Archive for May, 2010

70% of published papers are a waste of time.

Thursday, May 27th, 2010

Most published antenna related articles are a waste of time when it comes to verifying the findings. When confronted with an unfamiliar antenna the first step is to find reliable published research and the problem is not finding papers. It’s like doing a Google search with a popular key phrase and trying to get the information you’re actually looking for. I remember a while ago one of our engineers was struggling with an antenna and in desperation he cried out and threw his hands in the air. Immediately the Marketing guy (that being me) said, “why don’t you just use Antenna Magus?”. Everyone in the office burst out laughing because he was busy working on a new antenna that still needed to be implemented.

Very few articles contain sufficient information in order to reproduce new findings. Surprisingly these are all papers that have been peer reviewed (sort of like an engineering stamp of approval) but most of them lack critical information like a physical parameter or material property or measurement distance. I’ve often wondered if it’s due to the authors negligence or is it done on purpose?

Here’s an example of an excellent reference article found by one of our engineers while implementing the Rectangular Capacitive-disc-fed Patch (which should be in version 2.1) G. Mayhew-Ridgers, J.W. Odendaal, and J. Joubert, “Efficient full-wave modeling of patch antenna arrays with new single-layer capacitive feed probes,” IEEE Transactions on Antennas and Propagation, Vol 53, No. 10, 2005, pp. 3219-3228.

It took him 10 minutes to figure out how to design and model the reference antenna design and reproduced the results in less than an hour.

Fortunately for our users Antenna Magus has all good published references listed in each antennas information document.

Looking for the needle in the haystack.

Looking for the needle in the haystack.

Author: Robert Kellerman

When measured and simulated results do not match?

Thursday, May 20th, 2010

Most engineers at some stage get that sinking feeling when their measured and simulated data do not match, especially after they’ve tried and tested everything to find the root of the problem. Some engineers just give up and write off the discrepancies due to cable/connector loss or some phase shift that is not accounted for in the simulation.

One of our engineers recently had a similar occurrence with the validation of the Yagi-Uda Monopole Array with Folded Feed antenna soon to be implemented in the Antenna Magus database. He manufactured the reference design in order to have a trusted measurement to use in the validation process. Unfortunately the impedance measurements and simulated results had some differences especially around the resonant frequency. After examining and comparing the parameters of the manufactured antenna with the simulation model using a vernier he found that the brazing rods weren’t exactly 3.2mm as specified and the thickness varied as well. Not expecting these differences to have a huge effect, the simulation model was modified to take these into account and surprisingly the new simulated results almost matched the measurements. Later one of the other engineers confessed that the rods fell off his motorcycle three times on his way back to the office.

So next time before blaming the computer or the network analyser have a closer look at the built model – you might be simulating something that doesn’t exist.

Simulation model and manufactured model of the antenna before any modifications.

Simulation model (a) and manufactured model (b) of the Monopole Yagi with folded feed array antenna before any modifications.

Comparison of S11 for measured vs original and new simulated results.

Comparison of S11 for the measured vs old and new simulation models.

Author: Robert Kellerman

Story: How we added the Array synthesis tool in Antenna Magus (final: part3)

Friday, May 7th, 2010

In the previous two blog entries (part 1 and part 2) I showed two different prototypes of the array synthesis tool and summarised what we learnt from various user tests.

Just before we started to plan the third prototype we all sat around the table once again. Something was wrong, but I couldn’t quite put my finger on it.

It turns out that it is devilishly difficult to design an array while taking array element pattern into account. More importantly, the more elements you have, the less important the individual pattern becomes. Work that had been done quite a while before I had even started working at Magus included a design algorithm that designs arrays by assuming isotropic element patterns.

Furthermore, this is what antenna engineers also feel most comfortable with and why there was that unwillingness to start off with an element pattern each time.

Our terminology also had to be re-thought. “Array Calculator” became “Array Synthesis Tool” for the first time.

The three components: the array layout and excitation, element pattern and synthesized array were split up into three separate tabs.

Array tool: Final prototype (click to enlarge)

Array tool: Final prototype (click to enlarge)

Once again, a prototype was drawn up and put in front of five users (click on the image above to enlarge the final prototype). This time, we had a winner. The only biggish mental hurdle was apparent in users who had tried the previous prototypes and who now wondered, briefly, where their element pattern was coming from.

After tweaking a few last things, the array synthesis tool could be specified and implemented. Click here for a more detailed description and screen shots of the final array synthesis tool.

Author: Carien Fouch

Story: How we added the Array synthesis tool in Antenna Magus (part2)

Monday, May 3rd, 2010

I want to pick up the story on how we implemented the Array synthesis tool in Antenna Magus. The previous part1 of this blog series can be found here. In part1 I asked readers to look at the first prototype and to write down or comment on their first reactions.

This is what we learnt from the original usability tests:

  • The design confused people!
  • Fitting antenna parameters into the current Design Mode palette is a problem. Fitting array parameters into the palette would be impossible.
  • People don’t expect to be able to export a model of an array to a simulator. Most of the time, they are simply too big for full-wave analysis.

In the second design phase, we decided to go modal. The Array tool was separated from Magus’s Design Mode. One of the reasons for this was the sheer amount of information associated with arrays that had to be visible to the user. Another reason that emerged, after much discussion, was this: “Magus would not be taking mutual coupling into account.” By keeping the array calculator separate, we would not create that expectation. Parameters would be displayed in a large table in the main workspace, instead of trying to squeeze them in on the side-palette.

However, users would still start off by specifying an elemental pattern or dragging a pre-designed pattern of an antenna into the Array tool. After this, the excitation distribution and positions of elements, or some other objectives could be specified.

Array tool: Prototype 2 (click to enlarge)

Array tool: Prototype 2 (click to enlarge)

The prototype was drawn up in Balsamiq and five more users were tested.

This time, the usability tests went much more smoothly, however:

  • Users (as we expected) had trouble finding the array calculator
  • They seemed unwilling to start off with an element pattern
  • One tester began explaining the intricacies of measuring array S-parameter matrices to me and the difficulties of matching impedances. When I told him that this would not be taken into account by Magus, he said: “But this is useless!” It’s just a starting point!? Which is when he realised the intended value of the array calculator.
  • The design still felt a bit flaky, somehow.

Before designing the third prototype, we all sat around the table once again. Something was wrong, but I couldn’t quite put my finger on it.

In the 3d (last) part of this blog series, I will explain what final changes were made to the above prototype before we implemented the Array tool. Have a look at the above prototype so long (click to enlarge) and write down (or comment) what you think needed to change.

Author: Carien Fouch