Wednesday, May 05, 2010

Catching up....

A bunch of Random thoughts... Business is picking up and if we can manage the many parts shortages I should be able to fill in the LLC $ hole I dug in 2009 and even do some modest development.

Toward that end I've been working on several things.
The four motors shown in the back of the rocket two blog posts ago are meant to be 3D printed with built in cooling passages. I had a plastic version I was showing at Space Access, the Metal version I ordered arrived today and it looks really good until you try to pass fluid through the cooling passages they are all plugged with sintered metal. Its a really pretty $400 paperweight.
I may have to build in a conventional way or change to a more expensive 3D printing process.
I used a supplier that uses the Prometal R1 and would really like to have the work done on a
EOSINT 270 M, but that process is about 5 or 6x as expensive.

Last weekend I went out to the San Diego Tripoli club launch and Joined Tripoli and successfully flew my Level 1 and Level 2 Qualification flights using the poorly painted Liberty 2 I discussed earlier. I'm in the process of building some test hardware to evaluate several lower cost IMU and GPS solutions that I will fly on the Liberty next FAR weekend (the 15th)

I'm working at doing some development work to lower the overall cost of things. One of the areas I'm working on is a lower cost IMU based on MEMs sensors. The Mems Sensors are getting much better and lower cost. The area I'm sort of stuck on is doing the drift bias correction stuff. The guys at www.diydrones.com have been working on IMU drift correction for aircraft and have generated some really good stuff. I just started a discussion topic over there that I hope will help me make some progress in extending some of this work to highly accelerated rockets. (http://diydrones.ning.com/forum/topics/rocket-imu-thoughts)

9 comments:

ian said...

But Paul, just use a Kalman filter. This has been done before already many times. I don't buy it's not intuitive or much harder than an ad-hoc method. It really is pretty straightforward.

ian said...

What cheap GPS and MEMS are you looking at? The Analog Devices ones?

Paul Breed said...

I'm going to fly a ADIS16400 the analog devices 3 Gyro, 3 Accel, and 3 Mag sensor.

Also I'm evaluating the much lower cost sensors on the Spark Fun 9DOF IMU.

Paul Breed said...

Ian, I'm open to help and suggestions, every time I sit down to read a Kalman filter paper ...I wake up hours later feeling well rested. ;-)

Kidding aside at some level I need to see a block diagram where I have an intuitive grasp of what is happening in each box. The output from the raw gyro sensors are rotation rates. These clearly need to be integrated to get some concept of orientation. maybe one does not need orientaion, only rates?.


rates,

C. Scott Ananian said...

Paul: I agree--I know enough to know that Kalman filters are the Right Thing, but I've never found a readable reference describing theory and implementation. Ian: do you have any suggestions?

heroineworshipper said...

For a short duration, high acceleration rocket the IMU can be just 3 gyros.

People who sell IMU's don't like us talking about it because they depend on 6 & 9 DOF sales, but the 3 DOF IMU has been the most reliable.

ian said...

Paul,
I just sent you an email with a small initial explanation. The best book I've seen on GPS/INS Kalman filtering is Applied Mathematics of INS by Rogers, and I've seen a lot. I like it because it covers everything you need including examples, and it includes a good filter (although it is missing implementation details).

I have other books that go into more details about implementation but I wouldn't recommend those unless you really want to get into navigation.

Ian

Felix said...

Do you have some pics of the motors that you could publish on your blog?
It's unfortunate that unbound powder got trapped into the cooling passages. Was the machine operator aware of the nature/function of this part? From what I could see about this printing process on Youtube, it looks like he wasn't able to vacuum unbound powder out of these cavities before putting the motors into oven.

Monroe King said...

Paul I have the Kalman filter we use on the X-Prize lander we made for Frednet. I can send it to you if your interested or just hand it to you when I see you at FAR.