Re: [Autopilot] PWM and Edge Counting Input
Status: Alpha
Brought to you by:
tramm
From: Mark N. <m_l...@ya...> - 2002-05-27 14:08:19
|
HI Tramm >Tramm Hudson wrote: >Mark Newell wrote: >> I am new to the list but am trying to do basically >> the same thing using a BasicX24 processor. >Welcome! What sort of project are you working on? . . . . My original idea was to build a "Roswell Flyer" clone. I have been collecting parts for about a year. I felt the resolution for PWM input and output would be ok. I go alot for free samples and one can not always dictate the packaging. I have thought about a 3 engine flying platform, some how using thrust vectoring, possibly including GPS. I have a lot of time, not too much money, some neurological imparement from cancer treatment, so I don't get things done too quick. I look forward to following your progress and will offer input if I can. Thanks, Mark > Since it is somewhat limited I have tried> to off load some of the processing. In particular i am using a PAK-VII> from Al Williams to help with PWM input processing and edge counting.> The chip has 8 channels of input and the interface to a micro is easy.> There is info at> http://www.al-williams.com/awce/pak7.htmI'm a little worried about its resolution for this application.According to the web page, it has "16 bits at 5uS resolution".If I'm reading that correctly and it means 1 bit == 5 uS, thenits minimum acceleration or servo pulse that it can sense isfairly long.For the servos, the pulse is between 1000 us and 2000 us wide.At the stated resolution, the PAK-VII would only be able to resolve200 positions. That's less than 8 bits of precision and I knowthat is insufficient for our application.The ADXL would have a similar problem. You mention that you'reusing the 10 g version, which will have 5 times less resolutionin the small acceleration ranges. Depending on your application,that may or may not be acceptable.> I have the processor, gyros and Analog ADXL210 accelerometer. I was> wondering if it might help with the noise problems.We actually managed to eliminate most of the noise from the systemthrough redesigning the mainloop on the realtime board. You can seeplots of the accelerometer samples as the source code evolved and wetuned things to deal with the real-time tasks of handling thesampling:http://autopilot.sourceforge.net/accel.htmlHowever, we're considering moving away from doing it with the MCU.Brett has convinced me that using an FPGA for all of the real-timetasks is the best way to offload the CPU and let it get back tothe stabilization, navigation and other more interesting tasks.> I have not actually> tried it because I am some what "steady challenged" ie slight tremmors> in my hands and have not come up with an easy way to mount that little> "blank" ADXL210 chip.I can't solder very well, either. So I used the -EB version, whichis mounted on a small PCB with five through-hole leads. You can seea picture on the above page.Trammell-- -----|----- hu...@sw... H 240-476-1373*>=====[]L\ Tra...@ce... W 240-453-3317' -'-`- http://www.swcp.com/~hudson/ KC5RNF_______________________________________________________________Don't miss the 2002 Sprint PCS Application Developer's ConferenceAugust 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm_______________________________________________http://autopilot.sourceforge.net/ Development mailing listhttps://lists.sourceforge.net/lists/listinfo/autopilot-devel __________________________________________________ Do You Yahoo!? Yahoo! - Official partner of 2002 FIFA World Cup http://fifaworldcup.yahoo.com |