A NEW LOOK AT PICOLOG AUTOMATIC RATIOMETRIC CORRECTION
THIS INFO ALLOWS YOU TO RUN THE BARO CHIP BAREFOOT FROM THE USB BUS ELIMINATING THE REGULATOR & EXTERNAL PSU!!
The already described barometer based on the KP-235 chip is more a prototype rather than a complete, self standing unit for two reasons:
a) It requires the Universal DrDAQ interface to download data to the PC.
b) It requires an external PSU working off A.C. mains. This is a serious drawback because the risk of losing/corrupting the data already recorded in consequence of a mains failure is very real. The only way to overcome this problem is to use the +5V of the USB Bus. This voltage is not stable nor accurate, as it can be anywhere between 4.75 and 5.25 V.
There are two ways to solve the problem of the USB Bus voltage drift: (I) adopt an improved Ratiometric Error Correction based on the powerful capabilities of Picolog or (II) use a special, extremely low dropout PSU, capable of delivering a very stable Vcc supply voltage to the PK235. In this section we investigate the first option.
In PART 4, Parameter derivation, we obtained parameter r by empirical means, testing two different barometer chips in the voltage range of ± 200 mV from 4.8 to 5.2 V (see Excel file KP-235 Math 2) and then taking the average of the three readings. This is a rather crude compensation method, because the measurement should be done on many devices and furthermore taking the centre value obviously leads to over-compensation or under-compensation at the extremes of the range. In short this compensation is valid only for very small deviations of Vcc, as supplied by a stable and accurate PSU.
In the enclosed Excel file KP-235 Math 5 (5V) which is built on the previous one, parameter r is now mathematically derived (cells in black rectangle) firstly in the same voltage range of ± 200 mV and secondly for ± 250 mV. The first derivation shows an excellent correspondence with the empirical data: r = 0.75, which agrees very well with the previous values of r = 0.7083 and r = 0.755. But even a more accurate parameter is not enough, if the mean value is used over the voltage and the pressure ranges together. The Excel file for ± 250 mV computes the values of r needed at min. mid. and max. baropressure, respectively finding the values of: 0.63, 0.75 and 0.87. If we use r = 0.75 in all cases, we end up with an error of + 14 mmHg (18.6 hPa) at 677 mmHg (900 hPa) and an error of – 14 mmHg (-18.6 hPa) at 842 mmHg (1120 hPa). This result is of course not acceptable.
The solution is to make the parameter r a function of raw baropressure, or r = f(P). This is done in the Excel sections in light blue. The plot of required r for varying baropressure is the linear function:
r = (0.001067 *P) – 0.3265 
The new column (in light blue) gives the optimum value of r for every value of baropressure. This calculation is of course done automatically by Picolog when  is introduced in position 8 of PART 3.
Writing this function in Picolog will provide the spot on value of r at every pressure measured by the sensor, thereby minimising the ratiometric error for any value of Vcc at any possible baropressure. To use this notably superior correction, just modify the Picolog instructions as follows:
8a) In Calculated Parameter: PSU Error. This is really the calculation of the Ratiometric Error factor h due to deviations in the 5V USB Bus supply (Vcc) and is calculated from parameter r, given by  above.
Variable C: 5V USB Output (Vcc)
Variable D: Raw Sensor Output NOTE: This entry wrong. Pse see next reply. Amended 29.11.2015
h = (5000 – C)*r = (5000 – C)*((0.001067*D)-0.3265)
Parameter formatting: mV
Digits width: 2
Decimal positions: 1
Min. Value: -20
Max Value: +20
Once this mod. is done, the Voltage Regulator chip can be eliminated and the K-235 fed directly by the USB Bus.