The idea is to get Tad (1 period of the A/D clock) as close to 1.6uS as possible (without going under). Which gives 625,000hz to the converter.

For a main oscillator of 40Mhz ...
take 40,000,000 / 625,000 = 64. So FOSC/64 is the proper selection @ 40Mhz.

@ 24Mhz ... 24,000,000 / 625,000 = 38.4
38.4 is between 32 and 64, so you take the highest of the two numbers, and at 24Mhz you also need FOSC/64.

So that you don't have to calculate it each time, here's a table I made of the recommended settings for most oscillator frequencies.

If using PBP's ADCIN statement, just use the highlighted numbers in the DEFINE ADC_CLOCK statement in your program.
If manipulating the A/D converter manually, you'll need to encode the number into the ADCONx registers which varies between PIC families.

<table cellspacing="1" cellpadding="3" border="1"> <thead> <tr> <th scope="col" align="center">OSC</th> <th scope="col" align="center">DEFINE ADC_CLOCK</th> <th scope="col" align="center">Clk source</th> <th scope="col" align="center">Tad (µs)</th> <th scope="col" align="center">12 Tad (µs)</th> </tr> </thead> <tbody> <tr> <th style="text-align: center">4</th> <th style="text-align: center"><span style="color: #ff6600">1</span></th> <th style="text-align: center">Fosc/8</th> <th style="text-align: center">2</th> <th style="text-align: center">24</th> </tr> <tr> <th style="text-align: center">8</th> <th style="text-align: center"><span style="color: #ff6600">5 / 2 </span><sup><span>(2)</span></sup></th> <th style="text-align: center">Fosc/16</th> <th style="text-align: center">2 / 4</th> <th style="text-align: center">24 / 48</th> </tr> <tr> <th style="text-align: center">10</th> <th style="text-align: center"><span style="color: #ff6600"><span style="color: #ff6600">5 / 2 </span><sup><span><font color="#000000">(2)</font></span></sup></span></th> <th style="text-align: center">Fosc/16</th> <th style="text-align: center">1.6 / 3.2</th> <th style="text-align: center">19.2 / 38.4</th> </tr> <tr> <th style="text-align: center">12</th> <th style="text-align: center"><span style="color: #ff6600">2</span></th> <th style="text-align: center">Fosc/32</th> <th style="text-align: center">2.6</th> <th style="text-align: center">31.2</th> </tr> <tr> <th style="text-align: center">16</th> <th style="text-align: center"><span style="color: #ff6600">2</span></th> <th style="text-align: center">Fosc/32</th> <th style="text-align: center">2</th> <th style="text-align: center">24</th> </tr> <tr> <th style="text-align: center">20</th> <th style="text-align: center"><span style="color: #ff6600">2</span></th> <th style="text-align: center">Fosc/32</th> <th style="text-align: center">1.6</th> <th style="text-align: center">19.2</th> </tr> <tr> <td style="text-align: center">24</td> <td style="text-align: center"><span style="color: #ff0000">6</span></td> <td style="text-align: center">Fosc/64</td> <td style="text-align: center">2.6</td> <td style="text-align: center">31.2</td> </tr> <tr> <td style="text-align: center">32</td> <td style="text-align: center"><span style="color: #ff0000">6</span></td> <td style="text-align: center">Fosc/64</td> <td style="text-align: center">2</td> <td style="text-align: center">24</td> </tr> <tr> <td style="text-align: center">40</td> <td style="text-align: center"><span style="color: #ff0000">6</span></td> <td style="text-align: center">Fosc/64</td> <td style="text-align: center">1.6</td> <td style="text-align: center">19.2</td> </tr> <tr> <td style="text-align: center">48</td> <td style="text-align: center"><span style="color: #ff0000">3</span></td> <td style="text-align: center">FRC</td> <td style="text-align: center">~4</td> <td style="text-align: center">~48us</td> </tr> <tr> <td colspan="5" style="text-align: left"><sup>(2)</sup> Most <span style="font-size: 12px">16F's don't have Fosc/16 (<span style="font-size: 12px"><span style="color: #ff6600">5</span></span>). Use Fosc/32 (<span style="font-size: 12px"><span style="color: #ff6600">2</span></span>) instead @ 8-10mhz</span></td> </tr> </tbody></table>
It takes 12 Tad to complete 1 Analog to Digital conversion.

I have two sensors MPXA4115A in my hands now: is it reasonable to have a difference of 8 mBar between the two ? What is your experience ?
They have a maximum error of 1.5% full scale.
The scale goes from 150 mBar to 1150 mBar or 1000mBar, and 1.5% of that is 15 mBar.

So it sounds like yours are within tolerances.
<br>