Monitoring the SDI stream


Figure 1. As the signal degrades in long cables, noise distorts the waveform. Figure from SMPTE 259M-1997. Click here to see an enlarged diagram.

For decades, we have all been used to looking at a single test instrument to decipher almost anything about a composite video signal. You can see frequency response issues, differential gain, video levels, signal timing and many other parameters that identify the health of the signal and a lot about the quality of the picture itself. In many cases, the same scope can be used to look at 625-line signals. With slightly different models, you can look at component analog video and HD signals with the same clarity and simplicity.

When video is digitally sampled and then transmitted, things get a bit murkier. First, it is impossible to look at a digital signal on a simple analog waveform monitor. The digital signal (SMPTE 259M for SD and SMPTE 292M for HD) must be deserialized, decoded and converted to analog. After that, you can return to looking at the same familiar, albeit component, signal.

But what if the digital signal doesn't seem to be working? Bit errors and complete loss of picture can happen, yet you might still have a signal on the cable. Underneath, what makes SDI so different?

First, SMPTE 259 and 292 are really analog signals representing digits. Think about it. SMPTE 259M is a pseudo square wave at a center frequency of 270MHz. SMPTE 292M is the same structure at 1.485GHz. An analog signal would be unusable at just a short distance in coax at those frequencies. The SMPTE standards set the launch conditions for the signal with known characteristics in the coaxial cable medium it is transported. If the characteristics of the cable are known, you can work out the signal loss in the cable with simple math. It turns out the SD standard was reverse-engineered to achieve acceptable bit error rates at a distance of about 1000ft in precision coax. Beyond that, anything would be a gift. At about 1200ft, almost all signals would be lost.

The standard established a launch signal of 800mV amplitude with tightly specified rise times.

As the signal degrades in long cables, due to frequency response and loss issues, the noise on the signal begins to distort the waveform, and eventually the data transitions can no longer be recovered. (See Figure 1.) Noise added to the signal has the effect of adding jitter to the decoded data since the zero crossing is no longer as easy to identify precisely. (See Figure 2.)


Figure 2. This Tektronix WFM-700A display illustrates jitter added to the decoded data. The zero crossing is no longer easy to identify. Click here to see an enlarged diagram.

It is important to note that all this kind of display is showing you is the likelihood that the bits can be decoded successfully. There are plenty of other parameters that one must keep track of with digital video. It is vital to think of the digital signal as a sort of unidirectional network carrying a multiplex of signals in a tightly controlled structure. Both 259 and 292 define ancillary data, which resides in the time slices that would carry vertical and horizontal blanking in an analog signal.

Because digital signals can represent sync with a small number of bytes, lots of space is left over. Plenty of uses have been standardized, with the most commonly used one being embedded audio (up to 16 mono tracks, eight AES pairs in four groups of two channel pairs). Also common are the carriage of closed-caption data and the use of SMPTE 259 as a data transport called serial data transport interface (SDTI). SDTI generally carries either multiple compressed streams on one wire or one signal transported at up to four times real time.

When SDI is used for such purposes, a simple matrix or spreadsheet is often the best way to show the data being transported in a tabular fashion. This allows the syntax to be reviewed to ensure that the data is present, has properly announced its presence and is behaving appropriately. One can identify the number of channels transmitted (the presence of bit activity) by looking at the channel status bits in the AES stream, whether the content is data (for instance Dolby E) or linear audio. This can be highly useful. Such tabular display of the SDI data can also be used to look at the picture data. One might find stuck bits or gamut errors out of range for the standard.

Clearly, a waveform monitor type of display is still useful, and some products include these advanced alternate displays in waveform monitor products to increase their usefulness for troubleshooting purposes. To the video signal displays, one might add audio level, phase metering and picture confidence monitoring. It is still useful to know the content after all!

But just as a waveform monitor is still important, measurement tools that do not show waveforms as a primary display are also important. There are SDI monitoring systems on the market intended to measure jitter in the data and display tabular data for in-depth analysis without providing convention waveform or vector displays. This can be useful when checking for bit errors.

If you are testing a system for conformance to the standard and tracking the quality of the data itself, it is useful to have a display that shows checksum errors in picture or ANC data, as well as the status of the EDH check words specified in SMPTE RP-165. While these displays might be combined with a waveform monitor, measurements can be done by instruments intended for much more specialized testing and monitoring of signals for transient errors. This can be helpful in locating subtle problems in a system, such as illegal signal levels associated with one input picture source or intermittent bit errors.

If you think of the SDI (or HD-SDI) signals as both video and data and pick monitoring strategies appropriate to your system, the monitoring equipment you choose may well be unconventional.

John Luff is senior vice president of business development for AZCAR.

Send questions and comments to:john_luff@primediabusiness.com