mjb/altos/src/util, branch aprs

altos: make check-avr-mem utility executable

2012-10-23T04:39:12+00:00

Signed-off-by: Keith Packard

altos: Parameterize altitude table access and initialization

2012-10-12T20:59:50+00:00

This allows projects to store the altitude data in different representations or with different access modes. By default, altitude data is stored in meters, but the initializers include decimeter values so those can be used instead if desired. Signed-off-by: Keith Packard

altos: Shrink Pa to altitude table

2012-09-21T11:29:17+00:00

This improves the computation of the table enough that errors from a 470 entry table are almost all < 0.5m. Signed-off-by: Keith Packard

altos: Software implemenation of CC1111 radio encoding

2012-06-17T23:17:00+00:00

Add CRC, whitening, FEC and interleaving routines for transmission path to allow cc1120 to send telem packets to cc1111. Signed-off-by: Keith Packard

altos: Add conversion between Pa and meters

2012-05-19T03:19:01+00:00

To be used with the MS5607 which generates data in calibrated units. Signed-off-by: Keith Packard

altos: Remove debug printf from cc1111 check-stack script

2012-04-14T21:04:29+00:00

Signed-off-by: Keith Packard

altos: Check for cc1111 flash overflow

2012-04-10T05:25:13+00:00

The linker is supposed to do this, but it ignores the static initializer data added after the code. Signed-off-by: Keith Packard

altos: Check pdata+xdata memory usage during build

2011-11-13T02:37:53+00:00

The sdcc linker doesn't check the sum of pdata+xdata memory usage, it only ensures that xdata itself is small enough. This doesn't keep xdata below the end of usable ram on the cc1111 though (0xfe000). Fix up the check-stack program to also make sure all of xdata fits in available memory. Signed-off-by: Keith Packard

altos/avr: Must leave space for init stack in ram

2011-10-12T00:57:32+00:00

The stack used during system initialization lives at the top of RAM, so leave some space for that. Signed-off-by: Keith Packard

altos: Improve TM v1.0 apogee estimate

2011-10-09T04:23:16+00:00

v1.0 boards have noisy accelerometer data caused by interactions between RF transmission and the accelerometer measurements; this noise generates a negative bias in the accelerometer readings. The net effect is that the estimated speed is lower than the actual speed, causing early an apogee estimate. By increasing the sigma value for accelerometer data, the kalman filter 'trusts' the acceleration data less, putting more weight on the barometer data. This causes the estimated time of apogee to be closer to the correct value. This reduces the response to changes in acceleration. This new value is applied solely to TeleMetrum v1.0 boards. v1.1 boards correct for this error, and hence can use the correct sigma value for the accelerometer. Signed-off-by: Keith Packard