Molex 2.4GHz / 5GHz Antenna Kit - Review

Table of contents

RoadTest: Molex 2.4GHz / 5GHz Antenna Kit

Author: 14rhb

Creation date:

Evaluation Type: Development Boards & Tools

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?: I wasn't even aware of these Molex antennas!

What were the biggest problems encountered?: To test antenna systems properly requires several key pieces of expensive test equipment, although as these antenna were for the popular WiFi frequencies some slightly cheaper alternatives could be utilised. Equipment required would include a Network Analyser, Spectrum Analyser plus possible RF Signal Generator and ideally an Anechoic Chamber.

Detailed Review:


1.    The Roadtest Task

This roadtest was quite different from most that are offered up on Element14. It consisted of three component based 2.4GHz/5GHz antennas: one small adhesive plate antenna and two SMT chip antennas. The roadtest started some really good discussions on the Element14 Community pages regarding the amount of specialised equipment that would be required to test them, test methods and reasons for not applying. A brief overview of the antennas in this roadtest are:


{tabbedtable} Tab LabelTab Content

Antenna A



2.4GHz / 5GHz Wi-Fi PCB Antenna for Metal Surfaces (Molex 206995 Series)



All the detailed drawings and antenna plots can be found here at the Molex website:

Antenna B



2.4GHz SMT Ceramic Antenna (Molex 206513 Series)




All the detailed drawings and antenna plots can be found here at the Molex website:

Antenna C



2.4GHz SMT On-ground MID Chip Antenna  (Molex 47948 Series)



All the detailed drawings and antenna plots can be found here at the Molex website:

Antenna D


Buffalo Router

(my reference antenna)


The latter two SMT antennas were provided for this roadtest pre-mounted to a piece of PCB along with a length of semi-rigid and a SMA jack, which made testing much easier.


2.    My Background and Reasons For Applying

I have undertaken antenna measurements in past careers and I am familiar with some of the approaches and pitfalls, although I am very far from being an expert. I immediately saw this roadtest as something different and using antenna designs that I was not familiar with; if I were to use a chip antenna for a design I would have likely looked up the manufacturer's specifications and chosen a suitable part, ensuring it was mounted in accordance with any recommendations. After reading the roadtest I initially decided not to apply but later on there was a second call for applicants. Element14 have supported me in the past and I didn't want to see a product fail due to lack of applicants, so I submitted my very basic application. I was also motivated to apply for this roadtest because if successful it would give me the opportunity to brush up on my antenna theory, which is always a good thing (and have a bit of fun along the way).


3.     Equipments Required for Testing

Equipment required for professional antenna testing would include most of the following items:


  • A Vector Network Analyser (VNA)
  • RF Signal Generator (ideally tracking)
  • RF Spectrum Analyser
  • An Anechoic Chamber - with RF absorbent materials inside to attenuate reflections, whilst often shielded on the exterior to eliminate environmental signals.
  • A motorised turntable - to be able to rotate the antenna under test in both azimuth (the ground plane) and elevation (one of the vertical planes)
  • A screened room
  • A calibrated antenna
  • RF Cables and Connectors (low loss and ideally very flexible, SMApreferred to N type )


4.    My Application

4.1     Overview

My approach to this roadtest has changed more times that I can mention. I use some of the items listed above in my day job but I was unable to get permission to use them for this task. So I decided to change my approach and started to look for cheap alternatives that could be used at home to show that even without the expensive equipment some results could be obtained. I looked at how I could utilise a RF Voltage Controlled Oscillator (VCO) and how I could calibrate and use an RF diode or Detector Log Video Amplifier (DLVA) to measure received power levels. I also thought about using a WiFi router/dongle; and it was that approach that I later focused on, although I decided to use an Arduino MKR1010 to measure the signal strength at a set distance from my antenna under test.


My 'disclaimer' on my testing is that Molex spend a huge amount of money investing in high quality test equipment and therefore their datasheets should be read first and utmost and will state the actual specifications for these antennas. My tests will be using a multitude of unknowns such as power levels, antenna gains etc. Nothing is calibrated in my setup and will only be relative to previous measurements and change with the variations of the receive system as well. The testing environment for my experiments will also be far from ideal.


4.2     Test Location

Without the use of an anechoic chamber to create a screened environment, external noise could be a huge issue. However it would be mitigated to some extent as I would be connecting to the SSID of my router and as long as that was the strongest signal then I would be measuring that only, other routers using the same channel but at lower power levels would not register. There is a local company that hires out old shipping containers for site storage and I considered asking if I could hire one on their site for the day in order to make my measurements but RF reflections would mean the slightest movement would make the results fluctuate in excess. I thought about making the whole setup portable and conducting my experiments in a farmer's field as far from everyone as possible but there are logistical issues (making sure the power source is adequate, you take every single piece of equipment along) ,the issues with looking suspicious !and everyone who comes over to talk to you would likely have a mobile phone (with WiFi on).


I finally decided to just go to the end of my garden and do the experiments as I had electrical power, a shed of bits if required, I would be far enough away from neighbours' routers that my router would be at a higher power level and I would have a plentiful supply of hot tea (being British tea is important to ensure experimental success) not far away. That plan also didn't last too long because, as winter draws in, the weather has deteriorated and it seems to be raining every day I'm off work and looking at doing my testing. Finally I decide to setup my test equipment in the house, get the best unobstructed run that I could, with minimal metal objects nearby (fridge/freezer) and at a height approximately halfway between ceiling and floor. I achieved a 5m 'range' with a metre behind the power measurement/receive (rx) antenna and an opening into another room behind the transmit (tx) antenna (antenna under test).


5.    Equipment Setup


5.1   The Router

Using a router with a detachable SMA-RP antenna would allow me to add the Molex antenna under test. The only router I had in my box of bits that would suffice was a Buffalo WHR-G54S that ran off 3.3v ( enter my new Duratool D03232 Power Supply  ). This router unfortunately only operates in the 2.4GHz band so I cannot stretch my experiments to look at the 5GHz responses on one of the antennas. I planned to use the router's web interface to manually command the router to switch between the 11 WiFi channels as required. However, there might be benefits in using an older router as it likely has less clever processing such as beam forming.


The router page to change the frequency channel looks like this:



And the WiFi channels in the 2.4GHz band look like this:


[Source:  ]


5.2     Arduino MKR1010 - Power Meter

The software I wrote for my Arduino power meter was based on an Arduino example from the MKR1010 library and my 'scrappy' code can be found on my Github page here ;perhaps sometime I will get around to tidying that code up. A typical screen shot of the Power Meter running is shown below ( edited to remove MAC/SSID ).


I spent a while trying to add a timestamp to my code before realising that the Serial Monitor had one built in...the tick box at the bottom of the page. That worked really well and in the example above shows time as 08:50:33 seconds. However as my main cycle started it stopped printing the time. Instead I implemented code to print the elapsed millis on the screen. It was still a useful reference that I could note down on my paper results sheets and determine in the subsequent analysis which channel I had switched to and when. I also added plenty of semi-colons into this output to enable me to delimit the data into columns of the spreadsheet in the analysis phase.


5.3     Antenna Rotator

With my initial roadtest application suggesting plotting the radiation patterns for these antenna in azimuth (and perhaps elevation) I realised there were two ways to do this (1) my manually rotating the antenna under test using a protractor scale or (2) using a stepper motor to speed up the move/measure cycle. It was my application to do this Molex roadtest that spurred me on to get a stepper motor running as I thought it would be useful to achieve these gain plots, this resulted in my following post My First Stepper Motor . I also utilised that turntable for my sharethescare, entry in 2018 Peter the Pumpkin (part 5 - final thoughts) .


The stepper motor is on a 4:1 geared belt to increase torque provided to the turntable.


I used an Arduino Uno to rotate the stepper motor over a pre-determined azimuth angle and to do so in a known period of time. I would then be able to measure the antenna power measurement over that time period and therefore plot the azimuth gain. In my initial application assuming I could use actual RF test equipment I was planning on writing some code using SCPI to  control the RF, rotate the turntable and read a power measurement. However as that approach was derailed I had to think of a simpler approach. Instead I would power up the turntable and as it reversed up to the 'zero position microswitch' I would walk briskly to the power measurement end of the setup. As soon as I saw the turntable reverse direction to start its 180 degree sweep I would note the time on the power measurement system, and I would also note the time when the end of travel was reached. That should give me a reasonable idea of antenna gain against angle i.e. a polar plot or azimuth beam plot.


5.4      RF Transparent Support

To make the antenna under test electrically isolated from the interference of the metallic motor (on the scanning experiment) I had been hoarding pieces of polystyrene and planned to use my Nichrome hot wire to craft them into a nice stable pillar. This pillar would sit on my turntable and be topped off with the antenna under test. I saved old packing to make this but never had to form the pillar as my plans changed...someone (me) tidied up and, without thinking, threw the polystyrene away. Luckily I did have a few leftovers and glued them together to make a RF transparent tower, which I fixed to the turntable with hot melt glue. I'm thinking of making a submission for the with this !


5.5     Wooden stand and tripod.

For the Arduino MKR1010 power meter I also made a small wooded stand to give it some RF isolation from the metal tripod that it was mounted on. If it looks like a piece of broom handle, that is because it is. I did make an aluminium insert of 1/4" UNC thread and glued that into the bottom end of the wooden rod so I could mount it to a camera tripod. And the top of the wooden rod has a small piece of wooden baton attached, partially recessed with a 20mm spade bit and again glued. The aim was to minimise metal parts - as I write this I'm now wondering why I mounted my Arduino on my breadboard - which is a 'sea' of metal contacts! The Arduino was fixed to the wooden stand with some Velcro tape.


My plan was to now: Use an old WiFi router to transmit basic channel SSID data on each of the WiFi channels and to feed that signal into the antenna under test. I would then use an Arduino MKR1010, with its inbuilt WiFi, to detect the SSID and list the perceived power level.


5.6     Connectors and Adapters

There were a few additional RF connectors I had to purchase for this roadtest. These were:


{tabbedtable} Tab LabelTab Content
SMA jack to U.Fl Jack adapter

Essential for coupling to the MHF antenna (Antenna_A), This small connector relies on the mechanical properties of the material to click into place and has a very limited number of mating cycles before it fails. Product LinkProduct Link

SMA plug to SMA.RP jack

My Buffalo Wi-Fi router uses a SMA-RP connector. These adapter will convert the router and its antenna to SMA. Product LinkProduct Link

SMA jack to SMA.RP plug

As previously but the other gender. To enable me to use the existing Buffalo antenna in rotational tests. Product LinkProduct Link


6.    Tests Conducted

There are a multitude of test parameters that can be varied such as antenna orientations, mounting positions and frequency of interest and quite quickly a huge number of test combinations can be created. Therefore I tried to simplify my testing to keep to a manageable number of experiments (and therefore time to conduct) whilst still providing a valuable insight into using these Molex antennas.


6.1     Experiment 1 - Frequency Responses


6.1.1     Aim

Plot the variations in power received as the transmit signal frequency changes. I would be able to crudely sweep the router frequency by manually going into the settings and adjusting the channel it operates on (from Ch 1-11). The downside is that each change takes about 45 seconds to register and another 30 seconds for the output power to level out but that would be ample time for me to walk back to the measurement end and await the signal to settle down before taking a start time reading. I then gathered some measurements over approximately 1 minute before noting the finish time and walking back to the router end to select the next channel.


6.1.2     Setup

The following setup was used to test out the frequency response of each antenna.


Distance Tx to Rx:5.00m
Height Rx:1.50m
Height Tx:1.50m
Router Channels:1 - 11


The measurements were taken on my Toshiba NB510 Netbook (LUbuntu) and I uploaded them to my  Google Drive and then back down onto my Windows PC. I then opened these frequency response files in Notepad, copied the data and pasted it into LibreOffice Calc. I then highlighted the readings that corresponded to each channel using the timing notes I had handwritten. I ignored the outer 1-2 points as sometimes the signal was still settling (it was quite obvious looking down the set of values). I then used the AVERAGE function to extract a value from the remaining 10-12 readings and it is this value that I finally plotted against the channel frequencies for the WiFi band. The results I obtained are shown below for each of the four antennas.


6.1.3     Results

{tabbedtable} Tab LabelTab Content

Antenna A



Antenna Bimage
Antenna Cimage
Antenna Dimage


6.1.4     Molex Specifications

All three of the Molex 'Application Specifications' show the VSWR dipping at 2.45GHz and that is likely the frequency that they have been designed to operate best at. So for best response you would want to select channel 7 or 8 on your router.


6.1.5     Discussion

If the specifications show the antennas are tuned to 2.45GHz then I would have expected my best results (highest power) at that frequency, instead that frequency is where I actually see a dip in power received, which is odd. It is at this point if I look at the published specifications for the three Molex antennas I see their respective peak gains of: 2.6dBi, 3.6dBi and 3.3dBi respectively. Although I cannot get a direct dBi correlation from my results, the specifications show that antennas B and C are very similar while antenna_A has a lower gain by approximately 1dB.


A very broad observation could state my Antenna_A received -53dBm, Antenna_B -60dBm and Antenna_C -58dBm.


What I don't know is how the Arduino MKR1010 antenna performance affected those results or if my Buffalo router was competing against other nearby WiFi channels or interference. From my Power Meter output I was able to monitor surrounding WiFi signals and which channels they were on. It appeared my neighbour's WiFi was on channel 1, and I wonder if that were somehow responsible for the lower received signal on Antenna_D, although it should therefore have affected all four antennas equally.


6.1.6     Conclusion - Frequency Response

Antenna_A actually seemed to perform better than the two chip antennas, which were 'similar' in performance although this should have had a lower gain than the other two by approximately 1dB. This assembly was manufactured in the Molex factory whereas the two chip antennas had been mounted for this roadtest, looking at the documentation there is suggestion of balancing components but my provided boards only had a single in line SMT component. Perhaps these two chip antennas are therefore not properly matched to the 50 ohm feed. This cannot be verified without adding the components or taking VSWR measurements.


Power measurements taken on the MKR1010 did fluctuate by several db at times, I tried to minimise that by not walking around during measurements, turning off all my gadgets' WiFi and my home router. I suspect that some neighbouring routers adapt power either with Automatic Gain Control (AGC) or with active antenna beam forming and could therefore change their transmit power levels according to the situation in that house. Even though those were of low power compared to my settings they could cause my setup to adjust power in response to the external levels.


6.2     Experiment 2 - Antenna Patterns


6.2.1     Aim

Determine what the antenna pattern looks like in the Azimuth plane. This is actually provided by Molex and for these antennas is fairly omnidirectional (same power in all directions ). This experiment can see if that is true in a real situation and also because this is one of the more fun experiments to do with an antenna.


6.2.2     Setup

I would need to measure the power received at my set distance of 5m. I would concentrate on a single frequency and from looking at other competing signals I chose WiFi channel 8, as other (neighbouring WiFi) signals were on 1 and 4 albeit at very low powers. I would rotate the router antenna 180 degrees using the stepper motor setup. Issues I could see would be that the feed cable to the antenna would get fouled on something and damage the stepper mechanism by stalling it or ripping the antenna system off the turntable. I used the reset microswitch and a 12v Sealed Lead Acid (SLA) battery for the whole turntable mechanism. At the end of each run I would disconnect the battery and wind the turntable back by hand to very near the start. Once powered up again it would give me enough head start to get to the measurement end and log the start/end of scan times so I can correlate to my power measurements.


The setup is similar to the frequency response experiment except that I used an antenna rotator (shown in red below) and fixed the router to a single channel (ch 8).


Distance Tx to Rx:5.00m
Height Rx:1.38m
Height Tx:1.37m
Router Channel:8


This is a video of Antenna_A being rotated. Watching it back has made me think - perhaps I need to do less roadtesting and more DIY (as in paint my wooden door frame?) ....but then again this is more fun !


6.2.3     Results

These experiments were a bit easier to undertake as I selected WiFi channel 8 and did not have to wait for the router to reset (45 -60 seconds +30s levelling out). I had to repeat a few of the runs as the RF feed cable occasionally snagged on the corner of the turntable plinth and for another run the antenna pinged into a different orientation (fixed with a larger lump of BluTak). I repeated three azimuth polar plots for each of the four antennas. I averaged those results and plotted them out against a nominal 180 degree arc - this assumes the travel was constant and being driven by the stepper, this was the case although if the cable snagged that time v. angle correlation was lost - hence the need to repeat those runs. The measurements were effectively taken every 20 degrees. Below is an azimuth gain plot of all four antenna systems under investigation.



6.2.4     Molex Specifications

Looking at the Molex 'Application Specifications' for each of the antennas, and taking into account the XYZ detailed for each, I can extract out a 180 degree slice of the XY plots for 2.45GHz that should match my antenna's orientation. The datasheet results give approximate variations between max power and min power of 12dB, 7dB and 10dB respectively for antennas A,B and C.


6.2.5     Discussion

The Arduino MKR 1010 would have been trying to monitor the WiFi signal as it fluctuated rather than take an accurate instantaneous power measurement. On reflection I should have slowed the rotation speed down some more to gather more data samples and hence average them out better. Hopefully the three runs for each antenna went someway towards that and looking at the data it was fairly consistent on each run. There is strong correlation between the max and min differences of my results and those in the Molex specifications. e.g. 2,2 and 4dB difference.


6.2.6     Conclusion

With this limited equipment setup it is uncertain if the fluctuations in received power were real and caused by the transmit antenna gain or indeed by the Arduino MKR1010 receive antenna gain. It could have been multipath generated as the antenna rotated or even a cable working loose during the rotation. However the experiment still shows how such an experiment could be conducted - having a better test environment (anechoic chamber or wide open space) and a calibrated receive antenna would have helped ensure more accurate results. I am also quite pleased that the large fluctuation I can see in my results is not too dissimilar to those looking at the official polar plots at 2.45GHz.


I only rotated the antennas in one plane and could have repeated for the other two planes as shown in the manufacturer's datasheets. Perhaps if my results were more conclusive I would have gone back and conducted those additional tests.


6.3     Experiment 3 - Mounting Considerations


6.3.1     Aim

The Molex documentation, Webinar and related Blog post all stressed the importance of mounting these antennas properly and that some of the antennas did not perform very well when mounted onto a metal case. I thought I would do some experiments first and see for myself. For this experiment I would use a variety of backing materials and distances and record the power measurements for each. I would not be testing antenna_d during this experiment as the design is completely different.


6.3.2     Setup

I selected six tests to conduct on each of the four antenna and, as for the polar plot testing, I kept the router fixed on channel 8. The setup was changed slightly using a different support for the transmit antenna (Molex) as shown in red/green below:


Distance Tx to Rx:5.00m
Height Rx:1.30m
Height Tx:1.30m - 1.38m
Router Channel:8


The six tests conducted for mounting materials were:

{tabbedtable} Tab LabelTab Content
Wood Frame

Mounting using velcro band or small piece of BluTak to hold against a wooden coffee table leg. Photo below shows antenna_a.


Copper Plate

Mounting in direct contact with a 160mm x 100mm copper plate (actually a failed PCB layout). Photo below shows antenna_c.

Aluminium Plate

Mounting in direct contact with a 250mm x 250mm aluminium plate. Photo below shows antenna_a.

20mm Airgap

Using a 20mm thickness of polystyrene to airgap off the aluminium plate. Photo below shows antenna_b.

40mm Airgap

Using a 40mm thickness of polystyrene to airgap off the aluminium plate. Photo below shows antenna_c.

80mm Airgap

Using a 80mm thickness of polystyrene to airgap off the aluminium plate. Photo below shows antenna_b.



6.3.3     Results

As previously I analysed the data from the Arduino MKR1010 serial monitor log file by pasting it into LibreOffice Calc. I identified the times when I was using each of the backing materials by time and averaged the power measured results accordingly. The results obtained, after extraction from the raw serial data and some analysis, were:



6.3.4     Molex Specifications

The mounting specifications for Antenna_A details it on various sizes of metal plate. The specifications for antenna_b and antenna_c detail the effects of a nearby metal 'can' but not of the substrate: as the substrate is taken to be the metal PCB that it was supplied on.


6.3.5     Discussion

The results were not as I would have expected. Unless my understanding of RF v. polystyrene is incorrect, I thought it would behave more as an airgap and it is only on antenna_c that the 80mm airgap achieves a better result. These results are very mixed and I am disappointed that nothing conclusive can be deduced. If being very generous I could state that having a metal plate adjacent to the antenna is not good (which I know is true from the Webinars) but that only holds true for antenna A and B. The results for antenna_a are confusing as well as the Molex specification details the antenna being mounted to a metal plate whereas I get a few dB of extra transmitted power (unless an artifact of the rx antenna) when mounted to the wooden frame.


6.3.6     Conclusions

I suspect that my adhoc and rough approach to this test has been its downfall. Strange and unpredictable behaviour often occurs in lab experiments with RF signals and sometimes the slightest movement in the setup can have big impact on the output. Perhaps there are multipath issues and raising the height of the transmit antenna during each test caused these changes. Even changing the lay of the feed cable could have influenced the measurements. I would not say the experiment is a complete failure though, as it shows how such a test could be achieved, and with better equipment, a calibrated source and tx antenna and a anechoic chamber the results would have been far more conclusive.


7.     How Antenna Measurements 'Should' be Taken (note I am no expert ! )


7.1     Environment

There are two important aspects for taking measurements aside from the equipment. These are to eliminate external RF interference (such as other WiFi sources including mobile phones, microwave ovens, ZigBee, Bluetooth etc). Also poorly constructed devices radiating on harmonics that fell into the 2.4GHz band or even broadband noise sources (such as electrical arcing or poorly suppressed HT in automobiles) would also add noise into any experiment. To eliminate those would require conducting our experiment within a shielded box - large enough to keep the two antennas undertest in their far-fields. However, placing the source and receive antenna in a metal enclosure would result in the signals reflecting around causing huge problems with multipath. This is where the RF anechoic chamber steps in; it is lined with a radar absorbing material and design which attenuates reflections (in the same way that audio in a bathroom echoes whilst in a fully furnished room it is far less echoy). Large anechoic chambers will be expensive but there is a nice Youtube video of someone making their own; he experiments with different compounds and finally settles on one for his lining material:



[An anechoic chamber from Google image search: Anechoic Chamber: Anechoic Chamber ]


Another key point in using the anechoic chamber is that our test signals will not interfere with anything outside of the test chamber.


7.2     Equipments

An RF signal generator and RF Spectrum Analyser would be my next important items. With these some really accurate measurements could be achieved. Lastly a Vector Network Analyser (VNA) could be used to measure the VSWR of each antenna, this would show how well they were matched to the feed line (and would be a good measure of how much power they could radiate).


7.3     Frequency Response

I would use a CW signal at a known power into a calibrated antenna suited to the band being investigated. I would probably use the calibrated antenna for transmitting a signal, which could then be compensated with the gain at each frequency, although it probably doesn't matter if it is used for the Rx or Tx end. The RF source could easily be swept between two frequencies and the sweep time set as required. I would probably settle for a weep time of a few seconds for the 2.4GHz WiFi band. I would set the spectrum analyser to sweep that same band and place the mode into 'Peak Hold' - quite quickly the frequency response would be displayed. Compensating for the calibrated antenna could then be achieved in the analysis, although I am sure some spectrum analysers would allow for this offset to be added in. The cables and connectors could also be calibrated out at the same time as this antenna leaving the changes in power received down to just the path loss and the antenna under test. Path loss could be calculated (inversely proportional to wavelength and proportional to square of range)....we would at that point have a really good measurement of antenna frequency response.


7.4     Polar Plot

The equipment would consist of a rotation platform like I built, and the RF synthesiser and RF Spectrum Analysers plus a calibrated antenna and cables. If this type of test were conducted frequently it would be a good idea to automate the process using GPIB/Ethernet/SCPI & Matlab/Labview etc to control the turntable position whilst taking power measurements. Automating would speed the test up and therefore a series of plots could be measured at different antenna frequencies.


7.5     Backing Materials

This test would be conducted in a  similar way to that already shown. However a full frequency response could be taken for each backing material using the RF synthesiser/Spectrum Analyser as they are so quick. The important point here would be the use of an anechoic chamber and some additional though on mountings. To have 'live' feedback would enable the user to explore in realtime what makes the signal fluctuate - is it there hand nearby? a slight change in antenna orientation/polarization? a shift in cable position?


7.6     Other Tests

I would have used a Vector Network Analyser to determine the matching of the antennas to the 50 ohm load, it would show how much energy was taken in by the antenna and how much is reflected back onto the feed. More more modern measurement than VSWR and Smith's Charts seems to be the antenna efficiency  - and modern VNA will measure this. The VNA is the equipment that would enable the fine selection of the matching components and line lengths on the PCB to ensure maximum radiated power.


8.     Molex Kit of Antennas

I was given a really nice set of similar antennas at the Coventry Engineering and Design Show, as shown below. I'm already using a couple of those Molex antennas in my Elderly Person in the Community Care: Arduino MKR 1300 WAN


9.    Overall Conclusions


9.1     Expected and Unexpected

Even though most of the results do not conform with what I think they should, when I think about it more, I am not too surprised. I would have liked to have seen flatter frequency responses and responses that followed some similar changes across antennas or with external influences rather than my mixed set of inconclusive results. However - I've had a lot of fun doing this roadtest in the most basic way I could. The multitude of unknowns by not using bespoke test equipment has led to my inconclusive results but I believe the biggest issue was the lack of a suitably 'RF quiet' environment - using an anechoic chamber would have drastically improved my results.


9.2     Conclusions on Molex Antenna (this is what the roadtest was about)

I didn't even know about these types of antenna before seeing this roadtest and undertaking this roadtest has introduced me to a range of different connectors. It was in doing this roadtest that, when I saw the Arduino MKR1300 LoRaWAN, I realised what type of antenna connector I would require. I also learned that the router connector is a SMA-RP and how that differs to the standard SMA (i.e. the centre pin is opposite to that in a standard SMA). The pre-assembled Molex antenna are very small, well made and have ample supporting documentation. By taking a system's antenna off the PCB and using one of these cabled Molex antennas it is possible to mount it in a much better position within the enclosure, orientating if necessary to the optimal configuration. The antenna_a has an adhesive backing and so is really easy to setup. The chip antennas take up a minimal amount of PCB space and much less than a similar stripline equivalent.


Personally, if I wanted to add an antenna to a system, unless I was making my own, I would not undertake any measurements other than perhaps the VSWR to ensure the matching was correct. I would refer to the manufacturer's detailed test sheets to find out everything regarding the performance and select a suitable antenna accordingly.


I scored this product 60/60 as some categories did not apply and I did not want to mark it down on those. The areas that did apply, I could not fault the product or the quality of Molex documentation.


9.3     My Future Hobbyist RF Plans

This roadtest has got me interested in undertaking some more RF tests at home. Whilst rummaging around my boxes of bits I've found some old RF amplifier ICs which I never got around to using. I would like to get some PCB and SMA connectors and make up some small blocks that can be connected together as required. An RF source would be another item high on that list - something that covers 100kHz to 3GHz would make a great start.


One of the most useful pieces of test equipment would be a spectrum analyser. Whilst my past career I've used R&S, Agilent and HP spectrum analysers that cost more money than my house I had a look at the Farnell site and was surprised to see there are much cheaper alternatives for sub-3GHz. One item that really caught my attention was this Product LinkProduct Link which is £711 (UK as of Nov 2018)....that is really good value. It perhaps isn't as capable as the higher priced models but for getting an insight into what affects the signal it is really important.


Ideally I would like to construct my own anechoic chamber, perhaps not even as expensive as that shown in the Youtube link, but something that could enable measurements and setups to be conducted without any interference.


10.    Thanks and Reflection

As always, thank you to the Element 14 team and Randall Scasny in particular for selecting me to try these Molex antennas. Thank you to Molex for supplying the antennas for this roadtest and to their sales and engineers on the Engineering Design Show stall for the interesting discussions; although I haven't taken a professional approach to testing this product I am hoping my simple approach will prove interesting.


Molex also provide design advice on a whole range of if you have  a product you are manufacturing, and it needs an antenna, I would suggest you contact them for help.


Thanks for reading,



  • The ADALM is a great find, so thank you for linking it; I have just read about the specification. It also has a transmit channel so you could make a RF Spectrum Analyser and an RF Signal Generator in the same unit. And it is a really good price as well image so I might have to try and buy one. Shame it doesn't quite stretch to 2.4GHz/WiFi.

  • i just saw your post and it is inspiring to see you stretching the low cost gadgets to do the antenna test. This also arouse my interest in RF. i am even further from expert than you although in my past career i create RF calibration software using the same list of equipments which are collectively more expensive than my houseimage . The idea to use Arduino and hobbyist stuff to do the experiment i think is applicable to at least undergraduate lab (it is costly and not so practical for undergraduate lab to have equipments more expensive than my house to do some tests that they hard to apply in study).


    also, it is a very pleasant surprise from the link you post about this very affordable spectrum analyzer >>Product Link

    i am not able to find a cheap signal generator so i think your wifi dongle seems good way

    i do have Adalm Pluto, a zynq powered sdr but i can only reach 2.1GHz (using provided antenna), but the onboard ADI RF chip can cover from 325 MHz to 3.8 GHz


  • Yes, but initially you want to isolate the antenna from any outside emissions.

    With the shield grounded, you will get some suppression of emitted signals from the antenna, but you are correct, to get full dissipation you need some radar absorbing material.


    I am not aware of any good cheap way to make RAM.



  • Hi Shabaz,


    Thank you for your kind comments. I often find it can be quite rewarding making one piece of test equipment to build another and use a cycle of iterations to improve the other - then again, a fully loaded, professional piece of test equipment usually cannot be beaten image.  I'm looking forward to seeing your testing of these antenna - hopefully all is going as planned for you.



  • Hi Rod,


    Very comprehensive review! So cool that you automated the testing with an automatic rotation system : )

    I think your results are very valid - it's what you measured. Like you, I caught the interest in RF and wanted to acquire some test equipment. I didn't have a lot to spend for ages, so I used some home-made equipment, and some used equipment (including an awful boat-anchor spectrum analyzer) for years and the latter was truly awful! - barely usable. And especially when you're at the cutting-edge for consumer wireless at these frequencies, modern tools, and from what you've found modern routers and dev-boards can make very handy test-tools : )

    Also as you say the manufacturer's documentation is extremely useful, the documentation is heavily relied for antennas where most engineers don't have access to test chambers and procedures : (

  • Hi Gough,


    Thank you for your knowledgeable thoughts and comments. I think at times I was 'willing' the results to fit my expectations rather than to step back and look at what I was actually seeing. Jury is still out on the RF transparency of polystyrene - perhaps I'll do a another test on it vs. airgap.


    Luckily most of the WiFi module on the MKR1010 was hanging off the edge of the breadboard but I was a bit annoyed with my mistake - especially as I didn't actually need the breadboard for functionality.


    Re the power levels, I think I was fortuate enough that either neighbours were away (and WiFi was off) or my home is just far enough away to attenuate most of them. One neighbour's WiFi was definately visbile as their SSID is set to their name.


    I bought the U.FL adapter from Farnell…  there was a link in my blog but I now see it doesn't work, so I'll correct that. I saw your approach which would have been my other route. It is a handy adapter to have for any future work (assuming I can find it next time ! )


    This roadtest has enthused me and I'm looking forward to doing some more RF work at home.



  • Great effort , although I guess it was not surprising that collectively, none of us yet have really been able to quite get the kind of results we were dreaming of. I did like the British humour though - Tea is of vital importance, just as coffee is to PhD students image. I did see your review a day or so ago, but didn't quite have the time to go thoroughly through it until now.


    Just a few thoughts - given that Wi-Fi transmissions themselves occupy 20/22Mhz or so, and each channel is spaced only by 5Mhz, testing at every channel might not really be as useful as it first seems. I wonder whether the results might have been impacted on partial-overlapping Wi-Fi transmissions, especially if everyone follows the 1-6-11 rule and you have a few that do not. That might actually trip up some RSSI readings from time to time. Also consider the beacons are very short transmissions sent every 100ms (by default), so I suppose the variance in RSSI readings that you might see are not entirely unexpected. Device to device RSSI correlations are generally quite poor.


    You did make a note of the MKR mounted on a breadboard - I had exactly the same thoughts, it's in a "sea" of metal, but worse still, the lengths of the metal strips could well be just about the right length for the wavelength we're dealing with. I'm not sure which end the antenna is on the MKR in the way you've mounted it on a breadboard, but if it is over this "sea", it will affect the radiation pattern notably. Using it as a receiver, the radiation pattern of its antenna may actually interact with your "rotating" transmit antenna as some of the signal contribution could come from "off angle" reflection which may-or-may-not be well received depending on where the reflection lands.


    The other thing is that the styrofoam itself is something you often use as an air gap, but I'm not sure you can quite consider it that way. Even though it has a lot of air in it, it almost certainly has a dielectric constant and resting things on it may turn it into a capacitor of sorts which could detune antennas or affect impedances of transmission lines etc. But the way it interacts is not something I really understand well - to me, a lot of microwave stuff is "magic" as doing very subtle things can anger it and make your signals evaporate (e.g. bending the coax slightly).

    You did mention the azimuth testing seeming to show some correlation to the Molex data, which sounds promising. One thing I normally do when I'm not sure if it's a chance-correlation or not is just to repeat the experiment more than once and see whether the results are consistent or not. If you can't achieve consistency, at least you can then understand the limits of detection for your methodology (e.g. whether seeing a +/-1dBm change is actually meaningful or just noise). In a sense, characterising the test process with some reference tests may be equally important as characterising the product when the non-idealities can "bite" so frequently.
    The most interesting part was to see you managed to get an SMA adapter to the (claimed) MHF connector on the PCB antenna - where did you find that? I was looking for ages and gave up, hence desoldering an SMA tail from the other evaulation board to borrow for testing.
    - Gough
  • Thanks Gerrit,


    That is a great idea  - if I joined a local group one of the members might have test setups they allow the other members to tune their circuits on etc. Also there would be the kind of experience within the group that would never be found in a textbook.



  • Thanks DAB,


    I like your suggestion for a cheap test chamber - would it not still need some kind of RF absorbant material inside to attenuate the reflections to make it useful; I could use my metal garden shed for future tests?



  • Hi Rod,


    very nice review given the limited test equipment, I was in the same boat.

    Regarding starting to explore RF electronics, I would advise to get a ham radio licence, this will open you a new world. ( )

    There is even a not very active Amateur Radio Group here on Element14.