Video Baluns - Some Designs Give the Product a Bad Name
July 1998 (Revised product references, 09/2000)
With the development of structured wiring systems and high performance unshielded twisted pairs (UTP), many people are looking for devices that prepare various signals to travel on these "wire highways". In the last few years we have seen several companies, including the largest purveyor of structured wiring, offer passive boxes that prepare the baseband video signal for transmission over UTP. These boxes come in various flavors depending on the application. For instance, packaged with two audio channels, it is useful for television based video. Another variant has three video channels to support remote Red, Green, & Blue (RGB) Displays.
ETS also makes these devices and many more. Just have a glance at our products pages for more information. But our boxes are different. We took the systems approach to the design of the balancing circuits. You see, television and computer display signals are typically designed to run over a shielded cables. Examples are coaxial cables (with F connectors) to go from the VCR to the TV, and the colored cables (with the RCA jacks) you buy at Radio Shack to connect up your hifi. In order to send signals over UTP, the signal has to be modified so that whatever signal is sent on one wire has to have its mirror image on the other. This process is called balancing the signal. A well balanced signal will propagate down the UTP with minimal loss through radiation. Furthermore, the balancing device also blocks noise signals being picked up by the unshielded wire.
So, the secret to sending signals over UTP is to balance them well - in order to limit both radiation and noise pick up. However, mere balancing is not enough, we have to be concerned about interaction between the various elements. If we look at competitive units we find that they did not worry about the interactions (they also did a poor job of balancing, leading to other problems that we will not address here).
Chart 1 shows the level of undesirable interaction. If we send a signal down one pair, the amount of signal that leaks to another pair is called crosstalk. There are two flavors of cross talk, near end cross talk (NEXT) and far end cross talk (FEXT). Which matters is a function of the application. If I try to send a message while I am also receiving one, I'm mostly concerned about NEXT, but if I send several signals down a cable and want them received ungarbled, I am concerned with FEXT. With a RGB balun you have three signals, Red, Green, and Blue traveling down the cable. Cross talk will muddy these signals.
In the charts, the vertical axis is either NEXT or FEXT measured in dB. Complete interaction is at the top of the chart. Each line indicates 10dB better isolation, or about a factor of three, while 20 dB is a factor of 10 and 40dB is a factor of 100. You need about 40dB of isolation to make the crosstalk imperceptible (digital circuits can usually stand more cross talk). The horizontal axis is frequency. The chart starts at 1,000 Hz and goes up to 100,000,000Hz (100 million cycles per second) Note that the scale is proportional, every doubling takes up the same space. Thus, 2000 to 4000 is the same physical length as 200,000 to 400,000.
Chart 1: Far End Crosstalk (FEXT) vs. Frequency
Now that you can read the chart, note the traces labeled A, B, & C. A peaks at about 80,000Hz. In a TV set, the time for one scan line is about 50 microseconds. The time for one cycle at 80,000Hz is about 12 microseconds. The means that the frequency where the crosstalk peaks is about a fourth of a single scan. Consequently, the patterns that are about that long are particularly susceptible to crossing over and polluting the other signal. Trace A was created with two units from competitors.
Now note trace B. Trace B was created by putting one of ETS' boxes at one end while using the competitor's boxes at the other end. Quite an improvement! Now look at trace C, both boxes are from ETS. The very round peak is at 10,000 cycles and is below 50 dB. There is no chance of crosstalk causing a problem.
Remember, I said that at least 40dB was needed. The baseband video baluns have isolation better than 50dB until 30MHz and better than 40dB up to 70MHz. Note that at high frequencies much of the crosstalk is due to the cable, not the design of the balun. A use of a very good cable like Belden's Media Twist would give better performance if that were a requirement.
Chart 2: Near End Crosstalk (NEXT) vs. Frequency
I have measured NEXT as well. Chart 2 Shows NEXT. I eliminated trace B to contrast the differences between an all ETS and all competitors channel.
By the way, the peak exhibited by the competition does not occur at a constant frequency. It depends on the UTP length. At 75ft. it is 100KHz while at 450ft. it is 50KHz.
I have measured products from both of my main competitors. It is obvious that one copied the other, they are like two peas in a pod.
I also tested the units with live signals. I ran a Tiger Woods tape from the VCR through 300 feet of Cat5 cable to a TV. On another pair I ran a different signal through the cable to a termination. You could see the second signal superimposed on the Tiger Woods picture on the TV. When I substituted one of our boxes for the competitors, the other signal became weaker but was still there. When I substituted both boxes the interference completely disappeared.
So, if you need to send baseband analog video a distance, think of using UTP and also ETS for the finest video baluns made.
© Energy Transformation Systems, Inc.