[This article originally appeared on the web at http://www.qsl.net/k7on/contest/stubs.htm. Since it has vanished from that location, we are offering it here. This text was obtained by OCR-scanning a hardcopy of the original; while I have tried to proofread the numbers, a few errors may remain. -- VE3RHJ.] Using Coaxial Stubs In A Multi-Rig Station Received 6 or 8 suggestions regarding ways to improve rig isolation in a multi-setup including the following: 1. Bandpass filters between rig and amp. 2. Coaxial stubs at the amplifier outputs. 3. Antenna/Rig selection. 4. Separate towers for each band. 1. BANDPASS FILTERS: These come in XMIT or RCV-only types. Xmit filters are available from ICE (and elsewhere) or can be home-brewed -see K4VX' articles in NCJ and Sept'88 QST, and the-article in May?'94 QST; these designs include both 3 and 5 pole versions. Rcv-only filters can be built per the NCJ article by W3LPL and are a 5-pole design I think; these must be installed in the"receive antenna" line of your xcvr since they won't handle any power. Some stations use both types but it seems like the 5 pole Xmit filters would give equal performance to the 'LPL ones (ICE filters are 3 pole and have 10-15 dB LESS rejection than 5 pole types); filters can be cascaded but results unknown. 2. COAXIAL STUBS: Coax stub filters on "tees" (type N recommended) directly at the amplifier output on each band; sometimes these work better 1/8 wave down the line or a few feet after the power meter. Since my antennas are remotely switched at the tower I'm thinking of trying stubs on each line BEFORE the relay so they can be left PERMANENTLY mounted (isn't rf leakage from good coax at HF pretty low?). You can use either 1/4 OPEN or 1/2 SHORTED stubs (I recall seeing an old NCJ article on this by K2TR). Separate sets of stubs tuned for CW and SSB should be used for best results, giving 25 dB attenuation or so. Some stations use DOUBLE stubs at the second harmonic to increase the attenuation while others recommend one stub at both HALF and TWICE frequency (e.g. stubs for 40/15 and 10 on the 20m station). I tried coax stubs before I got the ICE filters and they DID help but haven't tried them yet in combination with the filters. Another interesting suggestion was to measure the receive power on one band while transmitting on another. Supposedly TENS or even HUNDREDS of wafts have been found in some situations! N4KG installs protective diode-resistor networks at the receiver inputs to guard against burnout. I plan to use my Bird to benchmark my station and then to track improvement? after each step. 3. ANTENNA/RIG SELECTION: Different combinations of antennas and rigs can have a big effect on interference. Using multiple tribanders versus monobanders opens the door to more severe interference and should be avoided if possible. I use both tribanders and monobanders but plan to confine them to two separate towers. I'm also thinking of trying some remotely located receiving antennas (e.g. trap vert. and inv. vee or small tribander). Some rigs "talk" to each other more than others (e.g. two TS-940s are purportedly worse than two TS-930s) and combinations of "synthesized" vs. "non-synthesized" (TS-830) can be tried (or try a Drake R4C for receive). 4. IF ALL ELSE FAILS: Several reported of having operated up to six stations pretty much interference-free (except for 2nd harmonic) using the above techniques with a single tower or rooftop. But this degree of success has not been universal. In these situations Separate Towers (widely spaced) for each band were the way to go. I have the space but not the energy to put up and maintain all this hardware (two towers are enough!). Nothing dramatically new came out of this except to prove that it CAN be done and there's still room for a lot of experimentation. Now where is that 5OW slug for the Bird -------------------------------- John, G4SXW, is correct that there are many more designs for transmission line filters than the shorted 1/4 wave stub. And some of them may work even better, depending on the design criteria. But I only addressed the shorted 1/4 wave stub, since that is the most common design for HF work currently in use by the ham community and was the design presented in RadCom. (I understand that there is a follow up article in RadCom with a correction in the November issue, but the colonies are far and the boats are slow, so I haven't seen it yet.) BTW I was not "proposing" the 2-stub shorted 1/4 wave design as G4SXW stated in his note, only responding to the series of messages that have recently appeared on the reflector with quantitative data. Unfortunately John hadn't seen the previous messages so didn't have the context in which my note appeared. There are reasons for the popularity of the shorted 1/4 wave stub type filter over designs that use open stubs with reactive compensations. 1) they are extremely simple to design and measure requiring no work on the Smith chart 2) they provide a DC short for static drain (quieter receiving and some additional lightning protection) 3) they don't require attention about what to do with the open-circuited end, which can have significant voltage, and needs to be insulated and shielded 4) they are typically good enough, except for odd harmonics (15M suppression of 40M signals) where they have no effect I am addressing some other designs in the current work I am doing, but didn't want to publish an entire volume on the reflector! G4SXW's November article in RadCom promises to provide a look at more complex designs and I am looking forward to the new insights I am sure it will provide ----------------------------- I'm gonna build another set of stubs as described by Fred Lass, K2TR, in his 1970's NCJ article. We've already built one set, and they work great! (I've appended the dimensions below.) I don't think it's that simple. The PERCENTAGE b/w is proportional to the length of the stub -- in wavelengths -- on the operating frequency. E.g., on 15m Fred recommends a shorted 3/4-wave stub, and on 80m he recommends a shorted 1/4-wave stub. So, the 15m stub will have ONE-THIRD the b/w of the 80m stub in percentage terms, but TWICE the b/w in absolute kHz. Note: This assumes that the stub coax has the same impedance as the line coax, and that the line SWR is reasonably flat. It's been my experience, too, that stubs work better. Here's an educated guess about why. BOTH do a good job on the primary problem -- rejecting one rig's fundamental TX signal in the other rig's RX line. But once that's solved, there's still a *big* problem with "broadband" noise from one rig's TX/amp on the second rig's RX freq. Low-power filters just can't touch this 'cuz much of the noise is generated in the amplifier. Once the noise escapes from the amp to the antenna, nothing you can do on the other rig will help. Stubs are great for this because they take high power and can be put on the output of the amplifier. But it's important that they be designed to notch the other rigs' bands, not just pass their own rig's band. That's why an antenna tuner can help in this regard, but usually not much as stubs. (Any of you who used to be Real-Old-Time "repeater gods" will remember that a set of notch cavities worked better than pass cavities ... in the days before everybody could afford pass/notch cavities.) I'd also guess that phase noise from modern PLL rigs makes the broadband noise problem worse. Even if you band-filter the signal between the TX and the amp, it still widens the harmonics generated in the amp. Anyway, plans at AG6D are to use both stubs *and* DoomStar filters. ---------------------------- P.S. Here's a description of the K2TR stub system. Thanks to WM2C for FAX'ing a copy of the NCJ article. Some bands use two or three stubs in parallel. All lengths assume a velocity factor of 0.66 -- FB for any solid-polyethelyene dielectric coax like RG-8A -58 -59 -213 -214 etc. but not foam coax like RG-8X or 8214 etc. 10m 11'6" open 1/2-wave nulls 20m 23'open full-wave nulls 15m and 40m 15m 23'shorted 3/4-wave nulls 10m and 20m 20m 23'open 1/2-wave nulls 15m and 40m 11'6" shorted 1/4-wave nulls 10m 40m 23'shorted 1/4-wave nulls 1Om and 20m 15'3" shorted nulls 15m 7'8" open reactance compensation 80m 46'shorted 1/4-wave nulls 1Om 15m 20m and 40m 160m 92'shorted 1/4-wave nulls 1Om 15m 20m 40m and 80m On 40m, the 7'8" open stub cancels the reactance of the 15'3" shorted stub; they should be joined with a T-connector. If the 40m and 80m stations still have serious interaction, a fourth 40m stub can be added: 40m 46'open 1/2-wave nulls 80m The 80m stub looks like it's cut for CW, but other bands seem be compromises between CW and phone. If you adjust the lengths to be either CW-or phone- specific, I'd recommend adjusting according to the desired null frequency rather than the pass frequency. ---------------------------- While preparing for my first single operator two radio venture, I ran into some severe RF interaction problems which at first looked overwhelming, but which proved amazingly simple to completely eliminate, with the help of transmission line stubs. Since I live on a small 70 x 100 foot city lot, and have only one small 48 foot tower, all of my antennas are very close to each other. In fact, the feedpoints of all the antennas are no more than six feet from each other. With this intimate physical proximity, there was enough stray RF coming down the feedlines to produce painful RF burns to the finger ("RF detector"), when a transmitter was operating at only 50 watts into one of the nearby antennas. I had purchased Dunestar bandpass filters, but I was afraid to damage them to so much stray RF when the amplifiers were operating at the 1500 waft level. I had already smoked an Autek Antenna Analyzer by inadvertently leaving it hooked up to an unused antenna while transmitting on another antenna. Having seen some recent correspondence regarding transmission line stubs on the contest reflector, I decided to try some experiments in that direction. Since I was primarily interested in operating two radios on adjacent bands (i.e. 10/15, 15/20, 20/40, and 40/80), I took notice of an interesting property of quarter wave stubs for 40 meters. These stubs, which are about 23 feet long, will alternately pass/reject the amateur bands 40, 20, 15, and 1 0 meters as shown in the table below: BAND LENGTH SHORTED 23'STUB OPEN 23'STUB 40 .25 wave pass reject 20 .5 wave reject pass 15 .75 wave pass reject 10 1 wave reject pass I built a pair of these stubs, one open and one shorted, and tested them in the "reject" mode sequentially on each of the unused transmission lines, and on all of the possible band/antenna combinations. The same feedlines that produced burns at 50 watts were now totally cool to the touch at 1500 watts! Operationally, the shorted stub is placed in the feedline anywhere after the amplifier on 40 or 15 meters, and the open stub is used on the other amplifier, which is on an adjacent band, 20 or 10 meters. For example, to operate stations on 15 and 20 meters, put the shorted stub in the 15 meter feedline, since it will pass 15 and reject 20. Put the open stub on the 20 meter feedline, since it will pass 20 and reject 15. 1 tuned the stubs to 7075, which makes them resonant also at 14150, 21225, and 28300. 1 made the stubs out of RG-8 and tuned them with an old tube-type grid dip meter, using the station receiver for accurate frequency determination. I sure wish I hadn't smoked the Autek analyzer! The concept of alternate pass/reject can be extended to 80 and 160 meters, but more stubs need to be built. To operate stations on 40/80 meters, build two quarter wave 80 m. stubs (about 46' long.) Leave one stub open, and put it in the 40 m. feedline. Make the other stub shorted and put it in the 80 meter feedline. Similarly, for 80/160 meters, make two quarter wave stubs for 160 meters. Put the shorted stub on the 160 feedline and the open stub on the 80 feedline. I recently used the 40 and 80 meter stubs in the CW Sweepstakes. on adjacent bands from 15 through 80 meters. Dunestar bandpass filters were also used between the transceivers and amplifiers. At no time did I hear even the SLIGHTEST HINT that there was a 1500w transmitter operating on the adjacent band. The second harmonic from 80 could be heard on 40, and the second harmonic from 40 could be heard on 20, but they sounded like just another very loud station on the band, and produced no desensing or phase noise outside the normal IF bandpass. If the "wrong" stub is placed on the transmission line, received signals go down by four to five s-units, indicating a rejection in the order of 24 to 30 dB. Not bad for a few bucks worth of coax. Note - DO NOT TRANSMIT ON THE WRONG STUB! I don't know what will happen, but it will probable involve your local fire department! The folks who have lots of real estate and/or big towers to physically separate their antennas may not need the extra rejection provided by transmission line stubs, But where the antennas are close, the combination of stubs and lumped constant bandpass filters definitely does the job. ---------------------------- I have read the information on stub filters that has recently been published here on the reflector with great interest since I have been building stub filters for multi-op contest efforts for more than a dozen years. I have been doing some quantitative measurements on stub filters on a low key basis for the last 6 or 8 months as part of a book that I have been working on and would like to add some information. I feel this information is important because of some misinformation that has been published in a number of places (most recently in RSGB's RadCom, October 1994 p. 56 "2nd Harmonic Filter for 50 MHz). This article, and others, show a pair of shorted 1/4 wave stubs installed 1/4 wave apart on the feedline in order to maximize suppresion of the 2nd harmonic energy (which is usually a worse offender than the higher order harmonics). In fact, for optimum suppression of the 2nd harmonic the shorted stubs should be placed 1/8 wave apart (or 1/4 wave AT THE HARMONIC). Two filters add no more loss to the fundamental than the 1/8 (or 1/4) wave section of cable that separates them. And, like multi-pole filters, the stubs can be stagger tuned to increase the bandwidth. Here is some data taken on stubs constructed in typical amateur fashion, ie with RG-213 cable and PL-259 connectors. The data were taken on an HP-4195 Network/Spectrum Analyzer and have been confirmed with modelling. The filters were built for 14 MHz as a compromise between the most useful frequency for a filter and the amount of coax that I would be using. Cable losses will be lower at lower frequencies so the performance should be somewhat better for 80 and 40 meter filters. CASE 1 A single shorted 1/4 wave stub for 20 meters, mounted with a UHF Tee connector, provides the following performance. Ultimate 2nd harmonic attenuation at 28.230 MHz -31.5 dB -20 dB bandwidth 910 KHz 3.2 % -10 dB bandwidth 3.1 MHz 11 % Ultimate 4th harmonic attenuation at 56.720 MHz -29.0 dB -20 dB bandwidth 1.0 MHz 1.8 % -10 dB bandwidth 3.5 MHz 6.2 % CASE 2 A pair of shorted 1/4 wave stubs for 20 meters, mounted with UHF Tee connectors 1/4 wave apart, provides the following performance. Ultimate 2nd harmonic attenuation at 28.240 MHz -53 dB -40 dB bandwidth 480 KHz 1.7 % -30 dB bandwidth 1.0 MHz 3.5 % -20 dB bandwidth 2.16 MHz 7.6 % -10 dB bandwidth 7.68 MHz 27.2 % Ultimate 4th harmonic attenuation at 56.640 MHz -52 dB -40 dB bandwidth 700 KHz 1.2 % -30 dB bandwidth 1.40 MHz 2.5 % -20 dB bandwidth 2.72 MHz 4.8 % -10 dB bandwidth 8.12 MHz 14.3 % CASE 3 A pair of shorted 1/4 wave stubs for 20 meters, mounted with UHF Tee connectors 1/8 wave apart, provides the following performance. Ultimate 2nd harmonic attenuation at 28.20 MHz -76 dB -40 dB bandwidth 1.3 MHz 4.6 % -30 dB bandwidth 2.3 MHz 8.2 % -20 dB bandwidth 4.0 MHz 14.2 % -10 dB bandwidth 6.8 MHz 24.1 % Ultimate 4th harmonic attenuation at 56.7 MHz -44.5 dB -40 dB bandwidth 300 KHz.53 % -30 dB bandwidth 900 KHz 1.6 % -20 dB bandwidth 2.15 MHz 3.8 % -10 dB bandwidth 7.98 MHz 14 % Conclusion: Although the case (#3) where the stubs are separated 1/8 wave provides somewhat poorer suppression of the 4th harmonic than the case (#2) where the stubs are separated by 1/4 wave, the suppression of the, typically most important, 2nd harmonic is much better. (-76 dB versus -53 dB) And the -40 dB bandwidth of the 2nd harmonic is also much better for case #3 by a factor of 2.7. (1.3 MHz versus 480 KHz) Of course this information can be scaled to the other HF ham bands where the cable losses are similar and the performance should be similar. ---------------------------- A while ago I was looking for info on making/using coax stubs for filtering between two stations. Thanks to WlPH, VS6WO, KY1H, K3NA, K1VR, WOCP, K3LR, KROY, WA60TU, WM2C, ON6TT, N40GW, (and maybe a few others who got lost while cutting and pasting this together) for the hints and comments. Several cited the K2TR article in the May - June 1984 NCJ. Following is a collection of the various comments received (in no particular order) beginning with the original post. ----------------------------- I'm looking for information on using coax stubs for filtering between the two stations. Some questions that come to mind include: 1) Lengths required for nulling various bands - open/shorted? 2) Can stuff like RG58 be used or should we stick with RG8 (will be running about 1 KW on each station?) 3) Do they need to be made with 50 ohm coax or can 75 ohm be used (have some of both laying around?) 4) Can we put these on a coax switch and switch in the ones we need depending on band? 5) Is it worth all this or should we just get out the MC/Visa and buy the Dunestar filters ----------------------------------------------------------- 1. Lengths required: This depends on the velocity factor of the coax being used to make the stub. In addition, you will have to have an appropriate serial coax impedance transformer between the point where the stub and antenna feed attach and the radio. Otherwise the radio will see a wierd impedance and potentially high SWR. The lengths of the serial matching coax sections will be different for each combination of antenna and band-stub. (If you are getting the impression that this is complicated, you are correct!) Either shorted or open-ended stubs can be used. However, shorted stubs are easier to build properly. Open stubs tend to leak at high powers (remember, that will be a very high voltage point at the end of the open stub!) 2. The quality of the stub as a filter depends on the loss of the coax. RG58 coax has high losses and you will lose many 10s of dBs of filtering. Even RG-213 is pretty marginal... (This is another problem for the use of stubs in an expedition...) 3. Effective stubs can be designed using any impedance of cable, although the lengths of the matching sections will change. 4. Coax switching of stubs is very complicated because the matching sections must also be switched. 5. It is simpler to buy the filters. You will get similar performance to reasonable-quality stubs (altho not as good as excellent stubs, and not at 1500 W power levels). Switching is a breeze. And Dunestar or ICE filters are smaller and lighter to carry with you to the site. ----------------------------------------------------------- The stubs are easy and cheap. I use RG62A/U that I bought by the pound from a metal salvage yard. Impedance of the stubs does not have to match the transmission line. RG62A/U is 93 ohm coaxial cable. Small cable is fine as it only has to carry the spurious signal. I use 6 dedicated feed lines for the 6 bands and have stubs permanently attached to each line. There is no reason not to use a switch to select the right stub. But if you forget to switch, the amplifier may be looking at a short circuit with the wrong stub. Stubs are not real high Q devices and seem to cover an entire band easily. ----------------------------------------------------------- An effective combo of stubs and filters allows almost any band/antenna combo to be on the air at the same time. ----------------------------------------------------------- Yes, stubs are worth it even if you use ICE or other tx filters I think. Most of the other filters are only exciter level so you put them between the exciter and amp, but they don't do anything for amp produced harmonics. In general the stubs are 1/4 wave shorted pieces of coax, except on 40m to null out 15m harmonics. If you are running kw's you want to use at least rg8x, preferably rg8 or rg-213. We melted down a stub using rg-59 here one contest. No, the impedance of the coax doesn't matter, but the lossiness of it does. Going from rg-58 to 3/4" hardline increases nulls from about 15 to about 40db. Using decent quality rg-8 gives good results. Yes, you can switch stubs, just remember that the size of the switch and any connecting coax pieces add to the length of the stub so shorten up the stub to compensate for it. ----------------------------------------------------------- As most expeditions involve dragging too much stuff through reluctant airline channels, I'd recommend buying the DuneStars. Yes, you may bandswitch stubs in and out. Yes, you may make them out of 75 ohm cable. But to make stubs that work, you MUST have a testing machine that will read out frequency (AEA, MFJ, network analyzer, etc.) RG-58 vs. RG-213: The less loss, the better the stub performance. Remember, DuneStars go between xcvr and amp. Stubs of RG-213 can go after amp. I use ICE bn xcvr and amp, stubs of .750 inch diameter between amp and antenna. ----------------------------------------------------------- Coax stubs are best at reducing broad band phase noise from nearby transmitters. They will provide about 20-25db attenuation on the next nearest band. The lower the loss of the coax used on the stub, the deeper and narrower the notch will be - take your pick from any coax that will handle the power. Note that stations have had stubs melt in the 'heat'of battle!. Better stick with 50 ohms... Many stations don't bother with stubs in multi-single - just good rx filters. ----------------------------------------------------------- There is a middle way, which is to use an ant tuner as a "tunable stub". This doesn't work in a m/m operation, but for what you guys are doing, it might be the best approach. I have used MFJ 16010 tuners for this, which cost about $30. 1 can get about 30 db of rejection on the offending band. Operationally, you simply switch to the offending band and tune the device for minimum band noise, and put little notation marks on it for reference during the contest.