Hi,

I am using AD7663 in my Multi-Channel DAS. My input signal is Bipolar, +/-10V. I need to estimate the Offset and Gain error associated with AD7663. These errors are not explicitly mentioned in the datasheet. I tried to attain those parameters as following;

Bipolar zero error : 0.06% of FSR - It is not mentioned that this error is inclusive of Gain error or not. If it is inclusive of Gain error then how to find the Offset error.

Bipolar Full scale error : +/-0.25% of FSR - It is not mentioned clearly whether it is inclusive of Offset error or not.

Pg:11, TPC-15, has information about Offset voltage. Please confirm that the TPC-15 illustrates the input referred offset voltage of AD7663.

If suppose I derive Offset voltage error from TPC-15, it comes around 0.0046% at +85DegC(i.e 3 LSB error). After the estimation of Offset error the Gain error is calculated as +/-0.245%. (+/-0.25% - +/-0.0046%)

Therefore, the maximum Gain error can be translated to 161 LSB. [0.245 * 20V/(100 * 1LSB) = 161 ; where 1LSB = 20/2^16].

Please confirm the calculation.

And also if there is any test results related to the external reference please share with me. It would be useful in error estimation.

If possible please explain the graph TPC-15. I couldn't understand the exact illustration of that graph.

Thanks,

Jebas.

Hi, Jebas.

The bipolar zero error spec does not include gain error. It will serve as your offset error in bipolar mode (it’s referred from midscale code rather than 1st transition). To get the gain error, you need to subtract the offset from the bipolar full scale error spec.

Your process of computing based on TPC 15 is correct however, at 85degC, the offset is 1LSB. Using this value, you’ll get a gain error of 0.248474%FSR or ~163 LSBs. If you notice on TPC 15, the arrow for the graph for offset points to the thinnest line. The thickest line is for the –FS graph while the other is for +FS graph. TPC 15 illustrates how the part drifts over temperature. Use the values on the spec table for calculations at 25 deg C.

Any errors coming from the external reference you intend to use will have to be added to overall total unadjusted error as they are not included in the specs provided in the datasheet.

Regards,

Karen