Testing & Reviewing LoRa® Antennas
Selecting a LoRa antenna is tough. Measuring an antenna well is a hard task and almost no one does it. Many antennas available via the hobby channels report gain or vswr, maybe at a frequency, that they clearly often pretend to measure.
So, since I have some test equipment and I’ve spent quite a while picking LoRa antennas… I decided to review a bunch of them.
I have a product that requires uniform high-quality LoRa communication and deciding on how to send that data has taken a very long time. Start by analyzing the various available antennas.
I have an earlier post just like this. It suffered (greatly) because I had no way to capture response curves so photos of a chart from my small handheld antenna tester were it. Not great. Now I have a usb-controllable tester with on-screen results and a half-dozen more antennas. Hence the update.
What make a good antenna
Radios supply a voltage to an antenna to send and read a voltage from an antenna to receive. We care about how far away a receiver can be and still receive the signal, or how well the radio can receive a signal from far away.
The most efficient power transfer happens when the antenna impedance matches the radio impedance (usually 50 ohms).
If an antenna is not 50 ohms at the frequency you’re using, then power is wasted (sometimes as heat) and component stress may increase. If an antenna is 50 ohms at frequencies you aren’t using then nearby radios could interfere with your reception or the radio could send noise on the wrong frequency (which the FCC really doesn’t like). We want the antenna to be a perfect match in our band and an open circuit outside our band.
The antenna impedance is measured at each frequency and we calculate how much of the radio power is being transferred and how well guarded we are from interference/interfering.
The radiation pattern shows where in 3 dimensions the radio power is going. A ‘perfect’ antenna sends power in a sphere — which is actually wasteful unless the receiver is in the center of the earth or on the moon. All antennas send more power in some directions and less in others — how that is done is critical.
Good commercial antenna manufacturers provide radiation patterns for their antennas. Most do not. Take a look at the very directional pattern shown above to see how important this is.
For my applications, I have two preferred patterns.
There’s a device that gets carried around with an antenna. It needs a uniform XY pattern so no matter how it’s pointed it still gets a good signal.
There’s a base station that communicates with the mobile. The base station is far enough away that a directional antenna can be pointed at the mobile area for additional gain over an omnidirectional pattern.
Nearby Objects and Ground Planes
Any metal (or RF reflective or absorbent) object near an antenna will affect its performance. So, don’t take these measurements as necessarily reflective (!) of how an antenna will work when mounted on your metal device.
In the picture above are two ways to measure this whip antenna. The long coax and the raised platform give some distance from any metal surfaces so the left-hand image is ‘clean’. The right hand image includes a piece of aluminum foil that provides a ground plane (if the foil is a bit smoother).
The reason for including a ‘ground plane’ is that vertical whip antennas are designed to work best with a reflective floor (ground plane) to bounce off. Without that ground plane the design frequency and impedance all wander off.
Even non-metal things that are nearby matter. The Molex flexible dipole is supposed to be stuck on a plastic wall of the chassis and it was measurably better when that was the case.
In my case, there’s a nice plot twist at the end of this article.
Who is being tested
Here’s a picture with the various antennas labeled->
I started by testing each antenna’s impedance through the ‘entire’ RF range (200MHz — 3GHz). This shows if things are totally mismatched or there are also WiFi or Cell bands being covered by the antenna. Each antenna’s chart is in their specific section.
Then, I retested in a narrower range to just look at performance around the 915MHz spot I’m using. The full US ISM band is 902–928 MHz. The below charts were imported into Excel so I could compare the various antennas at one time.
I ran two narrow tests for the antennas that ‘require’ a ground plane. I tested without a ground plane and with a ground plane, so those tests are separated.
The Return Loss chart above shows the how much power each antenna reflects back to the source (figuratively). The return loss is the amount of power wasted — the bigger the number the greater the waste. A returns loss of -10dB (0.1) means that 10% of the power is wasted, -3dB (0.5) means 50% of the power, and so on (power x in dB = 10log10(x)). When you see a huge dip in the charts above that’s a good thing.
Wasting half the power loses 30% of the range.
Amateur radio operators usually use a measurement called VSWR which is based on return loss but is best at 1 and gets progressively worse as the number increases. VSWR is a good indicator of the additional stress on components caused by bad impedance matching. Here’s a comparison of both types of charts.
Trademark: LoRa is a registered trademark of Semtech.