Careful connections keep signals moving
As I am writing this month's Digital Journal, the staff at Iowa Public TV is preparing for the Iowa State Fair, which occurs every August. Since 1998 we have been shooting much of the coverage in HD and this year, we are shooting almost exclusively in HD and will be editing on our own HD nonlinear editors.
I'll go more into the Iowa State Fair in my next article. I bring it up this month because as we have been preparing to install our new HD editing systems, the issue of how we wire the plant has once again been brought to the forefront.
Wire is still the path that signals use to get from point A to point B. And along the way, they will pass through at least two connectors. If you throw in patchbays and patchcords, there are more than a few potential pitfalls. In the analog world of composite video and audio, we really had it pretty easy. In my earliest studio construction projects I can remember using 8281 for video and RG-59 for sync. The smaller size of RG-59 made it easier to tell what cable did what. Obviously, this was in the days before the advent of colorful cable jackets. I remember buying white RG-59 for H-drive, V-Drive, blanking and sync and using the black RG-59 for black-burst.
(click thumbnail)Catastrophic failures such as bullet burnouts are easy to spot, as in this system at Iowa Public TV's Cedar Rapids public TV facility
The truth is that if you looked at those individual signals after a few hundred feet of travel, they were starting to show the signs of degradation from the relatively poor performance of the cable. Pulses had rounded edges and were down in level but they were still useable by the hardware they were feeding. I'm sure that if I had looked closely at the specification chart for the cables I would have noted that they had specs for "pull strength" and "bend radius," but this was RG-59 and it was being installed by cable guys who were yanking on it with all their might, bending at right angles to make corners and occasionally putting a staple through it, and it would still work. I was a little more cautious with 8281, mostly because it was so expensive and stiff that I didn't want to have to buy more or redo runs.
So why have I taken this trip down memory lane? At the time I was doing all of this cable abuse, I knew from my work in RF that I was impacting the impedance and performance of the cable. Intuitively however, I knew that the impact was minimal since these factors are mostly a function of wavelength. At 4.2 MHz, free-space wavelength is over 230 feet, and most of my cable runs weren't that long. Even in low band, VHF wavelengths range from about 30 feet at Channel 2 to about 11 feet at Channel 6. It wasn't until you got into high band VHF and UHF frequencies where wavelength became a "critical" factor.
Well folks, if you're building a plant to handle HD at 1.485 Gbps (SMPTE 292M) you're dealing with a frequency of 750 MHz and a free space wavelength of about 16 inches. Now your video cabling is transmission line and those specifications that I was ignoring become "critical" factors. Those of us who have worked on transmission line, especially on tall towers, have seen time-domain reflectometer (TDR) plots with clearly defined pips at each bullet connection between sections of transmission line.
We recognize that each of those pips in essence represents an impedance mismatch that generates signals (main and harmonics) that reflect back down the transmission line and show up as a voltage standing wave ratio (VSWR). In transmission systems where we are dealing with many thousands of watts of power, systems are designed to shut down when the amount of signal reflected back exceeds a safe limit.
A catastrophic failure is generally pretty easy to find since there was probably an explosion and a hole where the mismatch occurred. It's not always that easy, but from this picture (opposite page) of a recent bullet burnout in our Cedar Rapids facility.
The signals that will be traveling on the high-frequency coax used in a digital plant will be nowhere near the levels that are in the transmission line, so a complete reflection of signal due to a bad connector will not generate any obvious outward signs. The signal will simply not get from point A to point B. On top of that, because this is a digital signal, there is still the cliff effect to contend with. There will probably be no visible indication in the HD video that there is a problem until the video stops working.
So what are some of the considerations that we need to be aware of? Take, for example, a bundle of cable 20 feet long going from a bank of distribution amplifiers to a switcher. For the sake of neatness the installation engineer has used tie-wraps to bundle and secure the cable bundle to the racks. He has crimped these cable ties down every 12 inches and has torqued them tight enough that he has compressed the dielectric in the cable. If you ran a TDR on the cable you would see pips at equal intervals that are reflecting signal back down the cable. All of these signals traveling back are adding and subtracting with the desired signal. A frequency sweep of the entire pass band would show a nonlinear ripple. The effect, at best, is that the error correction in the hardware has to work harder. At worst the signal is gone.
Connectors are another potentially huge issue if they are not handled correctly. Like the bullets in transmission line, connectors need to match precisely to the line to minimize the return loss. Virtually everyone uses a crimp tool to install connectors on cables. I am not aware of anyone doing crimp connection on transmission line, especially the large-diameter rigid cables. I don't doubt that a tool could be made to do it, but the process of crimping deforms the center conductor and the outer conductor, which changes the impedance of the cable and will result in an increase in return loss - and probably an explosion in a high-power transmission line.
As I noted above, mismatches in SMPTE 292M are not going to generate explosions, so a small amount of return loss is an acceptable trade for the convenience of crimping the connector. However, there can not be any significant deformation of the conductors. Crimping tools need to be carefully used and adjusted to provide a secure connection to the cable with minimal deformation. And use true high-quality 75-ohm connectors. It is unlikely that you are going to be able to see, with the naked eye, a deformation that still may result in an unacceptable return-loss figure.
The sagest words of wisdom that I can offer are, don't go cheap on this part of your plant. Replacing rackmounted hardware is easier than replacing rackmounted wiring. And remember that return loss is a cumulative problem, so your best bet for a reliable infrastructure is to ensure that each individual path is as pristine and transparent as you can make it.
Careful connections keep signals moving