The Lindenblad: The Ultimate Satellite Omni Antenna
The Lindenblad: The Ultimate Satellite Omni Antennaby Howard Sodja, W6SHP
[email protected]
based on articles previously published in The AMSAT Journal
July 1990 and November-December 1991 and OSCAR News August 1989
and June 1991.
Maintenance problems with my satellite beam system because of its heavy use on the Microsats motivated me to experiment with omnidirectional antennas. My first try was with a 70cm ground plane. The results were disappointing. Print was not 100% until the birds were 10 degrees above the horizon, and even then deep fades when the birds were high left me with several periods of no printing on every pass. J-pole and discone antenna users I spoke to were similarly disappointed.
As I often changed polarization on my satellite beams to keep the downlink signal strength up, I reasoned the solution would be a pair of switchable left and right hand circular polarized omnidirectional antennas. In searching the amateur antenna literature the only circular polarized omnidirectional antennas I found were in the Satellite Experimenter's Handbook. Ease of construction and tuning made the Lindenblad antenna the natural choice.
From the Satellite Experimenter's Handbook. Not to scale.
As can be seen in Figure 1 above, the Lindenblad antenna consists of four half wave folded dipoles slanted 30 degrees to the horizon, oriented 90 degrees to each other in azimuth, spaced 0.3 wavelength apart. They are tied together by four half wave 300 ohm twinlead lines that divide the folded dipole's impedance by four where they connect to the coax feedline (see Figure 2 below). A remote coax relay switch selects either the RHCP (Right Hand Circular Polarized) or LHCP (Left Hand Circular Polarized) antenna which then goes to a GaAs FET preamp.
From the Satellite Experimenter's Handbook. See Tables 1 and 2 for dimensions. Not to scale.
My 70cm Lindenblads delivered a signal that gave me solid print when the 1200 baud PSK Microsats were 5 degrees above the horizon, as long as I changed polarization during deep fades. I often got intermittent print below 5 degrees. The overhead null seen in Figure 1 (B) is offset by the increased signal strength as the birds get higher. Exhilarated by the performance of these antennas, I shot off an article for AMSAT Journal (footnote 2) to share my discovery with other frustrated omni antenna users.
Unknown to me, my article contained an error that offset an error in the Satellite Experimenters Handbook, giving me a 1.1:1 SWR when it appeared I should have had a mismatch (footnote 3). The handbook incorrectly stated the Lindenblad's folded dipole feedpoint is 300 ohms. In calculating the impedance for my coax matching section, I made a calculator key entry error. My 93 ohm coax matching section should have been 61 ohms. As I got the SWR I expected, I never recheck my matching section impedance calculation.
Stu, WD4ECK/7, undertook an in-depth theoretical investigation of this apparent anomaly with MININEC (footnote 4), a powerful antenna analysis program. His research disclosed that the close overlapping spacing added an inductive component to each folded dipole's feedpoint Z, raising mine from the expected 300 ohms to 573.5 + j431.1. When fed through my phasing lines this unexpectedly high Z matched my theoretically too high 93 ohm matching section, giving me my low SWR. Stu revealed that the mismatched reactive component in my "serendipity" match gives me a low SWR at the expense of storing some power in the near field of my antenna (footnote 5).
To obtain a non-reactive resistive feedpoint impedance the Lindenblad's folded dipoles must be shortened from their "free space" resonant lengths. These shortened dimensions are a complex function of the folded dipole's diameter to length to spacing ratios. This relationship could not be reduced to a simple formula. Using MININEC, Stu constructed Tables 1 and 2 which give the dimensions and resulting impedances for folded dipoles in the Lindenblad configuration for the 2m and 70cm satellite bands.
Stu then tested theory by building a MININEC-modeled 70cm Lindenblad which performed as predicted. I then built a LHCP and RHCP 2 meter MININEC modeled pair of Lindenblads and likewise confirmed his findings. My SWR for both antennas are as predicted by MININEC, 1.35 to 1. A quarter wave coax balun (footnote 6), tuned with a dip meter and spaced 7/8 inch from the coax feedline, connects the balanced phasing line to my coax. Without the balun my SWR was almost 1.5 to 1.
Playing with various MININEC configurations Stu decided to build another Lindenblad with dimensions for a 1:1 SWR. MININEC indicated he'd have a 200 ohms feedpoint with folded dipoles 29.5 cm long with a 1/4 inch diameter lower element and #14 wire upper element spaced 1.3 cm. This would result in a perfect 50 ohm match after the parallel phasing lines divided this by four.
But Stu's 200 ohm "Lindy" had a 1.3:1 SWR at the design frequency. About this time Ron, W6XY joined our efforts to build the perfect Lindenblad, but using a different approach (footnote 7). He too found the 1:1 SWR elusive. As we experimented and tested various Lindenblad configurations several points became clear.
As we suspected, my comment in the July 1990 AMSAT Journal to experiment with the routing of the phasing lines until you got the lowest SWR was in effect changing the electrical length of the phasing lines, which changed the impedance seen by the 50 ohm coax feedline. Some builders reported needing to slightly trim their phasing lines to get their SWR below 1.5:1.
Stu and Ron tried using coax phasing lines to avoid this suspected twinlead field interaction problems. Stu used a coax sleeve balun at each element feedpoint and Ron did not. But that 1:1 SWR still eluded us. Tests revealed that at 437 MHz a millimeter change in length in the 50 ohm coax phase lines made a noticeable difference, more so than with 300 ohm twinlead phasing lines.
Around this time QST published an article explaining some of MININEC's limitations (footnote 8). Stu also established contact with a professional antenna design firm that used MININEC. He learned that even the professionals must trim and fine tune their MININEC modeled antennas to account for "real world" factors. Stu and Ron were actually within the "real world" range to be expected. Stu found his resonant frequencies to be consistently between 1 to 2 percent higher than expected, while Ron found his to be lower by the same amount, using different construction materials and designs.
Some of the variables that were found to effect impedance that are not accounted for by MININEC are
Our struggle for the golden 1:1 SWR was more because of our trying to be design purists and validate MININEC rather than from need. The SWR builders are reporting is operationally insignificant as even a 1.5:1 SWR only results in a 5 percent (0.2 dB) power loss in a lossless line (footnote 9). With the low loss lines we run to our satellite antennas, your SWR signal loss should be under 1 dB (less than the human ear can perceive). A bit of fussing with the phasing line routing and maybe their lengths should reward you with an SWR below 1.5:1.
The 1200 baud Microsats' downlink is RHCP when using normal PSK and LHCP when using raised cosine. This is because the turnstile phase lines are changed when different transponders are used. But when there is deep QSB you must switch polarizations to bring your signal back up to maintain solid print. Therefore it is essential that you be able to switch Lindenblads (polarization) during a pass.
I also use a pair of 2 meter Lindenblads for my uplink and find that they outperform my linear (groundplane) antenna. I determine which uplink polarization to use by observing if the byte count PG displays increases with each transmit burst.
As getting organized and moving on a project is often the most time consuming part, I recommend building both RHCP and LHCP antennas at the same time. You will find everything you do on the second antenna will take a fraction of the time it did on the first one. And in my case the second one was built better. Practice makes perfect.
I used some scrap quarter inch thick plexiglass for my 70cm Lindenblad (see Figures 3 and 4). Wood or any other nonconducting material will work fine.
LHCP 70cm Lindenblad Antenna
Layout of cuts made on 1/4 inch Plexiglass 8 1/8 x 8 1/8 inches square. Drill four 1/4 inch dia. corner radius holes first. Then cut to side of holes. Drill 1/16 inch dia. folded dipole mounting holes to fit folded dipoles after dipole mounts are made. Loop No 20 bus wire around folded dipole and through mounting holes and twist tight on other side to secure.
Note 1: Mounting holes for folded dipole. 1/16" diameter.
Note 2: Dipole mounts are secured with 1/2" by 1/2" angle bracket or other means to end of each arm, on top of arm end, as shown. All tops slope in the same direction.
See Figure 5 for element mounting details.
Dipole mounts are attached to the base with 1/2 inch square angle brackets cut from extruded aluminum, and attached with brass hardware. You can glue or screw the dipole mounts to the base if your materials permit. Elements are tied to the mounts with bare wire and secured by soldering.
For my 2 meter uplink "Lindy" (see Figure 6) I used 3/4 inch white PVC pipes reinforced with wood dowels bonded inside with PVC cement. This reinforcement is essential where you will be clamping or drilling. Use a length of dowel reinforced PVC (or other non-conducting mast material) to support the "X" booms that goes beyond where your phasing lines connect together to your coax. See Figures 7 and 8 for details.
LHCP and RHCP 2 meter Lindenblads
PVC 'T' fittings with flat plugs provides a surface for bolting the plastic spacers. Elements are attached to the spacers with heavy gage wire.
2 meter Lindenblad balun and phase line attachment detail. The SWR will vary in windy conditions unless they are secured to spacers or the mast.
I recommend making a jig with dowels (or other round objects) mounted to a board or workbench for forming your elements. The extra time making a jig will pay for itself when you make your 8 folded dipoles. Filling 1/4 inch tubing with sand while bending it will minimize tube deformation or collapse. I used solid 1/4 inch aluminum rod to avoid this problem.
Ron's AMSAT Journal article describes his use of an outdoor electrical utility box for the center hub of his 70cm "Lindy", with coax phase lines inside the four 1/2 inch PVC pipe booms which support the folded dipoles. Folded dipoles made from TV twinlead are sealed inside PVC pipes and attached to the ends of the PVC booms with PVC "T" fittings. This gives a neat rugged weatherproof structure (footnote 7).
I have decided to stay with twinlead phasing lines as they are easy to build and easy to use for tuning. Also I wish to transmit as well as receive on my "Lindys". Cliff, W6HDO reported smoke coming from his RG-58/U coax sleeve balun when he fed 100 watts to his Lindenblad.
Figure 9 shows the approach used by the FAA's Lindenblads. Parallel rods serve as both element support booms and open line transmission lines. This approach may get around the interaction problems discussed above.
Lindenblads on top of a Federal Aviation Agency Airport Control Tower in Merced California
Your polarization will be RHCP (clockwise) or LHCP (counter- clockwise) from the perspective of viewing the 30 degree slopes from the center of the base. Figure 1 (A) is LHCP.
The dimensions of the power divider phasing twin lead lines in Tables 1 and 2 are calculated for ordinary 300 ohm TV twin lead with a 0.82 velocity factor. If you use a transmission line with a different velocity factor, adjust these lengths with the formula:
14998.7 x velocity factor
1/2 wavelength (cm) = -------------------------
Freq. in MHz
The routing of the 4 halfwave phasing lines effects the common feed Z where they connect to the coax. Experiment with the routing of your phase lines until you get the lowest SWR, keeping all 4 lines as symmetrical and separated as practical. You may need to slightly shorten these lines to get a good match.
To weatherproof the quarter wave coax balun and main feedline I rolled small pieces of COAX-SEAL (tm) into gummy threads to wrap around all exposed shielding and wires. I then molded them with my fingers to form nice neat solid moisture proof seals. Beware of using silicon as I can testify to the corrosive effects this has on coax shields.
I was anxious to test my 70cm Lindenblads as soon as I finished building them so I attached them to a couple old broomsticks and fastened them with pipe clamps to joists in my attic eight feet apart. Their performance was so good I decided to keep them there away from the ravages of the weather and large feathered birds.
I got solid print on the 1200 baud PSK Microsats when my FT-726R's S meter indicated at least 1.5 (my noise floor is S1). Whenever the S meter goes down to S2 or 3 I switch Lindenblads (polarization). This always brings the signal back up as much as 6 S units as long as the birds are above 5 degrees. I have found that I usually need to switch near the horizon and when the birds are near overhead. If I do not hear a bird at AOS time, changing polarization will bring it out of the noise.
Unfortunately, no omni antenna will give an acceptable signal to noise ratio for the broad band 9600 baud FSK satellites. I was able to keep my directory updated, but downloading files was painfully slow, and impractical, as you can get higher throughput on the 1200 baud PSK Microsats. Getting into the queue is difficult because of the lower EIRP. Stations located in areas with low uplink contention report better results on the 9600 baud FSK satellites.
When I tried running 100 watts to my 2 meter omni antennas, severe desense blocked my 70cm downlink receiver. I separated my uplink and downlink omni antennas by 30 feet, which reduced my mode JA desense. I was surprised that I could further reduce my desense by pointing my disconnected 2m/70cm beams at a right angle to my Lindenblads.
A weekend spent building up a Mode J cavity desense filter (footnote 10) that I inserted between by remote polarization switching coax relay and preamp close to the antennas eliminated the remaining desense problem. There was no noticeable insertion loss (less than 0.5 dB), but the 2 meter uplink was attenuated over 50 dB at the filter output. You cannot transmit through this filter, so it must be removed or bypassed for transmitting on 70cm. Construction details are available.
When comparing my 2m and 70cm Lindenblads with my 2m and 70cm ground planes and discone (for both uplink and downlink), at no time were the GP or discone stronger than the strongest of the Lindenblads on the Microsats, FO-20, AO-21, RS-10 or RS-12. The Lindenblad was usually better. But often the GP and discone were stronger than the weaker Lindenblad. Therefore you need to build both a LHCP and RHCP pair with a remote coax relay so you can always have the strongest Lindenblad on line.
I compared my Lindenblad polarization switching with my KLM crossed yagis to see if some Lindenblad fades that I assumed were polarization changes were perhaps actually nulls in the Lindenblad's radiation pattern. My yagis consistently confirmed these were actual polarization changes as indicated by the Lindenblads. As print is solid when the PSK Microsats are above 5 degrees, any Lindenblad nulls are operationally insignificant.
Using Lindenblads for both uplink and downlink on FO-20 (mode JA) I can begin to copy myself when the bird reaches 5 degrees with 100 watts EIRP. Higher in the pass only 10 watts are needed. I can also digipeat through the Microsats above 5 degrees.
My experimentation on FO-20's mode JA shows polarization changes are unpredictable. I have had up to 14 polarization changes on one pass, with no observable correlation between the uplink and downlink polarization changes. After the pass with 14 changes I monitored 10 meters and found it wide open to Japan, Australia, South America and the USSR. Other passes had far fewer polarization changes. Only one pass had no change from AOS to LOS. These preliminary tests suggest ionospheric and/or geomagnetic conditions effect polarization. More experimentation is needed to see what might be causing this phenomena (footnote 11).
For terrestrial 2 meter FM reception my GP and discone are generally superior to either Lindenblad. This is expected as most stations are vertically polarized. But the Lindenblad often outperforms my GP and discone if the stations are distant and not line of sight.
This project is a bit more complex than other omni antennas for satellite use, but it is much easier and cheaper than a standard tower mounted circular polarized azimuth and elevation rotating beam system. They are ideal for portable or apartment operation where beams are impractical. And Lindenblads definitely outperforms the simpler omni antennas, plus they are virtually maintenance free.
Feedback from other Lindenblad builders confirms my observations. Several builders reported good performance without baluns, which simplifies construction. Except for the SWR change mentioned previously I could not detect any operational improvement when I added my baluns. Of course, "theory" says you should use a balun to feed a balanced load to coax.
The Lindenblad has proven itself to outperform linear polarized omni antennas on all the active low earth orbit satellites. Once the word gets out I expect the "Lindy" will become the antenna of choice for omni satellite antennas.
146 MHz Lindenblad Antenna Characteristics:
Dimensions given in centimeters, impedances in ohms.
#12 Wire 1/4 inch dia.
----------------- -----------------
Spacing Length Impedance Length Impedance
------- ------ --------- ------ ---------
4 87.3 299.3 84.9 288.1
5 86.4 296.3 84.1 284.8
6 85.6 293.5 83.3 281.5
7 84.7 289.9 82.5 278.0
8 83.8 286.0 81.5 273.3
9 82.9 281.9 80.5 268.4
10 82.0 277.5 78.8 265.0
15 77.2 251.5 74.8 238.5
20 71.8 219.8 69.2 207.7
Phase line twinlead length (.82 vf) 84.2
.3 wave element spacing 61.6
437 MHz Lindenblad Antenna Characteristics:
Dimensions given in centimeters, impedances in ohms.
#12 Wire 1/4 inch dia.
----------------- -----------------
Spacing Length Impedance Length Impedance
------- ------ --------- ------ ---------
2 27.9 282.4 26.7 264.0
3 27.0 269.3
4 26.0 252.8 24.9 231.9
5 25.0 235.4
6 24.0 217.7 22.7 193.5
7 22.7 195.8
Phase line twinlead length (.82 vf) 28.1
.3 wave element spacing 20.6
Updated 7 September 1995. Article by Howard Sodja, W6SHP. HTML conversion by and feedback to KB5MU.
