devices without fees or any questions asked until I get something that actually works with my system?
Plug aside, I again found myself in the market for a new hardware I/O unit, and this time I decided to go with a PCI-based solution. Both the MotU and Presonus units relied upon a Firewire connection, and in both cases at least some of my problems were clearly due to communication problems. I feared that if I bought a third unit that used Firewire I might have the same sorts of problems.
Thus, I did some further research and quickly narrowed the playing field to three competitors. My needs are pretty specific. I need a device that has at least four, balanced, analog inputs and outputs; at least one optical input and output, which must support at least S/PDIF and ADAT protocols; at least one MIDI input and input; and at least one headphone jack. At least two microphone inputs and preamps would additionally be quite useful for recording vocals and the like, but they aren't really essential given my primary interest in orchestral music.
My starting point was the RME Multiface because of all the good things I had heard about it. From reading posts on Steinberg's Cubase Forums and ProRec, I had become curious about the nearly rabid fans of RME equipment, so I had to give that company's products a look. Sure enough, the Multiface is a very capable unit. It sports eight balanced, analog inputs and outputs, optical I/O, MIDI I/O, a headphone jack, and can be used in conjunction with other RME devices through a single set of drivers. Better still, it's powerful enough that it introduces no substantive load on the CPU, even when streaming data through all of its channels at once. That's pretty impressive. Unfortunately, the Multiface doesn't have any microphone preamps and is somewhat more expensive than either of the two units I'd already purchased (around $900 compared to $700 for the Firestation), so I decided I would leave it on the "short list" and investigate further.
The second unit I found quite interesting was the Aardvark Q10. It has no less than eight discrete, microphone preamps; eight balanced, analog inputs; optical I/O; MIDI I/O; and even a special high-impedence guitar input for direct recording. That's an awful lot of I/O capability for a unit so reasonably priced (around $800 at the time of this writing). I wasn't wild about the bizarre, purple case—whatever happened to the basic black/gray/silver of more "traditional" audio hardware?—but I could probably live with that "feature". Unfortunately, I found some reviews of the Q10 that mentioned questionable driver support for Windows XP, a complete lack of any bass-cut switches for the microphone inputs (which is an odd omission), and some questionable audio specs for the preamps. I could live without the bass-cut switches, but given the problems I'd had already I wanted rock-solid drivers and clean audio.
I really thought I had found the unit for me when I next investigated the WaMi Rack 192X. Not only does it feature enough inputs and outputs to keep me happy, it features sampling rates up to an incredible 192 KHz. I was somewhat disappointed with the Firestation because its highest sampling rate is 48 KHz. Both the Multiface and the Q10 would sample at up to 96 KHz. at 24-bit precision, but why not go for a device that supports 192 KHz. if you can get it? I can't imagine myself ever using that super-fast rate, but it's always nice to have a little "headroom", right? Better still, the 192X was also cheaper than any of the other units (around $600 at the time of this writing).
Perhaps best of all, given the hassles I had previously with drivers, was that the 192X comes with "Enhanced Windows Driver Model" (E-WDM) drivers. The E-WDM drivers are really slick, providing a one-stop solution for ASIO 2.0, DirectSound, GSIF, and MME formats. What's more, each of those four formats allows multi-client connections, which means that the user can run any number of software applications using each of those protocols simultaneously. Amazingly enough there's still more: the 192X ships with a "DirectWire" utility that allows the user to digitally route data from any of those four formats to any of the others. So, that makes it possible for the user to take the audio output from any application and route it directly to any other application, whether the two "understand" the same format(s) or not. That's powerful. The driver sophistication alone was practically making me salivate.
But alas it was not to be. My ignorance was temporarily quite blissful, but I'm glad that I didn't stay ignorant. A little more digging revealed that the 192X has two, very significant Achilles heels: its audio quality and crosstalk. Various web sites conducting testing with the nifty and soon-to-be-ubiquitous RightMark Audio Analyzer all show the same thing: the 192X has questionable audio quality and absolutely terrible crosstalk. For sake of comparison, the other units I've examined have crosstalk at roughly -110 dB or better, whereas the 192X has crosstalk at roughly -50 dB or worse at some frequencies. Relatively speaking, that's a lot of bleeding between channels, and I don't need that kind of crosstalk fouling up my mixes. Mixing is hard enough without dealing with channel-bleed artifacts. Thus my ignorance was shattered, and as painful as it was after getting so excited I simply couldn't buy the 192X.
My choice was clear: I would buy the Multiface and live without any microphone preamps. I consoled myself with the knowledge that any high quality, stand-alone microphone preamp would probably give better results anyway and headed back to my local Sam Ash store. In a surprisingly short time I had exchanged the Firestation for a Multiface and returned home. I had already read through the manuals on-line, so I knew how to install it. I made sure to download the latest driver first, powered down my computer, installed the card, hooked it all up, and powered on my computer. The driver installation seemed to finish without any issues, and after another reboot I was ready to get to work.
What happened then surprised me greatly, for it highlight's RME's brilliance in writing drivers. After my computer finished re-loading Windows XP I saw that I had two new icons in my system tray. I clicked the one that would let me configure the Multiface, but it resulted in a dialog box which explained that the firmware on my Multiface was too old to be accessed by the drivers I had installed. Imagine that! No crashing or other ridiculous behaviors! Way to go, RME! Better still, the dialog box provided me a URL which when clicked initiated a download of the proper firmware-update utility. That's exactly how this stuff should work, and I doff my cap to RME for getting it right. After a quick download I ran the update tool, powered the system down, powered it back up, and I had a working Multiface.
Sort of. My first test of the device was to launch Kontakt, my software sampler, and test the Multiface MIDI and audio quality by playing the piano. Unfortunately, Kontakt wasn't receiving any MIDI data. The MIDI input light on the front panel of the Multiface indicated that it was indeed receiving MIDI data from my keyboard, but it apparently wasn't forwarding that data to the computer or the computer wasn't "listening". Fear of further configuration hell stalked me at the periphery of my awareness as I launched Cubase SX to try playing back some audio instead. My defenses against that fear were buttressed when Cubase played back audio from several of my projects without incident. So, the Multiface was working beautifully at streaming audio but MIDI wasn't working; what could that mean?
I'll tell you what it means: Microsoft consistently makes some of the the stupidest design decisions in the history of writing software. This one is a real beauty, so bear with me as I explain. As I discovered from a quick search of Google—how did I ever deal with this crap prior to the web?!—this is a known problem with a FAQ entry on the RME web site. When Microsoft created Windows NT, some genius decided that it would be a good idea to limit the number of MIDI devices that could be installed to ten. Such arbitrary limits are pretty silly to begin with. After all, does anyone else remember the 16-bit resource limitations of Windows 1.x that basically persisted all the way through Windows ME more than a decade later?
Still, a limit of ten MIDI devices isn't really so bad because even the most elaborate systems probably don't have more than three or four at most. But some genius, perhaps the same genius, decided that Windows NT would make a new entry in the table of installed MIDI devices whenever any of the associated hardware was moved or when any of the relevant drivers were updated. In other words, a user with a MIDI interface on a USB device can fill up his table completely in a mere minute or two by plugging and unplugging the device a few times. As if that weren't bad enough, merely updating the drivers causes the same thing to happen!
Even this wouldn't have been fatal were it not for the efforts of some genius at Microsoft, again perhaps the same genius as before, who decided that once the table is full, no other MIDI devices can be installed no matter how many of the entries in the table are duplicates. In other words, Microsoft did three increasingly stupid things: (1) they arbitrarily limited the number of devices to ten, (2) they made sure that the table could be filled up ridiculously quickly thanks to the almost unbelievably stupid behavior of Windows NT when moving devices or updating drivers, and (3) they made no efforts whatsoever to clear the table of any duplicate entries. This is one more proof of why I say Microsoft couldn't code their way out of a wet, paper sack. Seriously, it's a bloody miracle that Windows even runs in light of the stupid design decisions that permeate so many of its components.
At any rate, once I cleaned up the table—for the record I had a whopping two unique entries and eight duplicates—and reinstalled the drivers, a quick reboot rendered the Multiface fully functional. Its MIDI interface was working as well as its audio. I tested the device with several applications and messed around for a couple of hours without running into any of the problems I'd seen with the Firewire-based I/O devices. It didn't lose sync even once, and I heard no clicks or pops whatsoever once I found a good latency setting for my system.
Which brings me to the main thrust of this essay: I should have known better! I still remember when USB first came into vogue when Windows 95 was still shiny and new. Microsoft's "Plug and Play" was jokingly referred to by almost everyone in the know as "Plug and Pray" because that better expressed how well it worked. I learned my lesson then, deciding to eschew all USB hardware for as long as it took for the technology to mature. Sure enough, that was a smart call on my part. I've had some USB problems, but by the time I started buying USB devices most of the kinks had been worked out. I saved myself a lot of hassles simply by avoiding the new technology from the outset.
So why on God's green Earth did I think I could get away with using Firewire now?! Apple first introduced it into their computers in the year 2000, if memory serves, and that makes it less than three years old on the PC. What the heck was I thinking?! I shouldn't go anywhere near Firewire until at least 2005. For the record, I think Firewire is probably trustworthy today for the vast majority of its applications, which involve nothing more demanding than high-speed data transfer. But I was trying to use it for high-speed synchronized data transfer. Consistent, single-sample accuracy at 96 KHz. requires a reliable timing accuracy of around ten millionths of a second. That's a relatively large interval in terms of contemporary CPU speeds, but it's pretty brief when it comes to the timing of devices that make up the majority of the Firewire market (e.g., external hard drives, digital video cameras, etc.).
Thus, I've yet again learned the truth of Santayana's famous adage, and I intend to avoid Firewire like the plague for all but the most basic of applications, at least until it has been around for a good five years or so. In the meantime, the RME Multiface is working beautifully. Its drivers are rock-solid from what I've seen, its audio quality is absolutely pristine, and it functions exactly as promised without any hassles beyond those generated by Microsoft's seemingly infinite supply of stupidity. Every time I use it I gain new confidence in its abilities and trust in its reliability. I begin to think that my home recording studio is finally assembled, and that I can finally stop fighting the technology and start using it. That's a really nice feeling.
10/10/2003