I have questioned the data sheet for the MPX5700 recently and Motorola is at
a loss as to how the temp info should be interpreted. I found significant
variation (>2.5%) when going across a threshold at somewhere around 70
degrees F. in a number of units tested so far. In my app, the error amounted
to about 3 ft. of depth, in fact my system would turn off, thinking it was
no longer submerged when used in colder water. Initially I submerged the
unit from 70 degree F ambient into 53 degree water. It turned on as soon as
it went below 1 ft, as it was supposed to. After cooling from the new
environment, it would shut down. This drove me crazy searching in all the
wrong directions, because I first trusted the data sheet. Several ap
engineers have responded, and apparently can't verify the data sheet info
yet. It seems they can't give an exact interpretation of the listed error
figure. They have me waiting as they investigate further. It is most likely
that their full-scale error percentage figure means that at a single
temperature, there should be no more than that amount of error across the
full range, sort of a linearity figure. I'll wait to see what they come up
with before re-writing my algorithm.

I appreciate the additional thought on the temp of the in-hull air and
pressure on a differential sensor. You can tell I'm stuck in absolute and
gauge modes. Good point.

Dave, what temperature range have you tested your design in?

Chris
As for temperature errors, looking at the Motorola datasheet for the
MPX5050 it appears the sensor and its integrated electronics will have a
temperature-induced error of  1% of full-scale or less from 0-60 deg. C
(i.e., +/- 0.075 PSI maximum error, equivalent to a depth error of 2"
H2O).  And you're hardly likely to encounter this much temperature
change.

A much bigger error source will be the change in air pressure inside the
hull caused by changes in the air temperature.  With a fixed mass of air
trapped in a fixed volume, the pressure will be proportional to absolute
temperature.  To compensate for this you can use a PTAT temperature
sensor like National's LM335 and measure its output with the PIC A/D,
and do the compensation in software.

Frankly, I think it would be a whole lot easier to just use an AP sensor
like Motorola's MPX4200A (3-29 PSI absolute) instead; no fuss, no muss.
I'm using these things in a product and have found them to be very
stable with temperature.  Great little devices.

Hope this helps a bit...

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.