Windows '95 Resource Kit; MIDIman; PC Future

Taking over the PC Notes hot seat, Martin Walker looks at the future of PC audio, and exposes a few problem areas for soundcards.

During April, San Francisco will be hosting the Windows Hardware Engineering Conference (WinHEC), and one item on the agenda is the way forward for a new PC sound hardware standard. All software developers are heartily sick of having to cope with the support problems caused by running the dozens of different soundcards available for PCs; without a standard, few manufacturers are prepared to go out on a limb to support the PCI buss. The biggest problem is that mainstream PC audio is still largely driven by the games market, and this means maintaining SoundBlaster compatibility, as well as supporting the ageing ISA buss. Creative Labs were the first company to combine FM synthesis (as previously used by Adlib) and sampled capability. All subsequent cards, whatever their individual strengths, stood or fell by whether you could use them with games, and this meant compatibility with (or at least emulation of) SoundBlaster.

In June 1996, a group of companies including Intel, Creative Labs and Yamaha announced support for an audio component specification. The specification of AC97 was designed to ease the transition of audio onto the PCI buss, and was expected to result in products appearing in the second half of 1997. It demands a 20Hz‑20,000Hz frequency response, a minimum 75dB signal‑to‑noise ratio and THD (total harmonic distortion) of 0.02% maximum. Most significantly, the spec involves a two‑chip design, separating the analogue and digital parts of the audio circuitry. The digital controller handles the movement of audio data across the PCI buss, and the audio codec (compression/decompression) chip converts the digital audio stream back into an analogue waveform. The two devices are connected via a 5‑wire serial buss.

The beauty of the two‑chip approach is that the audio chip can be kept well away from electrically noisy circuitry, giving potential signal‑to‑noise ratios of around 90dB, even if it's incorporated onto the motherboard. However, high‑end manufacturers would like to use the Universal Serial bus (USB) or the IEEE1394 (Firewire) buss instead of the AC97's proposed 5‑wire serial solution, as these would allow the digital audio to be sent to external peripherals. Unfortunately, this pioneering attitude is not shared by mainstream manufacturers, who are only interested in smoothing the transition to the PCI buss. All we can hope is that WinHEC will thrash out the details and resolve the SoundBlaster question, so that everybody can get on with producing a range of PCI soundcards that will finally improve the tarnished image of PC audio.

Soundcards And Unsound Cards

You may have noticed several reports in SOS over the last few months of musicians with a PC and soundcard that refuse to work together. In an effort to help those in trouble on the PC, I intend to cover problem areas on this page — not from individuals, I hasten to add, since manufacturers' help lines and email facilities are the best way to address these (that saved the postman from a hernia!), but known problems that have been acknowledged by the manufacturer, along with any fixes or advice that are available.

To kick off, Digidesign have had a lot of problems with their Audiomedia III card and PC motherboards that use PCI Burst mode — this isn't the burst rate of hard disk drives, nor the DRAM Burst read and writes used by EDO RAM, but the way that data is addressed via the PCI buss. In Burst mode, large amounts of data can pass between the CPU and a PCI card in one continuous... well, burst. To avoid inexplicable clicks or DAE errors, Digidesign instruct all Audiomedia III users to disable PCI Burst mode via the BIOS. I can't help thinking that, whatever the performance boost, asking users to delve into BIOS settings may be asking for trouble, as changing some settings other than this one can stop your PC booting up at all. Besides, disabling PCI Burst mode to achieve better performance with the Audiomedia III would seem to compromise all other areas of PCI hardware performance, including graphics cards and SCSI host adapters.

The problem is compounded if your PC has the Triton 430HX, 430VX, or 440VX chip sets (and these account for most bought nowadays), since it's more than likely that a setting to disable PCI Burst mode will not be available in your BIOS, and in this case the Audiomedia III card will NOT work properly with your computer. Have a look at the Audiomedia III FAQ internet page (http:/www.digidesign.com/) to see details of likely problems caused by PCI Burst mode. Digidesign are currently finalising a chip upgrade that addresses this, and, to quote from their web site: "The PCI Burst Mode Upgrade should be available no later than 31 March 1997. The modification is FREE within the first 45 days of ownership or during the initial 45 days of availability from Digidesign. After the initial 45 days, the cost is $59 for those customers who wish to modify their card for increased compatibility."

In essence, this means that it's up to you to contact Digidesign Technical Support and get your name on the list for a free upgrade. When this new chip arrives, they'll contact you so that you can send your Audiomedia III card back for the upgrade. The chip will not add any extra performance or features to the card: all it does is enable the card to work with those PCs that cannot disable PCI Burst mode — so if yours works anyway, please don't clog up the system for those people who have genuine problems. All credit to Digidesign for managing to resolve this issue once and for all, we hope — their product support lines will probably get rather quieter as a result!

The second subject on the problem agenda is one about which I knew little until very recently: incompatibilities between hardware or software and certain processors. Back in December I upgraded my PC and, among many other improvements, it now includes a Cyrix 166+ processor. You can pick one of these up from a mail‑order retailer nowadays for about £120 inc VAT, whereas a straight Pentium P166 is more like £250, and the Pentium MMX 166 is currently even more, at £370. The Cyrix P166+ is generally held to out‑perform the equivalent Pentium P166 (non‑MMX) in the speed stakes, by up to 10% in many areas. Its only weakness is in its floating‑point performance, which in normal use doesn't usually matter a great deal, but does of course have an effect on the final performance of audio number‑crunching programs such as Sound Forge and WaveLab, which use a lot of floating‑point maths. However, I heard a few days ago that certain software/hardware combinations refuse point blank to work with a Cyrix processor. The new AWE64 card from Creative Labs works fine, but its WaveSynth/WG will only run with genuine Intel Pentium or AMD processors, despite the best efforts of CL to produce a software patch. Apparently Cyrix have acknowledged the problem, and say that it will be addressed in the next upgrade, but this isn't a lot of use if you've just bought a P166+ or an AWE64. I'll be reporting any other processor/software incompatibilities that I come across in my travels.

Quiet Please!

Papst Variofans have already been mentioned in SOS, in my roundup of PC utilities (see last issue), and fitting these as replacements for the cheap and cheerful fans inside PCs will give you a significant reduction in noise level. If you have tried this successfully, and want to achieve an even quieter machine, the likely cause of remaining noise is the sheet metal casing itself, thrumming along to the vibration from the hard disk. Try tapping the side panels lightly — they will probably sound like drums (you could always try sampling them!). If you are the tweaky type, you could try re‑attaching the drives via rubber grommets, but it's often difficult to get at them without completely dismantling the PC, and you may still find studio monitors 'setting off' your case at certain frequencies. In fact, contrary to what you might think, removing the cover often makes the computer quieter, simply because there's then less to vibrate in sympathy. A very cheap way to cure the problem once and for all is to attach a self‑adhesive bitumastic sound deadening pad to each side panel (on the inside!). These have been used for years to deaden vibration in cars for a quieter ride, although it's difficult to find them in car accessory shops nowadays. I got some from Russ Andrews, who runs a company specialising in audiophile accessories (RATA Ltd, on 01539 823247). At only £2.47 each including VAT, they are well worth a try, and it's easier than building your PC into a cupboard!