ArduPilot EU Dev Call 2025-01-08

Attendees (max): 7

UTC0705

Peter: Better log nan’s when DesPitch isn’t valid.
Michelle: Can the log plotters cope?
George: I can test.
P: Also check what currently happens re. reinitialization now, with navpitch


UTC0715

Sid: Temperature shouldn’t be affected by imu rates.
Andy: It was more a matter of consistency: we dont’ run fast imu on the initial temperature calibration.

  • RAM and CPU are not an immediate concern.
  • We can leave this for now, then.

UTC0720

A: I had baro issues with a Whoop, which I chased down to this driver.
P: It would be nice if you had split the EXT_SENSOR change into another PR.

  • Could you test some more boards, if possible?

UTC0733

Sid: This issue can bring down the whole system, if the conditions align, and can’t be turned off before boot.

  • It could affect 4.6 and needs backport and affects default firmwares.

UTC0736

P: Could Michelle test it?
M: I have to be away this week unfortunately.
G: Could this be a side-effect of the ground-effect compensator?
A: It does work a lot better in real flights.
P: Okay, needs a confirmation by the code owners.

  • Besides some other review comments.

UTC0750

S: Fixing a logic level during boot could very well bring down other hardware, while fixing some.

  • It’s preferrable to add external pullup/down resistors where needed, per case.
  • It’s not easy to fix this via a user option

A: except for a custom build server build.

  • A new wiki page on servo connection good practices would be nice.

UTC0809

A: This is ready to go now.
P: Needs rebase on master, can’t be done automatically.
A: Done
A: This feature is enabled by default on NAND-chip boards.

[Code review underway]

P: Why is there another JEDEC driver? We already have some.
A: None of the existing drivers are factored enough to be usable, unfortunately.
P: The filesystem driver shouldn’t be handling the SPI transfers.
A: Agreed; We are doing the same in the other backends currently, though.
P: We already test other JEDEC backends in SITL. We should be able to test LittleFS as well.
A: Thomas mentioned he’d do the simulation after this feature is done.

[Discussion on logging throughput. Seems slower than regular SD, but the actual performance isn’t very clear. Mostly depends on our calling code overhead.]

[Discussion on binary size changes. Nothing too unexpected, but a couple of cases need to be looked at closer.]


UTC0909

P: Build failing because of binary space.

  • We’d better freeze what Pixhawk1-1M can support, at least in terms of compasses.