Hi,

There are a lot of scale manufacturers that claims to have calibration without actually putting weight on to the scale. How are they doing such a thing?

For me calibration is;

1-) Suppose you have a loadcell with a dead weight. Dead weight is unknown. Can change in time because of environmental factors.

2-) When weighing hopper is empty suppose 16 bit A/D conversion results in decimal 10000.

3-) You put a weight of 500gr. Now A/D conversion result is 60000.

4-) For 500gr weight decimal 50000 change. store that value.

5-) Put a unknown weight.

6-) If A/D is 40000. 40000-10000 = net 30000 change is A/D

7-) If 50000 changes for 500gr what weight changes A/D 30000. Solve equation and calculation result is 300gr.

For me this is how to calculate weight of a unknown product.

Suppose there is a problem on weighing result. Or when you put weight 500gr on scale it shows 500.5gr. How can you correct it without a reference?

As far as I remember AD7730 has internal full scale, internal zero scale, system zero scale, system full scale calibration methods. Where does these calibration methods stands with MY calibration method.

Is it possible to make a calibration without weight with AD7730 for example?

Regards,

DC

Hi DC,

Refer to CN-0102 which discusses the AD7190 in a weighscale application. This may provide you with some additional information.

If the excitation voltage is 5 V and the sensitivity is 1.987 mV/V with the max weight being 10 Kg, then a 10 Kg weight will produce an output voltage of 5 x 1.987 = 9.935 mV. The loadcell will have some offset and FS errors though. I thought your previous questions referred to this. With no weight on the loadcell, a measurement should be made. The voltage/weight measured is the offset. This voltage needs to be subtracted from all readings. The ADC is capable of performing this offset calibration (system offset calibration). The offset of a loadcell varies with time and temperature, so the zero weight error needs to be calibrated periodically. With the 10 Kg weight on the loadcell, record the voltage or weight generated by the loadcell. If, for example, the weight recorded is 10.1 Kg, then there is a gain correction factor is 10/10.1 = 0.99. All further readings should be corrected by this amount. The ADC cannot perform the gain error calibration as the signal applied to the ADC (9.935 mV) is too small. So, the gain calibration needs to be performed in the processor.

The accuracy achievable from the loadcell is dependent on the ADC noise also. So, for example, if the ADC has an rms noise of 8.5 nV (this is the rms noise of the AD7190 at a gain of 128 and an output data rate of 4.17 Hz), its peak to peak noise is 8.5 x 6.6 = 56.1 nV. If the FS signal is 9.935 mV, the number of counts possible is 9.935 mV/56.1 nV = 177090. So, for a 10 Kg weight, the ADC can measure 10 Kg/177090 = 0.06 grams. The loadcell itself will have some noise which will degrade the accuracy – my calculations are based on the ADC only.

So, for an unknown weight, lets say the voltage read back is 5 mV after the offset and gain error has been removed. This corresponds to a weight of (10 Kg/9.935 mV) x 5 mV = 5032 grams.

So, performing the offset and gain calibration is fine. However, I do not know how the offset can be removed in an application such as a silo without emptying the silo.

Regards,

Johnny