Molex 2.4GHz / 5GHz Antenna Kit - Review

Table of contents

RoadTest: Molex 2.4GHz / 5GHz Antenna Kit

Author: gpolder

Creation date:

Evaluation Type: Passives

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?: Other antenna's as mentioned in my review

What were the biggest problems encountered?: Lack of faraday cage for proper measurements, as well as lack of a tracking generator on my spectrum analyser. An in the end my noise source rf-bridge combination did not support the WiFi frequency band.

Detailed Review:

As had to deal with too less applicants for this roadtest, I promised him to send an application.

Since I was a lucky roadtester of the (3GHz Spectrum Analyzer - R&S® FPC1000) I have a spectrum analyser available, which is a nice tool for testing antenna's. Unfortunately the FPC1000 is not equipped with a tracking generator. Therefore I will use the noise source as explained in my FPC1000 review.

For proper quantitative measurements of antenna's a faraday cage (https://en.wikipedia.org/wiki/Faraday_cage) should be used, as well as  well defined structures, as explained in the Molex application reports:

https://www.molex.com/pdm_docs/as/AS-146153-100-001.pdf

https://www.molex.com/molex/products/datasheet.jsp?part=active/1461530100_ANTENNAS.xml

 

https://www.molex.com/pdm_docs/as/2065130001-AS.pdf

https://www.molex.com/molex/products/datasheet.jsp?part=active/2065130001_ANTENNAS.xml

 

https://www.molex.com/pdm_docs/as/AS-47948-001-001.pdf

https://www.molex.com/molex/products/datasheet.jsp?part=active/0479480001_ANTENNAS.xml

 

Therefore my tests are more indicative, by comparing different antennas. Furthermore note that I don't have equipment for 5GHz, so all my tests are at 2.4GHz

In the end I was planning to do two tests with the antenna.

  1. Comparing signal strengths using a embedded WiFi controller, an old Particle Core (https://www.particle.io/) in this case.
  2. Measuring the VSWR using my FPC1000, a noice source and a RF bridge.

 

Here is a picture of the three items I received:

Two antennas mounted on a PCB, connected with a SMA socket, and the other antenna with a piece of wire and a uFL connector.

On the PCB close to the antenna you see some additional components. These are matching components that according to the data sheets improve the return loss. For the ceramic antenna details can be found on page 11 of the data sheet (https://www.molex.com/pdm_docs/as/2065130001-AS.pdf ):

On the PCB the capacitor is lacking, the inductor is too small to measure.

 

For the MID antenna its about the same story, circuit can be found on page 11, return loss on page 12 (https://www.molex.com/pdm_docs/as/AS-47948-001-001.pdf ):

Here both the capacitor and inductor are placed on the PCB, I measured the capacitor around 10 pF, I think this much larger value is due to the PCB traces and the rigid coax.

I tried to measure the inductor, but that didn't work:

 

 

Problem for the signal strength test is that the Particle Core is equipped with a uFL socket while only the balance antenna has a uFL plug. The other ones are equipped with a SMA socket. Therefor I needed to order quite a lot of adaptors.

For the first test, a uFL connector to SMA socket adapter and SMA-SMA cable for connecting the ceramic and the chip antenna.

 

For the second test I ordered a RF-Bridge, 50 Ohm SMA terminators, a SMA 10dB attenuator, which together with the noise source helped me to do the VSWR measurements.

In order to connect the standalone antenna to the RF-Bridge a uFL to SMA adapter is needed. So I ordered some uFL sockets and made an adapter myself. Hopefully this will not influence the measurements to much.

 

Part 1: Signal strength measurements

Signal strengt measurements are done with a Particle Core WiFi based IoT controller as can be seen on the picture below.

This controller is equipped with a uFL socket, so different antenna's can be connected and tested.

On the picture is a small piece of wire of 32 mm, which is exactly a 1/4 of the wavelength of 2.4GHz (300/2400/4).

I made a small program that does a WiFi scan, and publishes the number of networks found, and for each network the signal strength on the Particle cloud.

 

// Define the pins we're going to call pinMode on
int led2 = D7; // This one is the built-in tiny one to the right of the USB jack
WiFiAccessPoint aps[20];

// This routine runs only once upon reset
void setup() {
    pinMode(led2, OUTPUT);
  }


void loop() {


    int found = WiFi.scan(aps, 20);
    String f = String(found);
    Particle.publish("number of networks",f);


    for (int i=0; i<found; i++) {
        WiFiAccessPoint& ap = aps[i];
        String rssi = String(ap.rssi);
        Particle.publish(ap.ssid,rssi);
        digitalWrite(led2, HIGH); // flash number of access points found
        delay(200);               
        digitalWrite(led2, LOW); 
        delay(500);
    }
    
    delay(20000);               // Wait for 20 second in off mode
}

 

Here is how this looks like:

 

 

As radio signals are influenced a lot by surrounding objects I tried to standardise the setup as much as possible, still care must be taken when interpreting the results.

Six antenna's are tested; the aforementioned 1/4 wave wire, a 2dB external whip WiFi antenna, a home-brew bi-quad antenna (Old meets new, the 1-Wire Weather Station on the SPARK Core. (part 7) ) and the three Molex antennas.

For the external WiFi antenna I soldered a loop on the side of the Molex PCBs, in order to mount this antenna at a similar position as the Molex chip antenna's.

During the test the PCBs are placed at a fixed position at a distance of 2 cm above the IoT controller. Here some pictures of the setup:

 

 

For each antenna I logged the output of the scans for a couple of minutes and noted the mean number of networks found (rounded) and the mean signal strength for two of the networks that are at a distance of about 50 meter away from my home:

 

AntennaNumber of networks foundNetwork 1 RSSINetwork 2 RSSI
1/4 wave11-39-38
2dB whip antenna17-41-43
Bi-Quad

8 N

10 NW

13 NE

-35

-35

-37

-35

-35

-33

Molex Stand Alone Balance Antenna15-40-37
Molex SMT Ceramic Antenna17-37-41
Molex SMT On-ground MID Chip Antenna16-36-45

 

As expected the 1/4 wave antenna has the worst performance regarding the number of networks found, although the signal strength of the two networks is surprisingly high, compared to the other antennas. The Bi-Quad is a directional antenna, it only sees networks in the direction it is placed. The signal strength for the two networks is clearly the highest, which proves the high gain of this antenna. Network 1 is likely located in the North East direction, while the direction of Network 2 is more to the North/North West direction.

The whip antenna performs the best in the sense of the number of networks scanned, but the signal strength of the two networks are surprisingly low.

Then the Molex antennas. Given their size, and compared to the whip antenna I have to say they perform very well. The number of scanned networks is the same, or almost the same and the signal strengths are mainly better, while the whip antenna is much larger.

Regarding the numbers, be aware that RSSI is a coarse signal level indication that differs in linearity and offset over different chipsets, so the values mentioned are only indicative and serve as comparison between the antennas (https://en.wikipedia.org/wiki/Received_signal_strength_indication).

 

Part 2: VSWR measurements

The second test I planned is measuring the VSWR of the antenna's using my FPC1000 spectrum analyser. Unfortunately in contrast to the FPC1500, the 1000 doesn't have a tracking generator and rf-bridge. Therefore my plan was to use a broadband noise source and a cheap rf-bridge.

The video below, by James Eagleson nicely demonstrates how this can be used, also in comparison to a more expensive directional coupler.

 

 

The circuit diagram of the rf-bridge:

 

My test configuration starts with the noise source (more info can be found in my 3GHz Spectrum Analyzer - R&S® FPC1000 - Review) followed by a 10dB attenuator and connected to the input of the rf-bridge. The output of the rf-bridge is fed into the spectrum analyser. The reference of the bridge is a 50 ohm terminator, and the device under test (DUT) is the antenna.

 

In contrast to James Eaglesons rf-bridge mine is rated till 3 GHz.

Before testing the antennas I tested the setup, by measuring the return-loss for a perfect match by putting a 50 ohm load at the DUT port.

In that case we expect zero return loss. When removing the 50 ohm load from the DUT port, resulting in an open end the return loss is max.

And here are the spectrum analyser signals over the full 3GHz range:

And that was a big disappointment, as you can see, there is no measurable signal above 2 GHz.

Although both the noise source and the rf-bridge are rated at 3 GHz, the combination is useless above 2 GHz.

 

For lower frequencies this method works quite nice, I tested a 144 MHz ham radio antenna and a 868 MHz LoRa antenna which test I shortly will add to my  3GHz Spectrum Analyzer - R&S® FPC1000 - Review.

 

This unfortunately concludes the VSWR measurements.

 

Conclusion

This roadtest was quite different from most other roadtests offered at Element14. Since I'm a lucky owner of a spectrum analyser I promised who struggled with too less applicants to give it a try.  The 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 should have warned me. In the end I was only able to do some indicative comparisons, From these results I conclude that the performance of the three provided antennas are quite good.

I'm very disappointed that I wasn't able to measure VSWR at these frequencies with a broadband noise source and rf-bridge.

Anonymous
  • Hi Gerrit,

     

    Nice review! I'm glad you showed signal strength measurements using WIFi, I didn't get a chance to explore that method.

    Your procedure was good for return loss measurement, it's just unfortunate that directional coupler/bridge was not as capable as advertised : (

    Still, all the information in your review will help anyone who wishes to test, using a similar setup but with the coupler/bridge replaced.

    I don't have a noise source, but have been trying to use several different devices to generate spot frequencies. One candidate that I didn't get around to trying, was to use a synthesizer dev-board from TI. It can only generate a limited range, but it needs some code written because I don't have the motherboard that the smaller dev-board is intended to be used with.

  • Good job Garrett, these are very tough to test without a well established RF lab.

     

    DAB

  • Hi Gough,

     

    thanks for your very valuable remarks. I fully agree on the RSSI, as mentioned in my review the numbers are indicative only. I did an average over a number of samples, but far less than 1024. I observed that the variation mainly was within 1 RSSI unit.

    I also agree on your comment on the number of visible networks. In fact I show the effect of a directional antenna like the biquad.

    Reason for the 10dB attenuator is to protect the rf-bridge and SA. The output of the noise source is 0 dBm in some peaks, which is rather high. Also a 20 dB attenuator was placed between the bridge and the SA. I'm pretty sure removing the 10 dB attenuator will not improve the situation.

    I followed the approach of https://pa1ejo.wordpress.com/2018/05/23/swr-bridge-with-noise-source/  a bit.

     

    Sure it is a hard roadtest. I am very curious about the response from Molex.

     

    Gerrit.

  • Dear Gerrit,

     

    Good to see your RoadTest of the antennas - interesting to see you used the RSSI values as a comparison point. In the past, I published a paper as to the variation in RSSI reported by different devices/chipsets as it turns out that it is only used as a "coarse" signal level indication and many devices are not well calibrated or necessarily linear with each other. So you can probably make a relative comparison between antennas but (sadly) not get any absolute values. Resolving small differences is also problematic as many chipsets report RSSI values which vary by several dBm continuously, so the best approach is to take something like 1024 samples and average it. Of course, the longer you take, the more likely some surrounding device might be switched on/off (e.g. your neighbour's microwave) and skew your results.

     

    The number of visible networks is an interesting choice of metric - it's one I did think of but did not choose because it often really depends on a mixture of factors - primarily the signal-to-noise ratio which can be impacted by noise floor, beam angle of the antenna and positioning (whether the antenna sees/receives noise sources). A directional antenna might see less because of the beam angle, but the throughput to the wanted network is likely higher because of a better signal-to-noise even if its absolute signal level is low due to losses in coaxial cable.

     

    It's a hard RoadTest to do - so nice to see your efforts. It's a bummer that your bridge could not deliver any signal at 2.4Ghz - what is the reasoning behind the use of the 10dB attenuator from the noise source? Will it overload the bridge or your spectrum analyzer? Adding 10dB more signal might not really change the situation much looking at your spectrum analyzer plot either way, but just a thought.

     

    - Gough

  • Hi Gerrit,

     

    A great review and very informative both from your text and the photographs you have included. It is really interesting to see how you and approached this task; I've learnt a lot from both of your posts, thank you. Your bi-quad antenna is a work-of-art .

     

    Rod