jg2u#@andrew.cmu.edu (Joseph Gerard Gottlieb) (03/18/87)
ReSent-From: postman#@andrew.cmu.edu ReSent-To: nntp-xmit#@andrew.cmu.edu Return-path: <jg2u#@andrew.cmu.edu> To: outnews#ext.nn.sci.electronics@andrew.cmu.edu Newsgroup: sci.electronics Well all of this talk about wireless data links has gotten my mind in motion. Here is my situation: I am the Chief Engineer at Carnegie-Mellon Universities' campus radio station WRCT. Our studio is located 600 feet (line of sight) from our transmitter, which is located atop a near by building. I would like to devise a a way of relaying audio at 15-16khz between the studio and transmitter. I have looked at STL's (Studio to Transmitter Link's which work via microwave). There are two problems I have encountered with STL's: 1) Expensive ~$6500 + $1000 for the antenna's. 2) It is almost impossible to get a clear frequency in Pittsburgh. You might be wondering "How do you relay the audio now ?". Unfourtunatly our audio is run through 2000' of 9 pair shielded "mic" cable (through a conduit along with the power feed for the building (none of this was my idea it was like this when I got there)). As you may imagine our air signal is less then optimal (read not very transparent). Another option is the use of two 15khz balanced broadcast lines rented from the phone company. But at $4000 per year - no thanks. Here are my requirements: Bandwidth- 15khz Signal to Noise- above 60db (can be lower if necessary) Realibility- Works under all conditions and almost never dies Am I asking for too much? Should I go beg for an allocation and buy the STL? Better yet is there a way to send an audio signal of very good quality (transparent, high S/N, 15khz bandwidth) over a 2000' run of mic cable (for a relitivly low cost) ? If someone could figure out a way of doing this they could make a killing in the broadcast industry. Many thanks. Joey Gottlieb Chief Engineer WRCT-FM Pittsburgh
braun@m10ux.UUCP (MHx7079 mh) (03/19/87)
Remember that an equalized phone line from the phone company
will be considerably longer than 2000 feet.
(And not even shielded!).
The tricks used by the phone company to preserve the signal quality
are to make sure the line is correctly balanced (with a transformer
at each end) and that it is terminated with the correct impedance.
Also, equalization is added to correct the frequency response.
The most cost-effective thing for you to do would be:
Make sure the cable is of a balanced construction (twisted pairs).
Use good 600 ohm balanced transformers at each end.
To improve S/N, increase the signal level. You must make sure
that the amplifiers driving the lines are not being overloaded.
10 volts into 600 ohms is a reasonable signal level. Also
make sure the receiving circuits are not overloading.
Attenuation may be necessary.
Use a good equalizer at the receiving end. A parameteric
(as opposed to graphic) is probably better.
This could be done for maybe 50-500 dollars, depending on whether
you have to buy a lot of new stuff. Make sure the transformers
will handle the signal levels without saturating. The little ones
meant for microphones are probably no good. You might try
begging off a local radio or TV station; this is the sort of stuff
that piles up in their back rooms. For a driving amp, you could certainly
use any reasonably good amp (>25 watts) meant to drive speakers.
You might have to add source impedance (0-600 ohm resistors in series with
the output).
In any case, you will have to fiddle a lot with the various resistance
values, gain, and equalization to get the best results.
The reliability will be better that a STL, of course.
--
Doug Braun AT+T Bell Labs, Murray Hill, NJ
m10ux!braun 201 582-7039larry@kitty.UUCP (Larry Lippman) (03/19/87)
In article <174@m10ux.UUCP>, braun@m10ux.UUCP (MHx7079 mh) writes: >> Our studio is located 600 feet (line >> of sight) from our transmitter, which is located atop a near by building. I >> would like to devise a way of relaying audio at 15-16khz between the studio >> and transmitter. I have looked at STL's ... >> You might be wondering "How do you relay the >> audio now ?". Unfourtunatly our audio is run through 2000' of 9 pair >> shielded "mic" cable (through a conduit along with the power feed for the >> building (none of this was my idea it was like this when I got there)). As >> you may imagine our air signal is less then optimal (read not very >> transparent). Another option is the use of two 15khz balanced broadcast lines >> rented from the phone company. ... > Remember that an equalized phone line from the phone company > will be considerably longer than 2000 feet. If this radio station is experiencing poor transmission quality on only 2,000 feet of shielded, twisted-pair cable, then something is seriously wrong! First of all, assuming your cable is 22 AWG (even if it's 24 AWG, it won't be more worse for this distance), the attenuation at 15 KHz will be around 2.0 dB. That ain't very much. In fact, since most radio program lines are equalized to +/- 1.0 dB, you might not even need equalization at all in this application. Should you wish to equalize, a simple series-resonant equalizer should be adequate. These devices are available from both telecommunication and broadcast suppliers. However, with no equalizer you claim to have a present problem. As I see it, there are four likely causes: 1. You have a defective cable, possibly with leakage from conductor-to- conductor and conductor-to-ground. Such leakage could be rather high resistance and caused by water permeating a damaged section of cable. Since this may be high-impedance leakage, it may not have been recognized as such; i.e., any attenuation and noise may have been attributed to equipment and not the line. Many installations of conduit which run underground will fill with water. This is an accepted fact-of-life and generally not a problem, unless the sheath of the cable is damaged. How did the cable get damaged in the first place? Most likely by the electricians who pulled it! I have had firsthand experience with electricians who yanked the hell out of communication cable, causing internal damage, by thinking it was just like a piece of (more robust) 250 MCM wire! 2. The audio transmission line is incorrectly connected at the audio control board and/or transmitter ends. 2000 feet of cable between two buildings is too far to run without full transformer isolation at both ends. Perhaps one or both ends have no isolation transformer; if so, one should be provided at _each_ end. (600:600 ohm 15 KHz line transformers are common ). Also, be certain that the cable shield is connected at ONE end only; this situation is ripe for _real_ ground loop trouble (unlike the "imagined" ground loops 5 feet away in the same room as claimed by some twits in rec.audio :-) ). 3. You have correctly connected, but improperly adjusted equipment at one or both ends of the transmission line. As an example, perhaps you are sending from the audio control board at an abnormally low level like -20 dBm, and the transmitter audio gain is cranked up enough to also pickup hum. Use a properly terminated audio sine wave oscillator to send a test signal into the audio control board, and measure the level at both ends of the line using a properly terminated (bridging impedance to start) transmission level meter (i.e., AC voltmeter calibrated in dBm with proper isolation and termination impedances). This should give you a clue as to your problem area. NOTE: My personal opinion is to forget about VU's here, don't rely on VU meters for this type of measurement, and deal strictly in dBm using telecommunications transmission measuring equipment. 3. You have correctly connected, but defective equipment at one or both ends of the transmission line. > The tricks used by the phone company to preserve the signal quality > are to make sure the line is correctly balanced (with a transformer > at each end) and that it is terminated with the correct impedance. > Also, equalization is added to correct the frequency response. Equalization, as in the case of a series-resonant equalizer, will add overall line attenuation, but flatten the attenuation-vs-frequency curve to an acceptable deviation (usually +/- 1.0 dB for program lines). > To improve S/N, increase the signal level. You must make sure > that the amplifiers driving the lines are not being overloaded. > 10 volts into 600 ohms is a reasonable signal level. Also > make sure the receiving circuits are not overloading. > Attenuation may be necessary. 10 volts into 600 ohms is a pretty large signal, like +22 dBm. Signals this large can cause crosstalk on adjacent circuits. A more reasonable level would be +8 dBm. > ... For a driving amp, you could certainly > use any reasonably good amp (>25 watts) meant to drive speakers. No, no, no! We're talking about SIX MILLIWATTS (+8 dBm) here. Your audio control board should certainly be able to provide this type of signal to drive this transmission line - even allowing for insertion loss through a transformer. Fewer interposed amplifiers will result in less distortion and fewer other problems. > In any case, you will have to fiddle a lot with the various resistance > values, gain, and equalization to get the best results. There should be little need for any "subjective" fiddling here. All you need is a variable frequency sine wave oscillator and a levelmeter with proper terminations and designed for telecommunications measurement. Start at say, 50 Hz, and send tones in 1,000 Hz increments at the _same_ reference level through the audio control board and over the transmission line. At the transmitter, measure the level at each frequency, and make a frequency/attenuation chart over the full 15 kHz range. If you don't like the slope, then crank in a series-resonant equalizer. For each equalizer setting, plot a new graph; quit when you like the slope. Don't forget, as you add equalization, the overall attenuation increases, but curve gets flatter. When you get the correct slope (i.e., correct equalizer settings), then you can make a final overall level adjustment. > The reliability will be better that a STL, of course. Agreed! <> Larry Lippman @ Recognition Research Corp., Clarence, New York <> UUCP: {allegra|ames|boulder|decvax|rocksanne|watmath}!sunybcs!kitty!larry <> VOICE: 716/688-1231 {hplabs|ihnp4|mtune|seismo|utzoo}!/ <> FAX: 716/741-9635 {G1,G2,G3 modes} "Have you hugged your cat today?"
jpexg@mit-hermes.UUCP (03/20/87)
In article <MS.V3.18.jg2u.80021112.easton.ibm032.3336.0@andrew.cmu.edu>, jg2u#@andrew.cmu.edu (Joseph Gerard Gottlieb) writes: > Here are my requirements: > Bandwidth- 15khz > Signal to Noise- above 60db (can be lower if necessary) > Reliability- Works under all conditions and almost never dies > Is there a way to send an audio signal of very good quality > (transparent, high S/N, 15khz bandwidth) over a 2000' run of mic cable (for a > relitivly low cost) ? If someone could figure out a way of doing this they > could make a killing in the broadcast industry. Many thanks. > Joey Gottlieb How about digitizing your signal to whatever standard is required (bits/sample and samples/second) and sending it via a good electronic link (could it be done via twisted-pair using balanced line drivers?), maybe a coax cable, or via a fiber optic link, or via an over-the-air optical link? --John Purbrick
dsi@unccvax.UUCP (03/20/87)
In article <174@m10ux.UUCP>, braun@m10ux.UUCP (MHx7079 mh) writes: > Our studio is located 600 feet (line > of sight) from our transmitter, which is located atop a near by building. I > would like to devise a way of relaying audio at 15-16khz between the studio > and transmitter. I have looked at STL's ... > audio now ?". Unfourtunatly our audio is run through 2000' of 9 pair > shielded "mic" cable (through a conduit along with the power feed for the > building (none of this was my idea it was like this when I got there)). As > you may imagine our air signal is less then optimal (read not very > transparent). Another option is the use of two 15khz balanced broadcast lines > rented from the phone company. ... No. No. No. No. No. Do ** NOT ** rent two broadcast grade telephone lines from the phone company under any circumstances. Their program equipment is unbelieveably archaic (although BellSouth have finally gotten with the program, so to speak, and are now providing active equalisation on their program audio circuits). I am Director of Engineering for a nationwide radio network, and have more problems with program audio circuits from BellSouth by far than any single other thing. 95 % of our network failures are caused by deterioration or loss of our 15 kHz and 8 kHz loops. Even double redundant circuits to our uplink still gives us unacceptable downtime (about 24 hrs/year). We are installing microwave feeds as a result. Now, over 2000 feet of twisted pair, even as the original poster described, it should be entirely possible to be flat within 2 dBm from 30-15,000 kHz PROVIDED THAT THE LINES ARE PROPERLY BALANCED (the usual Western Electric repeat coils are very good for this purpose, but not as good as say, Jensen Transformers repeat coils) AND TERMINATED. We have 12,000 feet of copper to the first CO, and about 8000 feet from there to our uplink facilities, and can guarantee a flat frequency response to our Board of Directors within 0.5 dBm from 30 to 15,000 Hz. Total harmonic distortion is 1.0 percent, though, at an injection level of 8 dBmV. Principally, we feel that this is due to the equipment at the central office. I prefer an STL, even over such short distances. The problems of pulling cables through a college campus are unbelieveable. Mechanical mishandling is exceedingly common in this environment, as are dufuses who habitually bridge everything on a punchblock to see what is going on. Also, you are using remote control (presumably), and I find the implementation of remote control over the SCA to be desirable as well. However, over 600 feet, this may not be a consideration. If you are not getting "transparent" audio quality over 2000 feet, even without equalisation, you are probably experiencing grossly mismatched levels in the program audio equipment (again, having a composite STL, when engineered correctly, allows you to virtually remote the connection to the transmitter modulator into your studios). Try using the station's modulation monitor, an ordinary oscillator, and remove ** all ** the program audio equipment and make a frequency response check. Be sure that the 75 uS preemphasis curve is correct (Royal and Aincent transmitters used LCR circuits for this with -- gakk -- waxed paper capacitors). If you don't have a decent modulation monitor, get one. Put the station on monaural for a few days and see if the problems change or go away. Linearizing circuitry in some direct-FM exciters needs periodic calibration for optimum THD and IM distortion characteristics. This should be done by an experienced consulting engineer, however. Are you sure that the VSWR at the final amplifier plate circuit is acceptably low, and that you don't have gross antenna problems, excessively low loading in the driver stage (meaning high circuit efficiency but also high "Q") or high incidental AM? Find the manual on your modulation monitor and learn how to use it (the McMartin and Belar monitors are good, and the QEI is exceptional). Are the intermediate stages in your transmitter tuned correctly (yes, people try to yank 110 kHz-ish crystals around still and multiply yp to the carrier frequency) with correct stage loading and tuning? Does the modulated 38 kHz subcarrier cross the time axis with a rising slope at the same time as the 19 kHz pilot ? Just 5 degrees of change will cause stereo distortion to rise sharply and separation to decrease. If it were ** my ** station, the first thing I would do is take my trusty CD player (no flames from the analog recording tech weenies, as I am an analog engineer by profession), and connect it to the monaural input of the transmitter with no compression or limiting. Try this also at the input of the stereo generator, and listen critically. This will at least give you an indication of how rotten your transmitter is. If you have older equipment, replace all of the electrolytics, particularly if they were made by Sprague. Parallel them with some decent polypropylene caps for better transient response. I hope this helps...nothing could be more satisfying than taking an underdog station and making it the best in town. David Anthony DataSpan, Inc .l
jg2u#@andrew.cmu.edu.UUCP (03/20/87)
<1. You have a defective cable, possibly with leakage from conductor-to- < conductor and conductor-to-ground. Such leakage could be rather < high resistance and caused by water permeating a damaged section of < cable. Since this may be high-impedance leakage, it may not have < been recognized as such; i.e., any attenuation and noise may have < been attributed to equipment and not the line. Many installations < of conduit which run underground will fill with water. This is an < accepted fact-of-life and generally not a problem, unless the sheath < of the cable is damaged. How did the cable get damaged in the first < place? Most likely by the electricians who pulled it! I have had < firsthand experience with electricians who yanked the hell out of < communication cable, causing internal damage, by thinking it was just < like a piece of (more robust) 250 MCM wire! DING! DING! DING! The bells and whistles are going off. As for your other points - I have already looked into those problems. But this one escaped me. So I have just gotten off the phone with the electrian that ran the wire. He remembered pulling the wire well (from three years ago) - why ? Because they used a 4 ton come-along and a cable griper to take the slack out of the run (since they couldn't access the pull boxes along the way). Consequently the Chief Engineer at the time tried (and failed) to make them replace the entire line (it was all one piece). And to top it off the conduit is always filled with water! AAAGGGHHHH!!! Thanks for the response. But now what? Since there are 18 wires and 9 shields could I find enough good wires an possibly go digital (as many people have suggested) Joey
larry@kitty.UUCP (03/24/87)
In article <MS.V3.18.jg2u.80020d03.newcastle.ibm032.3867.1@andrew.cmu.edu>, jg2u#@andrew.cmu.edu (Joseph Gerard Gottlieb) writes: > <1. You have a defective cable, possibly with leakage from conductor-to- > < conductor and conductor-to-ground. Such leakage could be rather > < high resistance and caused by water permeating a damaged section of > < cable. Since this may be high-impedance leakage, it may not have > < been recognized as such; i.e., any attenuation and noise may have > < been attributed to equipment and not the line. Many installations > < of conduit which run underground will fill with water. This is an > < accepted fact-of-life and generally not a problem, unless the sheath > < of the cable is damaged. How did the cable get damaged in the first > < place? Most likely by the electricians who pulled it! > > DING! DING! DING! The bells and whistles are going off. As for your other > points - I have already looked into those problems. But this one escaped me. > So I have just gotten off the phone with the electrian that ran the wire. He > remembered pulling the wire well (from three years ago) - why ? Because they > used a 4 ton come-along and a cable griper to take the slack out of the run > (since they couldn't access the pull boxes along the way). Consequently the > Chief Engineer at the time tried (and failed) to make them replace the entire > line (it was all one piece). And to top it off the conduit is always filled > with water! AAAGGGHHHH!!! > Thanks for the response. But now what? Since there are 18 wires and 9 > shields could I find enough good wires an possibly go digital (as many people > have suggested) On the assumption that the cable is the cause of your transmission problem, the first thing to do thoroughly test the cable to ascertain the extent of the damage. The simplest test is to disconnect the pairs from the apparatus at both ends, and using an ohmmeter test the open-circuit resistance of each pair itself (i.e., conductor-to-conductor of given pair), and of each pair conductor to ground. I would venture that each of these measurements should be well above 1 megohm for any of the above measurements, since the inherent insulation leakage resistance of any polyethylene or PVC insulation on your cable for such a short distance is pretty high (probably several megohms for 2,000 feet). A few words of caution on the making these resistance measurements: 1. Allow sufficient time for the ohmmeter excitation voltage to "charge" the capacitance of the line under test (you'll understand what I mean the first time you make a measurement). 2. If you have a conductor-to-ground fault close to the opposite end of the cable && there is a ground potential difference between the two buildings (a common situation), you may have AC and/or DC foreign potential across your measurement leads. This foreign potential can introduce a serious amount of error in your resistance measurements. Always take a measurement using both forward and reverse polarity on the ohmmeter leads; both measurements should agree - if not, you have a foreign potential problem. You can also measure the voltage from ground to each conductor to verify the presence of foreign potential. If you do have a foreign potential problem, then you need a resistance bridge set up to do cable-fault testing in order to get useful measurements. The ideal cable-testing device for your application is a time-domain reflectometer (TDR) designed for cable-pair testing (most TDR's are used for coaxial cable testing; that type of TDR won't work properly on pairs). A TDR display will instantly show the precise location of any resistive, capacitive, or shield-related cable faults. Unfortunately, your institution probably does not have such a TDR. I would suggest that you try contacting your local telephone company or electric utility company to see if they would loan you a TDR and a person to run it as a "public relations gesture"; all it will take is 10 minutes or so. While electric utility companies don't deal much with twisted pairs, many of the TDR's that they use for fault-locating (like those made by Biddle) happen to also handle pairs. Once you locate your fault, you should replace the section of cable between pull boxes. Be sure to use an encapsulated splice kit (like those made by 3M that use Scotchcast (tm) resin) to maintain the watertight integrity of the cable splice. The ideal situation of to replace the ENTIRE cable; who knows what stress may have been applied in excess of the safe mechanical strength rating that will suddenly become a problem as the cable insulation ages? If you find some good and some bad pairs, you could use only the good pairs - but I view this as only a stop-gap measure, prior to replacement of the entire cable. I view cable faults (especially where moisture may be involved) as a type of "cancer"; while the fault may go into remission, it will invariably come back - worse. This also goes for presumably "good" pairs in a faulted cable eventually going "bad". <> Larry Lippman @ Recognition Research Corp., Clarence, New York <> UUCP: {allegra|ames|boulder|decvax|rocksanne|watmath}!sunybcs!kitty!larry <> VOICE: 716/688-1231 {hplabs|ihnp4|mtune|seismo|utzoo}!/ <> FAX: 716/741-9635 {G1,G2,G3 modes} "Have you hugged your cat today?"