Servers by jDrones

Failsafe Changes

I’m using crossfire, and now w/ rc2, I get constant GCS failsafes. I get them at times when I have not lost telemetry. The signal hasn’t changed, my MP screen is still updating.
FYI, the telemetry signal level is always detected quite low when using crossfire… About 20% is actually the max. It has something to do w/ Crossfire not sending as many packets (or frames or something) as MP is expecting… but it has always been this way.


OK, thanks for the report. So it seems like the telemetry failsafe timeout might be too short for the radio you’re using. I’ll add this to the to-do list but for now you might want to disable the GCS failsafe using the FS_GCS_ENABLE parameter.

1 Like

IIRC, the GCS heartbeat timeout is 5 seconds. That is already a fairly long time for something that’s expected at 1hz. But probably could be a little longer to account for less than ideal telemetry links. We just can’t keep making it longer or it defeats the purpose. At some point, we can’t work around poor hardware with code.

So is that what triggers the failsafe? No heartbeat for over 5 seconds?
Telemetry is definitely updating faster than that… but they’ve changed the rate at which everything is sent, and they don’t send the full set of mavlink data that a traditional telemetry modem would send. I’d imagine it’s something they can change.

Crossfire is great in general… I can fly a little 130mm copter 8km away and lose video before any sign of RC signal loss. Get telemetry over the same link, no additional modems to add weight, etc… But it was obvious from the beginning, mavlink users are not their primary market and their implementation of it is a bit odd.

I’ve already got a ticket open with them. I’ll mention the issue is by design on this end, and maybe they should look in to the changes that were made.

Yep. Just double checked and it is indeed 5 seconds. So if they’ve mucked with the rates, that could certainly be causing this. Typical heartbeat is 1hz.

Thanks. I’ll see if they can make any changes on their end.
It’s not really an issue for me, as I didn’t need GCS failsafe anyway. I’m using sbus for control. But they actually recommend mavlink users use RC over mavlink for control, and people doing that need GCS failsafe…

@Pedals2Paddles, I guess we could also consider changing the failsafe so that the timer is reset by any message coming from the GCS instead of just the heartbeat.

My concern with that is we’re potentially allowing a very faulty GCS link. There could be 90% packet loss, but one got through so no failsafe.

1 Like
Servers by jDrones