paparazzi-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Paparazzi-devel] Fwd: Re: controller board, gps


From: nisma
Subject: [Paparazzi-devel] Fwd: Re: controller board, gps
Date: Thu, 13 May 2004 13:43:43 +0200 (MEST)

--- Weitergeleitete Nachricht / Forwarded Message ---
Date: Thu, 13 May 2004 13:08:21 +0200 (MEST)
From: address@hidden
To: address@hidden
Subject: Re: controller board, gps

Hi to all,
i plann to make a new PBC design.
Can you give me some hints, what sensors you want to include in future?
The design goal is to be less than 75 gram including ARM7 CPU, GPS, Radio,
digital webcam, backup battery, servo controller, sensors.

> Hi Chris.
> 
> Sorry that I didn't answer more quickly to your previous mail. I am very
> busy at
> the moment.
> 
> Maybe you could expose your project on the list
> 
> 
> regarding your previous mail:
> 
> nb of ADCs: currently mega128-> 2x10bits for infrared 
>                       mega8  -> 3x10bits (current, battery, regulated
> supply)
> when we tried the rotomotion unit, we needed 6 ADCs on the mega128.
> I think that for pressure measurements, 10 bits is not enought. Maybe a
> programable gain amplifier would be nice for absolute pressure. For
> differential
> pressure, we need a very small scale (100Pa ~ 10m/s)
>
I use a 8052 MCU with integrated PGA and 24 bit ADC (cheaper than equivalent
external ADC/PGA/... circuit) for magnetometer/pressure.  
I suppose, you don't need it.


> I think I would buy only the PCBs + chips (maybe assembled) as I already
> have
> GPS camera etc.. You should ask people on the list. Maybe ask on the
> autopilot
> list too.
>
The IMU is not really designed for helicopter (too slow GPS rate and other
sensor configurations)
 
> For sensors, I don't really know. Using separate boards (like SPI ADC +
> amplifier + pressure sensors) allows more flexibility when fitting the
> system in
> an airframe. Why not 3 boards (analog / MCUs / Linux)
The cost is to high. the current design is 2 boards, 
sensors (analog/MCUs/gps/modem/radio) and CPU (ARM 16/32bit -or- MSP430
16bit -or-
MEGA232  8bit -or- PIC18xxx 8bit -or- optional LINUX).
> Power consumption : for now, our system needs about 1-2W. This is very
> little
I assume, that this is with the BEC PSU.
> compared to motor current consumption, except for the smallest airplanes.
> The linux board will probably need much more power.
I'm able to include linux with 1-2W. If the power consumation is not a
problem, then use the AMD ELAN586 CPU. This have a true FP coprocessor.

> For external buses, I think SPI and i2c are very important. Today we have
> 1MHz
> SPI and we use less than 1% of this bandwidth. For controller board only
> uses,
> rs232 is convenient.
> 
> For dimensions. We are wanting to test very small airframe (300g) with the
> controller alone. So the weight/size of this board is very important. Our
> v1.2
> board (2mcus+modem+power supply+servo driver) fits in the case of a futaba
> r136
> receiver. GPS, infrared sensor and radio transmitter are external.
This design is slightly larger. The weight can be similar.
This is due the fakt, that the OEM gps module, OEM radio module should be
plugged on the board using smt connectors. The size is a bit larger then a
64 pin dip carrier and the underside of the PBC is used (in part) for the
rfpic505ad and PBC antenna in order to provvide low cost communication for
the base station, when the RC is on ground and collision warning for
formation flights.
Other usage is to collect data from remote sensors using low power low 
cost RF PIC designs in combination with very small solar panels.
This must not be fitted, the PBC layout simply include this.
> 
> logging: we have no logging capabilities now. We could add one with SPI.
> I'm not
> sure this feature is really needed at the controller level. On the linux
> board, 
> it would be nice. Maybe a memory card or usb memory stick....
OK, you don't need logging of all sensors and RC servo responses.
I need it to evaluate the quality of control algorithm or it's parameters.

> For now, the 4K eeprom in the mega128 seems more than sufficient.
How many eeprom you need ? the mega128 is not used. the mega232 can be used,
but is optional (cpu socket) and exclusive, only one CPU module can be
attached. If you make annother PBC, you can use the plain I2c and SPI or
RS232 signals using a dip64 PBC connector on one side, and 10pin header to
the other side.
> 
> For now we have 4800bps telemetry. We started with 1200bps and this was
> really
> too low. 4800 is okay. Of course we could use more 
No problem. 19.2-64K is possible including error correction and shuffling.
> I think we don't need DGPS in the controller board. In the linux board
> this
> could be nice. The microblox tim-lp unit we use can use RTCM data to do
> differential corrections. 
> 
> Price range: Well i really don't know. 
> The controller board can probably be very cheap if it includes only a
> minimal
> number of sensors.
Lower price is the primary concern.
> The linux board will probably be a little bit more expensive, depending on
> peripherals. 

> 
> For servos (mixing, response course etc..) + rc (PPM) decoding , we have
> found
> that the 16MHz mega8 is sufficient. We want PCM decoding and this task
> seems to
> be CPU intensive, but I hope that the mega8 will still be enought.
I sample PPM from the trainer port and use the modem to send it. 
This give minor jitter and gives the possiblility to use licenced radio 
equipment. (amateur radio licence as example).
Making the same thing on the UAV is possible using a 
dedicated micro for link validation on the RC TX (trainer port).
> 
> For servo controlling, we use 40Hz pulses. I measured this value on a rc
> receiver. I don't know if more is needed. The guys at autopilot use a
> "high
> speed" servo for their heli tail. 
On certain (cheap) servos, it's possible to use 400 HZ for the time to
update
the position and later 50HZ to keep the position. Conrad sources cheap 5
EURO
servos supporting this.
> PSU: we use only one battery onboard. Servos are supplied by motor
> controller
> bec. All the rest is drawn on a switching power supply. We had 5V/400ma.
This is not acceptable for the analog parts. You have to include one battery
for the analog part. One single 1.2 V chell can be enought.
If you use 14 bit, only 8 bit is usable resulting from tests having two
batterys, one for the motors and one for the servos/mcu. The filtering 
circuit cost more and lowers the resoltuion than a secondary battery.
The tests was made measuring high precision voltages references with the AD
converters modulated with a constant sinus period.
For the IR, it does not matter, but for the pressure sensors.
> This
> was to little with some "high power" video (transmitter + camera). We now
> have 5V/1A
> 
> We have active PSU for servos thanx to the becs. I don't know if this is
> needed
I need it in order to have the same reaction to the control loops
regreadless
of the battery voltage.
> and to what extend "variable speed" servos would interfere with control
> loops.
These are abstractions layer in order to have simpler control loops.

> As electric powered aircraft have a lot of EMI, I think a 5V analog
> interface is
> good for "distant" sensor (our infrared sensors are on top of the rudder
> and
> MCUs under wings).
> 
> We don't reprogram our system in flight (you mean flash ?). We have code
yes
> to
> adjust gains and
> neutrals in flight when tuning for a particular airframe.
> 
> For now we don't have JTAG nor ICD. I think there is a Linux version that
> would
> be a nice feature.
I prefer a deterministic controller instead a highly dynamic (unpredictable)
scheduler. 
> 
> We have rs232 today for ground interface. USB would be nice as rs232 tend
> to
> disapear on recent laptop.
> 
> I don't know what to tell you. Our current system is very light and it
> will fit
> in all the configurations you describe.
> 
> We plan to make a more sophisticated version, but it's difficult to forsee
> our
> needs.
I know this, but i must have a feeling on what are you needs in order that
the board is usable for you. 
> We will need more sensors, processing power, memory, bandwidth. But
> we
> must also keep things simple and cheap :) 
Open issues:
 o Camera 
    - do you want to use the analog camera you have,
    - digital using the  modem for trasmission
    - usb using the linux board
 o Do you want use the RC transmitter for uplink (including data) and video
   trasmitter for downling (including data) or using dedicated radio 
trasmitter 
 o What type of sensor your plann for the future. If i have PBC space,
   i can integrate it, eventually as option between RFPIC and xxx sensors.
> 
> maybe you could make a diagram of your proposal and we can discuss on this

> 
> regards
> 
> Antoine
> 
> 
> 
> Quoting address@hidden:
> 
> > Hi Antoine,
[cut]
> > 
> > Some links  of interrest.
> > www.tinyarm.com 
> > the msp430 MCU from texas instrument
> > developer.axis.com  etrax mmc SoC (System on Chip)
> > in particular
> > http://developer.axis.com/products/mcm/ds_etrax_mcm_low.pdf
> > http://www.fredericks.com  cheap electrolytic tilt sensor
> > www.cypress.com for the SCoC MCU (Programmable System on Chip).
> > PGA from microchip (Programmable Gain Amplifier)
> > Depending on application, the Maxim part can be interresting.
> > 
> > Later
> > Chris
> > 
> > 
> > 
> > 
> > 
> > -- 

-- 
"Sie haben neue Mails!" - Die GMX Toolbar informiert Sie beim Surfen!
Jetzt aktivieren unter http://www.gmx.net/info





reply via email to

[Prev in Thread] Current Thread [Next in Thread]