The Arducopter/Gremsy Solo Operator Thread

[viperhawk (Mark Townsend)]

The time has come to hash this out…

I’ve flown unmanned systems for a living for 18 years now. I built my first Arducopter octo/X8 (VulcanUAV Raven) in 2017 and began heavily using Ardupilot builds for mapping and inspection work around 2020.

That said, around 50% of my work is still centered around creative/photography/filming. I do not use dual operator setups for anything and all filming is done with two Inspire 2’s that still ride with me everywhere I go. I need to get rid of them but still do not have an Arducopter/Gremsy solution that can replace them yet.

The number of good videos that I’ve seen from solo operators using Arducopter/Gremsy for film work is “0”

I’ve only seen a few good examples that came from the Airpeak, Astro, etc.

Please review my example from today and let me know what is wrong. The shutter speed is 1/400 as I’m still waiting for an ND to arrive for this particular lense. That said, I’ve filmed on MANY occassions with shutter speeds in excess of 1/400 without any of this warbling, etc.

If it’s a gimbal gains issue, what should they be for this camera?

If it’s an issue with the stock T3V3 dampers, which should I use?

I’m not able to perform smooth arcs around the subject since my yaw tends to overspeed the camera. I’m sure this is an expo/deadzone thing and I’ll need to experiment. They are incredibly easy to do with the I2 though so I’ll need to get that figured out for these builds.

Arducopter machines should gain more traction in the days ahead and for all of the obvious reasons (proprietary drone problems, RemoteID, cost, etc.).

If you have an Arducopter setup in conjunction with Gremsy…and it is butter smooth for filming…please comment and share your knowledge.

Thanks in advance

Here is the video from today’s test (custom build with a Gremsy T3V3/Sony A7RIV/24mm):

Here is an example of the work I typically do with the Inspire 2:

[viperhawk (Mark Townsend)]

Here are my T3V3 settings:

[Minh_Le (Minh Lê)]

Hi @viperhawk

Sorry for your inconvenience with these gimbals!

There are some Gremsy gimbals’ footage with Sony a7rIV 35mm and 55mm.

Please contact support@gremsy.com for more information.

[viperhawk (Mark Townsend)]

Thanks. Your footage examples look good. I’ve written an email to support.

[viperhawk (Mark Townsend)]

What dampeners and settings?!

I wrote to them and have not received a reply

[tridge]

@viperhawk moving my responses here from facebook.
As I mentioned before, I think you could take advantage of the new filtering options in 4.3 to get a better tune.
Looking at the log you posted ‘2023-04-12 15-22-36.bin’ you can see this in the FFT for the pre-filter gyro:

and this is post-filter:

and this is the actual frequency of the notch during the above:

If we load your parameters into FilterTool we see this:
https://firmware.ardupilot.org/Tools/FilterTool/

The bit of interest is the width of the notch, and the fact it doesn’t quite cover the actual vibration.
The first thing to do is to get the notch properly centered. That would be:

• INS_HNTCH_FREQ = 52.5
• INS_HNTCH_BW = 35
• INS_HNTCH_REF = 0.27

Notice that the frequency is quite a bit higher. My guess is the original tune was (quite correctly!) done without the payload, so the reference throttle and frequency was lower. That is not as important with 4.3 as it was with 4.2 as we now have INS_HNTCH_FM_RAT which allows us to scale the notch down to lower throttle levels. I notice you have that at 0.7 which is fine. I’ve also set a bit wider bandwidth, which does slightly increase phase lag, but you’re not after a really sharp tune, so I think this is worth it.
Next, let’s do a wider notch by using the new triple notch option:

• INS_HNTCH_OPTS = 16

we would end up with this:

which is a better fit for your actual noise.
I’ll be interested to know if that makes a noticible difference. I suspect the difference won’t be large.
Once you post a log with LOG_BITMASK=65535 then we will be better able to look into the PIDs and see if there is an oscillation there that could explain the results.
Also note my suggestion of a neutral density filter. I suspect that would make a bigger difference than the tuning.
The other thing worth trying is EK3_PRIMARY=2 to move to the 3rd IMU. It looks like your 3rd IMU is a bit cleaner than the first two, possibly because of resonance on the isolation foam. Again likely not a big difference, but worth a try.

[viperhawk (Mark Townsend)]

Awesome! Thanks for taking the time to look at this. I’m on the road but will implement and report as soon as I get back.

[viperhawk (Mark Townsend)]

Will I get the “lane switch”/descend/do weird behavior by the IMU change?

[tridge]

based on the data in the one log I have access to, the 3rd IMU is fine, and won’t cause any issues