MultiWii

Bye bye to my EzUHF with 1.2GHz 3rd harmonic and expensive RX.

Removed: 'ACK' only for MSP_SET_RAW_RC (useless byte and not in TDD sync)
Extended MSP_SET_RAW_RC timeout to 2 or 3sec. with a uint16_t counter (all ch to 1500 is GPS hold failsafe !)
I added a more *TDD efficient* telemetry streaming task (TDD sync)
I can use original MSP with any software and over modems (it temporaly and automaticly disable the simple telemetry streaming.)

code : https://github.com/Serveurperso/multiwi ... eam_shared
from last MultiWii 2.3 master

Short test flight with telemetry streaming enabled, almost no latency : https://www.youtube.com/watch?v=U8N6o8dga6w
I use a SBUS to MSP_SET_RAW_RC relay embedded inside my RF relay for long range flight (house roof/car roof...)

SiK last unmodded firmware 1.9 master @ 33ms TDM window

Because I use a modded 2W shireen amp on 878-888 (900MHz saw removed, and external 888MHz murata bandpass) I get more range vs. EzUHF and no harmonic to 1.2GHz and less noise floor on all quad.

Pascal

Hi Pascal,
It looks great. Perhaps you can try my MW modded SiK firmware, which tries to align radio packets with MSP_ packets, and injects radio status packets into the stream. (https://code.google.com/p/sik-multiwii/)
BTW, could you share some info about your octo diversity ?
Regards,
EOS

Hi EOSBandi,

Sorry for my bad english. I just read your message here and cloned your repo to check your work.

The 8 diversity is basically a "solid state" drone tracker with high speed analog switch between 8 tuners(+LNA, filters and now with 8 circular wireless CPATCH12 to get advantage of two world, long range tracker / urban or three non LOS flight).
I make a sort of "adaptive diversity" with "servos calibration" for RSSI level and video level -> the high speed switching is not noticeable on video (not opensource at this time)

I already make my own version of the SiK firmware because I don't like the embedded TDM (TDD).
With embedded TDD we loss a lot of RF duty cycle -> a lot of bandwitch and add a lot of latency because it's not smooth and not in-sync.
TDM+golay get counterproductive on a 8051, it need a real RTC for efficient timing and all calculation during TX/RX.

I returned to a simple and more reliable low level transparent serial link, external TDD, for maximum performance, I get near 2 * 50% TX/RX duty cycle.
In pratical I get 2* more framerate (30Hz) and/or air-rate/3 !!! and I get >2* the range and less packet lost.

Now there is no difference with any commercial radiocontrol (ImmersionRC EzUHF...) + I get telemetry (my own frame reply to msp raw serial) with ALL interesting data from copter.

I added 2*2 tail bytes to monitor remote/local noisefloor+rssi.
I can draw the 3D UAV attitude on ground : all is very smooth at 30Hz I don't need OSD anymore.
Working : I can use few HM-TRP to use multiples frequency in same time (reliability better than FHSS because it's a 3 frequency diversity, but it need 3 UART on a mega)

Work/test in progress :
I make a bi-directionnal amp for my LRS with the 7W module and RF switch (cheaper and better than Shireen amp for this usage)
I must add 1 header byte to control the EzRADIOPRO hopping channel, if I want remake a real FHSS with external timing control with a real RTC (and GPS time for very long random jump pattern)

Pascal

Edit https://github.com/Serveurperso
modded SiK and Multiwii inside my branch

tdm.c



There is a 2uS hole between radio_preamble_detected() and radio_receive_in_progress() this can make false negative.
I had to take this into account in my implementation. (TX during RX issue, this make rare random freeze)

https://www.youtube.com/watch?v=w3RbhVvTGFI (french with english bubble)

Now it's ready for FHSS:D