Ever since the introduction of the DJI FPV system, people have been trying to get extra range from them by installing directional antennas. More recently, this became simpler & cleaner with the introduction of tailor-made antennas like the iFlight Crystal HD.
The stock goggles come with four antenna ports, each sporting a stubby, omnidirectional antenna. So on which port should you install the directional antennas? There are several videos on YouTube where people are testing & comparing different configurations (primarily directional on top/omni on bottom, and directional on bottom/omni on top) and trying to reach some conclusions. There tends to be some relatively weak evidence in these informal tests for the directional on top/omni on the bottom leading to better long-range performance. At the same time, advertising pictures for the iFlight Crystal HD usually show them connected to the top, with omnis on the bottom.
I’ve recently purchased some Crystal HDs and went through a thought-process to decide where to install them on my v1 DJI goggles. What I’ll attempt to do here is to document that thought-process, exploring & reasoning about each option based on publicly known information about the goggles, and a healthy dose of educated guessing. I’ll then explore whether what I come up with is consistent with what Joshua Bardwell saw in his tests. While some of this post is specific to the Crystal HDs, the general idea applies to other similar products.
Caveat: I am not affiliated with DJI, nor do I have any insight into the design of the goggles beyond what is publicly available. As such, please consider this post a hopefully entertaining & interesting thought-exercise, not a source-of-truth. This is also not a recommendation to use non-stock antennas on your DJI goggles, nor is it an endorsement for the iFlight Crystal HD product.
What are each of the four antenna ports on the DJI FPV goggles for?
According to documents publicly posted as part of DJI’s FCC filing, the V1 goggles feature two RF chains, only one of which is active at a given time. Each chain uses 3 of the 4 antennas, and the goggles can switch between each chain. Two of the 4 antennas are common to both chains. The following table shows at a very high level what each antenna does as part of each RF chain. Note: The antenna numbering I’m using corresponds to the figure on page 23 in the “SAR Evaluation Report” that is part of the FCC filing. I’m arbitrarily referring to the RF chains as A and B.
RF Chain | Ant 0 (Bottom Right) | Ant 3 (Bottom Left) | Ant 1 (Top Right) | Ant 2 (Top Left) |
A | Transmit & Receive | Receive only | Transmit & Receive | (not used) |
B | Transmit & Receive | Receive only | (not used) | Transmit & Receive |
iFlight Crystal HD antenna arrangement
Next we’ll look at the Crystal HD antennas themselves. iFlight advertises a fairly wide horizontal beam width of 109 degrees (9dBi). How is this achieved? They do it by pointing each of the two directional antennas slightly offset from each other. In other words, the narrow antenna patterns from each of the two antennas combine to give a 109 degree combined beam width. Individually, each of the Crystal HD’s antennas (left or right) is probably a bit over half that beam width. This becomes relevant when thinking about which ports on the goggles to connect them to.
Configuration 1: Crystal HD on top; omnis on the bottom
So let’s see what happens if we connect it this way. This happens to be how they are connected in the product pictures on iFlights website, so it’s natural to start there. Let’s update the table above, adding the antenna assignments.
RF Chain | Ant 0 (Bottom Right) | Ant 3 (Bottom Left) | Ant 1 (Top Right) | Ant 2 (Top Left) |
A | Omni: Transmit & Receive | Omni: Receive only | Directional Right: Transmit & Receive | Directional Left: (not used) |
B | Omni: Transmit & Receive | Omni: Receive only | Directional Right: (not used) | Directional Left: Transmit & Receive |
As you can see, each RF chain (A and B) gets only one of the Crystal HD’s directional antennas. This means only one of the two directional antennas in the Crystal HD is being used at any given time, because only one RF chain is active at any given time. What this implies:
- In order to switch from the slightly right-pointing Crystal HD directional antenna to the slightly left-pointing Crystal HD directional antenna, the goggles need to switch between RF chain A & B.
- Since both left & right Crystal HD directional antennas are connected to ports that can not only receive but also transit, the reverse link (Goggle to Air Unit) can benefit from having access to a directional antenna regardless of which RF chain (A or B) is active.
Based on this, we could say this configuration is basically symmetric — we should be able to expect similar long-range performance from each RF chain. Whether A or B is better would depend on whether the quad is within the beam width of the left or right directional antenna. As the quad moves from left to right (or right to left), a switch in RF chain would at some point be required to get optimal performance.
Configuration 2: Omnis on top; Crystal HD on the bottom
So let’s look at this configuration next. Let’s update the table again, adding the antenna assignments.
RF Chain | Ant 0 (Bottom Right) | Ant 3 (Bottom Left) | Ant 1 (Top Right) | Ant 2 (Top Left) |
A | Directional Right: Transmit & Receive | Directional Left: Receive only | Omni: Transmit & Receive | Omni: (not used) |
B | Directional Right: Transmit & Receive | Directional Left: Receive only | Omni: (not used) | Omni: Transmit & Receive |
As you can see, RF chain B now gets access to BOTH of the Crystal HD’s directional antennas simultaneously (one on the receive-only port, the other on a transmit & receive port), while RF chain A gets access to only one of them (on the receive-only port). What this implies:
- When RF chain B is active, the entire combined 109 degree beam width of the Crystal HD is available. There’s no need to switch RF chains as the quad moves left to right (or right to left) across the beam to maintain optimal forward link (Air Unit to Goggles) performance.
- When RF chain B is active, the reverse link (Goggles to Air Unit) can benefit only from one of of the Crystal HD’s directional antennas, since the other one is on a receive-only port.
- When RF chain A is active, only one of the Crystal HD’s directional antennas is available to receive (similar to configuration 1 with the directional antennas on top), and no directional antennas are available for the reverse link (Goggles to Air unit)
Based on this, in a long-range scenario, I would expect the goggles to maintain RF chain B active most of the time, with little need to switch to chain A. With the quad in the distance, it’s hard to imagine a situation where chain A would ever yield better performance than chain B.
Comparing Configuration 1 & 2
For the long-range scenario, configuration 2 (antennas on the bottom) effectively trades potentially reduced reverse-link performance in RF chain A, for a wider combined beam width in RF chain B. With the wider beam width, at long range, I expect less frequent switching between RF chains. On the other hand, configuration 1 maintains a balance in performance, but probably requires more frequent RF chain switching.
My guess is that there is some downside to RF chain switching being required in order to switch between left & right directional antennas. While I don’t know the criteria DJI uses to determine when to switch, it’s often the case in other RF systems to prefer the already active RF chain while it still yields acceptable performance. Eventually when the estimated performance of the other chain gets significantly better compared to the active chain, only then the switch happens. If this is the case for the goggles, then as a result, I would expect configuration A to yield less consistent performance as the quad crosses from left to right (or right to left) in the distance, since the switch to the other antenna is effectively delayed until whatever the condition is for RF chain switching is met. So I’d expect video link performance to drop to lower levels for configuration 1 when the quad is meandering across the middle of the combined beam. In configuration 2, as stated above, I’d expect the goggles to stay on RF chain B most of the time. With the entire combined Crystal HD directional antenna beam available with no RF chain switching required, I’d expect more consistent performance when the quad is meandering across the middle of the combined beam, and I’d expect performance to remain above that of configuration 1 in that scenario.
What about the reverse link (Goggles to Air Unit)? Without knowing more about how the DJI system works (or doing some elaborate testing), it’s hard to say what the impact of configuration 1 vs. 2 is on performance. Having said that, I would guess that the reverse link is very low bitrate compared to the forward link, and as such, might have enough margin even at long range to work on the omni antennas, so that it really doesn’t make much difference. But again, that’s just a guess.
Comparing to Joshua Bardwell’s test results
I think what Joshua Bardwell has seen in his test is potentially consistent with the theory presented here. Both configurations performed fairly similarly at times, but configuration 1 (directional on top, omnis on bottom) dropped lower at times. It would be interesting to plot the angular position of the quad with respect to the antenna beam alongside the bitrate, to see if the drops correspond to times where the quad had just moved from one side to the other. This might be hard to do, unless Joshua maintained his head in a perfectly fixed position for the duration of the test 🙂
So which configuration is better?
I’d be inclined to favor configuration 2 (directional on bottom, omnis on top) if I was strictly doing LOS long-range over open terrain. That seems like it might give the most consistent performance.
On the other hand, I like the symmetry inherent in configuration 1 (directional on top, omnis on bottom). This configuration might conceivably (I have not tested this yet) do better when flying in closer proximity outside the directional beam (and potentially taking better advantage of a multipath environment), with 2 of the 3 antennas on each chain being omnis, while still doing pretty well at longer ranges.
Of course it’s possible that there are other factors at play that I’m not considering that might tip the balance. I’d say try both, and if it works well enough for you, stick with it and focus on enjoying FPV flying 🙂