Designing and building a great entry-level patch antenna for FPV

When we ambitiously set out to develop the best FPV goggles World has ever seen (Orqa FPV.One), we figured that they should at least be bundled with decent enough antennas.

Given the privilege of having an amazing RF engineer in our team (we call him The Doctor), we’ve gotten to work of testing and measuring the performance of the existing patch antennas on the market. Dissatisfied with what we could find in the entry-level segment, we have set out to develop our own. And this is a story about how Orqa FPV.P1 was born.

About patch antennas

Patch antenna is a microstrip antenna originally described by Howell in 1972, a type of directed radio antenna consisting of two sheets of metal mounted on top of each other.
 
A smaller flat rectangular sheet called “patch” is positioned over a larger sheet of metal called “ground plane”, forming a resonant piece of microstrip transmission line with a length of approximately half the wavelength of the radio waves.

The patch antenna is used at frequencies between 300 MHz and 300 GHz, as the wavelengths are short enough for the patches to be small enough to be practical.

Patch antennas are widely used because, in their simplest form, they can easily be fabricated as specially designed printed circuit boards (PCBs).

Due to the elegance of their design, and simplicity of fabrication, these single-PCB patch antennas are currently considered to be the entry-level category of patch antennas.

Single-PCB antennas are very simple to manufacture by using a connector and just one 2-layer PCB. They are designed so the patch is fabricated on one layer of the PCB (the top one), and the reflector on the other (the bottom one).

What makes a good antenna?

When we started designing Orqa FPV.P1 we set out to make the best patch antenna available in the entry-level segment. But what makes a good antenna?

First of all, you want to optimise the design so the antenna performs well across the entire frequency range you might be using it in. In our case, that’s between 5.658 MHz (R Band / CH1) and 5.945 MHz (E Band / CH8).

A well performing single-PCB antenna will have a Voltage Standing Wave Ratio (VSWR) of well below 1.4 across the frequency range you will use it in, and a good cross-polarisation attenuation. Let’s go over these two aspects one by one.

VSWR

You can think of the VSWR as a proxy to power transmission efficiency (how well the antenna delivers the RF energy to the receiver), with 1 being the ideal case (i.e. 100% of RF energy is delivered to the receiver). It also affects the noise level of the low noise amplifier (LNA), which is important because the LNAs (the first transistor after the antenna) used in the Rx modules are designed to have the minimum noise at 50 Ohm (VSWR = 1), so you want to be as close as possible to that, otherwise you have more noise, and less sensitivity.

The VSWR values below 1.4 are considered to be within the optimal range, 1.4 to 2.0 is considered to be usable, while everything above 2.0 should be avoided.

Cross-polarisation attenuation

There are many advantages to circular polarisation of radio waves, as opposed to linear polarisation, which is why most of the FPV antennas are designed so they emit and receive on one of the two circular polarisations: left hand (LHCP) or right hand (RHCP).

One of the advantages of circular polarisation is greater resilience to multipathing: a phenomenon in which the primary signal and the reflected signal reach the receiver (nearly) simultaneously, and thus interfere with each other.

With the linear polarised antennas, the signal reflection is in the same polarisation, and thus can cause serious multipathing issues. This is much less of an issue with circular polarised antennas, as the signals in circular polarisation change polarity on reflection, which makes circular polarised antennas more resilient to multipathing, as they can attenuate cross-polar signals (signals of opposite polarisation) quite efficiently.

Typically, radiation pattern graphs of circular polarised antennas show the radiation curves for both polarisations (co-polar, for the antenna’s designed polarisation, and cross-polar, for the antenna’s opposite polarisation). For good multipathing resillience, you want the cross-polar pattern to show sufficiently larger attenuation compared with the co-polar pattern, at any given angle.

The difference between the attenuation of the two patterns at a given angle is given by Cross Polar Discrimination (XPD), illustrated on the image below.

Cross Polar Discrimination [Credit: Halberd Bastion Consulting]

So what’s a good enough XPD for an FPV antenna? Across the frequency band typically used in FPV, signal-to-noise ratio for NTSC video needs to be at least 10 dB. Given that the wave reflected off the ground can lose even 90% (10 dB attenuation) or more of its power, this implies that, as a rule of thumb, a 5 dB worst-case XPD should be more than enough for a solid multipathing resilience (unless the signal reflects off a flat metal surface, in which case the loss will be significantly smaller). This means that an FPV antenna with a 5dB worst-case XPD is quite resistant to multipathing.

Enter Orqa FPV.P1

The RF design of P1 is finely tuned so that the optimum range in terms of VSWR is exactly between 5.65GHz and 5.97GHz, with the best results between 5.7GHz and 5.9GHz, and peak performance at 5.8GHz (5.803 GHz, to be exact).

At 5.803GHz, the P1 measures an exceptional VSWR of 1.174 with a good level of worst-case XPD of 8.7 dB.

Frequency vs. VSWR for the Orqa FPV.P1 patch antenna

As we can see on the P1’s Frequency vs. VSWR graph above, between 5.7GHz and 5.9GHz, the VSWR is well below 1.2. Towards the edges of this band, VSWR goes even below 1.1, but with weaker XPD compared to 5.803GHz: at 5.7 GHz we have VSWR of around 1.1, but with the worst-case XPD of around 4.6 dB (measured at the -60º angle, i.e. the very edge of the coverage area), while at 5.9 GHz we have a VSWR of around 1.05, and the worst-case XPD of 5.3 dB (also measured at -60º).

Across the entire intended frequency range, P1 measures VSWR well within the limits of the optimal range (below 1.4), with very good XPD levels.

Antenna production

This is all good when you’re simulating the performance of the antenna’s design, but how does this design translate to a real-world product?

As we said earlier, single-PCB patch antennas are very simple to manufacture: you basically need to fabricate the PCB, and solder the connector. What could possibly go wrong, right? Well, it turns out there’s always something that can go wrong…

When we were designing the P1, after we have carefully chosen a good and reliable PCB fab, we took what specs of their PCB material we could find and made assumptions about the specs we could not, and The Doctor then plugged all of this in his calculations while designing the antenna.

When the first prototypes arrived, we happily took them to the lab, expecting to measure a perfect antenna performance, as the calculations would suggest. But (surprise, surprise), instead of showing optimum performance at 5.8 GHz, as designed, the antenna behaved as if we designed it for 5.6 GHz: imagine the entire Frequency vs. VSWR graph above shifted 200 MHz to the left — this antenna would perform poorly on all frequencies above 5.8 GHz.

Damn. What’s wrong? Turns out that the spec about PCB thickness that the fab provided was off by 0.1mm from the actual thickness of the PCB that arrived, and that our assumptions about the PCB material properties (which we had to make, since our fab did not provide a precise enough specs) were also off.

However, the measurements that we made managed to give us enough data to calibrate our assumptions about the actual material specs, and redesign the antenna which will perform as intended. Doctor went back to the drawing board, made those changes, we did another prototype run, and voila! This new prototype worked exactly as we expected it to work.

Conclusion

Before we set out to build the P1, we were confused as to why some of the best brands in the hobby were selling the single-PCB patch antennas that were simply not performing as well as we would expect them to.

For example, when we measured one of the products sold under the brand that we have utmost respect for, the measurements implied that they have designed it for the 5.6 GHz, which defies all logic. Surely, they would know how to design a simple patch antenna?

But after the experience with the design and manufacture of P1, it’s much clearer how this happens: you design your antenna, you may even measure the performance of the prototype, and do a run or two. But then all of a sudden, your PCB fab simply changes the supplier of the PCB material without notifying you, and before you know it, your superbly designed antenna is now an underperforming weak link in your customers’ RF stack.

The moral of the story is: with each production run, check with your supplier whether there have been any changes of the materials, and verify this by measurements. Better spend a bit more time in the lab to be sure, than to disappoint your customers.

If you want to buy Orqa FPV.P1, you can find it here.

Tenzor Hardware Innovation Lab is a full-stack innovation and hardware engineering lab serving large corporate clients and startups.

Orqa is a technology company on a mission to build premium FPV products, and take FPV tech into the 21st century.