Well I completely removed the stock Y-UFO electronics (except for speed controllers of course) and am now doing all the mixing and stabilization with the PicPilot. Here is a pic of the install.
It's been too windy to fly outdoors so I've been doing what testing I can inside. I have a little instability at the moment. The Y-UFO will begin to oscillate in sort of a circular motion reminiscent of a coin being dropped on a table. I'll try to put up a video soon. Part of the problem might be ground effect as I'm only about 6 inches off the ground, but I'm looking closely at all the sensor data to see if everything looks as it should. I thought I'd share one of the plots that demonstrate the filtering effect of the DCM. This plot compares the unfiltered Y acceleration data with the seventh element in the DCM Matrix. Both terms essentially represent the sin of the roll angle.
The plot represent me holding the Y-UFO lightly and rolling it approx. 20 degrees one way then 20 degrees the other. This plot shows one of the advantages of using gyros. If you attempted to filter the Accelerometer data to be as clean as the DCM data you would introduce considerable delay or lag to the signal. This lag would then limit how fast of a perturbation the stabilization loop would be able to remove. Since the DCM is integrating the rates instead of position (essentially what the accelerometers are giving you by virtue of the gravity vector) the noise is reduced by the magnitude of the sample freq. Before the rate has time to integrate to a significant position, it reverses polarity and begins to cancel itself out. Of course huge amounts of rate noise or extremely slow sample times would start to be an issue. For slow lumbering airframes the accelerometers by themselves would probably be fine, for faster airframes not so much so.During this test I start out with the three motors turned off. When I began to throttle up that's when most of the noise showed up in the accelerometers (and the gyros, but because of the above mentioned filtering effect the noise is not apparent in the DCM data.). All indications are the noise is simply vibration from the three motors, but I'm looking closely at the data to make sure it's not EMI/ power supply noise.One thing worth noting is this level of noise/vibration is not having an adverse effect on the ST LISY300AL gyro I'm using for the Z axis rate sensor ( the other two axis are covered by an IDG 500 which was not expected to have vibration issues). I do have the PicPilot isolated through three layers of foam tape. Maybe that's enough to mechanically filter out the higher frequencies that will give the LISY300AL problems.Let me know if your interested in seeing more plots such as this from time to time or if I'm just filling up the server with useless data. It takes a few minutes for me to format things for posting, but I don't mind if others find it useful, and there's always a chance that someone will have an observation that helps me get over the next hurdle or problem I'm struggling with as well.Brian
I have been using Bill's UDB with the Red Board gryros swapped out with ADXRS610 gyros for less vibration sensitive versions and with 300 deg/sec max rates (Green was only 75 I think) on my TREX 450. Works OK but I think I still see vibration effects...working on better isolation. But to really get a stable hover I need a magnetometer or compass...so checking on your progress. I will keep following.
Nice tactic with the shopping spree, but I think they know that card so be prepared for a quick come back;)
John,
I haven't had a chance to put the PicPilot in my T_Rex yet. When I can find time to play I've been focused on getting features fleshed out on the AXN Floater (Easy Clone). I'll be sure to post when I get back to working on the heli. I may wait for the next revision of the board. That board will replace the ST gyro with an InvenSense and will have a remote mountable magnetometer. The design is ready to go to the board house, just waiting for the wife to go on a cloths shopping spree so I can play the "If you bought something I get to buy something" card. Also planing on getting a web site up, but not there yet. I'm finalizing my design for the 28 hour day at which point everything else will quickly fall into place.
They're so mechanically simple that many people have built their own from scratch. Without the battery, Rx, and brains a quadrotor can be put together for $100 or so. Pretty much just 4 outrunner motors, 4 props (fixed), 4 escs, 4 cf tubes, and some sort of center to hold it all together.
Comments
I have been using Bill's UDB with the Red Board gryros swapped out with ADXRS610 gyros for less vibration sensitive versions and with 300 deg/sec max rates (Green was only 75 I think) on my TREX 450. Works OK but I think I still see vibration effects...working on better isolation. But to really get a stable hover I need a magnetometer or compass...so checking on your progress. I will keep following.
Nice tactic with the shopping spree, but I think they know that card so be prepared for a quick come back;)
Best
John
I haven't had a chance to put the PicPilot in my T_Rex yet. When I can find time to play I've been focused on getting features fleshed out on the AXN Floater (Easy Clone). I'll be sure to post when I get back to working on the heli. I may wait for the next revision of the board. That board will replace the ST gyro with an InvenSense and will have a remote mountable magnetometer. The design is ready to go to the board house, just waiting for the wife to go on a cloths shopping spree so I can play the "If you bought something I get to buy something" card. Also planing on getting a web site up, but not there yet. I'm finalizing my design for the 28 hour day at which point everything else will quickly fall into place.
Have you used your PICpilot on a heli? If so can you give me the link?
Thanks
John
There was a link in my previous post, I didn't think to include it again. It's a Tri Rotor helicopter like thingy. Here's a link -> Y-UFO