Prof Hidetsugu Yagi

I recently discovered a re-publication paying homage to a paper published by Mr Yagi in 1928 titled, “BEAM TRANSMISSION OF ULTRA SHORT WAVES”. [Reference: James. E. Brittain, “Yagi on a Microwave Communication System”, PROCEEDINGS OF THE IEEE. VOL. 72, NO. 5, M A Y 1984]. It was fascinating to see the discoveries they made almost a century ago long before the existence of transistors. Prof Yagi and his student Shintaro Uda (hence the name “Yagi-Uda antenna”) discovered that positioning an element having a slightly higher natural frequency next to a single driven element acts like a director shaping the radiation pattern to be more directive where a slightly lower resonant frequency element acts like a reflector. Up until then most experiments had to be done at > 100 cm wavelengths (lower than 300 MHz) simply because they couldn’t produce stable oscillations at higher frequencies. A quote from the same paper: “Mr. K. Okabe, assistant professor at the Tohoku Imperial University has succeeded in generating exceedingly short sustained waves by introducing certain modifications in the so-called magnetron.” Could this be the first microwave signal generator?

These guys didn’t have the luxury of ordering components and connectors from a catalogue, design and build something and plug it in to a signal generator, simply turning the knob to the preferred frequency. They probably had to start preparing for a measurement by making the cable!

As Isaac Newton once famously remarked in a letter to his rival Robert Hooke in 1676:

“If I have seen a little further it is by standing on the shoulders of Giants.”

How many things do we – as ‘modern’ antenna engineers – take for granted, when infact we are truly standing on the shoulders of Giants…

Author: Robert Kellerman

We recently investigated the “Skeletal wire biconical antenna” (often refered to as the wire biconical) as a useful EMC antenna for Antenna Magus. As part of the validation process we wanted to compare the simulation data with measured data but we faced a slight problem. In order to balance the feed of the physical antenna a balun is required which has to be placed in the center between the two cones, physically separating them. This cannot be accounted for in the simulation as it has to be fed using a single port. We wanted the physical model and the simulation model to be as similar as possible. Another alternative that we considered was to feed the antenna from the side through the center of one of the arms – but that introduced other uncertainties and non-symmetry issues.

Image of the Skeletal wire biconical antenna

Eventually we decided to build a wire monocone using the same dimensions as the biconical antenna, replacing one of the cones with a ground plane. Skeletal wire biconical antennas are usually designed to operate between 30 – 300 MHz and are used in EMC applications but we decided to frequency scale the design by 3 times in order to reduce it to a more practical size. Once we finished building the wire cone section, we relised that in spite of our efforts to reduce the antenna’s size we still needed a +- 3m diameter ground plate if we wanted to simulate an infinite ground. Someone came up with a clever plan to instead of using a solid metal ground to use six 10m wires which are placed in the symmetrical plane perpendicular to each element of the cone. This would have the same effect as using a solid ground plane. The following picture shows the final antenna being measured on the roof of the Stellenbosch engineering faculty.

A few risked their hands to help with constructing the antenna while I’m holding the blow torch : )

Rooftop measurement with the 10m wires running along the roof.

The conditions weren’t ideal as it was pouring with rain while we had to do the measurement. We were quite surprised how well the 6-wire ground plane worked! Perhaps the wet roof had enhancing ground plane effects?

The graph below shows measured and simulated s-parameters of the antenna. It was interesting to see the resonance at 850 MHz is not a computational discontinuity but is picked up by the measurement as well. Some investigation shows that the resonance can be moved out of band by adding a short between one of each cone’s wires and the center wire.

S-parameter comparison of the Skeletal monocone antenna.

Author: Robert Kellerman

A few posts ago I showed some pictures and results of the Cantenna (also known as the Pin-fed Circular Waveguide Antenna) made from a Pringles potato chip container compared with a copper replica. Click here to read the post. We also compared the gain patterns of the copper cantenna (mentioned in the previous post) with a Coaxial cavity horn using the same wave guide dimensions as the Cantenna. Well actually Wilco (a 2nd year engineering student who did some vacation work for us) came up with a clever idea. He used the same copper Cantenna for the waveguide section and made the cavity section to slide over the cantenna forming the Coaxial cavity as shown in the next few images.

Copper cantenna and slideble cavity section which forms the coaxial cavity horn.

Coaxial cavity horn

Measurements were done in an anechoic chamber

It was interesting to see the effect of the added cavity compared with the waveguide (Cantenna) by itself. Most of the back radiated power is directed toward the main beam which now has a much wider, almost flat shape which is ideal for being used as a reflector feed.

Comparison between the Cantenna and Coaxial cavity E-plane pattern measurements

Author: Robert Kellerman

As something of an antenna enthusiast, I was delighted to discover a source of reliable information about the antennas in many commercial devices – including the iPhone. The Hong Kong Applied Science and Technology Research Institute (ASTRI) (www.astri.org) has taken apart many commercial phones, done measurements on them, and built 3D CAD models of them. Accurate impedance and pattern measurements are also done in anechoic chambers and published on their website for free. It is great to see a reliable, scientific, study of these devices, rather than anecdotal evidence from an enthusiast in their garage with 50% speculation, 50% measurement resolution error and 50% fact.

It is worth blocking time in your schedule to go to their website and look through some of the documentation. Follow this link to view their antenna R&D gallery.

Author: Sam Clarke

You’re probably aware of the iPhone4 antenna saga that’s been in the news the last couple of days. I find all the responses from different bloggers, news reporters and Apple quite amusing.  Something in particular that caught my attention is that most smartphone designers position the antennas at the bottom of the phone. My immediate response is, “Why do they keep on putting the antenna at the bottom?  Why don’t they simply move it to the top of the phone where users’ hands would least likely interfere with the signal?”

I think Spencer Webb hit the nail on the head with his explanation in his blog entitled. Apple “iPhone 4 Antennas…”. He explains why in spite of the fact that it’s the most impractical position, 99% of modern cell phones are manufactured with the antenna placed at the bottom, back of the phone. The FCC (Federal Communications Commission) puts strict limits on the amount of energy from a handheld device that may be absorbed by the body. The absorbed power is determined by measuring something called SAR (Specific Absorption Rate). All new phone models have to pass SAR tests which, surprisingly, are done in the vicinity if a human head model (or phantom) but without a model of a hand holding the device! Cell phone manufacturers are required to print the maximum SAR in the phones user manual and the test results have to be less than 1.6 or 2 watts/kg (in the US and Europe respectively). So in order push down measured SAR, the antennas are moved as far away from the head as possible – the bottom, back of the phone.

It seems to me like the issue that needs to be addressed isn’t the fact that Apple slipped up or how to improve mobile phone reception when the antenna is shielded off by a human hand.  Obviously SAR needs to be regulated but it seems like there’s a lot of hype generated by uninformed individuals who claim that RF radiation from mobile phones is a massive health risk. It seems that this is more of a marketing issue. The SAR value that is printed in a device’s user manual needs to be minimized to keep the public happy in spite of the fact that SAR in the hand is ignored during measurements. In practical situations more power needs to be radiated to compensate for signal loss due to the vicinity of the hand – increasing the actual SAR and reducing the phone’s battery life.

I wonder what would happen if the FCC includes a mandatory hand model in SAR tests – how many smartphones would pass these tests…? Or if they decide to double the allowable maximum SAR limit what would the response be…?

Another solution for the iPhone 4 would have been to make the phone work when flipping it upside down (so the screen automatically rotates 180 °) while on loud speaker or browsing the internet. This would in effect reposition the antenna in its most logical position – at the top of the phone. Maybe they could then call it the “!phone” : )

"Hold it like this."                                                    by cartoonist, Konrad Brand

Author: Robert Kellerman

Pringles Cantenna in anechoic chamber

We also built a replica of the Pringle version out of copper as show in the image below. The graphs below compare measured S11 and gain patterns between these two antennas.

Pringles and copper cantennas

S11 simulation vs measurements

Pringles vs copper cantenna pattern comparisons

Author: Robert Kellerman

Wire zig-zag antenna

It is not always apparent when modeling wire structures whether a thin-wire model approximation will still give reliable answers. We recently included the Wire zig zag antenna in Antenna Magus and had some trouble validating designs of this antenna with many cells. There are various methods to model and physically construct a wire zigzag and we found that the corners of a >10 cell zigzag have a noticeable effect on the performance.

Thin-wire model of a zig-zag corner

Detailed triangle mesh model of the zig-zag corner

The thin-wire model is obviously the quickest and easiest way to model the antenna but the simulation results differed from the more detailed triangular mesh model. To see which of the two simulation models could be trusted we built and measured a 12 cell zigzag (shown in the first image above) from copper brazing rods and filed each wire at the correct angle to ensure neat flush corners as shown in the zoomed photo below.

Zoomed view of the zig zag corners

The following image shows a comparison between the measured data and two simulation models which proved that a high gain wire zigzag has to be modeled with more detail.

|S11| Comparison between measured and simulated results

Author: Robert Kellerman

Cantenna made from a Pringles potato chip can.

A couple of months ago I blogged about how hard it is to find reliable measured data for a Cantenna (also known as the Pin-fed circular waveguide antenna) which we wanted to use to verify our Antenna Magus design. You can read the blog here. We decided to make and measure one ourselves.

After designing a Cantenna at 2.45 GHz, Wilco (a 2nd year engineering student from the university, doing vacation work for us) went to the local grocery store to see if he could find a can with the right dimensions. All the cans with the right diameter at 2.45 GHz were too short so he bought a can of large Pringles chips. After we ate the chips he adjusted the design to fit the inner diameter of the Pringles can and cut it to the right length. The new design worked at 2.85 GHz. The inside of the can is lined with a thin aluminium sheet with a thin wax coating which saved a lot of effort – not having to stick conducting material on the inside ourselves.

He soldered the sma connector onto a small copper plate (shown in the next image) and attached it to the waveguide with two small screws after sanding off the wax coating on the inside to assure good contact with the aluminum.

Sma connector soldered onto a small copper sheet

The following image shows the inside of the antenna with the connector and pin. We measured s11 and compared it with the simulation (see graph below) which had pretty good correlation. Most industrial cantennas have a thin dielectric coating on the outside of the waveguide to prevent unwanted surface currents. We didn’t measure the dielectric properties of the Pringles can but it seems like a very cheap, practical antenna with 8 dBi gain and 20% bandwidth at 2.85 GHz.

Inside of the Pringles cantenna showing the connector and pin

|S11| measured vs simulated results

Cantenna sketch with physical dimentions.

For those who might want to try to make this antenna, I added the above sketch showing the required physical parameters that will work using a Pringles can operating at 2.85 GHz center frequency.

Author: Robert Kellerman