Re: [Paparazzi-devel] BMP085 sensor temperature wiht large error

[Martin Mueller]

Hi Helge,

yes, should be like you said. Committed. Seems they try to get some 
extra accuracy if the resulting number would still fit in 32 bits.

Thanks,
Martin

On 05.08.2011 21:47, Helge Walle wrote:
> Hi Martin,
>
> Thank you very much, that really made a difference! I am not sure if I
> would have found that one.
>
> While trying to analyze the code I noticed something in:
>
> void baro_bmp_event( void ) {
> .....
> /* compensate pressure */
> .....
> if (bmp_b7 < 0x80000000)
>     bmp_p = (bmp_b7 * 2) / bmp_b4;
> else
>     bmp_p = (bmp_b7 * bmp_b4) * 2;
> .....
>
> According to the BMP085 datasheet, should the last statement rather be:
> bmp_p = (bmp_b7 / bmp_b4) * 2;     ?
>
> Thanks again,
> Helge.
>
>
>
>
> 2011/8/5 Martin Mueller <address@hidden <mailto:address@hidden>>
>
>     Hi Helge,
>
>     the extra noise is from reading the lower three bits of the pressure
>     result wrongly. They are in the higher bits of the XLSB, not the
>     lower, fixed in master:
>
>     @@ -130,7 +130,7 @@ void baro_bmp_event( void ) {
>            /* get uncompensated pressure, oss=3 */
>            bmp_up = (bmp_trans.buf[0] << 11) |
>                     (bmp_trans.buf[1] << 3)  |
>     -                bmp_trans.buf[2];
>     +               (bmp_trans.buf[2] >> 5);
>            /* start temp measurement */
>            bmp_trans.buf[0] = BMP085_CTRL_REG;
>            bmp_trans.buf[1] = BMP085_START_TEMP;
>
>     Thanks for looking into that.
>
>     As the max conversion time is 25.5ms it could be run at 30Hz without
>     changes, maybe having the temp reading only once per second as
>     suggested in the spec. Some averaging could be done depending on the
>     frequency you need.
>
>     Martin
>
>     http://paparazzi.enac.fr/wiki___images/Bmp085_scp1000.png
>     <http://paparazzi.enac.fr/wiki_images/Bmp085_scp1000.png>
>     http://paparazzi.enac.fr/wiki___images/Bmp085_scp1000_fixed.__png
>     <http://paparazzi.enac.fr/wiki_images/Bmp085_scp1000_fixed.png>
>
>
>     On 04.08.2011 11:36, Helge Walle wrote:
>
>         Hi,
>         To make sure that the sample rate is not set so high that it
>         interferes
>         with the conversion time of the BMP sensor, I would need to
>         check the
>         data flow with my logic analyzer.
>         I did not do that yesterday evening. However, Increasing freq
>         from "8"
>         to "20" in .../paparazzi/conf/modules/__baro_bmp.xml increased
>         the data
>         flow and did not create any problems.
>         The code I used with my Arduino board that made better results
>         was from
>         Sparkfun. I am quite determined to work through the code since I
>         feel
>         sure that this is the right way to go.
>
>         Regards,
>         Helge.
>
>
>     _________________________________________________
>     Paparazzi-devel mailing list
>     address@hidden <mailto:address@hidden>
>     https://lists.nongnu.org/__mailman/listinfo/paparazzi-__devel
>     <https://lists.nongnu.org/mailman/listinfo/paparazzi-devel>
>
>
>
>
> _______________________________________________
> Paparazzi-devel mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/paparazzi-devel