PERFORMANCE OF THE NEW r PARAMETER FOR THE RATIOMETRIC ERROR CORRECTION
First of all a mistake was found in the calc. Excel File KP-235 Math 5(5V). Luckily this mistake did not impair the relevant computations. The amended File KP-235 Math 6(5V) is enclosed.
The performance of the proposed correction method works unbelivably well, reducing the ratiometric error to less than 1 mmHg over the full barometric pressure range, while meeting a power supply variation Vcc from 4,75 to 5,25 V! The following table compares the different situations.
Correction....Vcc = 4.75V............Vcc = 5.00V......................Vcc = 5.25 V
......... Max. error hPa/mmHg..... Max. error hPa/mmHg.......Max.error hPa/mmHg
No corr................± 64,1/± 48,2..........± 69,6 / ± 52,3.............± 75,1/± 56,5
With r = 0.75.........+29,34/+14....................0.....................-29,34/-14
With r = f(P)..........±1,33/±1......................0.....................±1,33/±1
In fact I am of the opinion that performance is even better, because the test was done while continuous changes in baropressure where happening. The PLW file with all measurement data is enclosed.
To eliminate PSU transients produced while changing Vcc and their effect on recorded baropressure, data has been copied and processed in an Excel file which is enclosed along with the generated graph. Of course no transients will be experienced with the chip fed by the USB Bus, as variations there will be extremely slow.
There area few gliches with this graph. Pse download the Excel file & browse it there.
This work has shown the great potential of contemporary software like Picolog in efficient handling of ratiometric type sensors, as the correction methods shown are of general interest and can be profitably applied to a great variety of devices. In the next instalment the second solution mentioned in Part 6, namely the application of a special minimal dropoff PSU will be described. This solution provides even better performance, zero error throughout, at the cost of more hardware.
Last edited by Glovisol on Sat Dec 05, 2015 12:59 pm, edited 1 time in total.
FINAL CONSIDERATIONS ON FEEDING THE BAROGRAPH WITH THE USB BUS VOLTAGE
Another analysis of the dynamic r parameter correction method is shown below, which again shows that error is negligible and not additive.
Dynamic r parameter correction analysis
Please carefully note the "cleanness" of the Baropressure waveform, in view of what is coming next. Up to the PC screen shown above, the Barosensor has been fed with variable voltage (4.75 to 5.25 V) coming from a lab PSU. But our aim is to feed the system from the USB bus. Here is the result.
Here at last, Barograph working off the USB Bus!!
Precision and correction are excellent, but unfortunately the USB Bus voltage is "dirty". In other words the short term stability of the PC derived USB voltage is far from good (contrary to what I had previously written " PSU Bus voltage variations will be extremely slow") so that the system records the small voltage change, then corrects: therefore the Barograph trace is not smooth, but has some "lumpiness", which however does not impair the precision of the reading.
Finally the "barefoot" schematic diagram of the Barograph, with direct, buffered connection to the DrDAQ (which eliminates the need of the Universal DrDAQ Interface) is shown below.
[The extension bmp has been deactivated and can no longer be displayed.]
It came to my mind that perhaps the short term instability I had found was peculiar to the laptop PC I was using (IBM / LENOVO X61 equippend with Windows XP O.S.). I connected the Barograph to another X61 and found a much better short term stability of the Bus voltage (as was perhaps to be expected). Will post the results here soon.
The PK235 Sensor was connected to a different IBM/Lenovo SX 61, equipped with Windows 7 O.S.
The short term stability of the USB Bus voltage was found to be notably better, 0.3 mV pk compared to 1 mV pk, as shown below:
PC screen using exactly the same time & amplitude scales as USB FEED 3
The "lumpiness" of the Barograph display is barely noticeable here. The next PC screen shows Barograph operation going from Mains to Battery supply and vice-versa. There is no significant error caused by the transitions.
A.C. Mains feed & Battery feed compared
To conclude, each PC has an USB Bus voltage with a different short term stability. Therefore with Parameter r compensation only, the Barograph display quality is somehow influenced by the characteristics of the PC used. In any case this effect does not noticeably impair the precision of the measurement.