[Paparazzi-devel] Re: Paparazzi UAV Project

[Hugo Vincent]

Hello again Antoine,

I forgot to ask you what your main flight sensors are now, with the 
CerfBoard configuration? Are you still using the Infrared sensors, or 
have you moved across to the inertial type, gyros and accelerometers?

On 16/04/2004, at 7:42 PM, address@hidden wrote:

> Selon Hugo Vincent <address@hidden>:
>
> Hi Hugo
>
> > Hi Antoine,
> >
> > Thanks for your prompt reply. I will check the CVS again in a week or
> > two, and look forward to seeing what you have been up to.
> >
> > Would it be fair to say that you are moving more to the AUTOPILOT
> > topology - I mean with the Linux board and so on?
> >
> > Do you think there is still a need for two AVRs when you have a
> > CerfBoard? What sort of interface are you using to talk to the AVR 
> > from
> > the ARM CPU. (BTW, XScale == (ARM + extensions))
>
> I Like the idea of having two avrs:
>
>  - For robustness : the code in the small avr is short, well tested 
> and frozen.
>    If you put a bug in the big avr where the code is bigger and more
> complicated, you will still get manual control, health supervision and 
> failsafe.
>
>  - the parameters of the small MCU are tuned during the first test 
> flight of an
> airframe and never changed after that. So, no risk of trashing these 
> vital
> settings when playing with autopilot gains and flashing the big MCU
>
>  - another advantage is that it gives alot of spare "power" in the 
> small avr to
> implement PCM decoding ( I think we really need that ) and lets you be 
> less
> "realtime" in the big avr.
>
>  - we want to keep developing the avr-only system because is is 
> simple, cheap,
> light and has a low power comsumption. It is very well suited for the 
> smallest
> aircraft. We intend to fit one in a 300g airframe.

wow, 300g!!
Like i said, i am new to RC, so: what is the advantage of PCM? just 
better range and signal quality? or something more subtle...

> We are using RS232 to talk between the big avr and the cerfboard.
>
> > I am going to buy a cheap airframe real soon -- do you recommend the
> > Twin Star? I am kind of put off by the twin motors.
>
> The twinstar is a good airframe. It flies well and is very "heavy 
> duty". The
> "high wing - twin engine" configuration has advantages and drawbacks.
>
> advantages : it leaves space in the fuselage - namely in the nose of 
> the
> aircraft - it is very handy to put the camera and the video 
> transmitter.
> it protects the motors and propeller during belly landings.
> it's easy to hand launch
>
> drawbacks : if you want to use high end motor, you'll have to buy 
> two...
> if you want to use brushless motors you'll need two controllers. 
> Generally
> speaking, one big motor has a better efficiency than two small.
>
> > Approximately what
> > is the payload weight capacity of that airframe?
>
> well... it depends. With a 8*2400mAh nicd, it weights about 1300g. We 
> have flow
> one up to 1600g.
> I would say payload is between 200g and 500g depending on battery 
> weight.

ok, how much of the 1300g is batteries?

> > Also, what
> > approximately is the battery life like?
>
> What do you mean life ? for flight duration, with the stock electric 
> motor, we
> get 17 minutes with a 8x2400mAh nicd packs. We also have a 8x3300 nimh 
> (same
> weight as previous pack) which give 25 minutes flight.
>
> We have buit a new fuselage with 2P3S 2000mAh lipo. It shoud give much 
> longer
> flight time. We are also working on a new wing with brushless motors 
> to further
> improve flight time.
>
> For how the battery behave in time. Nicd tend to be robust and last 
> long. NiMh
> tend to be more fragile (last less than a year, maybe 100 cycles). We 
> have very
> little experience with lipo cells.

I did a project at work with Lithium Ion cells on a handheld computer 
design we were working on -- a charger and power supply unit. At least 
with the cells we were using, they prefer constant, lower current draw, 
i.e. not 30A, and not on-and-off, but if they had to, they can supply 
fairly the high currents. I am not 100% sure about the differences 
between LiIon and LiPo, but have heard LiPo have a shorter life cycle 
(# of charges) but have a better energy density. I have heard that 
after around 1000 charges for LiIon or 500 charges for LiPo, the total 
capacity is about 1/2 of what it was new. I would guess that high 
discharge rates would decrease them somewhat.

For comparison, the RCtoys Predator 
(http://www.rctoys.com/predator.php) apparently runs 1 hr 30 min on 
LiPo, and 16-20 min with the NiCad. There battery pack is 7.4V, 
4200mAh, for US$139. A quick look at www.jelu.se which sells (among 
other things), surplus Nokia LiIon cells of 3.6V and 1200mAh capacity 
for about US$13. So 6 of them for about US$80 would give 10.8V and 
4400mAh!!! Maybe too heavy though.

The charging is a lot more complicated for Lithium, than for Nickel 
cells. But you can get nice single chip solutions to do it (we used the 
MAX1649 + an ATtiny micro).

> > I am definitely keen on getting
> > started with an electric model.
> >
> > Would I really need a test pilot? Can I just teach myself to fly it :)
>
> It's you to see, but it may take a little time before you are 
> confortable with
> flying.
> I think an experienced pilote will better handle the emergency 
> situations that
> might occur during initial test flights. Furthermore, it is very handy 
> to be
> two. One will keep his eyes on the airplane, the second one on the 
> laptop screen.
>
> If i were you, i would buy two aircraft. A small one like a multiplex 
> pico cub
> to teach you flying and a second one like a twinstar to test flight 
> your autopilot.
Are there any other airframes that you have been eyeing up, thinking, 
"that would be perfect for this project"?
> I'm sure you will be less stressed when you learn to fly if you don't 
> risk your
> precious electronic toys (GPS camera cerfboard).
>
> Anyway, I think it's a good idea to have contacts with a modele 
> aircraft club.
> People there know well how to build/fly model airplanes and it will 
> save you
> time and money.
I agree. I am working on it...
> > If you are offering, I would really appreciate a blank PCB -- they are
> > unreasonably expensive to get made over here in NZ.
>
> I will send you one when its done, in a couple of week. It will be a 
> panel with
> several PCBs (ground_modem, programmer, ir_sensor, ctl_board, 
> power_supply, GPS).
>
> Maybe you'll want to review the routing/design before i send it to fab.
>
>
> Maybe we could continue this discussion on the mailling list so that 
> other can
> participate
>
> Regards
>
> Antoine
>
> > Regards,
> > Hugo Vincent
> > address@hidden
> >
> > On 15/04/2004, at 7:49 PM, address@hidden wrote:
> >
> > > Hi Hugo
> > >
> > > Selon Hugo Vincent <address@hidden>:
> > >
> > > > Hi Antoine,
> > > >
> > > > I am a university student in New Zealand, currently in third year
> > > > Electrical/Computer Engineering. I am planning a UAV project as a
> > > > hobby
> > > > project, and part of it would be my thesis. I have been writing a
> > > > document describing some of the aspects of it, that I have attached.
> > > > Lots of things in it are wrong or incomplete or subject to change, 
> > > > but
> > > > you'll get the general idea. I am looking at using an AVR and an ARM
> > > > processor, with a similar topology to your paparazzi system (AVR for
> > > > real-time stuff- RC reception, servos, etc., and the ARM for
> > > > navigation
> > > > and stabilization). I am new to RC aircraft and don't really know
> > > > where
> > > > to look, but I have experience with the electronics and software 
> > > > sides
> > > > of things.
> > >
> > > That is a very good news. I have some experience with RC aircraft but
> > > less with
> > > electronics :)
> > >
> > > If you are new to RC aircraft, you will need a test pilot for your
> > > project.
> > > Maybe you should look for a local RC aircraft club.
> > >
> > > Regarding the topology of our project, we use a 3 levels/processors
> > > architecture.
> > >
> > > We have a small avr mega8 mcu which is responsible for radio control
> > > decoding,
> > > servos actuating and low level monitoring (like battery). With this
> > > MCU alone,
> > > you are able to fly the model in manual mode.
> > >
> > > We then have a second MCU, an avr mega128, wich is responsible for
> > > measurments
> > > (ADCs, compass etc...), low level control loop, telemetry 
> > > transmition,
> > > etc...
> > > With both MCU's our aircraft if capable of fully autonomous 
> > > navigation.
> > >
> > > We intend to expend the system with a third high level processor 
> > > which
> > > would
> > > allow intensive computation and network communication. We have an arm
> > > cerfboards running Linux that is almost ready to test flight. We
> > > intend to
> > > connect it to a 800MHz radio modem (coronis wavecard
> > > http://www.coronis-systems.com/produits/PageProduit.php?
> > > IdCategorieProduit=21)
> > > and to our real time controller board.
> > >
> > > > I just found about your project today, and have been reading the
> > > > mailing list archives etc. It seems you project might be headed 
> > > > where
> > > > my one wants to begin, at least in terms of electronics and
> > > > processors... I am more interested in a longer duration (therefore 
> > > > gas
> > > > powered) airframe, but an electric drive is so appealing for
> > > > convenience reasons...
> > >
> > > The airframe is not really important at first. Use something cheap 
> > > and
> > > realiable
> > > to devellop your autopilot. When it's done, you can then mount it on 
> > > a
> > > sophisticated airframe.
> > > We use the twinstar because it flies well, is cheap, can fit in the
> > > car, doesn't
> > > need a runway etc... Someday, when i have time, i will mount a
> > > paparazzi
> > > autopilot in one of my gas aircraft.
> > >
> > > > I would be most interested in your views and opinions about my 
> > > > ideas,
> > >
> > > I think your project is heading in the good direction.
> > >
> > > > and maybe the possibility of working together. I look forward to
> > > > hearing back.
> > >
> > > Working together would be great!
> > > We have spend past weeks writting documentation, reorganizing code 
> > > and
> > > writting
> > > user interface. I will update savannah in the following days and
> > > announce it on
> > > the mailing list.
> > >
> > > I am now designing a new version of the controller board. I intend to
> > > have PCBs
> > > build by eurocircuits. I could send you one of these when they are
> > > done.
> > >
> > >
> > > Regards
> > >
> > > Antoine
> > >
> > > > Regards,
> > > >
> > > > Hugo Vincent
> > > > address@hidden