In this blog post, I summarise my experience with FPV drones in reverse chronological order.

2026-11-08, Sunday

I flew my drone in fairly cold (5 deg C), humid, and foggy autumn conditions. Given the excessive cold, I could only fly three LiPo batteries before I had to abandon my trip and warm up. I did not notice any significant reduction in battery capacity, although the initial voltage drop seemed slightly quicker on older (more used) batteries. I did not test my LiIon batteries, only my LiPo ones. Apart from the first, I warmed up the subsequent batteries in my jacket pocket, which appeared to help a bit. I wore thin finger gloves while flying, which worked without dexterity issues. However, they are definitely too thin for colder weather. I did not find a straightforward way to wear my goggles over my warm hat. Perhaps I need a thinner or shorter hat for this specific purpose.

I tested mid-range flying and revamped my OSD configuration for this occasion. I added the rate profile name OSD element (“Free”, “Cine”, “Wind”; useful), total distance flown (an interesting statistic, but perhaps misleading in the long run), and tweaked the location of other OSD elements. The maximum distance per capacity efficiency is achieved at a flying speed around 50-60 km/h with my low-pitch (2.75-inch) props. The efficiency (capacity per km) OSD indicator is particularly useful for this type of flying. The efficiency changes dramatically when flying against the wind compared with flying with the wind. On this day, flying against the wind (a steady 15 km/h or so) resulted in a 60-70 % efficiency penalty. Fortunately, it was the outward journey that was against the wind, so the return leg was very fast and efficient.

On my third flight, I flew through the low-hanging clouds, which resulted in a misty lens and reduced vision. Subsequent high-speed flying did not dry the lens. I must find a solution for this when flying for longer durations.

2026-11-07, Saturday

I tested my new freestyle and cinematic rates. The higher maximum rate and lower expo make the freestyle rate snappier and more dynamic, which I like. The cinematic rate is noticeably smooth around centre stick (ideal for cinematic), even though its maximum rate is now higher. Overall, I am quite happy with my new rates.

The HQProp Freestyle 5043 V2S (5x4.3x3) props are significantly louder and deliver higher top speed, as expected. However, low-throttle authority is reduced, and perhaps total flight time too. I still need to try them more for actual freestyle flying. I did not like that their M5 shaft holes are very tight, which makes mounting and removing them a challenge.

The safest and by far easiest way to land the drone is indeed to hover at around 10-20 cm above the ground or landing pad and simply disarm the drone.

2026-11-02, Monday

I overhauled and tested my rate profiles in the simulator, including my freestyle, cinematic, and new “windy day” profile mentioned below. I noticed that I might have to use even higher-pitch props for more dynamic freestyle flying. I will try the HQprop 5x4.3x3 props next time. Switching from 2.8 to 4.3-inch pitch makes a significant difference in agility and top-end speed in the simulator.

2026-11-01, Sunday

I have been flying my drone in very windy (almost stormy) conditions recently, with wind speeds up to 35 km/h. I noticed that the recorded footage contains micro-twitches that the Gyroflow software fails to easily smooth out. As it turns out, these micro-twitches are partially caused by my own stick inputs, correcting the drone’s position as it is moved around by the wind, and partially by wind gusts, i.e. the actual wind-induced motion of the drone.

The easiest way to diagnose the cause of the twitches is to look at the gyro data recorded in the drone’s HD footage, as produced by the DJI O4 Pro system. Stick input micro-corrections are visible as steps in the quaternion traces, while wind-induced drone motion looks like a noisy random walk around a mean trend, visually comparable to a stock market trace. That is because many random gust eddies hit the drone from different directions, a motion that the PID controller tries to dampen.

The solution to the micro-twitches in the footage due to stick input micro-corrections is to use softer rates when flying in windy conditions. That means rates with lower centre stick sensitivity, maximum rate, and possibly higher expo. This way, the stick micro-corrections should be smoother, and jerkiness should be avoided. I have created a dedicated “windy” rate profile with significantly softer rates - to be tested next time. I also used that opportunity to overhaul my other rate profiles (freestyle & cinematic)

A possible solution to the micro-twitches caused by actual drone motion due to the wind gusts is to use the optical flow-based stabilisation implemented in newer Gyroflow versions - also to be tested. Using a low-pass filter (15-50 Hz) on the gyro quaternion data also works, but negatively affects the overall dynamics of the footage. So it is best to avoid it.

2026-10-24, Saturday

The AxisFlying BlackBird BB39 propellers come in two versions with different pitches. BB39 V2 has a 4.35-inch pitch and is made of carbon fibre, while the BB39 V3 (current version) has a lower 2.6-inch pitch and is made of polycarbonate. Hence, it is important to buy the newer V3 version.

For low-pitch props like the AxisFlying BlackBird BB39 V3 (4.9x2.6x3) or the Gemfan Proxy 5127.5 (5.1x2.75x3), a Dynamic Idle value of 40 (i.e. 4000 RPM) is ideal. For somewhat higher-pitch props like the HQProp HeadsUp FPV R38, a lower Dynamic Idle value around 35 (i.e. 3500 RPM) is optimal. The higher the prop pitch, the lower the Dynamic Idle value. On very windy days, you can increase the setting slightly to compensate. A static idle around 5.5 % (default Betaflight setting) is good. It is only used just after arming or when DShot telemetry becomes unavailable.

To improve flight performance on very windy days, increase the D-term (gust resistance) and I-term (steady wind resistance) slightly (+5 points) in the PID loop settings.

2026-09-28, Monday

Damaged LiPo batteries have significantly reduced capacity - about half, in my experience. If the average cell voltage is already down to 3.8 V after a minute of flying, something must be wrong with the LiPo. Damaged cells have zero voltage after some storage time.

2026-09-27, Sunday

The low-pitch AxisFlying BlackBird BB39 props are pretty quiet and handle strong wind well. Especially if there is decent wind and traffic noise around, the drone noise becomes a non-issue if you are at least 10-20 metres away. This excludes severe throttle punches, of course.

2026-09-26, Saturday

Today was the first time I flew with another FPV drone nearby, with the operator a few hundred metres away and outside verbal range. However, I avoided flying simultaneously to avoid interference, as you never know what control and video systems the other person uses. ELRS 2.4 GHz supports hundreds of drones in the air simultaneously, so I was not too worried about this. Things might be more complicated with the DJI digital video system and its 5.8 GHz channel allocation, which is limited.

I tested very low-pitch props, namely the AxisFlying BlackBird BB39 4926 (4.9x2.6x3). In fact, these are the lowest-pitch props I currently own. They have pretty decent flight performance and propwash handling. Perhaps there is a slight lack of thrust (could also be too high throttle expo). The top speed is reduced, as expected. They are noticeably quieter than the HQProp R38 Racing Props I flew before, which is nice. I get around 7-10 min flight time on 1600 mAh. This might be a bit lower than the 12 minutes I got with slightly steeper props before, but it depends on the flying style.

Not using a drone landing pad draws less attention and makes it less obvious that a drone is in the air.

Landing becomes very easy when you realise that you simply disarm the drone when you are low enough over your target area.

2026-09-05, Saturday

I upgraded my ExpressLRS control link firmware to version 4.1.0. The Dynamic Power functionality now lowers the emitted power reliably and quickly. It works very well in practice and feels very responsive. The lowering was too slow in the previous version (4.0.x).

The HQProp R38 5138 Racing Props (5.1x3.8x3) fly well and offer more top-end speed than my lower-pitch freestyle props.

Sometimes Betaflight does not seem to remember that I switched rate profiles (freestyle -> cinematic). It might be worth adding a rate indicator to the OSD.