How I once again made antennas at home


By Vlada, OK1VPZ, 2020-05-17 / English version edited and commented by Hartmut, DG7YBN

Intitially we had to come to a decision, what to do with the clubs (OK2A) 70cm antennas. I had built a vertical stack of 12 x 6 el. years ago. These Yagis still work more or less as expected. This time I undertook the task of upgrading the clubs 10 element WY7010 antennas by WiMo, which were once bought by David OK1RK together with OK1TEH. It turned out that these antennas, assembled into stacks of 8 x 10 el. Yagis do not "pull" as would be desirable. So we analyzed the question and came to the conclusion that the problem is complex, but probably the largest part was caused by too large vertical distance between individual antennas and phase inconsistency
DG7YBN: such as a lost contact on the braid side of the enclosed N-bushing. Which led to splitting the vertical lobe of the whole stack and operational loss due to unsuitability of such a system for likewise Airplane Scatter propagation. We wanted antennas with a consistent phase (which turned out to be the biggest problem in the end), with relatively good SWR that would not detune in rainy conditions but mechanical parameters similar to the existing antennas - ie. length approx. 150cm, formast mounting, boom 15 x 15mm (so that existing brackets can be used) and connector let out towards the rear.

So I turned to Hartmut DG7YBN, who is currently one of the best experts in antenna design, and asked him to cooperate. Hartmut accepted the collaboration (TNX!) and tentatively designed a 9 element Yagi for which I bought material (for 16 antennas). I didn't want a bent emitter that is "trendy" and has a lot of electrical benefits, but because I wanted seamless repeatability and consistent parameters, I asked DG7YBN for a straight, simple and unfolded dipole, as well known from DK7ZB designs. In order to achieve a strictly same phase of radiation, I wanted to use a microstrip impedance transformer etched on AD450 laminate as made for the 12 x 6 el stack - see - here. In the end, however, this laminate could not be obtained. So I was forced to do the impedance transformation and desymmetrization using a 35 ohm Teflon cable RG 141 which can be bought in Italy. So I started producing the first sample. The designer of the antenna also continued to work on the design and finally, in the given dimensions and gain, designed the same length and gain wise even slightly better antenna than the tentative 9 ele. Yagi, even with only 8 elements - and it is unnecessary to describe parameter values here - you can find them on the web here on www.dg7ybn.de.

DG7YBN: my first reply was "I do not do DK7ZB designs" (hi). But sure I can design a 28 ohms Yagi with a straight dipole. Be assured that the bent dipole designs also work for Yagi bays. A good example is the 128 x 11 ele. GTV Yagis at DL7APV and many smaller arrays used for moon bouncing were every 10th dB counts. But it must be also said that nothing comes for free. Having a second parameter in the bending angle of the originally introduced by K6STI bent dipole is a second issue that has to be reproduced on spot.

However, I feel honored to work with Vlada on this project. Being part of a larger scale Yagi array build for an experienced contest station like OK2A certainly is a field to gain more experience myself. What started as a modification of the WiMo WY 7010 with Vlada's encouragement and feedback became a self-containing design. I named the Yagi "YBN 70-8z".


The first design of the final antenna looked like this:

aaaNote: DE means emitter (dipole = Driven Element).


Comparison: Simulated Elevation Patterns in Free Space: Black = WY-7010, Blue = YBN 70-8z


Hartmut sent me a copy of the result of his design in SW EZNEC as well as a comparison of data and analysis of the existing antenna. Hooray - so I immediately ordered the material according to the yellow column (arrow "NO") of the electrical length of the elements. After about a week, the elements came and I tried to build such an antenna using the plastic clips of the WiMo antenna. It didn't work at all! Only then did DG7YBN sensitively warn me that it is necessary to take those lengths with a correction of 4.8 mm due to the size of the boom and plastic clips. So from the right column. So I had to order the elements and pay again ... :-)

I built the antenna as soon as the new elements came using the boom and parts of one of the original WIMO antennas and measured - the first measurement turned out completely wrong. I consulted with Hartmut day after day every step and sent him measured values, made recommended changes and slowly approached the required parameters.. I'm not equipped to measure gain and anteroposterior ratio and I just needed to adjust the antennas well so that they repeatedly have a reflection attenuation of at least 27dB and in the case of a wet antenna at least 22dB at 432.2MHz, while for a 1dB change they should be frequency wide min. 3MHz so that they are not extremely sensitive to negligible changes in size and surroundings. I learned a lot when setting up the antenna. In addition to the length of the dipole, position D2 is probably the most sensitive element. After two weeks of work, I recommended Hartmut to change the dimensions of the elements as follows (red numbers) with a tolerance of + 0 and minus 0.5mm:

8 el. Yagi as upgrade of WY7010 - by DG7YBN
el. fixed by PP holders by Konni on 15x15 mm boom
mm el length distance diameter
Refl 341 0 8
DE 306 129 10
D1 307 265 8
D2 299 444 8
D3 291 676 8
D4 287 937 8
D5 283 1198 8
D6 275 1425 8

The antenna made in this way (still with the original dipole box and the manufacturer's connector) finally looked relatively good, and so it was time to verify the repeatability of the antenna design when making a larger number of them. (min. 16pcs). Hartmut's answer was to return the position of D1 back to 257 mm (measured from the reflector) and that the rest, according to his analysis, is not the best, but that it could have been worse ..

Because the original WiMo antennas did not have N connectors that passed this test and would therefore have to be replaced, which was a relatively significant financial item, and because drilling the original, relatively weak 15x15 mm boom would further weaken the strength of the antenna, we decided that we would rather make new antennas than remake existing ones.

First it was necessary to choose the method of dipole production - after good experience from the past, I used the production of a compact dipole separately and its subsequent installation in a box. From a laminate rod with a diameter of 10 mm (manufactured as a rod for electric fences ), I turned the couplings for both halves of the dipole and pressed them into the inner ends of the radiator so that 18 to 20 mm remained between the ends of the aluminum tube. The 10x1mm aluminum tubes were both 143mm long. The two halves were joined into a solid unit and I smoothed the burrs with sandpaper. Subsequently, I drilled about 4mm from the ends of the AL tube with 3mm holes for the dipole screws. The whole operation can be seen in the pictures:

Another point of the work was the production of a drilling template for the position of individual elements on the boom. I used a steel U profile of 18 x 18 mm with a wall thickness of 1.2 mm. Because it looked like the WiMo plastic box had a significant effect on the antenna fitting, I spent a lot of time building a suitable box that would be strong enough (we carry these antennas for each contest), waterproof and electrically least affecting the Yagi. Finally, I proposed to use an aluminum L profile of 50 x 30 x 2 mm, cut to 30 mm slices, which is be inserted into the box and carries 2 x M4 screws that hold the box to the boom. I think everything is pretty good to see in the pictures below:


SCAME type 855 Box, downside

SCAME type 855 Box, inside

The connector hole will be on the opposite side

Cut off gland to fit in the N-bushing

The bottom of the box under the angle is leveled

The hole for the connector N is of 12mm in diameter

L profile 50 x 30 mm, width 30 mm

The holes for the screws are threaded and the distance between the M4 screws for attachment to the boom is 30 mm.

The bottom of the box under the angle is milled

Use stainless steel screws M4 x 25 mm


It should fit


M3 x 12 mm screws fastening the N-bushing


The screws on the dipole are positioned to an angle of approx. 45°

I have now ordered a complete set of material - at the height of the crisis I used mainly e-shops - such as aluminum parts (15 x 15 mm boom and individual construction elements, cut to the required dimensions, I bought at the Alupa e-shopu cut parts came with accuracy up to 1 mm - mostly + 0.5 to + 0.7 mm. I ordered plastic boxes for the antenna radiator (SCAME type 855) here - these Italian distribution boxes have waterproof glands that fit a dipole diameter of 10 mm and have proved very useful to us already during the construction of 12 x 6el. stack that uses antennas designed by DK7ZB. Well, we bought plastic element insulators for 8 mm elements from DM3MS .

I made inserts into the boxes from aluminum L profile 50 x 30 mm with a width of 30 mm, which reinforce the boxes and ground the (important!) N-bushing (see pictures above). Then I assembled the antennas and adjusted them. I have another job, so it took such a long time, that Hartmut asked if I would be alright. :-)
So a few words according a useful setting to test the antennas:

What will you need?

a) A suitable place where the antenna can be directed to the sky and where there is no obstacle in the area of ??the assumed radiation angle of the antenna (see simulations). I did this work on the balcony, where I attached a 20 x 20 x 2 mm square steel tube with a length of about 180 cm to the corner of the railing, into which the antennas were one-by-one inserted for testing.

b) Without measurement there is no knowledge. A good two meter measurment tape is the basis, a large 310mm drawer is more than needed. A decent vector reflectometer is a must. But beware: small Mini VNAs, with which the market is flooded nowadays, must be able to be calibrated accurately, including the measurment coax cable, about 2 to 3 m, which is not always a matter of course. That's why the elderly Wiltron Site Master S331 came to honor. But I got on with it myself, when I didn't check the calibration via the cable quite often and believed that it would be stable. This is not the case, if in April the sun shines for a while and the measuring instrument with the cable is hot for a while and cold again in a while, it also has a negative effect on the calibration and this must be respected in the measurement, otherwise the results will be misleading. It is necessary to realize that the cable always somehow transforms the impedance (real and complex part) and then the results do not correspond to reality. So calibrating the measuring system every half hour is definitely worth it.

And the results of my optimization? See the link at the end of the article. Based on a long measurement and "antenna pitching", I finally used the following dimensions for all antennas:

8 el. Yagi as upgrade of WY7010 - by DG7YBN

El. fixed by PP holders by Konni Antennen on 15x15 mm boom
mm el. length distance diameter
Refl 341 0 8
DE 304 129 10
D1 305 257 8
D2 299 445 8
D3 291 676 8
D4 287 937 8
D5 283 1198 8
D6 275 1425 8

 

I know it's not exactly the same dimensions recommended by DG7YBN. But Hartmut will comment. We are happy to publish his answer here

DG7YBN: The relatively large SCAME Box and also the offset between dipole and element plane that goes along with using this box need to be compensated. Usually this is done by trimming D1 in position and / or length alone. But here Vlada went a bit a different route which I checked in a NEC simulation to be fine also. And we must not forget that using the transformation line made from 35 ohm coax, as shown in explicite down the page, leads to a need for an impedance of 25 ohms at the dipole. So that a bit further adaption was needed and well fitted the experimental way by OK1VPZ which he all explains further down on this website.

What should you look out for when adjusting the antennas?

Building real antennas with a low impedance - see DK7ZB's idea of making antennas with an impedance of 28 ohms - is more complicated in that each piece of free wire at the antenna terminals has some inductance and it behaves as a more significant shift in complex impedance at a lower real circuit impedance. Therefore, for wiring a 28-ohm Yagis dipole, it is necessary to have the shortest possible terminals - resp. just such leads as are needed to compensate for the imaginary part of the impedance. I think it's impossible to hit right wiring here "spot on" without testing.

With the antenna according to this article, it is necessary to pay attention to the length of the elements, their distance and symmetry (so that the element is the same length on both sides of the boom). If we follow these dimensions with submillimeter accuracy, there is no need to worry that the antenna will not work in the band. If the elements need to be shortened, pay attention to the perpendicularity of the cut. The possible use of plastic seals at the end of the element has a marginal effect. Much more critical is the case with radiators: the use of a 35-ohm Teflon cable RG141 (can be purchased in the RF Microwave e-shop ) has a length of 1/4 lambda with a braid length of just 12 cm ( between the ends of the braid). If the cable is shorter, the reflection attenuation will be low - for example, at 118mm length, the reflection attenuation will be 17 dB only. If it is longer, the minimum reflection attenuation will break into two minima, which will broaden in frequency the more excess length the cable shows. It is the same with the inductance of the terminals at the antenna dipole. This course of reflection attenuation indicates that the inductance of the terminals at the dipole is too small and it is appropriate to increase it - for example, by bending the wires away from the dipole and this course shows the opposite. Of course - these are extremes - however, I recommend first setting up multiple radiators in one antenna and finding the optimization for each radiator individually by the length of the terminal at the dipole (or by shaping them). It turned out to me that with the length of the terminals listed here, I get the optimal fit if a small, narrow faston is inserted on the solder lug on the side of the dipole where the shield is connected - see the picture here - and only then check the reflection attenuation at placing the radiator in "its" respective antenna mechanism.

DG7YBN: For the 70 cm band element cutting accuracy within 0.5 mm becomes a necessity. That of coarse pends on the design, but as a rule of thumb that is a figure. It is no overdoing to buy and make plenty use of a large caliper gauge when cutting elements for 432 MHz Yagis. Wires inside the dipole box should in general be led, as good we can make them, in parallel to the dipoles arms and only be bent towards the feed line for the least possible length. The following photos of Ok1VPZ's wiring in the SCAME box are a good example how to get there.


high terminal inductance average inductance of terminals low inductance of the dipole terminal

If you set only one antenna (or make a change for all antennas), then within small limits, the frequency responce of a Yagi can be increased by moving D2 forward by a few millimeters. Shifting D2 is critical! However, the frequency response of the antenna is especially related to the length of the dipole. Interestingly, the length and position of the reflector has little effect only and the position and length D1 has a strong effect on the impedance of the Yagi. However, if the inductance of the terminals of the low impedance dipole is too low, it can be somewhat compensated by a displacement in forward direction. This is also the reason why the test antenna with high terminal inductance gave me a position of 266 mm instead of the correct 257 mm. By the way - this may be related to this whole "adjustment", because the calculation is made for an impedance of 28 ohms and the transformation cable is 35 ohms - the output impedance using this coax should ideally be less than 44 ohms. In oreder to "tune" to 50 ohms it is necessary to introduce additional correction elements into the circuit - which are the parasitic inductances and capacitances. It follows that the older circuit solution mentioned here make a deep sense for antenna repeatability and consistency of antenna parameters (which is important for antenna assemblies) with low emitter output impedance - only the AD450 laminate is unfortunately no longer available ... I will look for another solution to ensure full repeatability as of such a microstrip matching circuit.

However - to make matters even more complex: part of the production of these antennas was the need to address their construction resistance to rain. I liked the WiMo solution. They fill the radiator box with polyurethane (PUR) foam used for construction works. It is full of bubbles after hardening and thus its dielectric constant will be low. It also reliably clogs all leaks and is water-repellent. So I was tempted to use a PUR foam to seal the matching circuit of a SCAME box also, which in this use has one unused "port", i.e. a sealant filler plug. So I filled the ready made antennas with foam. - Make sure the pressure in the foam spray can doesn't tear the seal in this filler plug, hi. With the foam solidified (2 days), I measured the antennas again. For antennas with a reflection attenuation of 25 to 29 dB, the adaptation has improved by 2 - 3 dB. On the contrary, for antennas with super adaptation (37 dB and better) there was a certain splitting of one minimum to two. However, there was no significant deterioration in reflection attenuation to be observed. From the logical point of view - even if the foam is "thin" inside the box we find a certain reduction in the coefficient of propagation speed and thus a relative "elongation" of the line respecitvely of the inductance at the dipole wiring. In practice, this should have no effect. Only I was nicely "tangled" after the foaming of the antenna boxes ...

Not the last task was to determine the sensitivity of the antennas to detuning by rain. I used a common water sprayer for moistening houseplants. I have not tested all the antennas, only 30 percent of them. The best frequency response of the antenna drops by about 2 MHz, but for all tested antennas a limit of 21 dB reflection attenuation remains. I would consider that satisfactory. See pictures.

Conclusion: in addition to the above, I experimented a lot with measuring the radiator alone without incorporation into the waveguiding structure of the antenna using a pocketVNA. If you are interested in the lot of results, it is possible to download them here. But I say in advance that I did not read anything reasonable from those measurements. I measured the dipoles like shown here, but if you would like to correlate the results of the table with the final results of the antennas, I would like to point out in advance that the measurement of dipoles took place before the optimization of the inductance of the dipole connection. This is not to say that there is no correlation, I just don't see it. I still have a lot of work to do to complete the antennas (dividers, cables, holders ...) and the day is not endless. In any case, it would be worth paying even more attention to microstrip matching to ensure 100 percent repeatability of the design.

I would also like to ask Hartmut for a comment, in particular to assess whether the described method of optimizing the adaptation could not lead to such a large phase inconsistency of individual antennas that it would prevent their successful merging into groups.



DG7YBN: ok, lets calculate an expected maximum phase difference between the Yagis and simulate the impact in an 8 Yagi stack. The matching coax is a quarterwave line of a length of 120 mm. That equals 90 deg. of phase. Lets assume a maximum cutting and soldering tolerance of +/- 4 mm. If 120 mm equals 90 deg., then 4 mm equals 3 deg. of phase.

Now lets edit the sources of the EZNEC file of the 8 Yagi stack and set some of the 8 sources to +3 or -3 deg. by random, simulate and compare to the original elevation pattern for with all sources are fed with 0 deg. of phase lag.



EZNEC Sources Window: Random edit to +/- 3 deg. of phase lag



Original pattern, free space, all 8 Yagis in phase


Comparsion: with and without phase lag of +/-3 deg.

My conclusion: Neither gain nor F/B are influenced. Pattern shape is touched marginal by the simulated maximum phase lag that could occure due to wiring the dipole boxes in slightly varying length of coax or split ends of coax.





Well, for a last word: if you are interested in the measured results of reflection attenuation of all antennas, then here is the table:

Reflection Attenuation (dB)

#   Reflection after debugging and after foaming
---------------------------------------------
1.  30,17 ..... 30 ........ 30 (two indistinct minima)
2.  27 dB...... 29 ........ 27
3.  33,4 ...... 33 ........ 30
4.  30,2 ...... 30 ........ 29 ....... 30
5.  25,3 ...... 25 ........ 26 ....... 23
6.  29 dB ..... 29 ........ 30
7.  25,2 .....  25 ........ 26 ....... 25
8.  30 dB ..... 32 ........ 38 ....... 21 !
9.  28 dB ..... 28 ........ 32 ....... 32
10. 28,6 ...... 31 ........ 39 ....... 39
11. 28 dB ..... 28 ........ 32
12. 31,7 .....  32 ........ 37
13. 30,4 dB ...  37 ........ 43
14. 30,7 .....  30 ........ 37
15. 30,6 .....  34 ........ 35
16. 44 dB ....  41 ........ 38 (two minima) (could D1 probably move closer to the dipole?)
17. 26 dB .....  26 ........ 27
18. 30,2 .....  30 ........ 36 (two indistinct lows)
19. 30 dB .....  30 ........ 35 (two indistinct lows)
20. 29,6 .....  31 ........ 37
0.  26,9 .....  21 ........ 29 : OK1TEH's antenna (WiMo box), D1 moved from position 257 mm to 251 mm

Measurement conditions: Wiltron Site Master S331. The VNA was calibrated to the end of approx. 3m of cable (to the antenna connector) at 432.2 MHz in the location were the measurements took place.

73! and hear you on 70cm during the July 2020 Contest. Perhaps (but it remains to be seen) already on the new DG7YBN antennas stacks