Electronic – Why does the data sheet for A1362 have conflicting procedure to program sensitivity

The preferred way to set up the Hall Sensor for initial experimentation would be to acquire the evaluation board, and then follow the instructions to use its software for calibration and configuration.

If that is not an option, similar results can be achieved manually, albeit with a bit more effort and some guesswork / empirical data. This is how it goes:

Required parameters:

• Expected magnetic field strength - active range expected?
  • Intrinsic, earth's magnetic field at surface = 300 to 600 milliGauss
  • Small toy magnet or fridge magnet = 30 to 150 Gauss
• Expected rate of change of field: How many cycles per second?
• V_cc of MSP430?
  • MSP430 digitizes from 0 to V_cc, and can tolerate up to V_cc+0.30 Volts
  • Permissible V_cc is up to 3.6 Volts
• Expected / desired full-scale range from Hall Sensor (assuming ADC10 of MSP430 is being used)?
  • Both upper and lower references can be changed from default GND and V_cc, to external reference voltages, 0 to 1.2 Volts as lower reference, and 1.4 to V_cc as higher
  • Internal voltages references of 1.5 Volts and 2.5 Volts are also available, to use as the full-scale limits.

Once the above parameters are determined, a suggested way to proceed would be:

1. Program quiescent voltage to around midway between the MSP430 ADC10 reference values you select
2. Determine actual sensitivity of your sensor using the magnet of choice, and a good voltmeter: The manufacturing tolerance range is pretty high. For instance, if using strong fridge magnets, the highest sensitivity range, 4.5 - 16 mV/G, could give a maximum of 4.8 Volts full-scale North-to-South, way more than the MSP430 can handle.
3. Program sensitivity based on above tests, don't use the maximum sensitivity mode of the part.
4. Don't count on built-in clamping of the device, use an external clamping arrangement if at risk of exceeding MSP430's V_cc+0.30 Volts.
5. Opt for ADC reference voltages in between the rail voltages instead of full 0 to V_cc, that way there is margin for error in case the sensor overshoots expected full-scale voltage, e.g. if a mobile phone rings nearby.
6. Use a capacitor on the bandwidth FILTER pin such that your bandwidth is limited to perhaps 10 times the expected cycles per second mentioned above - otherwise the output gets too jittery, even if a person with coins in their pockets walks by.

That's what I can recall off the top of my head, but if additional details are required, please edit the question accordingly.

I think you are under the misapprehension that R3 and R4 define the voltage VpM when Sw9 is closed. That is not the case : R3 and R4 only define VpM as VpM = V1 * R4/(R3+R4) when Sw9 is open.

Close Sw9 and the voltage VpM is defined by R3 and the parallel combination of R4 and the output pin Vout. We don't know very much about that; the datasheet does say (on page 2) that it can sink a maximum of 10mA. Therefore you may have to assume a current source in parallel with R4 capable of sinking anything up to 10ma (i.e. sourcing up to -10ma) and designing your programming voltage sources with that in mind.

The programming instructions do refer to programmable power supplies as programming voltage sources, which suggests that low impedance sources are required.

There is also the remark about 300ma and 0.1uf being required to actually blow a fuse : the current required during the remainder of the programming operations is presumably much less than this; probably of the order of the 10ma output sink current specified.