MPX 4115 output smoothing

[Kevin Walton]

Hi

I have linked up a MPX 4115 pressure sensor

[http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?
code=MPX4115&nodeId=01126990368716]

via a MAX 188 12 bit ADC

[http://dbserv.maxim-ic.com/quick_view2.cfm?qv_pk=1070]

to a PIC 84A.

The ADC is using a 5v reference (supply voltage) rather than the
internal 4.096v reference as ground preasure output of the MPX 4115 is
greater than 4.095 volts.

The measurments are very noisy though, a summary of 5000 readings over
15 mins:

meters adc output
max 536.79 3,472.00
min -131.45 3,176.00
mean 97.74 3,368.43
median 87.29 3,373.00
range 668.24 296.00
std dev 69.73 31.11
variance4862.00 968.00

and the graph in Excel:

http://mach.unseen.org/~kevin/electronics/altimeter/altitude1.jpg

Using the sugested filtering circuit from Motorola at:

http://e-www.motorola.com/files/sensors/doc/app_note/AN1646.pdf

It smooths the readings pretty well (although still not giving the 1m
resolution I would like):

meters adc
max 78.36 3,391.00
min 47.18 3,377.00
mean 70.31 3,380.61
median 73.90 3,379.00
range 31.19 14.00
std dev 6.52 2.93
variance42.50 8.56

http://mach.unseen.org/~kevin/electronics/altimeter/altitude2.jpg

but causes a large skew in the data both as an ofset throughout and a
'charging' curve at the start - the picture explains a thousand words


Have I done something wrong, missed something?

Ideally I would love an answer of 'this is how you get better than 1
meter resolution', but I'll appretiate all the help avaliable to solve
the 'skew' problem that is causing me most grief.

Thanks
Kev

 

[Kevin Walton]

Hi

I have linked up a MPX 4115 pressure sensor

[http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?
code=MPX4115&nodeId=01126990368716]

via a MAX 188 12 bit ADC

[http://dbserv.maxim-ic.com/quick_view2.cfm?qv_pk=1070]

to a PIC 84A.

The ADC is using a 5v reference (supply voltage) rather than the
internal 4.096v reference as ground preasure output of the MPX 4115 is
greater than 4.095 volts.

Through additional supply filtering and some 'luck' I have reduced the
noise to the LSB. I now have a question on a slighlty different tack.
I would like to increase the resolution of the readings, trading off
range.

I don't want to change the A2D as I know this one works, have the code
sorted to read from it etc. I don't want to add an opamp due to the
additional complexity and chip space (this is going into a rocket) so
I am looking for more innovative/simple solutions.

The 'desktop' reading is around 4.1v, I would ideally like a range of
around 4.2 to 3.5 volts, but appretiate such specificness can probably
only be achieved by an opamp.

Any simple solutions or am I looking for the impossible and should
just pick up a rail 2 rail opamp and learn how it works?

Cheers
Kev