Most high-speed feeds we encounter in our daily lives are of the point-to-point variety. A USB peripheral connects to a USB-host port. An Ethernet client attaches to an Ethernet hub. In such cases, the line terminations that provide impedance matches ‘twixt source and destination reside happily (and invisibly to us) within the subject equipment.

Less common in our day-to-day experience are high-speed point-to-multipoint connections, which as the name suggests, are configurations in which multiple signal clients connect to a common source. Examples of this sort of arrangement include clock- and time-code distribution schemes within a media-production environment. In large permanent installations, clock distribution often takes the form of a clock tree formed of a low-jitter master clock source and the clocking equivalent of a distribution amplifier that provides multiple outputs characterized by low channel-to-channel skew.

To accommodate various clock-distribution topologies, it is common for clock-signal clients to require external terminating resistors. Small installations or portable rigs, for example, in which only a small number of pieces of equipment require clocking, often get by with less formal daisy-chain connections. This is particularly true when the devices are arranged to minimize the physical distance clocking signals must traverse. In this arrangement, a single-output clock source can drive several clients, with one line-terminating resistor attached to the last client in the chain.

The production suite I use exploits such a daisy chain-topology. An Apogee low-jitter clock serves as the master, with a single feed daisy chaining, in sequence, to a mixing console, an outboard signal processor, a stereo recorder, a multitrack recorder, and its transport controller—each described as a digital device but each necessarily a mixed-signal system.

In particular, note that all of the important attributes of the distributed clock signal exist in its analog-domain behaviors including phase noise (jitter), rise and fall times, pulse flatness, and ringing. Clock input circuits are, by their nature, analog interfaces, which must provide a degree of signal conditioning to compensate for cabling losses, line reflections, and other forms of signal degradation.

This came up during yesterday’s mixing session when I discovered that I needed to switch clock sources temporarily. Without going into the gory details, I needed a variable-frequency clock source—a function that the Apogee clock does not provide but which happens to be available from the transport controller.

Easy enough: I disconnected the clock feed from the Apogee source and switched the transport-controller connection from its external-clock input to its clock output. Everything else along the daisy chain could remain as was. Well… almost.

The setup hack worked exactly as expected… except for one leeetle problem: Every so often—as rarely as once in several minutes, the monitors issued a mighty POP!, which was accompanied by a sudden peak on one of the console’s VU meters. After some minutes of head scratching, it occurred to me that I had introduced an error in the clocking setup: Originally, the daisy-chain termination attached to the transport controller via a BNC T adapter. When I made the transport controller the clock source, I did so without removing the termination resistor. So the source had a termination resistor sitting on its output and the rest of the line was running without termination.

The console’s clock-input signal conditioning did a remarkably good job, under the circumstances, but would lose synch periodically, resulting in what I’ll describe as a framing error. Simply removing the terminating resistor from the transport control and connecting it to the console’s clock input restored the clock distribution topology and proper system operation.

Thankfully, the problem took only a few minutes to diagnose and resolve, but I’m certain this incident will inspire a new item on my setup checklist for sessions in the future.