RF Column 34 - August 1994 Copyright (c) 1994,1995 H. Douglas Lung ALL RIGHTS RESERVED TOPICS: Evaluation of the S.A. digital video compression system: Video quality excellent during initial testing Minor gripes about IRD operation and lip sync Brief tutorial on digital error correction What is "bit error rate" How are errors corrected? Real world results - acceptable error rates? Common problems encountered during Telemundo's digital tests Digital satellite system installation tips to avoid problems ---------------------------------------------------- Last month I mentioned that although Scientific Atlanta was able to get the software working with our digital video compression system we were missing an upconverter to test it through the uplink. Shortly after I finished the column we received an upconverter and started over the air testing on Galaxy 4, transponder 14. My column this month will be devoted to my real world experience with digital video compression, with focus on the problems that can arise when receiving digital signals via satellite. If you haven't been following my columns the last few months, here's a bit of background information on my work with digital video compression at Telemundo Network. I evaluated both the General Instruments Digicipher I system and Scientific Atlanta's Digital Video Compression system (no catchy name) last year. During the first part of 1993, the Digicipher I looked like it was my only choice - it was in service and video quality was acceptable. Later in the year, Scientific Atlanta made a strong pitch for their system, which appeared to be a bit more advanced than the original Digicipher. I decided to go with the S.A. system, in spite of the risks of buying into a system that was still "under construction". It offered some options essential to our network's broadcast operations that weren't included in the G.I. design. General Instrumentsder - 9 Megabits/second at 704 x 480 resolution. At 8 Mbs. compression/motion artifacts were visible on some soccer programming. They appear as a shimmering effect (like heat haze) on edges in the picture. A wide shot of the playing field that includes players on the field and lots of people going crazy in the stands causes the most noticeable artifacts. I'm told that MPEG-2 video coding (instead of the current MPEG-1+) will reduce these artifacts and may permit lower data rates. I've heard no complaints about the audio (provided it is kept out of clipping), however, on this system lip sync can move around a bit (+/- one frame), which can make it difficult to set. Installation of the S.A. D9222 IRD's (Integrated Receiver Decoders) is not difficult, however, S.A. could have made it easier by offering two video outputs and switchable LNB power on the L-band input. Based on the installations so far, I'd estimate 70% or more worked the first time. I'm sure this is low because I'm more likely to hear about problem installations. Of that 70%, most of them provided a good bit error rate. The rest worked, but had error rates too high to provide reliable reception under adverse conditions like heavy rain, wind or partial eclipse. If the high error rate isn't caused by local interference, there are simple ways to improve it. I'll describe the methods a bit later. First, however, let me explain bit error rate. If you haven't worked with digital receivers before, it's probably a new concept. Before I get E-MAIL and faxes telling me that all RF signals are analog, let me explain the terminology I'll use here. I'll use the word "digital" to refer to RF signals carrying digital data using QPSK modulation. I'll use "analog" or "conventional" to refer to RF signals carrying analog data using FM modulation. Unlike analog satellite signals which deteriorate gracefully, the digital signal will look perfect until the point where the error correction collapses and it becomes unusable. Getting a perfect picture isn't good enough with a digital system. You also need a low bit error rate. A hit of noise or a bit of interference will wipe out part of the digital data. Since we're compressing approximately 240 Mbs of digital component NTSC video down to 9 Mbs, each bit becomes even more important. We can't eliminate errors completely in over the air systems, so forward error correction (FEC) is used to reconstruct data lost by noise or interference. In the digital video transmission systems, forward error correction works by encoding the data stream with a convolutional code. This results in an output symbol rate that is greater than the input symbol rate. (Remember this!) On the receive side, the incoming data is compared with the same code used on the encoding side to generate the decoded data stream. Now, what happens if noise destroys one or more of the incoming data bits? Well, remember I said the output symbol rate was greater than the input symbol rate. The convolutional code in the encoder spreads the incoming data over several bits in the output data. Even if part of the incoming data is lost, the decoder can regenerate the original data from the remaining data. The strength of the error correction is proportional to the length of the data stream the incoming data is spread over and the number of additional bits. This is not a simple topic and I'll try to come up with a better explanation in a future column. For a more detailed explanation of error correction, check out Merrill Weis' Advanced TV columns. One of the best explanations I've seen recently is an article titled "Toward New Link-Layer Protocols" by Phil Karn (KA9Q) in "QEX" for June 1994. ("QEX" is published by the American Radio Relay League, 225 Main Street, Newington, CT, 06111.) What does this mean? First, a perfect, error-free digital data stream from the satellite isn't required to get a picture on the digital IRD. Second, any interference or noise will be show up as an increase in the number of errors. Because errors keep on coming, we can't simply count errors, we have to count the number of errors in relation to a given number of data bits -- in other words, ratio of errors to total data. This is the base of the "bit error rate". Because the number of errors can vary widely, a simple 0 to 100 scale isn't good enough. The S.A. D9222 IRD displays bit error rate with an exponent. Typical values might be 5.2E-6 or 2.2E-3. The first is pretty good -- 5.2 errors in 10 to the "6"th power (1,000,000). The second isn't -- 2.2 errors in 10 to the "3"rd power (1,000). Another way to picture this is as a fraction - the first is 0.0000052, the second is 0.0022. Note that in the first example the decimal point was moved 6 positions to the left and in the second it w as moved 3 positions to the left. In a digital system you want the least number of errors, so adjust the antenna and polarization for the smallest number on the left of the "E-" and the biggest number on the right. If you've followed me this far, you're probably interested in what can be done to improve the bit error rate for digital reception. Here's a few suggestions. First, look at a conventional signal with the same polarity and on the same satellite as the digital feed and optimize the dish using the traditional techniques. I recommend using a spectrum analyzer to adjust polarization to minimize interference from the opposite polarity. Adjusting for maximum signal is not sufficient. Connect the digital IRD and look at the bit error rate. If it is sufficiently low, you're done. How low it should be depends on the dish size. I found that it is possible to get an error rate as low as 0.0E-6 on the S.A. IRD using a 3 meter dish - in other words - too low to count. I found that error rates above 5.0E-4 were prone to occasional dropouts and error rates above 9.0E-4 were prone to frequent dropouts. Above 1.5E-3 it was difficult to obtain a picture. Sometimes a dish that is producing an excellent analog picture will not yield an acceptable bit error rate. If that's the case, here's another suggestion -- replace the LNB. Older LNB's have phase instability which will introduce errors in the digital data. Today I was at a site in Austin, Texas where we had a good analog picture, plenty of signal and the polarization was right on. Inspite of this, the digital IRD wouldn't even lock on to the signal. The AFC error was bouncing all over the place. The LNB was probably close to five years old and had a rated noise temperature of 30 degrees. The case was stamped with the name of one of the most popular home dish system manufacturers/distributors. When the satellite engineer I was working with removed the LNB, the connector and probe looked like they were in fine condition. I gave him a California Amplifier SlimLine 25 degree LNB (Part #31207) which cost $60 at ElectroTex in san Antonio. Did it make a difference? With no dish alignment other than those available from the polar mount positioner (one actuator) and Polar-Rotor remote control we were able to obtain an error rate that showed 0.0E-6 with an occassional bounce to 1.0E-6. You've probably seen ads for "Digital ready" LNB's that are phase locked. These cost several times as much as conventional LNBs. Are they necessary? For narrow band systems (9 MHz. transponder bandwidth or digital radio), I'd consider them essential. For full transponder MCPC (multiple channels per carrier) operation they offer little if any improvement over the current generation of conventional LNB's in most environments. I heard from one distributor/importer of "budget" phase locked "Digital" LNB's (Dawn Satellite) that they were necessary in earlier generation digital video downlinks in very cold climates. I've also heard phase-locked "Digital" LNB's recommended for areas where interference is a problem. I don't know if they would help though, at least with the S.A. D9222 IRD's I'm using. When the D9222 loses the signal, it starts a sweep over the AFC (automatic frequency control) range to reaquire it. If the LNB is anywhere close to the right frequency, it shouldn't matter if it is phase locked or not, provided its stable. If you can afford the phase locked LNB, you may see some improvement or you may look at it as insurance. In any event, if you are planning to install a digital receiver and the LNB on your dish is over two years old, buy a new LNB. For as little as $60 you will save a lot of headaches. A properly aimed dish and a stable LNB aren't the only requirements for low error rate digital reception. Because the data is spread over the entire transponder, the amplitude and phase response of your cabling must be flat over the transponder. The most common cause of problems here is an unterminated port on a splitter. I watched the bit error rate drop to half the original value once I terminated two ports on a splitter. Watch out for other imperfections in the receiver RF cabling that can affect response. Bad F connectors, a broken shield, crushed cable or a "loop-thru" connection can introduce reflections that increase errors. Bad cabling can exacerbate another potential reception problem - interference. Digital signals don't like interference. A short burst of radar can wipe out the picture for a second or more. A bandpass filter installed between the feed horn and the LNB will help eliminate out of band radar interference. Interference from terrestrial microwave is becoming less of a problem as phone companies switch to fiber for interconnections. If you are unlucky enough to have terrestrial microwave interference, the traps that help improve conventional reception won't work. The same notch that reduces interference will distort the digital signal. My recommendation is to move the dish. In most areas large dishes aren't required. While it may be difficult to hide a 7 meter dish from interference, it isn't hard to use a building or a screen to shield a 3 meter dish. We have some locations receiving our digital signal using 2.4 meter dishes. While microwave terrestrial interference seems to be declining, another type of interference is increasing. Modern LNB's operate in the 950-1450 MHz. range. This frequency range didn't see a lot of high power use. Ham radio operators used the spectrum around 1296 MHz., aircraft transponders and DME equipment used the frequencies around 1.0 GHz. and the military filled in the gaps. Times have changed and we're now seeing high power two-way radio and paging transmitters showing up in the 900 MHz. region. These signals can find their way into LNB's and interfere with reception. Better cable may help. If it doesn't, switch to an LNA (Low Noise Amplifier) and downconverter (C-band to L-band). Put the downconverter at a location where it is shielded from the offending transmissions. Use high quality RG-214 or better cable from the LNA to the downconverter. Here's a quick summary of what you need to know to prepare for receiving the digital video signals headed for your antenna. A perfect picture isn't enough -- a low "bit error rate" is important for reliable reception. Reduce the error rate by carefully aligning the dish using conventional techniques, paying particular attention to polarization. Replace LNB's over two years old. Be wary of off brand or consumer LNB's. A "digital" LNB is nice but isn't essential in most cases. If RF cabling is old, replace it. Use high quality splitters and make sure all unused ports are terminated. View loop thru connections with suspicion. Reduce out of band interference with a bandpass filter at the feed horn. Traps can't be used to reduce in band interference. Instead, shield the dish. It may be necessary to use a smaller dish that can be easily shielded. Interference at L-band frequencies (around 950 MHz.) may be a problem. Better cable or a separate LNA / downconverter configuration may help in t hese cases. That's it for this month. In future columns I'll cover some of the more technical details of digital satellite transmission. Also, I'll be reviewing some of the information I've received recently from manufacturers. Installing this equipment and getting our distribution system switched over has caused me to get several weeks behind answering letters. Please be patient. E-MAIL is much faster. I now have an Internet mail address -- dlung@gate.net -- please use it instead of the CompuServe address for Internet messages (CompuServe charges me for forwarded Internet messages). This should help those of you who would like to get copies of my programs (Cheap Remote Control Basic program, calorimetric RF power calculator, etc.). Let me know what program you want and I'll EMAIL it to you. The Cheap Remote program is an ASCII Basic program while the calorimetric power calculator/calibration logging program is for DOS machines. CompuServe users can reach me at 70255,460. Look for the files in B PFORUM. (Type "GO BPFORUM".) I try to answer questions by phone, but it gets busy during the day. The best time to reach me is between 6P and 7P eastern time at 305-884-9664. My fax number is 305-884-9661. If you must use mail, write me at 2265 Westwood Blvd, Suite 553, Los Angeles, CA 90064. Expect a 4-10 week delay on mail requests at this time. (California Amplifier can be reached at (805) 987-9000 / (805) 987-8359(Fax) in the United States, (33) 1 48 64 52 52 / (33) 1 48 64 52 55 in France or (55) 11 884-6411 (voice and fax) in Brasil.) Copyright (c) 1994,1995 H. Douglas Lung ALL RIGHTS RESERVED