Thierry,
I see you are "busy".30x30 mm very small octocopter.Oh, ho,.............you gave me an ideea.Can you design a setup for a micro tricopter....having one motor tilt with servo.
This is one I dream to make one but I don't have the board/program.

Another thing.If you have some spare time.
Can you investigate.....if I can use a delay routine(which time) in place of "wait for calibration end " on A7105.

Originally Posted by midelic

Thierry,
Can you investigate.....if I can use a delay routine(which time) in place of "wait for calibration end " on A7105.

"Yes we can..."
"IF Filter Bank Calibration" takes 2.3ms
"VCO Bank Calibrate channel 0x00" takes 90us
"VCO Bank Calibrate channel 0xA0" takes 180us

btw, I found a bug in my RX code so that channel 5 doesn't work (mask not well initialized)
replace
" movlwf 0x01, maskServo5Portc ;portc1"
with
" movlwf 0x02, maskServo5Portc ;portc1"

Thierry,Dave,
Finally my module is WORKING,BINDING!
Thank you very very very much.............both of you!

thierry, that is one cool looking pcb. in addition to octo would be great for tri, quad, or hex. 2502 is my favorite fet too. kreatures similar "krefly" multicopter board based on frsky protocol was very exciting. but looks like it was never finished. when do you expect to get that running?

Originally Posted by midelic

Thierry,Dave,
Finally my module is WORKING,BINDING!
Thank you very very very much.............both of you!

no problemo. i will pm a mailing address to send the check to.

Thierry,
How about a program for a mini tricopter with brushed motors?I have this Syma X1 microquad board.It uses flysky compatible.I can remove the original MCU and wired an dead bud style m8 mcu programmed for a minitri........too much dreaming.
Dave,
Don't look too much in the eyes of that funny guy ...you'll get ...dizzy.Lol!

Congratulation Midelic !

Well I have just ordered some L3G4200 gyro sensors (2.5$/piece at aliexpress, l3g4200 looks like a clone or perhaps better than itg3200) but I think it will take one month or more to get them (order processing + china post ...)

I have worked again on pcb to have something with better look.

pic18f14k22 has not enougth pwm outputs for 8 motors so I shall do it with software and interrupts... if it does not work I'll draw a new pcb with bigger pic, probably a 18f26j13 (10 pwm outputs, 28 pins). I have looked at H36 mini quadcopter motor output and pwm is at about 670Hz but I don't know pwm resolution.
I'll try 400Hz pwm frequency with 8bits resolution, that means a 400x256 = 102khz interrupt. With an instruction clock frequency of 16Mhz(max for this pic), I think each interrupt will take about 4us (about 64 instructions) , so pic will spend about 40% of his time (4us*102khz = 0.4s) to update pwm, so 60% is left to get tx packets, well, it should work.

taking into account ic delivery time, programming, tests, ...I think it should work in 2 months.
Then we can imagine different programs for tri, hexa, ...

Therry ,
Ask Dave about that,he made a program for converting PWM signal(brushed motors) to servo(brushless) for a V939/V929 board.So he studied all this and he can help you.
See also:
http://www.rcgroups.com/forums/showthread.php?t=1746379
About octo-copter Dave build one with a 8,90USD, V939 board.....You can use also a V929 board and I'm sure you can use the Vitality board for an octocopter.
http://www.rcgroups.com/forums/showthread.php?t=1710948
IMHO the real challenge would be a mini-tri because requires servo signal for controlling the 3 rd motor angle.

Good luck with your project.

I think a tri copter is easier to do because you only have 3 motors/ pwm (pic18f14k22 has allready 3 pwm outputs, so easier to program) and mixing is also simplified because yaw control and individual motor speed control are not really linked.
Yes, with a mini quad copter board you can do an octocopter with motors in parallel (btw this way, you can do copters with 12, 16, ...4xN motors) just like Dave has done but you cannot do counter rotating propeller octocopter because individual motor control is different and I think better than a quad:
with a quad to turn (yaw axis) you slow motors diametrically opposite, and accelerate the other ones... if you do quick and rapid turns you'll notice sometimes an oscillation on pitch/roll axis and I think it is related to the fact that 2 motors are really slow to have good control...
In a contra propeller configuration you just slow upper motors and accelerate the others (or vice-versa) so this can't happen.
What's more studdies and tests have shown that a contra propeller configuration is 5% to 15% more efficient than 2 separate propellers at same speed.
However with turboprop engines mechanical parts (gears) are complicated and does not last long, that's why you will rarely see that (russian have done pretty good planes however), but with independant electric motors we don't have these drawbacks

Thierry,
One question.I see in your code:
-------------------
;init table read channel sequence
---------------------------
movlwf 0x01, cptChannel ;init channel counter (goes from 16 to 1)
------------------------
You said the counter it goes from 16 to 1 and you initialized the counter from 0x01 not 0x10(16 decimal)
Is it a typo error or something else?Because the following related code will decrement the counter.

in procedure readNextChannelNumber, I decrement fisrt counter then test if equal to 0. if yes, I reset counter to 16.
So initialising counter to 1 in main procedure means it is then initialised to 16 in readNextChannelNumber procedure...
Sometimes to have more efficient code (assembler) you need to do things like that...

Thanks,
Mind blowing.
It was a curious code....decrement counter test with zero and after jump(skip reset channel sequence) to read table channels without even initializing table read channels pointers.
Why not call directly a normal table channels read routine?
Like I said mind blowing....pain in the brain.
Thanks for info!
I understand it now ,very smart move.

Originally Posted by midelic

Why not call directly a normal table channels read routine?

well, that's what I do: I call a "table read" but I initialize pointer (to the table) just once at the beginning of sequence. Then pointer is self-incremented each time I read in table

Yes I fully understand it now,ingenious.
It remains for me to test the PPM decoder code.The rest is ready and not complicated.If it is working, in few days is all over.

Hi,

I'm a bit ( a lot) slower than you. I'm still trying to catch a packet without success... I noticed a difference between the original RX and the one set by me. When I ask 0x22 register (0x62) it's return 0x05 rather than 0x07. Could it be the bad calibration or is it normal?

Edit : I finaly got a package. Register 0x22 still give 0x05. I assumed that it doesn't matter.

Thanks