|Product Performed to Expectations:||10|
|Specifications were sufficient to design with:||8|
|Demo Software was of good quality:||10|
|Product was easy to use:||10|
|Support materials were available:||10|
|The price to performance ratio was good:||10|
|TotalScore:||58 / 60|
Molex Antenna Review
This review examines Molex antennas intended for 2.4GHz and 5GHz radio communications, and in particular Bluetooth and WiFi (also known as 802.11 Wireless LAN). It’s becoming easier and easier working with wireless technologies, particularly when manufacturers provide a lot of useful information to get antennas connected to custom hardware. This review examines what information is available for the Molex antennas, and how to use the antennas and how to measure some aspects of the performance. It is very difficult to conduct accurate measurements without special test equipment and test chamber, but nevertheless it is still feasible at low cost to measure some very useful information.
The three particular antennas reviewed are shown below.
|MID||0479480001||2.4 GHz||On-Ground MID chip antenna||4x3x3mm|
|Ceramic||2.4 GHz||Ceramic chip Antenna||4x3x3mm|
|Patch||2069950150||2.4 GHz, 5 GHz||Adhesive PCB/Patch Antenna||20.5x20.5x4.8mm|
The first two antennas are quite small! But they are of interesting construction, so I was keen to see how they performed.
The third one provides for flexibility when mounting in unusual enclosures. A typical scenario for the latter could be a metal chassis, but with plastic trim, and the antenna could be hidden behind the plastic.
But first, I wanted to discuss why antennas are interesting!
For a long time RF was considered black magic, but modern technology and low-cost equipment provides a means for any engineer to get going with wireless effectively. We don’t always need to be reliant on off-the-shelf RF modules with integrated antennas – we can choose how we want the RF performance to be, and design accordingly.
For a commercial project, if better performance is required than what an off-the-shelf all-in-one RF module can provide, then there is a strong desire to use a discrete antenna such as from this Molex selection, or even custom designed antennas (requires a hefty budget!), with custom circuitry, designed and tested with the correct tools (some of the tools and procedures are indicated in this review). Otherwise, there is almost no chance that your commercial product is better than the competition when it comes to radio performance.
Working with radio technology was quite hard a couple of decades ago! Radio architectures were not very integrated, and required a fair amount of inductors and capacitors, and sometimes even different semiconductor materials, and precisely adjusted parts such as crystal filters. Since then, the state-of-the-art has moved on, and better digital processing and high speed analog-to-digital conversion (ADC) have led to previously-unusable techniques becoming feasible. These methods are very integrated, and don’t require many additional components.
As an example, here is the insides of an early 2000’s wireless LAN (802.11 WLAN) WiFI card. This photo shown just half of the circuitry (screening cans removed, and some rework done – some parts are missing); the other side is packed too, but I couldn’t easily desolder the screening cans off that side. The entire radio side is clearly demarcated, and there are a lot of components.
It was extremely hard for electronics equipment to incorporate WLAN capability at that point in time. Plug-on integrated modules were frequently used. I developed a computer board that had a PCMCIA card socket on it, so that an off-the-shelf WLAN card could be inserted if wireless capability was needed.
Compared to a modern WLAN part, the Texas Instruments CC3200, the difference is incredible. Now there are only four discrete components that are wireless related, if we exclude the antenna and the zero-ohm resistors.
There is still some experience and knowledge required for engineers to do that, however the manufacturers have provided lots of documentation and example board files to minimise the risk of total failure. I’d not worked with any Bluetooth chips until earlier this year, and it was pretty easy to get some communication going, because not many parts are needed, so there is less to go wrong.
However, getting good performance is another thing. It requires antenna considerations. It is a key thing that can differentiate a mediocre product to one that is really great. No-one likes Bluetooth headphones that drop out, or a laptop that cannot connect to the home network when it is not close to the router.
For the Bluetooth project, I used a PCB antenna. However, the performance was not as good as it could have been, because I wrongly dimensioned it in the prototype by about a millimeter.
Recently, I worked on a WLAN project using a part called CC3220MOD from TI. This is a tiny module containing a CC3220 chip, plus some memory and so on. I decided to experiment a little bit with this WLAN project, to explore the Molex antennas.
One concern was how much testing could be possible? For basic testing, anyone with a small lab and $100 can get going with this. My RoadTest colleagues have also been showing the same thing; none of us have expensive test gear for radio, but we aim to review what we can with the resources at hand.
An antenna provides the interface between radio waves traveling in space such as air (or a vacuum), to signals on a wire, and vice versa. Magnetic and electric fields occur close to a transmitting antenna, and the amount and shape is dependant on the antenna design. Further away from the antenna, the energy gets transmitted using waves, which have the magnetic and electric fields perpendicular to each other and to the direction of the wave. These waves function like light, and can be focussed, parallel or divergent, and can reflect off things and be absorbed. The antenna is designed to send the waves off in the desired directions. Sometimes you may want the available energy to be sent in many directions, or you may wish to concentrate it in a particular direction; it all depends on the antenna design, including shape and dimensions.
One thing that immediately stood out was the identical dimensional details for the ceramic and MID antennas. Not only are they the same physical size, but the documentation and PCB shows that the same space of 6.87 x 5.71mm is needed on the PCB for these antenna, not including the matching component footprint area. Clearance to large metal components such as a battery or screening can is identical too, at an extremely impressive 5mm from the edge of the antenna itself, i.e. less than 2.5mm off this 6.87x5.71mm PCB area!
I’m guessing the ceramic and MID material plays a significant role in the antennas too; the MID antenna is hollowed out, to reduce the dielectric material; that could be more expensive to achieve with ceramic material perhaps. The conductor is wider on the ceramic antenna compared to the MID antenna.
Despite these clear physical differences, it is impressive that Molex managed to standardise both of them to the exact same dimensions! It provides engineers with a choice between the two, with no PCB layout modifications.
The patch antenna appears to be made from a lamination of two PCBs, although I’m not sure. It has a brand adhesive on the back (Tesa 68537). There is 150mm of cable with a decent (this is Molex after all!) gold-plated U.FL connector on the end. The connector is compatible with IPEX MHF and MCRF connectors.
It is dual-band, but I only tested at 2.4GHz. It was nice that the antenna already has spacers attached on the cable, to keep it a controlled distance from the metal chassis. It saves costs – less, or zero plastic clips are needed in the design to hold things in place.
When using antennas, there are quite a few things we care about. One property that is useful for many purposes, is the antenna impedance. An antenna doesn’t look like a pure resistance; it has inductance or capacitance too. The resistance portion at the desired frequencies of interest is important, because it is hoped that the majority of that resistance is radiation resistance, i.e. any energy expended there gets converted into waves. In practice some of the resistance component will be due to the antenna wire (and the wires leading up to the antenna) and leakage too.
It is well-known that for DC, maximum power is transferred when the destination has the same resistance as the source. If the source happens to have 10 ohms internal resistance, then maximum power transfer will occur when the destination load is 10 ohms (this also implies that then the efficiency is 50%; at maximum power transfer, 50% of the energy gets converted to heat at the source too). The same thing almost applies for AC, except that if there is any reactance (inductance or capacitance) at the source or destination, then the other end needs the opposite reactance. So, if the source has some capacitance, then the destination needs some inductance. In math this is known as the complex conjugate.
Although the discussion here could be thought of as applying to transmitting antennas, the same principle applies to receiving antennas. In that case, the receiving antenna acts as the source, and with a perfect impedance match to the electronics, 50% of the received energy will be lost as heat at the receiving antenna, or re-radiated back.
A large volume of radio electronics circuits have standardised on an impedance of 50+j0 ohms, meaning a resistance of 50 ohms and zero reactance (no capacitance or inductance). Test instruments will expect this impedance. Formulas for things like transmission lines (which are used to transmit power with low loss from the electronics to/from the antenna) will be based on this impedance too. Common co-axial wires will almost always be designed to look like a 50 ohm transmission line too (one major exception is analog video which uses 75 ohm impedance).
Antenna structures will hardly ever have a pure 50 ohm resistance however. In fact, it could be very different from 50+j0 ohms. So, a conversion circuit (also known as a matching circuit) needs to be designed to make it appear as 50+j0 usually. Sometimes the conversion is needed to something slightly different to 50+j0, because the electronics may not precisely be at this value. As yet another example, you may match a receiving antenna to 50 ohms coaxial cable, and then match the other end of the cable to something else if the electronic circuit doesn’t have 50+j0 ohm impedance precisely.
The matching circuit usually just consists of a single inductor and/or a single capacitor. The steps to calculate the components is described in a vector network analyzer (VNA) review here.
For matching to 50 ohms, antenna manufacturers may sometimes specify the precise inductance and capacitance needed anyway, in their datasheet. But for the best performance, you’d want to match precisely to the attached electronics, and then you’ll need a VNA.
When the antenna matching (described in the previous section) is not perfect, then not all the power that should be transmitted will get transmitted. Some will get reflected back. It means that transmission distances will be poorer, and also the transmitter will run hotter. In high-power situations it could damage the electronics.
VSWR and Return Loss provide a measure of the ratio between the incident and the reflected signals. Either VSWR or Return Loss is fine to measure. The former provides the ratio of voltages, and the latter happens to use a ratio of the power. VSWR is popular with Ham (amateur) radio.
VSWR and Return Loss are a convenient, scalar way of representing how good or bad the antenna match is. One could use percentage, but the difference between the incident and reflective values ideally ought to be very large, and so the percentages would be bunched at one end. Return loss uses a power decibel ratio, making the numbers nicer to work with than VSWR – just personal taste.
There are several ways to measure VSWR/Return loss, and generally the technique uses a device called a directional coupler. VSWR meters internally contain this, sometimes with a forward/reverse switch to take two measurements that can then be converted to a ratio. The transmitter hardware sends a test signal briefly to the antenna, during which time the forward and reverse measurements are made.
Directional couplers can be very low cost surface-mount devices, however for a general-purpose one with wide coverage and RF connectors fitted, the cost rises a lot more though. For hobby use only, they are often available from ebay (some are highly likely to be fake or faulty or repaired with poorer performance). A new coupler from (say) Mini-Circuits is a nice investment if you’re planning to explore antennas on a limited budget - $100 should be adequate to allocate for the coupler. The photo here shows some example directional couplers; the larger one is for a lower frequency range.
Some spectrum analyzers have the directional coupler built-in; particularly the analyzers which are intended for site antenna work. For design engineers (designing the electronics or the antennas) then a vector network analyzer (VNA) as discussed earlier can be essential.
A signal source is needed to use the coupler. The procedure measures the ratio between incident and reflected signals so the signal needs to be created in the first place. Some spectrum analyzers can do that (look for ‘tracking generator’ capability), or an external signal generator could be used. The lowest-cost way is to use a wireless development board from a semiconductor manufacturer, and program it to create the signals, perhaps in a test mode.
The methods mentioned above all rely on test equipment such as a spectrum analyzer, or a vector network analyzer. There are lower-cost ways too, although today they are more suitable for the hobbyist. For any business use (even for a single project) it would make sense to obtain the correct tools. Using an SDR is feasible, but a lot of time could be spent setting it all up.
A typical low-cost SDR as described here won’t be able to replace a spectrum analyzer, since it lacks the accuracy, dynamic range, far lower noise floor, hardware attenuators and protection.
There are several options that could be suitable for a hobbyist. The new contender is the Analog Devices ADALM-PLUTO (also known as PlutoSDR) which is under $150. I have not not used it, so I cannot say if it is any good though. Another contender is the LimeSDR Mini board available to purchase from CrowdSupply. It is about the same price but needs a case (it could be 3D printed perhaps). I’ve used its bigger brother, the $300 LimeSDR (i.e. non-mini variant) since that was all that was available at the time I purchased it. They are nearly identical, the main differences are the sampling rate, and number of transmit/receive channels (unlike a spectrum analyzer, many SDR boards have more than one channel). I think the LimeSDR Mini is extremely good value, it is half the cost of the LimeSDR.
Amongst the many software applications available for SDRs, there is free software called QSpectrumAnalyzer, and I used it with LimeSDR. It is usable, but has some bugs and is lacking capabilities currently. Still, it is open source and could be improved. Overall I liked it, it is very worthwhile for an SDR user, despite its limited usefulness compared to a real spectrum analyzer.
The mentioned SDR devices (Lime SDR and ADALM-PLUTO) can also transmit signals at a decent power level (up to 10dBm). With the LimeSDR, it is possible to use free software to script up what you need to generate. It’s straightforward to generate single continuous waveforms, or to perform some modulation such as frequency shift keying (FSK).
If the home-user equipment budget was to be less than several hundred USD/GBP, then the SDR would be the first choice possibly. It beats all other options I think. It is far better than many boat-anchor used spectrum analyzers from ebay. It still pains me that I once spent $500 on an ancient 1.5GHz spectrum analyzer, when the SDR is more feature-packed and has better performance, and bigger spectrum coverage too.
If the equipment budget is sub-$100, then a really useful option is to use wireless development boards. Wireless microcontroller dev-boards are very low-cost, and some could have built-in test mode capability. Texas Instruments has a PC based application called Radio Tool that works via USB or serial attachment to some of their wireless microcontroller (MCU) development boards. I decided to try Radio Tool, not with a dev-board, but with my own custom circuit board that contains the CC3220MOD module. The CC3220MOD contains an ARM Cortex-M4 microcontroller, Flash and RAM, and WLAN capability. I had some problems running Radio Tool (it appears to not have been tested with the latest libraries) but after an awfully long time (several days of troubleshooting code : ( sufficient functionality worked for this review. If anyone wants to build such a custom board, I’m happy to provide the Gerber PCB files, parts detail and the modified code.
The Radio Tool software allows the user to transmit bursts of packets, or unmodulated RF, and also receive data. I used Radio Tool to select a WLAN channel, select the modulation and bandwidth, and then transmit packet bursts on that channel.
The modified code can also generate a continuous wave at a single frequency (2.428GHz) by default, at power-on. That makes it handy for range testing in locations where I may not be able to connect with my laptop. In future I may extend the code to provide a frequency sweep across the WLAN spectrum, and have adjustable power, to turn it into a more usable test tool. Currently it transmits the unmodulated signal at a high power of +17dBm (50mW) (see spectrum analyzer screenshot below) which isn’t good to use without an attenuator attached. It can run on two AA batteries but consumes 370mA transmitting this signal, so they won’t last for long.
Those extra spikes in the display may look visually high, but spectrum analyzers have a logarithmic vertical axis, and so the spikes are only about 0.001% of the power of the main signal.
In summary the wireless microcontroller option is extremely useful. The signal output is of a high quality, and it can be used to make outdoor and portable tests from battery power.
Putting some of the methods mentioned above together, I decided to try to measure the return loss of one of the Molex antennas using low-cost tools, to see if it was possible.
For this test, I used the following general setup:
For reduced cost, the signal source was the TI dev-board in transmit mode, with a 10dB attenuator attached to it to prevent transmitting at too high a power.
The receiving spectrum analyzer was actually a Lime SDR board with free software called QSpectrumAnalyzer.
The directional coupler was model Pasternack PE2202-10 from ebay, and cost about $50. The directional coupler is a device that has a central line (between the IN and OUT ports) that has low loss bidirectionally. It has a third port that receives a defined fraction of power that arrives from the IN port. Therefore, if connected as shown in the diagram above, if there is any power reflected back from the antenna, some of it will pass on to the spectrum analyser.
For the return loss measurement, it would be necessary to compare the forward power with the reflected power. So, before I set up things as shown in the diagram, I first removed the antenna under test, and set this up:
The aim of this was to just measure the amount of power that would make it to the antenna, by placing the receiver (the Lime SDR running the spectrum analyzer software) at the position where the antenna would eventually be placed. Directional couplers only work when the coupled port has a 50 ohm load, so that was terminated as shown in the diagram above. Note that these diagrams might look unusual compared to typical non-RF schematics since no ground wiring is visible. In the RF world, such diagrams frequently omit to show that for clarity, because the connections between modules are transmission lines that could be (say) coax or specific width traces on a circuit board. The lines in the diagram represent the transmission line with both conductors.
The SDR software showed the following output for this main-line measurement, using the Molex MID chip antenna:
For this test, the Radio Tool software was configured to transmit a continuous burst of packets on WLAN channel 2 (2.417GHz) at 1Mbps. It can be seen that the peak output level is at -85.4dB. This actual value is meaningless, because the QSpectrumAnalyzer software only approximately functions like a spectrum analyzer; it currently has no ability to indicate the actual power output (in dBm). Still, since return loss can be calculated from relative measurements, there is no need to know the precise output level. The test can be done using relative dB values provided no other settings are changed between the main-line measurement, and the reflected power measurement.
For comparison, the display with a real spectrum analyzer () is shown below; it can be seen that the display quality from the SDR was therefore not bad (i.e. it was reasonable for this scenario, although you can still see artefacts in the SDR software that are not present with the real spectrum analyzer). An SDR cannot easily replace a real spectrum analyzer for reasons mentioned earlier).
I moved the spectrum analyzer Lime SDR module to the coupled port, and attached an antenna to the port marked IN. Now the result looked like this:
The output at the coupled port peaks at -104.6dB. This is a difference of 19.2dB. The coupled port for the particular directional coupler I used is specified to contain 10dB less power than the mainline port, for power that arrives from the IN port. So, in reality the amount of power reflected back from the antenna was actually 9.2dB lower than the power the antenna received, and the Lime SDR saw a value 10dB lower than that, i.e. a total of 19.2dB lower.
So, at the frequency of 2.417GHz, we can say that the antenna had a return loss value of 9.2dB. A table can be consulted to see that this means the VSWR was about 2.06.
According to the datasheet, the return loss should be 10dB or more. My measured value of 9.2dB is very close, and well within the tolerance of the directional coupler specification, which states that the coupled port offers 10dB +- 1.25dB coupling (for a more accurate result, the precise coupling for that directional coupler would need to be determined - one way of doing that would be to remove the antenna, and connect a short circuit there, i.e. 0 + j0 ohms, so that all the power is reflected back, and then make a measurement, and compare with the main-line measurement).
In summary, it is possible to measure the antenna return loss with low-cost equipment. My approximate figure was within tolerance of the equipment I used, and it showed that the Molex antenna was correctly functioning by only reflecting back about 12% of power (see here for a PDF Return Loss, VSWR, Percentage Transmitted/Reflected Conversion Chart). It is hoped that the 88% remainder power was radiated, although a small fraction would be lost as heat.
One final point; when doing return loss measurements on an antenna, it is very feasible for any external radio waves to be picked up by the antenna, and passed into the coupled port too! Usually it’s not too bad, because the spectrum analyzer will show it up as an anomaly perhaps at a different part of the spectrum. For example in the screenshot below, in the spectrogram view, you can see some signal to the right of the frequencies of interest. That was from a WLAN access point that happened to be very close to the equipment while I was performing the test! I had to eventually unplug it to make it disappear. Still, it showed that the antenna was successfully receiving transmissions : )
This is the kind of thing that is very difficult to identify with older spectrum analyzers that do not have spectrogram capability. The spectrogram display is really handy.
The better way to measure return loss is with a real spectrum analyzer or a vector network analyzer (VNA). Although a spectrum analyzer is sufficient, the better tool would be the VNA for more flexibility. The reason is, the VNA can be used for more than return loss measurements – it can help determine the component values needed to match the antenna. So, any design engineer will be working semi-blindfolded with just a spectrum analyzer, if the task is to improve on the match between the electronics and the antenna.
If you’re using a spectrum analyzer, then the test topology is as described earlier (a few spectrum analyzers have an integrated directional coupler, so you may not need that). The signal source can be a tracking generator if the capability exists in the spectrum analyzer.
I used the FPC1500 which is a combined spectrum analyzer and VNA product. I used it in VNA mode. For best results, the automated calibration tool is required (I don’t have that, I used manual calibration for the test, which takes a lot of care and is error-prone). The amount of time saved with the automatic calibration tool would make it pay for itself after just a few antenna tests.
The manual calibration procedure involves connecting reference impedances to the VNA test port connector, and running a software routine. Once that is done, the antenna is connected to the test port. The VNA will automatically inject a signal into the antenna, and measure what gets reflected back, and plot the return loss chart.
Here is what the datasheet showed for return loss, for the ceramic and MID antennas:
I first tried the ceramic antenna. As can be seen from the screenshot, the return loss chart is almost identical to the datasheet (the lower the dip, the better).
The MID antenna according to the datasheet has better performance, with a far lower dip return loss (the return loss is technically a positive number, but many instruments/diagrams use something known as S11 plots, in which case it is negative). The results are shown below. Although it may look different (I only tested 2.4-2.5GHz although the datasheet shows the results for 2.3-2.6GHz) and the return loss didn’t quite reach the lowest dip value, it is actually almost spot-on. That difference in return loss at the dip corresponds to a difference of just 1%. This is well within experimental error, because between calibration and measurement, I had to introduce an extra adaptor to suit the relevant connectors I was using.
In order to crudely simulate a real hand-held device user, I placed my hand close to it (about 3cm away). The difference was not much! The return loss worsened by about 1-2dB which was impressive!
I also deliberately tried to de-tune the antenna by putting my hand extremely close (1cm) from the antenna. This caused a large dip in the return loss, but it’s misleading because this is due to my hand causing loss, with less radiation resistance as part of the match. Ordinarily a user would be unable to put the hand so close if it were in an enclosure.
In summary both antennas perform well when it comes to return loss, with the MID antenna being exceptional. At the lowest dip, the MID antenna is reflecting back only 0.5% of power! The ceramic antenna would be reflecting back about 4%.
Next, I explored the patch antenna. As mentioned earlier, it is ideal for mounting on a metal chassis (and typically protected by a plastic cover). The datasheet demonstrates the return loss with it mounted on a 100x80mm metal surface, at several different positions (the patch antenna stuck to the center, or close to one of the edges; whichever side of the antenna is close to the metal surface edge will impact the return loss in a different way).
I used a 100x100mm aluminium plate. Also, my plate had a couple of protruding screws on a couple of edges, but I felt that could be realistic. This was the measured result at the center position (location 1); the lowest dip is shifted slightly in frequency, but this likely, due to the slightly larger metal plate I used. I did not use the adhesive on the patch, so this too could have caused some differences.
Positioning the antenna at location 4 (one of the worst locations according to the datasheet) matched the datasheet to within 1dB:
Finally, I tried location 2:
The results from the patch antenna provided me with valuable information about where an antenna should be located on a metal chassis, and I was happy to see results that were in the same ballpark as the datasheet, considering I used a different sized metal surface and it had some protrusions.
Observing the radiation pattern takes some effort! Automated equipment and test chambers and multi-thousand-$ antennas are used to capture data. But the average user may not have access to any of this. I have never seen a test chamber or such an antenna : (
However, on a budget, it could be possible to purchase a low-cost log-periodic antenna (see here: Antenna Measurement with the R&S FPC1500 Vector Network Analyzer ) which has a couple of very useful properties; it has good directionality, and it also works over a very large spectrum!
By connecting up the antenna to the spectrum analyzer or SDR (or perhaps to a development board if it provides a received signal strength measurement), it could be feasible to perform measurements just-about! For this review, I used a very cheap log-periodic antenna.
I printed off a 360 degree design (basically a circular protractor on a piece of paper), stuck it down to a sheet of plastic to act as the platform. I then rotated by hand the antenna and signal generator (the wireless microcontroller board discussed earlier). To make it easier I fitted it in a card box to keep it level. A nicer design would use a ‘lazy Susan’ or similar (an even better method would be to fully automate it, because the manual measurement exercise took ages, and I was conscious that environmental changes during this time could affect the results). Anyway, I positioned it away from radio reflective surfaces as best as I could manage, by placing it in the garden, away from most metalwork. I positioned the log-periodic receiving antenna indoors, pointed at the window, and attached to the spectrum analyzer. Overall, there was a distance of about 20 meters between the transmitting and receiving antennas. Now I was ready to perform measurements! I manually rotated the transmitting equipment 20 degrees at a time, and took a measurement. For now I used the ‘max peak’ capability on the spectrum analyzer, and used the RMS detector setting, but it’s for further study what are the recommended settings on the spectrum analyzer for this.
The results, whilst not the same as the chart in the Molex datasheet, are still valid I think. The square shape of this Molex antenna is clearly reflected in the chart’s four lobes. With more care aligning the planes and more accurate rotation and measurements, I think this is a usable technique, in the absence of real antenna test chambers.
Even if the results were off by a dB or two, a testbed set up for this could still allow comparisons between antennas, and could provide quick verification if there was something wrong with range, such as antenna detuning related effects due to other electronics close-by on the circuit board.
In summary I was happy that the MID antenna appeared to function as expected according this test, despite the output not being a precisely identical shape; after all, I was testing outdoors with changing temperature and humidity affecting the electronics, and I had only eyeball’d the alignment and angles.
I did not test the radiation pattern of the ceramic and patch antennas, since it is now clear that I would need to automate and more accurately control the orientation and rotation. However the datasheets do have the radiation pattern information for these parts too.
Rather than write a separate blog post, I just wanted to briefly highlight that it is possible to simulate antennas too. Unfortunately there are not many open source or free software options. The commercial option are extremely expensive, although some of them will allow some very limited simulation for free. This is not relevant to the Molex antenna review, except to the extent that some time spent with a simulator can help familiarise people with what to expect from antennas.
A commercial package with limited free capability is called AN-SOF (or AN-SOF100), and it’s quite easy to use.
The screenshot above shows my crude attempt at a helical antenna, and its radiation pattern at 2.940GHz.
I’ve learned a lot about Molex antennas through this review, and I’m pleased that I now have a few new antennas in my mental toolbox to choose for projects. I was surprised how board-friendly the ceramic and MID antennas were; they don’t take up a lot of space, nor do they need much clearance; 5mm is enough. This is excellent for compact portable devices. I also liked that I finally have a good solution for metal enclosures, other than ugly protruding, screw-on antennas. The patch antenna can be totally hidden behind a plastic panel, and will work despite the enclosure being metal.
Thanks for reading!