Spanspek

Ask any engineer and you hear a few weird stories, from finding old gear to PA horror stories. I've had my share. One of the oddest was when I was recording a 10-minute jazz piece. The computer was running MacOS 9 and Pro Tools 5.1. It crashed about 30 seconds from the end. I mean, the computer froze even though the Pro Tools system kept passing audio. I rebooted but of course the files just recorded weren't registered with the session. So I imported them. Each was now 60 minutes long!

After the 9 minute mark, there were random bits of audio from the previous six months' sessions. After thinking about this for a little while I figured out that evidently, the end of file marker (EOF) that the OS needs to tell where the file ends hadn't been written and so it was just reading the drive sequentially!

Now I work with Simon Polinski, teaching sound production. We're lucky to have recruited him (as you can see if you visited the link above). I've learned a lot from him (and I barely spend any time with him in class as I'm always running around after the students). He tells some great stories.

But today we experienced some Pro Tools weirdness that neither of us had ever seen.

Simon was running a workshop about mastering, for which he'd brought in his Avalon VT747, and things hadn't been going so well. The most intractable problem was that after processing in Pro Tools, running the signal through the Avalon, and returning to an Aux track containing an L1+ limiter he wanted to record the result. So he'd bussed the output of the L1+ to an audio track. The L1+ peaked at -0.3 dBfs, yet the right channel of the audio track was clipping.

He closed and reopened the session. No change. He rebooted Pro Tools. No change. He restarted the computer. No change. That's when he called on me.

We tried everything. When we pulled the output of the Avalon, the right channel should have fallen silent. But it didn't. It read no more than 6 dB below the left (connected) channel. So it appeared there was cross-talk. We narrowed it down to the Pro Tools bus. We tried using another pair of buses and reassigning the right side of the bus pair to a mono bus not adjacent to the left-channel one. Nothing worked.

Then I thought, what would I expect to happen if I sent the L1+ channel down one or other of the buses? We tried both. Either side of the audio track showed clipping because it was now receiving left + right which were peaking at -0.3 dBfs, the output ceiling of the L1+. That's to be expected.

When I reassigned the limiter track's output back to the stereo bus, the problem fixed itself.

So it was a bug which produced one-way crosstalk in the bus. Very weird.