AN EXAMPLE OF SIMULATION SOFTWARE’S FAILURE TO ACCURATELY OPTIMIZE WIRE COLLINEARS
Most of Collinears posted on internet are flawed design. These collinears are shown to be high gain by posting the supporting results of simulation. Almost no one posts results of actual measurements of Gain & SWR, or results of comparison with a standard benchmark antenna of known parameters.
It is my experience that these simulation software give results with errors which may be of the order of 5% to 10%, which is too high for accurate optimization of collinears. To highlight this, I optimized wire collinears by simulation, and calculated parameters of these collinears and a 1/4 wavelength (non-collinear) benchmark antenna. I then made prototypes and put on trial run with reference to benchmark antenna.
Results after optimization by simulation software were:
Benchmark antenna - Standard Cantenna: Gain = 1.48 dBi, SWR = 1.28
Antenna Under Test 1 - Coiled Cantenna: Gain = 2.99 dBi, SWR = 1.17
Antenna Under Test 2 - Franklin Cantenna: Gain = 3.13 dBi, SWR = 1.9
The simulation results show that both the Antennas Under Test should give better performance than Benchmark antenna. However when I made the prototypes using optimized dimensions given by simulation, and put on trial, results were opposite. This proved the unreliability of simulation results. I will now try to find the optimum dimension by trimming the prototypes.
I have actually made 2 whips under test only, and simply pulled out the 1/4 wavelength (69mm) whip of existing Standard Cantenna and inserted in its place the Whip under test. This made all other factors identical for the benchmark & under test antennas.
Prototypes: Standard Cantenna, Coiled Cantenna, Franklin Cantenna
If the antenna isn’t matching the model then logically there is something that isn’t being modeled. Just a thought but plenty of transmitting antennas use ferrite’s on the coax below the antenna to kill off common mode currents on the shield which affect antenna performance, maybe the can on its own isn’t enough isolation.
If can did not provide enough isolation, how 1/4 whip gives good performance, better than the high gain coiled whip & franklin whip?
The very reason I have opted to use a can is that it is 2 in 1. It’s circular horizontal bottom provides ground plane, and it’s vertical cylinderical part provides decouplig sleeve.
Also I have adopted the method to swap only the different whips, leaving everything else as it is. This makes all other factors identical, and difference inperformance can be attributed totally to the whip, and nothing else.
I feel the dimensions given by simulation for minimum SWR have an error of say 10% or more. The SWR is actually greater than 2 or even 3 at these dimensions, resulting in poor performance. I will now try impedance matching th Franklin Cantenna (using Transmission Line + Capacitor). The simulation software seems to give less erronous SWR = 2, as no coil is involved, only hairpin (stub) is there. I have used the impedance values (126-j48) and Smith Chart, and worked out that antenna, followed by a coax piece of 0.7 x half wavelength (using velocity factor) then a 3 pF capacitor, then feed coax, With my RG6 coax having VF=0.84, the half wavelength = 114 mm, and 0.7 x 114 = 80 mm long piece of coax (including F connectors at both ends) will be good. I will make it tomorrow, an put to trial. I will also try coax piece slightly bigger & shorter than 80 mm to reach the best length. I will also try 2.5 pF and 3.5 pF capacitors as well.
Trial & error is the option left to me due to inaccuracy of software results. If I had good test equipment, things would have been much easier.
Fair enough.
Though the 1/4 wave element + ground plane is inherently a good match to the coax so cant cause issues, the others maybe not so much.
Anyway looking froward to seeing your results from the capacitor matching. I tried caps once with the AI COCO but couldn’t get any real improvement, If this works out it will be time to try again.
The “1/4 wave element + ground plane” is inherently a good match to the coax ONLY impedance matching wise. For unbalanced common mode currents, situation is same as for all other antennas.
To suppress the unbalanced common mode currents, various methods are used, none perfect, but give substantial improvement. These include use of balun (balance to unbalance), ferrite beads, decoupling sleeve 1/4 wavelength, and making a few turns coil of the feed coax near the antenna.
Anyway looking froward to seeing your results from the capacitor matching. I tried caps once with the AI COCO but couldn’t get any real improvement, If this works out it will be time to try again.
The 4nec2 & mmana-gal softwares give large error for complicated antennas involving curved radiating wires/surfaces like coils of the collinear whip & shield of coco. Therefore for these antennas, the impedance value given by simulation have substantial error. Hence basing impedance matching capacitor & TL value on these values again gives poor results, and large number of trial & error attempts are required, which is tedious.
The Franklin whip consists of only straight sections of wire, hence error is not large. In my opinion, the reason of poor performance of Franklin Cantenna is not wrong optimized dimensions. Rather it is due to SWR which is nearly 2, while 1/4 wave whip has SWR 1.3, and this difference in SWR results in Franklin Cantenna performing inferior to Standard 1/4 wave Cantenna.
Anyway I will make & try the “Transmission Line + Capacitor” impedance match shortly and come back with the results, may be success, may be failure.
Impedance matching calculated by Smith Chart, then followed by trial run and trimming
I have installed Impedance matching for Franklin Cantenna by “coax+capacitor” as shown in diagram in my last post above. I tried 3 lengths of coax: 85 mm, 80 mm, and 75 mm.
The left-most part of graph is for the period when Cantenna had 1/4 wavelength (69 mm) whip. I then replced 1h4 whip by Franklin whip, and you can see drasticdrop in performance. I then inserted Impedance matching 75 mm coax +2.7 pF capacitor, then after about 15 minutes, changed piece of coax by 80 mm coax, then again after 15 minutes, changed piece of coax to 85 mm coax. As you can see best results are are 80 mm coax piece, as calculated by Smith Chart, very promising. But one thing is still bothring me: why the results of best match are only slightly higher than 1/4 whip? It should be say about 150% or so. I did not have 3 pF capacitor, so I have used 2.7 pF capacitor available with me. May be using aslightly higher or lower value of capcitor will gurther improve the performance. Someday,Iwill visit our local electronics & hobby shop and purchase many different values of capacitor in 0.5 pF to 10 pF range, and prrform trial and error again, this time keeping piece of coax fixed to 80mm, and changing the capacitor value between 1pF and 4 pf insteps of 0.5 pF.
Below is rrd collectd graph which shows the results.
That would only happen if there are sufficient aircraft there to receive. If you are already reliably receiving the majority of aircraft above the radio horizon, then adding more antenna gain isn’t going to increase what you receive much. It will only add those marginal signals that would otherwise be below the noise floor. Measuring receiver performance by counting aircraft is only really useful if you already know how many aircraft that are actually there.
Perhaps a tool like obj’s polar plotting script that can give an overview of where the signal is weak would provide a better picture, but would obviously need running for a while to obtain sufficient data. I’d like a tool that can produce similar plots, but using the average signal strength at a location rather than message rate. I don’t know of any that can do this though.
@caius:
Thanks! I seem to be somehow distracted from the fact that I am already receiving planes upto 250 nm+ using my Standard (69 mm whip) Cantenna, and there are hardly any planes left un picked.
I will now attempt to use a DVB-T dongle as signal generator.
Keeping signal generating dvb-t’s (trasmitting) antenna at a fixed distance from the “antenna under test”, I will compare the signal strengths of 2nd dvb-t, to which “antenna under test” will be connected, using SDR# or some “spectrum” software. I will keep the setup away from window, deep inside my apatrment where ads-b signal is practically non-existant, so that real ads-b signal do not interfere with the test.
I bow to your logic there abcd but i still think the discrepancies have something to do with the feed line!
@caius:
Thanks! I seem to be somehow distracted from the fact that I am already receiving planes upto 250 nm+ using my Standard (69 mm whip) Cantenna, and there are hardly any planes left un picked.
I will now attempt to use a DVB-T dongle as signal generator.
Maybe a couple of in-line attenuators dropping the signal strength (like these 6db units) would make it clearer which antenna option is performing the best.
Haven’t tried the RTL SDR signal generator my self but I see blogs around that say the local oscillators don’t go much below1.8Ghz, only one way to find out.
I have now tried to suppress common mode current on the shield of coax by making a coil of feed coax just before it connects to antenna. Coil is 6 inch (15 cm) dia, 4 turns. I tried all three antennas with & without coil of coax, and did not notice any difference.
Maybe a couple of in-line attenuators dropping the signal strength (like these 6db units) would make it clearer which antenna option is performing the best.
Good idea, will give it a try. How about reducing gain setting of dongle from “max” to say 45 and then 40 and see the effect?
Haven’t tried the RTL SDR signal generator my self but I see blogs around that say the local oscillators don’t go much below1.8Ghz, only one way to find out.
The one you mentioned involves modification of circuitry of DVB-T dongle. What Martin suggested is something different and does not involve any circuit modification.
Thanks for trying the coax choke on the feed line. Arr-well back to the drawing board.
[quote=“abcd567”]
Probably the best idea of all, but it would be interesting to try some hardware attenuation to see just how accurate or otherwise the gain settings on the dongles are.
[quote=“abcd567”]
Will try this my self when i get home, Martin G8JNJ has great stuff in that web site of his. Slightly OT but must try out his HF active antenna design and RTL-SDR noise suppression one day
I ordered a Broadband noise source and directional coupler from Ebay the other day for some SWR measurements, hopefully it will last longer than the one you tried a while ago!
but i still think the discrepancies have something to do with the feed line!