Accuracy analog outputs

Hi,

i'm comparing the accuracy of some analog inputs

S5 95U 0-10V 10Bit

S7 314C-2DP 4..20mA 12Bit

Endress+hauser SWA70 4..20ma measuring error is 0,125% of span https://portal.endress.com/wa001/dla/5000373/0002/000/03/TI00026SEN_1310.pdf

Working with a pressure sensor and a volume of +- 40000dm³

I know i need to work with the product density

What can be the deviation

10bit = 0 - 1023 = 1024
40000/1024 = 39,0625

12Bit = 0 - 4095 = 4096
40000/4096 = 9,7656

= 40000*0,125% = 50

Am i right ?

denhondt said:

Hi,

Am i right ?

Click to expand...

No.

Like many before you, you are confusing accuracy with resolution. The number of bits the A/D conversion uses defines the resolution of the input. In other words it tells you how much the input signal must change before the data in the PLC register changes. For a ten bit input the resolution is +/- 1 part out of 1024, or +/- 0.1%.

On top of that you have accuracy - the variation of the signal or data from the true process value. It includes nonlinearity and zero offset and hysteresis and such. You will never get better than +/- 0.125% because that is the accuracy of the transmitter. Then you have the accuracy of the input card itself, errors from off spec resistors and bad calibration etc. That is often expressed in percentage on the card specs.

So, let's say you have a ten bit card with an accuracy or +/- 0.2%. You total accuracy would be:
Transmitter x Card Resolution x Conversion Error = 0.99875 x 0.999 x 0.998
= 0.9957 or +/- 0.43%

Therefore your maximum error would be +/-172 dm3

If you kept the same transmitter and had a 12 bit input with +/- 0.2% accuracy your maximum error would be +/- 140 dm3.