Re: Re : Re: accelerometer calibration
--- On Sat, 12/25/10, Timo Juhani Lindfors wrote: > "W. B. Kranendonk" writes: > > I found it after searching for accelerometer-dump, but (the FR) > Just capture the data on FR but run it on your PC. Like > > li...@sauna$ ssh neo cat /dev/accelerometer-top | > accelerometer-dump I get that far on my PC :-) > > Any idea what to do about the diversion? I bought that > FR second > > hand, and it has had this "feature" ever since I got > it. > > I guess you could fix it by software but such a fix is > probably not > going to be accepted upstream since it is so ugly :-) > My coding skills are about the right level for compiling accelerometer-dump.c, so I do not think anything I fix is going to get accepted upstream anytime soon ;-) Coming back to the initial question, can those accelerometers be calibrated or is it a black box that emits (timestamps and) coordinates? Boudewijn ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
--- On Sat, 12/25/10, Iain B. Findleton wrote: > The output from > the chip is quite noisy. The problem is not that the signal is not reliable: it is. It is just that it is off by a few degrees. > There is a script I wrote in TCL > on the > SourceForge web site under the fltkwish project page. It > may give you > some ideas about that can be done. In my experience, the > results vary > depending on which of the chips you use. I have used the (a?) waterpas before, when there was such a program in the SHR repositories. I just had a shot at gravity, but there are some missing dependencies. It complains about libtcl8.4.so: # gravity fltkwish: error while loading shared libraries: libtcl8.4.so: cannot open shared object file: No such file or directory I tcl, fltk and fltkwish installed and libtcl8.5.so.0 is available I symlinked it to libtcl8.4.so: # ls -latr /usr/lib/libtcl8.* -r-xr-xr-x1 root root663504 May 29 2010 /usr/lib/libtcl8.5.so.0 lrwxrwxrwx1 root root14 Dec 23 20:17 /usr/lib/libtcl8.4.so -> libtcl8.5.so.0 Now it aborts with just that: # gravity Aborted Perhaps another time :-) > > Note that on the iPhone, some versions of which used the > same chip, you > only get access to the 100/sec samples, On a sidenote: long live closed platforms > The other issues is there was at one time some confusion > about how the > driver reported through the event mechanism. I forget the > details now, > but working code depended on how up to date your kernel > was. If you have > a kernel of recent vintage, should not be an issue. It's the latest SHR: # uname -a Linux om-gta02 2.6.34.7 #1 Sun Dec 5 23:10:39 CET 2010 armv4tl GNU/Linux I wonder (..doubt..) if the issue is in the software, because it has been like this over the last half year or so, with different distro's and no problems on the other FR with the same version of a distro. Thanks for the insight anyway! Boudewijn ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
"W. B. Kranendonk" writes: > I found it after searching for accelerometer-dump, but (the FR) > didn't have gcc installed and don't know yet which packages to > install to make it useful. At least a reason now to dive into > figuring that out :-) Just capture the data on FR but run it on your PC. Like li...@sauna$ ssh neo cat /dev/accelerometer-top | accelerometer-dump > Any idea what to do about the diversion? I bought that FR second > hand, and it has had this "feature" ever since I got it. I guess you could fix it by software but such a fix is probably not going to be accepted upstream since it is so ugly :-) ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
--- On Sat, 12/25/10, Timo Juhani Lindfors wrote: > "W. B. Kranendonk" > writes: > > Would that mean that (one of) the meter(s) is > connected at an angle? > > Are they not calibrated in software? > > top and bottom accelerometers are oriented differently > anyway. Yes, now you say it I recall reading about it. I actually meant: could it be that it is installed at an incorrect angle, i.e. not perpendicular or paralell to the case, but slanted. > > Is the accelerometer-dump standalone, and do you have > a binary version available? > > http://iki.fi/lindi/openmoko/accelerometer-dump.c I found it after searching for accelerometer-dump, but (the FR) didn't have gcc installed and don't know yet which packages to install to make it useful. At least a reason now to dive into figuring that out :-) Any idea what to do about the diversion? I bought that FR second hand, and it has had this "feature" ever since I got it. Regards, Boudewijn ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
"W. B. Kranendonk" writes: > Would that mean that (one of) the meter(s) is connected at an angle? > Are they not calibrated in software? top and bottom accelerometers are oriented differently anyway. > Is the accelerometer-dump standalone, and do you have a binary version > available? http://iki.fi/lindi/openmoko/accelerometer-dump.c ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
--- On Sat, 12/25/10, Timo Juhani Lindfors wrote: > li...@ginger:~$ accelerometer-dump < event4.log | head | > cut -d ';' -f2-|tr ';' ' ' > x y z > 108 36 900 > 144 36 900 > 126 36 900 > li...@ginger:~$ accelerometer-dump < > /dev/accelerometer-top | head | cut -d ';' -f2-|tr ';' ' ' > x y z > 36 0 990 > 54 0 972 > 36 0 990 > 36 18 990 > 18 18 1008 > 36 18 990 > If I look at the top accelerometer and lift the bottom of > the phone a > little and tilt it to the right I get the same readings as > you. Would that mean that (one of) the meter(s) is connected at an angle? Are they not calibrated in software? Is the accelerometer-dump standalone, and do you have a binary version available? Thanks so far! Regards, Boudewijn ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
"W. B. Kranendonk" writes: > See attachment. li...@ginger:~$ accelerometer-dump < event4.log | head | cut -d ';' -f2-|tr ';' ' ' x y z 108 36 900 144 36 900 126 36 900 126 36 882 144 36 900 126 36 900 126 18 900 126 36 900 126 36 882 li...@ginger:~$ accelerometer-dump < /dev/accelerometer-top | head | cut -d ';' -f2-|tr ';' ' ' x y z 36 0 990 54 0 972 36 0 990 36 18 990 18 18 1008 36 18 990 36 18 1008 36 18 990 36 0 1008 li...@ginger:~$ accelerometer-dump < /dev/accelerometer-bottom | head | cut -d ';' -f2-|tr ';' ' ' x y z 54 54 972 36 54 954 54 72 972 54 72 954 54 72 972 54 54 972 54 72 954 54 54 972 54 72 972 If I look at the top accelerometer and lift the bottom of the phone a little and tilt it to the right I get the same readings as you. -Timo ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
"W. B. Kranendonk" writes: > Is that useful? This is while the FR is lying flat on its back. I > haven't compared it to output from the other FR yet. Raw binary output would be a lot easier to parse. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Re : Re: accelerometer calibration
> Le 23 déc. 2010 10:11, Timo Juhani Lindfors > a écrit : > >"W. B. Kranendonk" writes: > > > One of my FRs is dizzy: the accelerometers are off. > > Can you send a sample of the accelerometer data? (Preferably in > > the binary format that you can read from /dev/input/) > On Thu, 12/23/10, sylvain.p...@gmail.com wrote: > With a command line like this : > cat /dev/input/event4 | od -c (or -b or -d dont remember and i am > not on my laptop) Thanks for the hint, it would have been my next question I suppose ;-) I used od -b, which gives octal byte representation: 0017720 230 ] 022 M 371 H 005 \0 003 \0 001 \0 022 \0 \0 \0 0017740 230 ] 022 M \0 I 005 \0 \0 \0 \0 \0 \0 \0 \0 \0 0017760 230 ] 022 M 220 m 005 \0 003 \0 \0 \0 ~ \0 \0 \0 002 230 ] 022 M 260 m 005 \0 \0 \0 \0 \0 \0 \0 \0 \0 0020020 230 ] 022 M 035 267 005 \0 003 \0 001 \0 $ \0 \0 \0 0020040 230 ] 022 M K 267 005 \0 \0 \0 \0 \0 \0 \0 \0 \0 0020060 230 ] 022 M 234 \0 006 \0 003 \0 001 \0 022 \0 \0 \0 0020100 230 ] 022 M 305 \0 006 \0 \0 \0 \0 \0 \0 \0 \0 \0 0020120 230 ] 022 M 234 223 006 \0 003 \0 001 \0 $ \0 \0 \0 0020140 230 ] 022 M 273 223 006 \0 003 \0 002 \0 226 003 \0 \0 0020160 230 ] 022 M 277 223 006 \0 \0 \0 \0 \0 \0 \0 \0 \0 0020200 230 ] 022 M 310 270 006 \0 003 \0 002 \0 204 003 \0 \0 0020220 230 ] 022 M 330 270 006 \0 \0 \0 \0 \0 \0 \0 \0 \0 0020240 230 ] 022 M 231 335 006 \0 003 \0 \0 \0 l \0 \0 \0 0020260 230 ] 022 M 244 335 006 \0 003 \0 001 \0 022 \0 \0 \0 0020300 230 ] 022 M 254 335 006 \0 \0 \0 \0 \0 \0 \0 \0 \0 0020320 230 ] 022 M 352 001 \a \0 003 \0 \0 \0 ~ \0 \0 \0 0020340 230 ] 022 M 031 002 \a \0 003 \0 001 \0 $ \0 \0 \0 0020360 230 ] 022 M 037 002 \a \0 \0 \0 \0 \0 \0 \0 \0 \0 0020400 230 ] 022 M 4 p \a \0 003 \0 001 \0 022 \0 \0 \0 0020420 230 ] 022 M b p \a \0 \0 \0 \0 \0 \0 \0 \0 \0 Is that useful? This is while the FR is lying flat on its back. I haven't compared it to output from the other FR yet. Best regards, Boudewijn ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re : Re: accelerometer calibration
With a command line like this :cat /dev/input/event4 | od -c (or -b or -d dont remember and i am not on my laptop)Le 23 déc. 2010 10:11, Timo Juhani Lindfors a écrit : "W. B. Kranendonk" writes: > One of my FRs is dizzy: the accelerometers are off. Can you send a sample of the accelerometer data? (Preferably in the binary format that you can read from /dev/input/) ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
