Table 1. Recommended Serial Digital (SDI) Transmission Distances through Coaxial Cable

Figure 1. Standard reference level SDI signal conforming to SMPTE 259M

Figure 2. SDI signal after 100 feet of Extron Mini-HR coax

Figure 3. SDI Signal after 100 feet of Extron SHR coax cable

Figure 4. SDI signal after 700 feet of Extron SHR coax cable

Figure 5. Color enhancement shows

Did You Read the Last ExtroNews?

If you have a copy of the last ExtroNews handy, you might find the table on digital formats helpful for reference. As you scan that table, you'll see that most of the professional broadcast formats (SDI, SDTI, SDTV, and HDTV) are serial and use single coaxial cable with BNC connectors. So that you can't say I ignored the D1 parallel connection. I'm going to effectively skip discussion on it as it is really a short hop connection meant for close-in connected systems. Anytime you are involved with that parallel format, you'll need to look for a specific cable. In this installment, we'll look into details of cables and connectors for the serial digital format. Other formats from the table are the topics for the next article.

Cables and SDI

Cable loss specifications for standard SDI, SDTI, and uncompressed SDTV are addressed in SMPTE 259M and ITU-R BT.601. In these standards, the maximum recommended cable length equals 30 dB loss at one-half the clock frequency. Note that this high loss value does not correlate with losses normally accepted for analog video and graphics signals. This serial digital loss level is acceptable due to the serial digital receiver. Serial digital receivers have special signal recovery processing.

For HD SDI running at 1.5 Gbps, SMPTE 292M governs cable loss calculations. In that standard, maximum cable length equals 20dB loss at one-half the clock frequency. Due to the data coding scheme, the bit rate is effectively the same as the clock frequency in MHz. Similarly, high definition serial digital receivers have special signal recovery capability as well. See Table 1 for some examples of cable length calculations.

Recommendations among cable manufacturers will certainly vary, but it is good practice to limit your run lengths to no more than 90% of the calculated value. This provides leeway for cable variations, connector loss, patching equipment, etc. Table 1 includes this allowance. In all cases, your system must operate solidly before the "cliff region" where sudden signal dropout occurs. Recall that digital systems do not perform linearly to cable losses. Final performance rests with the cable and the type of receiver used. The bottom line in these systems is maintaining low BER (bit error rate). SDI signals are nominally 800 millivolts…not much different in level than analog video signals. Refer to Figure 1 for a standard-level SDI signal that conforms to SMPTE specifications.

What is different about SDI cable loss considerations? With SDI signals, the receiver is more complex in its ability to equalize and recover the signal. Signal recovery is a nonlinear situation. SMPTE 292M describes the minimum capabilities of what it calls a type A receiver (the better) and a type B receiver. Like RF receivers, SDI receivers are adaptive in their ability to amplify, equalize, and filter out the information. Selecting the best receiver will make a tremendous difference in the final performance of a serial digital system. Figure 2 shows the loss effect on an SDI signal after 100 feet of Extron Mini HR cable. Although rise time is significantly affected, all quality receivers can recover this signal. In fact, for this particular cable a class "A" receiver can recover a solid image after 425 feet (see Table 1). Figure 3 illustrates signal quality after 100 feet of Extron super high resolution (SHR) cable. Note that the improved signal waveform ensures that the signal can be conveyed much longer distances. For SHR cable, standard SDI can be transmitted over 1000 feet.

Cable Versus Receiver

So, how much of your system performance depends on the cable and how much depends on the receiver? It's a good idea to know this boundary as receiver and cable specs vary. The primary loss parameters that affect serial digital losses are risetime/falltime degradation and signal jitter. This is why serial digital signals normally undergo reshaping and relocking as they pass through major network hubs like matrix routers. Interestingly, viewing the SDI signal waveform on a scope will not really tell you much once signal level drops to a certain point. Only specific instrumentation made for testing SDI signals will yield the ability to receive a proper image transmission. Figure 4 shows a typical scope presentation after 700 feet of Extron SHR cable.

Although the SDI waveform is not discernable, a good receiver will capture it. By using color enhancement modes on a time domain reflectometer (see Figure 5), you can see a pattern in the data that is somewhat recognizable.

Table 2. SMPTE Serial Digital Performance Specifications

But, we don't live in an ideal world. The economy of distributing SDI and HD SDI lies in the ability of the serial digital receiver to recover a low-level signal much like TV receivers recover a complex television image from a weak RF signal. The extended capability of the serial digital receiver makes the run lengths in Table 1 possible with few exceptions. Just what is the receiver's contribution? Well, comparing the SMPTE loss calculation to the -3dB point used in regular video systems suggests an effect upwards of 10 times; i.e. -30dB compared to -3dB.

Figure 6. Step response comparison for 50-ohm verses 75-ohm BNC connector in a 75-ohm system

Figure 7. TDR of 50-ohm BNC connector in a 75-ohm system

Figure 8. TDR of 75-ohm BNC connector in a 75-ohm system

Figure 9. TDR of standard VGA connector in a 75-ohm system

Figure 10. Step response of standard VGA connector in a 75-ohm system

BNC Connectors -
Is There a Difference?

OK, I will not attempt to convince you to use 50-ohm connectors on 75-ohm cable, even though many of us did this for many years. No contest. We all know that proper impedance matching is the right thing to do. The hard truth is that for many years, the only cost effective version available was the 50-ohm BNC. Sure, you could get the 75- ohm type at a premium of 5 to 1. The 50- ohm version dominated because of the RF equipment industry. AMP and many other connector manufacturers were making 50- ohm versions by the millions. The real question is: Does it make any difference to AV system performance or image quality?

I'm not here to artificially raise the VSWR of your thoughts… we'll let the connectors do that. Why not look at some real measurements and then decide for yourself? We connected our test signal generator (VTG 200) through a twelve-foot length of 75-ohm cable utilizing, in one test, 75-ohm BNC connectors and, in the second test, 50- ohm connectors. The test signal is a step response. The transient response at the leading edge tells us if we will see any anomaly attributable directly to the mismatch. Any serious perturbations here could translate into image artifacts that affect high frequency details in an image. Compare the two waveforms of Figure 6. The yellow waveform illustrates the performance with a 75-ohm BNC connector. The blue waveform represents the 50-ohm situation. You'll see that no difference is visible. This is our experience overall.

Think in 1-D

The actual dimensions of BNC connectors are small enough that we will not see significant effects created by the connector in a system until we approach 3000 MHz (where the connector's physical length approximates one-quarter wavelength of the frequency of interest). Therefore, the reason that BNC impedance mismatch effects are not prevalent in systems we design is that the connector dimension is a miniscule part of the transmission line length at frequencies for which we are primarily interested. In the microwave industry, the connector dimension is significant. Now, refer to Figure 7, which shows the TDR (time domain reflectometry) presentation of a 50-ohm BNC. Compare this to the TDR image of a 75-ohm BNC in Figure 8. These images represent time domain measurements of the same hookup for Figure 5 where a 12-foot length of 75-ohm cable is used. Perturbations seen for either connection impedance center primarily on the connector crimp and contact interface.

Does this mean you can actually ignore coaxial cable impedance in system design? NO. The electrical length of cables is significant at the frequencies we encounter with graphics systems. This is why you see poor performance with some cables. They are not the correct impedance and, therefore, reflect much of the transmitted energy back to the signal source.

VGA Debacle?

Does anyone know the impedance of a 15-pin VGA connector? Ah-Ha! You hadn't thought about that, had you? I can assure you it is NOT exactly 75 ohms, nor is it coaxial. Yet, how many people are concerned about that? My guess is only those that might have an interest in marketing a new connector product. This ubiquitous connector is convenient, low cost, and most importantly, adopted by IBM. So, how does its performance compare (you now ask)? Thanks for asking.

Well, fortunately, it's not a lost cause, but its effect swings in the other direction; i.e., its impedance is as much higher than 75 ohms as the 50-ohm BNC is lower. This means that it's about 100 ohms. Internally, you must take a good 75-ohm cable and split its conductors so as to connect to the parallel pins in the VGA body. Then, after the interface, this parallel connection must return to the cable's symmetrical world. Take a look at the interface anomalies of Figure 9. Again, the primary issue centers on the limited length of the connector interface but does not significantly hamper performance in systems we most often deal with. The step response shown in Figure 10 shows no significant effects, hence the popularity of the VGA connector as a low-cost, general interface for the PC.

Gosh, we had a lot of fun making these pictures. It answered some questions for us and, I hope, for you too. Making good technical decisions in systems design is very important. Hopefully, these examples paint a realistic picture of SDI cabling as well as one of the most nagging little questions in the AV industry… the BNC question.