Sorting digital gremlins and rogue reflections is all in a day's work for the Studio SOS team...
Ray Parkes called us in primarily because he was suffering from a very odd digital audio gremlin that only started when he relocated his studio equipment from one room to another in his house in Walsall, near Birmingham. The problem was that whenever anyone switched on any appliances in the house, or when the central heating fired up, his old Yamaha 01V mixer would lose lock on the digital S/PDIF input from his Firewire interface and cut out for a few seconds. This same setup, comprising a Mac G5 1.8GHz machine running Cubase SX using an M-Audio 410 audio interface, had worked fine in an adjoining bedroom but steadfastly refused to operate reliably when reinstalled in Ray's larger studio room right next door.
Ray told us that most of the recordings made in his bedroom studio were co-written and performed by himself and his old friend Alan Jarvis, with whom he played in a band during the '70s and '80s. They had started out in a very small bedroom using a Tascam 4-track, then progressed to a Tascam 488 8-track, then, via Cubase VST on a Mac G3, to the present system described above. However, Ray now also records and produces some local bands.
When his daughter and her boyfriend decided to move into their first flat, Ray quickly bagged her room for his studio, which is when he first encountered the mysterious audio loss problem. When Hugh and I arrived, it only took a few flicks of a hallway light switch, or even the power switch on Ray's Focusrite Platinum Voicemaster vocal channel, to cause the audio to drop out.
The system was configured with the computer and M-Audio 410 as the clock master at all times, and the 01V mixer was clocked from that via its S/PDIF input. Ray was using decent quality Hosa digital audio cables, so poor impedance matching or RF interference didn't seem likely, but to be on the safe side we switched the cable with a known good one that Hugh had brought with him — but the problem remained.
The next test we tried was inserting a sample-rate converter in the S/PDIF path to act as a buffer, as this would help reveal where the failure was occurring. Was the M-Audio interface introducing a glitch on its output, or was the desk falling over for some other reason? Hugh brought along his Drawmer Master Clock because its SRC section generates a continuous clock output (referenced to its internal clock), regardless of the input. So the signal path was now M-Audio out into the Drawmer SRC, and the SRC output into the Yamaha desk. By repeating the light switching test, we were able to see that the desk now remained stable — so it couldn't be a problem with interference into the desk. However, we noticed that the 'lock' light on the Drawmer's SRC input blinked when the lights were turned on or off, which suggested that the problem was actually caused by the M-Audio unit interrupting its S/PDIF output signal in some way.
Looking more closely at the M-Audio interface, we noticed that Ray was running it directly from USB bus power, rather than the mains PSU. Sometimes USB power is a bit marginal, so we decided to install a mains wall-wart just in case the current drain on the USB bus was destabilising the system. This also produced no improvement!
Next, we tried a filtered mains socket for the M-Audio's power supply, in case the problem was mains interference upsetting the interface, but this produced no improvement either. However, in plugging things into the mains supply we spotted the potential problem. In fact, I'd guessed at the solution when we first came into the room but thought we'd try a few simpler options first. Though Ray said he was sure the whole room was supplied from the same ring of his domestic ring main, he had his various plug-boards connected to different wall sockets on opposite walls of the room, and I was worried we might either be getting feeds from two different ring mains or at least be getting some ground-loop problems between equipment running from different mains points. While ground loops between pieces of equipment that are connected digitally don't tend to cause audible hum problems, any induced hum can certainly affect the reliability of the digital data being transmitted.
Ray mentioned that he often ran out of CPU power, and that he also had some memory cards that could be used to increase the capacity of his machine from a sparse 512MB to 1.5GB, so we offered to fit them for him while we were there. While the amount of memory has no direct effect on CPU loading, it does help to keep the system running smoothly when running plug-ins that use a lot of RAM, such as software samplers, delays and reverbs. In this case, increasing the buffer size to the 2048 maximum when mixing also helped enormously in preventing the brief CPU spikes that had proved so frustrating for Ray.
If you suffer from the same CPU issues as Ray, there are some other things you can do to relieve the strain on your CPU. First, where your DAW provides it, there's the track Freeze button, which creates a temporary audio file with the effects added, and then bypasses the effects (or soft synths). Another tip here is that if you have audio tracks with long periods of silence in them, you can cut the audio into regions and discard the silent sections. Then you're not continuously streaming silent audio data from your hard drive unnecessarily — something that uses just as much drive bandwidth as useful audio. This will make an even greater impact if you have a DAW system that takes the load off its plug-ins when there is no audio region to process.
Both Hugh and I felt that, before proceeding any further, we should minimise the risk of ground loops by creating a 'star' mains distribution system, fed from a single power point in the room. To establish such a system you can use a multi-socket distribution board, with four further distribution boards fed from it, all with as short a cable as practical. Not only could this power the main studio system, it would also power the keyboard and Yamaha electronic drum kit connected to it.
Reconnecting everything took around 10 minutes and, once we were done, we tried the 'switching on and off' test again. This time, no matter what we did, we couldn't make the problem recur, and by this time we were pretty sure that some sort of ground-loop issue had been the problem, but we decided to play it safe and replaced the main power distribution board with one that contained some filtering and spike suppression. We also tightened the coaxial connectors at either end of Ray's digital cables, as the fit was a little looser than we'd have liked, and could possibly have affected the cable screening.
Having overcome Ray's main worry, we listened to some recorded music to see how well the room worked for monitoring. In fact, it was pretty good already. Ray had hung duvets either side of his listening position, and his Fostex PM05 powered monitors were placed on wall-mounted stands that projected over his desk, which meant that they were at a sensible height and distance for nearfield monitoring. However, the front wall (the wall behind the speakers) was completely untreated, and there was a lot of reflective desk surface exposed, which might affect the tightness of the image at the listening position, so we decided to try a few experiments.
To see what a difference could be made to the perceived sound, I felt that we should improvise some absorption on the reflective surface, so Hugh folded a duvet and draped it over the desk. This made a huge difference, but while it is good for diagnosis, putting a duvet over everything is clearly not practical! I initially suggested trying a folded bath towel to cover the exposed desk between the computer keyboard and the computer monitor screens. This improved matters to almost the same extent. It also proved that desk reflections can be a significant problem, so if you can use a smaller desk, or put absorbing material on those parts of the desk needed for standing things on, it could be very beneficial. As a more permanent solution, we cut a piece of Auralex foam to fit on the desk between the computer keyboard and screens.
Next we decided to apply a couple of Auralex panels to the wall behind the monitors, and also to hang one on the ceiling above the computer keyboard. As Ray didn't want to do anything too permanent to the room, we glued the foam to fretwork MDF panels and hung these up on picture hooks. The ceiling panel was fixed using six wood screws, as it is easy enough to fill and hide such small holes. A final small piece of foam was fixed to the side of the rack at the right of the desk, using double-sided carpet tape, to reduce reflections from that source.
Even before we did anything, the low end was reasonably consistent throughout the room, except for some build-up in the corner where Ray had his Yamaha drum kit set up, so we didn't feel that dedicated bass trapping was essential. With all the foam in place, the difference in stereo imaging and definition was noticeable and was definitely worthwhile. Ray said that the improvement was very obvious to him — and he expressed surprise that the difference was as noticeable as it was!
With the monitoring sorted, our final hurdle was Ray's vocal recording setup. In accordance with advice gleaned from articles in Sound On Sound, he'd arranged it so that the singer stood with their back to a duvet-covered corner, to cut down on reflections getting back into the front of the mic, and this worked reasonably well. But with so many other hard surfaces in the room, as well as fan noise from the computer, we felt that placing an SE Reflexion filter behind the microphone should improve things further. Sonic Distribution had kindly provided us with one for this purpose, which we were able to leave with Ray.
The Reflexion Filter is a small acoustic screen that keeps unwanted sound out of the rear and sides of the mic, and absorbs some of the vocal sound travelling past the mic, so that less sound escapes out into the room and, consequently, there are fewer unwanted reflections that can return to the mic.
Although the Reflexion Filter can be a wonderfully effective piece of kit, the mounting hardware is rather heavy and clunky, with all the weight to the rear of the stand fixing. This tends to make lighter mic stands unstable, so we decided to disassemble part of the Filter and then put it back together so that the supporting point was closer to the centre of gravity. This simply entailed unscrewing four cap-head screws with an Allen key and then replacing the component the other way around. Once we'd done this, the whole setup was far more mechanically stable, and a test recording showed that the vocal sound was now impressively dry. It is possible to achieve a further mechanical improvement by making and inserting a spacer block to move the stand clamp mechanism further away from the microphone slide bar, and we left Ray to consider this modification. A simpler solution, but one which requires two mic stands, is simply to mount the Reflexion Filter directly on to a mic stand and use a second mic stand to position the mic inside the curve of the filter.
To further minimise the computer fan noise reaching the microphone, we folded a rug over the computer to form a tunnel that extended out from the front of the machine. This made a useful difference without impeding the air flow necessary for cooling.
With all the improvements completed, we turned to discussing Ray's options for upgrading his system. His 01V digital mixer was now playing the role of a monitor controller and little else. We agreed that there was little point in mixing through it — not only are its A-D and D-A converters now a little long in the tooth, but its mixing capabilities are probably now less sophisticated than those inside Ray's Cubase software. If he were to replace the 01V with a low-cost monitor controller, such as a Mackie Big Knob or Samson C Control, he'd still have all the necessary monitor level and routing control and would also regain plenty of useful desk space. Ray felt that this was probably the way forward, but before doing anything else he wanted to spend some time playing with the newly reconfigured system to see how the improvements would affect his results.
When I moved the studio into the larger bedroom and the audio problem started, I emailed Paul, expecting just a few pointers. When he said that he and Hugh would come over to take a look, I couldn't believe my luck.
A couple of hours with these guys has saved many hours of trial and error, and it is the hours that are in short supply. Since reorganising the wiring I've had no more dropout problems.
The Auralex foam has improved the quality of the monitoring immensely, with better stereo imaging and a much tighter feel to the mix.
The SE Reflexion Filter has helped to keep out extraneous room noise and has dried up the vocal recordings significantly. I really can't thank Paul and Hugh enough. Their expertise and guidance has been invaluable, both from their visit and through the pages of Sound On Sound.