0
\$\begingroup\$

I have an ADC circuit that utilizes an ADC128S102QML-SP. The ADC is powered by a 5V power supply that has an output voltage error of +/- 0.075V.

I would like to take the PSRR error into account for my ADC error calculation, but the ADC datasheet does not mention PSRR error.

Is there a way to calculate the error at the output of the ADC as a result of a change in power supply voltage?

\$\endgroup\$

2 Answers 2

1
\$\begingroup\$

Unlike an Op-Amp with high gain and error feedback sensitive somewhat to supply noise, this is a Track & Hold MUX ADC which also sensitive to supply, Va , the analog & reference voltage.

  • good question, unusual result

I would estimate the PSRR as 0 dB for all Vref and Agnd supply signals on SAR ADC's.

They indicate \$V_A\$ ...

This pin should be connected to a quiet 2.7 V to 5.25 V source and bypassed to GND with 1-µF and 0.1-µF monolithic ceramic capacitors located within 1 cm of the power pin.

I suspect you will want to analyze Va with an AC coupled 50 Ohm Spectrum Analyzer from a pair of test points and define your spec for this noise.

I recall my 1st ADC (late 70's) was an X-RAY inspected MIL-STD 883B Burr Brown 12 bit ADC that had missing codes at boundaries of xxxxxx111111. I ended up having to use the industrial part. At that cost, difference, being a newbie, I was surprised, not thinking about buyer/supplier feedback and getting the supplier to fix the problem < I did it my way to save time.

Now that was an metal cased ceramic hybrid with an internal problem. I surmised it was due to internal logic current offset Vref or gnd shift of a couple bits. It was corrected by luck, but that was an internal crosstalk or ground shift or Vref PSRR problem.

\$\endgroup\$
0
\$\begingroup\$

The analog supply is used as the reference, so a 75mV change in the 5V supply voltage would represent a 1.5% change in reading.

That's a 'span' error so 61 LSB at ~5V in linearly decreasing to no additional error at 0V in.

If the input voltage exceeds the reference then you can expect the digital output may saturate at 0xFFF so you may not see that error reflected at the very top of the range. For testing it would be better to use something like 4.900V.

Rad hardened, eh?

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.