Well my back ordered parts finally came in so I built up my last three PicPilot boards. I plan on keeping one of them so I'll have two to play around with. One is in my Tri Rotor and the other will be installed in the Easy Star clone I just received. Of the remaining two, I have one potentially sold leaving one up for grabs. If your interested the cost is $375.Thanks,Brian
The zip file has two files, one with the init and processing code and one that shows how I'm reading from the device using the microchip library for I2C.
Hello Brian, interesting proyect.
There is any possibility that you share the routines for the bpm085? I am trying to read it (with not succes) with a 18f873.
I will be very gratefull if you could help me!!!
The BMA has a +-4g range. Don't you think it is on the low side ?
Making the magnetometer an external attachment is a good idea. Have you ever considered doing the same with the Gyro/Acceleros and also the Air Data (speed + altitude), and maybe also with the current/voltage sensors ? I'm not a HW specialist so my question may be dumb.
This way you have one central board with plenty of power and all sorts of peripheral devices.
So the components would be:
Central unit (somewhere where there is available space)
Magnetometer (tip of the wing i.e.)
Gyrometer/accelerometer (center of gravity)
Airspeed + Barometric altitude (wing)
Telemetry
GPS (top of airplane)
Current and voltage sensing
This way if one of the peripherals does not give satisfaction then you don't need to redesign the whole AP and if somebody does'nt want the magnetometer he just can forget about them ?
Paul,
For the most part the magnetometer is working satisfactorily although I do wrestle with noise issues coming mainly from the motors. In my current setup I only use the magnetometer for heading info when the speed is too low for a GPS heading fix, but as I get closer to doing wind estimation ect I'll need to address the noise/calibration chestnut. I'm am seriously considering making the magnetometer an external attachment on the next revision of the board so it can be optimally placed on any given airframe (for example on a wing tip). I would look for any feedback along those lines if anyone has any.
As far as the gyros are concerned, it was important to me from the beginning to have a flat board and at the time the only way to do that with lower cost mems devices was to use the ST gyro for the Z axis even knowing it had a low resonance point. I will be replacing it with the newer Invensense Z axis device in the next revision. It's a shame though because in some ways the ST gyros are better (better bias and noise characteristics) but they do have vibration issues on some platforms. I haven't had any issues on fixed wing platforms but have seen occasional problems on my tri-rotor and even my RC truck when hitting some heavy bumps. The higher resonant frequency of the Invensense devices makes it much easier to isolate from the mechanical vibrations and the DCM(I hate calling it that) and I'm sure the EKF your working on have no problems dealing with their bias issues.
I've been happy with the Bose accelerometer.
I'm still finalizing the design of the next board revision, but the parts cost will be around $220 or so (not including a GPS receiver, radio modem). The first revision was $260, but some parts actually got cheaper. These of course are low quantity costs with a big hunk being the cost of the PCB. The wild card is assembly cost. I built five of the first design and sold them for $375, but it takes me several hours to build (all by hand, I don't have an oven) and I'm not sure I want to take that on again for the next revision. That means contracting out assembly which will be a little pricey per board for small quantities.
Congratulation for your T3 run. You are absolutly right about making your gains vary with airspeed, that is what is done in real life. The dynamic pressure is factor #1 for a flying machine.
I found your first blog thanks. Are you satisfied with your magnetometers ? And with your gyrometers and accelerometers ?
I could be interested in buying one of your boards, but I don't know anything about export restrictions. Do you have an idea of the cost of a complete board ? (With components mounted on).
Hey Paul,
Of Course. One of my first blogs has a link to the schemo and parts list, but I'll be adding that to my web site soon along with the gerbers, code and details of various flights. If there's something in particular your looking for and don't want to wait on my glacier pace getting the web site set up let me know and I'll see what I can do.
I am looking to make some more boards too, so if anyone is interested let me know.
That's a great representation of the differences of GPS vs. a barometer. I'm researching the BMP085 (same as you're using) and it looks like a great solution. Apparently it offers resolution down to .01 hPa, which is equivalent to a 3.3" column of air. Given this resolution, along with some software filtering, it looks like this pressure sensor would be able to detect even the smallest of altitude changes. This could lead to some pretty tight altitude holds =).
Of course resolution is quite different from accuracy, but your GPS vs. barometer looks spot on in the areas where the aircraft is not making any changes (leading me to believe the baro is pretty accurate, but GPS is slow to respond).
I've been doing a lot of research and I am planning on building a similar board using the AVR line of MCU's. I'm heading towards a helicopter-based system instead of an airplane. If all goes as planned, the board will have a full "9DOF" IMU (gyros, accels, magnetos), barometer, uBlox 5, ultrosonic sensor for landing, and an XBee connector for the XBee 900 long-range units, as well as PPM input, an SD card slot, USB port, etc. This will probably be based on the ATMega2560.
Anyway, I look forward to reading up on your project. I've found it very inspiring so far!
BW,
Thanks for the comments. I do have tons of data and plan to set up a web page to document the development, but thus far haven't found the time to do it. I'll post here on DIYDrones occasionally when time allows. I also plan to build up (or have built up - I don't have the time to assemble more myself) a few more units in the coming weeks.
Here is a plot of altitude hold for the DIYDrones T3-2 contest that's going on this month. It shows the Baro altitude (which I close my loop around) vs. GPS altitude along with the target altitude bands. Originally I used only the GPS altitude but would have odd dives and climbs happen when they shouldn't. Looking at the data I found the GPS would often have significant momentary altitude errors particularly after high G- turns. I speculate it's partly from antenna orientation, but mostly from limitations in filtering/processing in high acceleration situations. Since using the Baro, all those problems are gone with respect to altitude. Still have issues with the Lat/Lon going bonkers during fast turns and dives, but I'm trying some things that I think will help significantly plus I just installed a u-Blox GPS receiver that I'm hoping will do a better job in dynamic situations than my Santa Barbra Research receiver did. Just need to wait for better weather to try it out.
Comments
I didn't know about that page of your project.
I will test your routines as soon as possible.
Thanks again!!! :)
charly,
Someone else had asked for help so I posted the code on my website.
http://desert-wolfe.com/PICPilot/default.html
The zip file has two files, one with the init and processing code and one that shows how I'm reading from the device using the microchip library for I2C.
Hope it helps.
Brian
There is any possibility that you share the routines for the bpm085? I am trying to read it (with not succes) with a 18f873.
I will be very gratefull if you could help me!!!
The BMA has a +-4g range. Don't you think it is on the low side ?
Making the magnetometer an external attachment is a good idea. Have you ever considered doing the same with the Gyro/Acceleros and also the Air Data (speed + altitude), and maybe also with the current/voltage sensors ? I'm not a HW specialist so my question may be dumb.
This way you have one central board with plenty of power and all sorts of peripheral devices.
So the components would be:
Central unit (somewhere where there is available space)
Magnetometer (tip of the wing i.e.)
Gyrometer/accelerometer (center of gravity)
Airspeed + Barometric altitude (wing)
Telemetry
GPS (top of airplane)
Current and voltage sensing
This way if one of the peripherals does not give satisfaction then you don't need to redesign the whole AP and if somebody does'nt want the magnetometer he just can forget about them ?
What is your opinion ?
Paul
For the most part the magnetometer is working satisfactorily although I do wrestle with noise issues coming mainly from the motors. In my current setup I only use the magnetometer for heading info when the speed is too low for a GPS heading fix, but as I get closer to doing wind estimation ect I'll need to address the noise/calibration chestnut. I'm am seriously considering making the magnetometer an external attachment on the next revision of the board so it can be optimally placed on any given airframe (for example on a wing tip). I would look for any feedback along those lines if anyone has any.
As far as the gyros are concerned, it was important to me from the beginning to have a flat board and at the time the only way to do that with lower cost mems devices was to use the ST gyro for the Z axis even knowing it had a low resonance point. I will be replacing it with the newer Invensense Z axis device in the next revision. It's a shame though because in some ways the ST gyros are better (better bias and noise characteristics) but they do have vibration issues on some platforms. I haven't had any issues on fixed wing platforms but have seen occasional problems on my tri-rotor and even my RC truck when hitting some heavy bumps. The higher resonant frequency of the Invensense devices makes it much easier to isolate from the mechanical vibrations and the DCM(I hate calling it that) and I'm sure the EKF your working on have no problems dealing with their bias issues.
I've been happy with the Bose accelerometer.
I'm still finalizing the design of the next board revision, but the parts cost will be around $220 or so (not including a GPS receiver, radio modem). The first revision was $260, but some parts actually got cheaper. These of course are low quantity costs with a big hunk being the cost of the PCB. The wild card is assembly cost. I built five of the first design and sold them for $375, but it takes me several hours to build (all by hand, I don't have an oven) and I'm not sure I want to take that on again for the next revision. That means contracting out assembly which will be a little pricey per board for small quantities.
B
Congratulation for your T3 run. You are absolutly right about making your gains vary with airspeed, that is what is done in real life. The dynamic pressure is factor #1 for a flying machine.
I found your first blog thanks. Are you satisfied with your magnetometers ? And with your gyrometers and accelerometers ?
I could be interested in buying one of your boards, but I don't know anything about export restrictions. Do you have an idea of the cost of a complete board ? (With components mounted on).
Thank you very much,
Paul
Of Course. One of my first blogs has a link to the schemo and parts list, but I'll be adding that to my web site soon along with the gerbers, code and details of various flights. If there's something in particular your looking for and don't want to wait on my glacier pace getting the web site set up let me know and I'll see what I can do.
I am looking to make some more boards too, so if anyone is interested let me know.
Brian
Are you willing to share the plans of the PicPilot ?
Thank you very much,
Paul
Of course resolution is quite different from accuracy, but your GPS vs. barometer looks spot on in the areas where the aircraft is not making any changes (leading me to believe the baro is pretty accurate, but GPS is slow to respond).
I've been doing a lot of research and I am planning on building a similar board using the AVR line of MCU's. I'm heading towards a helicopter-based system instead of an airplane. If all goes as planned, the board will have a full "9DOF" IMU (gyros, accels, magnetos), barometer, uBlox 5, ultrosonic sensor for landing, and an XBee connector for the XBee 900 long-range units, as well as PPM input, an SD card slot, USB port, etc. This will probably be based on the ATMega2560.
Anyway, I look forward to reading up on your project. I've found it very inspiring so far!
Thanks for the comments. I do have tons of data and plan to set up a web page to document the development, but thus far haven't found the time to do it. I'll post here on DIYDrones occasionally when time allows. I also plan to build up (or have built up - I don't have the time to assemble more myself) a few more units in the coming weeks.
Here is a plot of altitude hold for the DIYDrones T3-2 contest that's going on this month. It shows the Baro altitude (which I close my loop around) vs. GPS altitude along with the target altitude bands. Originally I used only the GPS altitude but would have odd dives and climbs happen when they shouldn't. Looking at the data I found the GPS would often have significant momentary altitude errors particularly after high G- turns. I speculate it's partly from antenna orientation, but mostly from limitations in filtering/processing in high acceleration situations. Since using the Baro, all those problems are gone with respect to altitude. Still have issues with the Lat/Lon going bonkers during fast turns and dives, but I'm trying some things that I think will help significantly plus I just installed a u-Blox GPS receiver that I'm hoping will do a better job in dynamic situations than my Santa Barbra Research receiver did. Just need to wait for better weather to try it out.
Brian
-
1
-
2
-
3
of 3 Next