Similar Dipole arrays are giving differnt patterns.

[phased]

Dear Engineers,

I have recently been pattern testing six separate dipole arrays in a remote part of Queensland. All these arrays are of the same design. The arrays are for use over 7 to 17GHz.

My problem is that strangely, different arrays have given different patterns. That is, some have dynamic ranges of 20dB and some have dynamic ranges of 30dB. (despite all being from the same design).

Each array consists of a metal tray of dimensions about 9cm by 22cm and 3mm thick. On this metal tray are mounted 12 dipoles. The dipoles are mounted in three rows of four along the tray. The dipoles are spaced “sort of” randomly along these rows. The dipoles themselves consist of 2 planar patch’s which are roughly 5mm squares and are wafer thin. The patch’s are adjacent to each other (about 1mm apart) –One patch ends up being connected to the outer of the feed coax and the other is connected to the inner of the feed coax. The outer of the feed coax is also connected to the tray itself. The “plane” of the dipole patch's are horizontal to the tray surface and all dipoles are orientated similarly.
-This metal tray is screwed in to the middle of a square metal plate (of side 55cm and 3mm thick). This metal plate has a "window" cut out of it which is about the same size as the dipole tray –and the dipoles are thus not behind the metal of the plate because they can “see” out of it. The plane of the tray and the metal plate are parallel -though due to a small "skirting" round the edge of the tray, the plane of the tray is about 1cm behind the metal plate.

Each dipole feeds into a (12 way) combiner so that the received power from the array can be fed out along coax to the receive amplifier. (During each pattern test the same combiner was used)

The problem (I repeat) is that even though all dipole arrays are of the same design they have given different patterns. That is, some arrays give patterns which have 30dBs of dynamic range and others have just 20dBs of dynamic range.

The test set-up has been on grassy ground in a very remote part of Queensland. (no interferers). The Dipole array was used as a receiver and rotated through 90 degrees either side of boresite. The transmit antenna was a rectangular ridged horn of aperture about 9cm by 15cm. Patterns were taken at 7GHz, 9GHz, 12GHz, 15GHz and 17GHz. Each antenna (transmit & receive) was 2 metres above the ground. The distance between the Rx and Tx antennas was never exactly the same for every array but was about 8 metres in all cases.

I am wondering if 8 metres is enough to get the far field distance for these patterns? [8 metres was based on the largest aperture of the “tray” (22cm)] –However, I am wondering if I should have taken the “largest aperture” as being the “55cm” length of the side of the metal plate(?)-This gives a far field distance of some 34 metres.
[Far field distance formula used was 2*D^2/lambda …where D = Largest aperture of antenna)]

I am just wondering whether I have been taking patterns in the near field and that the difference in dynamic ranges has been due to having slightly different distance between the (Tx & Rx) antennas when in a “near field” situation(?). (i.e. slightly different distances when each dipole array was pattern tested)

Alternatively, I am wondering if the dynamic range difference between the different array’s patterns could be due to different sized solder blobs on the surface of the dipole patch’s (-the dipole patch’s were soldered to the feeds and quite significant solder blobs are visible on these patch’s and the patch’s are themselves only small) –Obviously each array has a different “set” of solder blobs and maybe this accounts for the differences in dynamic ranges observed(?)

I really must apologise for the longevity of this question! Any thoughts would be most gratefully appreciated.

 

[VEBill]

Sensitive dependance. It means that small differences are magnified by the design. It could be anything.

Perhaps variations in the coax feeds to the individual elements. Sometimes you see the term 'phase matched cables'. Are your cables all phase matched? Probably not.

But it could be anything.

 

[phased]

Thanks for your reply,

The cables are phase matched. We used the VNA to compare the phases at the end of the cables and kept minutely adjusting cable lengths until we got a good phase match.
The pattern problem is worse at the highest frequency (17GHz).

I suspect the solder blobs must be having some effect. One antenna's patterns and gain has been deemed to meet the spec but others are deemed to be way off the mark.

Also, the far field distance is a mystery with this array antenna. If the far field is actually 34 metres then we will have problems getting adequate received power during pattern testing at this distance. We dont have an amplifier with enough gain and most of the pattern will be "drowned" in the noise floor of the spectrum analyser.

 

[VEBill]

You may want to define 'dynamic range' because I think you're using the term in an unusual manner.

Can you take all the individual elements with the visually worst (largest) solder blobs and re-assemble them into one known-bad array, and take all the elements with the visually nicest (smallest, tidiest) solder blobs and make another known-good array. Then compare the two arrays. If you can show that the bad one is really bad, and the good one is really good, then you might be able to prove your point to management.

How about using silver solder for your next build?

Disclaimer - stand by for other, perhaps better-informed, ideas.

 

[phased]

These are good ideas -much appreciated.
Ref the term "dynamic range"...apologies if i have mis-used it...i am taking it as meaning the decibel difference between the peak of the pattern and the lowest point of the pattern.

 

[VEBill]

The term 'dynamic range' is normally used to apply to a receiver's performance: the input signal amplitude range, min to max signal input, where max is limited by distortion (not damage).

Two antennas might have identical relative pattern plots (or 'dynamic ranges' as per your term) but one might be a very lossy copy of the other. Thus, the absolute gain value is normally a critical measure.

I suppose there might be some obscure applications where THE critical antenna parameter is the relative pattern, but most applications would need to confirm the absolute gain (if for no other reason than to make sure the antenna isn't too lossy).

Your solder blobs might very well affect absolute loss. But such loss might bring down the entire pattern and not be visible in just the relative measure.

Beyond the simple experiment I already suggested, the usual engineering approach would be to check the original design for 'sensitivity' (to variations). In other words, re-run the simulation many times over while varying the design to account for real-world variations. Ideally, such analysis is done as part of the initial design cycle for any mass production. In electronics, you may have heard about 'Monte Carlo analysis'. Same sort thing applies to some antenna designs where dimensions are critical.

 

[Higgler]

I'm fairly certain it's your range. At 2m high and 8 meter's long range that's a downlook angle of atan 2/4 or about 25-30 degrees. Your antenna pattern if the antennas are oriented with the 9 cm in the vertical makes a 50 degree beamwidth max. and 15 degree beamwidth min. That's way too much energy hitting the ground for any kind of repeatable antenna pattern. Greater variation at lower frequencies might be added proof that it's your range.

Why are you doing outdoor tests? no anechoic chamber?

If you need real antenna patterns you probably need to have a good antenna chamber (you only really need D^2/Lambda for most testing, alot of chambers have 15 foot range). If you just need to prove the antennas are all the same, you can do a simple probing in the near field, just make sure all mechanical parameters of each antenna test are identical, i.e. spacings and angles. I can talk you thru that if you only need to prove they are similar.

kch