search

parallaxinc / Flight-Controller

Quadcopter Flight Controller
Other
24 stars 20 forks source link

Improve telemetry communications #23

Closed MatzElectronics closed 7 years ago

MatzElectronics commented 7 years ago

Implement xon-xoff in FTDI chip, GS, and FC. This is a breaking change.

JasonDorie commented 7 years ago

XOn / Off is software only, so the FTDI shouldn't need to be involved. Not sure what the benefit would be though. At some point, packets will get lost. I'd rather they were the old packets, not the new ones. Reduce the XBee retry counts, and possibly switch to using API mode with the XBee instead of pass-through (if it can be done without requiring additional pins).

J

On Friday, September 16, 2016, Matthew Matz notifications@github.com wrote:

Implement xon-xoff in FTDI chip, GS, and FC. This is a breaking change.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/parallaxinc/Flight-Controller/issues/23, or mute the thread https://github.com/notifications/unsubscribe-auth/ANak_nbLo8ghIDhfQphw0LicSbZbI3dXks5qq3iUgaJpZM4J_gwa .

MatzElectronics commented 7 years ago

Open to suggestions, or even calling where it's at good enough...

JasonDorie commented 7 years ago

I just want to know what you expect to get from enabling flow control. It would mean not corrupting packets on slow delivery, but the packet loss itself is from the XBee failing transmission. If loss happens, flow control will make new data wait for old data to finish going through, but for realtime display, it's more relevant to have new data prioritized over old data. It's never going to be a lossless connection.

I was thinking it would be worthwhile breaking the comms and making new packet types that were packed. I'm using 16 bits to transmit each radio value, even though they're only 11 bits each. Ditto for motor values, and others could be reduced similarly. It would cost a bit of code time to pack them, but there'd be less to send, so more room for retries.

J

On Saturday, September 17, 2016, Matthew Matz notifications@github.com wrote:

Open to suggestions, or even calling where it's at good enough...

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/parallaxinc/Flight-Controller/issues/23#issuecomment-247757207, or mute the thread https://github.com/notifications/unsubscribe-auth/ANak_gyu9xAnekh4PBDJ3hKcQIDgm_Vbks5qq6Q0gaJpZM4J_gwa .

JasonDorie commented 7 years ago

Telemetry comm is now using less data per packet, and packet transmit rate has been reduced by 20% (coinciding with 20% reduction in update rate), giving more time for retries. In theory, this should improve reliability - can one of you check this?

MatzElectronics commented 7 years ago

I might be able to tomorrow, I'll let you know either way

On Oct 4, 2016 9:00 PM, "Jason Dorie" notifications@github.com wrote:

Telemetry comm is now using less data per packet, and packet transmit rate has been reduced by 20% (coinciding with 20% reduction in update rate), giving more time for retries. In theory, this should improve reliability - can one of you check this?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/parallaxinc/Flight-Controller/issues/23#issuecomment-251576354, or mute the thread https://github.com/notifications/unsubscribe-auth/AS0quClYUvbcLtXVGbqcg0p_6Z6nUq5mks5qwyDlgaJpZM4J_gwa .

MatzElectronics commented 7 years ago

Closing - addressed in #40.