Of Feedback & Wobbles Part Two

Deep Thought Image ©Archaeogaming

After nannying the blancmange beast along for the thick end of four years, the owners of the presentations, the recently-created  National Grid, decided to revamp the show and its smaller underground sibling in the actual pumped-storage station and update their content.

We decided that we should pitch for the contract as we had been left high and dry by our London partner [thereby hangs another tale altogether] and needed a decent-sized job to set things straight again. As we were techies and not content producers, we contacted Upstream Presentations from Windsor, who we'd previously worked with on a show for CADW at Caernarfon Castle and put in a joint tender.

We got a pitch meeting with the powers that be at the Grid and the Director of the museum itself and offered them our seemingly mad idea to keep the 35mm cine as the central feature of the new show. Apparently, this was exactly what they wanted to hear and it got us the contract, as none of the other firms would go near the thing. The only slightly worrying rider to the contract was a commitment to 99.95% reliability, well up from the existing show's woeful performance. Still, we had plenty of time to spec up a system based on Upstream's ideas for the show.

It was decided to keep the cine image static to avoid using the unreliable and cumbersome moving mirror system then used, so we turned the projector round to face centre-screen and reduced the number of slide projectors used and kept them to the cyclorama. Two video monitors were flown high above and to either side of the main cine image, to take videodisc footage of the main presenter; a well-known weather presenter who did the main introduction and the links between the content-sections shown on the cine. The sound system was 1,600 watts top & mid and 2,000 watts sub-bass. A smattering of other special effects completed the hardware roster.

At this point we had decided on the digital sound delivery medium to replace the CD players: the then state-of-art-and-a-little-bleeding-edge Tascam DA88 digital tape - something actually designed for the recording studio, but which fitted our purpose nicely as it could chase timecode natively and varispeed in micro-increments if anything should happen to drift. One potential issue, though, was the unknown factor of the optical soundtrack on the film used to drag everything else behind it to keep sync. Our experience with that thus far had been less than encouraging, and scoping the code output from the old show prints did nothing for our confidence.

So, the first order of business was to make the cine projector's speed as stable as possible. We asked a local electronics boffin and friend of ours to design and build us a phase-locked-loop controller for the otherwise free-running asynchronous three-phase motor. We also specced that it should have some way of indicating the stability of the motor's speed so that we could demo to client. A couple of weeks later, the unit turned up in the hands of its maker and was installed, tweaked and tested. Its final incarnation proved to be so stable that we could have used dead-timing from a known start trigger to run the show. However, Upstream had included a test for us in the cine content of the show. A single piece of lip-sync about two-thirds into the presentation. We would need to hit a single frame target [a single 'k'-sound] every time, around 14,500 frames into the show. Nice. So it would have to be chasing timecode, then.

We knew that the Tascam was very capable on timecode and testing it on the old prints seemed to bear this out. It was pretty forgiving of the odd frame-drop, being capable of timecode regeneration and having a decent buffer to hold it steady. We made the decision to not bother with our existing timecode reader/generator to do an initial regen as the Tascam itself was far better at the job itself. So between the stability of the projector and the timebase locking in the audio and control data on the DA88, we were pretty confident we would nail it.

I found out just how good a job we'd done one afternoon, working alone in the auditorium doing some sound mixing. I decided to try the old show on the new rig and see what happened. I dumped the audio from the main stereo pair CD onto the DA88 and calculated the timecode offset. The original presenter image on the Videodisc player I would replay via a small desk monitor and I laced up the 35mm with a single show reel of film from the old installation. With the DA88 in chase and the cue codes fed to the Videodisc from the data channel, I started the projector up. The result was perfect sync on both replay media for the entire 15 minute show - around 21,500 frames of film and 27,000 frames of video. Perfectly. Locked. This had never happened once during the life of the original show. I was the first person to see it in its intended form. A luxury not even afforded its original designers. Result. My celebrations were probably heard throughout Llanberis. I always like it when a plan comes to fruition, especially when re-purposing kit like the DA88. Immense fun and geek-heaven.

By the way, the readout on the speed controller was engineered to show what at first appeared to be a random figure, its stability being the point rather than the figure. A simple calculation gave me a number.  That number? 41.999 recurring. A deliberate? if technically oblique reference to The Hitch Hikers Guide to the Galaxy, slipped in by Alan in the build: 42.