|
Archives
 February 1999 Issue
Hranac - Notes for the Technologist
Headend Combining
By Ron Hranac
When designing a headend combining network, one is faced with several often conflicting objectives. First and foremost is to electrically combine a large number of channels into one signal pathsay, a cable feeding downstream lasers or maybe the systems first trunk amplifier.
Ideally, the combining should be done with little or no signal loss to maintain a good carrier-to-noise ratio (C/N) at the input of the first active device after the combiner.
To minimize interaction among various headend components, each of the combiners inputs must have high isolation from every other input. That way, a severe impedance mismatch on any given port will have little or no effect on other channels.
Reality check
In reality, combiners have insertion loss, usually 15 dB or more per combiner input, but a good headend design in conjunction with the relatively high signal levels from processors and modulators can accommodate the loss.
Most quality commercial headend combiners have reasonably good port-to-port isolation, as long as each input device provides a good impedance match to the combiner.
Unfortunately, each piece of equipment connected to a combining network generally has a nominal impedance of 75 ohms only within the channels 6 MHz bandwidth. Outside of the channels bandwidth, the return loss can be quite poor.
There are ways to deal with this, and thats the subject of this months column.
Traditional trickery
One trick thats been around for several years is to use quarter wavelength coaxial jumpers between processor and modulator outputs and the combiner inputs.
The reasoning behind this is the impedance-transforming properties of a quarter wavelength transmission line. Mathematically, the input impedance of a quarter wavelength line terminated in a resistive impedance is:
Zi = Zc2 /Zt
where
Zi is the impedance at the input of the quarter wavelength line
Zc is the quarter wavelength lines characteristic impedance
Zt is the impedance of the load, or termination at the other end of the quarter wavelength line
Using basic algebra, the above equation can be re-written as:
Zc = square root(Zi x Zt)
This latter formula demonstrates that any given terminating impedance Zt can be "transformed" into any desired input impedance Zi using a quarter wavelength line with an impedance Zc.
Translation: The square root of the product of the input and output impedances equals the impedance the quarter wavelength line must be in order to match the input impedance to the output impedance.
The really tricky part
Simply installing a quarter wavelength jumper between the processor or modulator output and the combiner input isnt the complete answer.
First of all, when you consider higher frequency channels, a quarter wavelength can be rather short, sometimes only a few inches. For example, using headend coax with 83% velocity of propagation, an electrical quarter wavelength at Ch. 78s visual carrier is 4-15/32 inches. (Note: An appropriate odd multiple of a quarter wavelength could be used to deal with the physical distance while maintaining the same electrical performance.)
Second, what frequency should a quarter wavelength jumper be based on: the visual carrier, the aural carrier or something in between? Remember, each TV channel is 6 MHz wide. At Ch. 2, an electrical quarter wavelength at the visual carrier is 3.694 feet, and at the aural carrier its 3.416 feet. The difference is about 3-1/4 inches, which is significant as far as obtaining proper impedance matching is concerned. At Ch. 78, the difference is only 1/32 inch. If longer jumpers based on an odd multiple of a quarter wavelength are used, the physical difference will be greater.
You might ask, "Why bother to match the output impedance of the processor or modulator to the combiner input impedance? Arent they both 75 ohms?" Well, Im glad you asked. The answer is "not necessarily, and in most cases probably not."
Consider a fixed channel processor such as Scientific-Atlantas 6150, arguably one of the best processors available and a common fixture in many headends. According to the spec sheet, the output voltage standing wave ratio (VSWRsee my column in the December 1998 issue of Communications Technology for more on this) is <1.25:1 over the 6 MHz channel.
Assuming a worst case of 1.25:1 VSWR, the return loss will be about 19 dB, which is a pretty good number. That means the actual impedance can be anywhere from 60 to 93.75 ohms. S-As 68-12TS combiner has an input return loss spec =17 dB (1.33:1 VSWR), which means the impedance can be between 56.43 ohms and 99.67 ohms. This range is what is meant when we say the nominal impedance is 75 ohms.
To use a quarter wavelength transmission line to match the impedance of the processor output to the combiner input, it will be necessary to know the actual impedance of each. Unless you have access to a good network analyzer, that might be a little hard to do. But lets make some assumptions here, just to see how this might be done.
Assume the processors output impedance actually is 75 ohms, and the combiner input is at its possible upper spec of 99.67 ohms. Plug these numbers into the previous formula, and youll find you need a quarter wavelength line with an impedance of 86.46 ohms to match the two impedances {86.46 =square root( 75 x 99.67)}.
The quarter wavelength jumper is then installed at the combiner input, and regular 75 ohm headend cable can be used from the jumper back to the processor output. Unfortunately, I dont have any 86 ohm cable lying around; do you?
A better way
There is an easier way. Install a 6 dB in-line pad at the combiner input. This will improve the combiners input return loss as seen by the processor output by double the value of the pad.
Heres why: Without the pad, the combiners effective input return loss will be 17 dB. Because the combiner input represents a slight impedance mismatch, some of the signal coming from the processor will be reflected back toward the processor at a level equal to the incident signal minus the combiners return loss.
For example, if the signal from the processor is +59 dBmV, the reflected signal will have an amplitude of +59 dBmV minus 17 dB, or +42 dBmV.
If you install a 6 dB attenuator at the combiner input, the processors incident signal (+59 dBmV) will be reduced to +53 dBmV at the combiner input. The reflected signal from the combiner input port still will be down 17 dB, but it will be further attenuated by the 6 dB pad, bringing the level of the reflected signal down to +30 dBmV (+59 dBmV - 6 dB - 17 dB - 6 dB).
The effective combiner return loss as seen by the processor will now be 29 dB, which is 1.07:1 VSWR. This means the processor output will see a combiner input impedance somewhere in the range of 69.86 ohms to 80.52 ohms. If your headend design can afford the additional signal loss, install a 6 dB pad at the processor output as well. The overall impedance match will be improved even more.
Forcing an impedance match in this manner makes coax jumper length for the most part irrelevant, except for the added attenuation of the jumper.
One more trick
There is one other trick you can use to improve headend combining performance. Instead of stacking channels vertically in each rack, for instance, Chs. 2, 3, 4, 5, 6, 14, 15, 16, 17 and 18 in the first rack, 19, 20, 21, 22, 7, 8, 9, 10, 11, 12 in the second rack and so on, try arranging channels horizontally.
That is, put Ch. 2 in the first rack, Ch. 3 in the second rack, Ch. 4 in the third rack, Ch. 5 in the fourth rack and Ch. 6 in the fifth rack. Then come back to the first rack and install Ch. 14 (below Ch. 2), put Ch. 15 in the second rack, Ch. 16 in the third rack and so on.
With a little creative work, you can have the combining such that no adjacent channels are connected to any one combiner, resulting in even better adjacent channel isolation. - CT
Ron Hranac is senior vice president of engineering for the Denver-based consulting firm Coaxial International. He also is senior technical editor for "Communications Technology." He can be reached via e-mail at
 Back to February Issue
Access Intelligence's CABLE GROUP
Communications Technology | CableFAX Daily | CableFAX's CableWORLD | CT's Pipeline
CableFAX Magazine | CableFAX databriefs | Broadband Leaders Retreat | CableFAX Leaders Retreat
|
