MSP430F2618: What's the calibrated ADC12 accuracy? Where is this documented?

Cameron:

> I checked the data sheet, and it doesn't mention the values being uncalibrated, those
> values should be for a calibrated ADC. What's giving you the impression that it's
> uncalibrated?

(I apologize for the delay in responding! I had to spend significant time working on our
older MSP430F2618-based product but yesterday, I was finally back working on the
ADC12 subsystem in both families.)

I guess what gave me this impression is that the only part of the Data Sheet that mentions using
the TLV Calibration Data is the temperature sensor associated with the ADC12:

All of the other paragraphs dealing with either the ADC12 or VRef are silent on the topic of calibration.

And when one looks at, say, the data for REF:

One sees VRef+ specified as +/-1%. Give that the calibration factor is specified in terms
of parts in 32768 (~30ppm), I'd have expected that the calibrated accuracy of the Voltage
Reference would be better than +/-1% (+/- 10,000 ppm). Instead, it should be about whatever
the accuracy of the instrumentation was in the chip tester. (I used to work for Teradyne, BTW,
and I'm proud of the accuracies we could achieve; for stuff like DC voltages, they were 'way
better than +/-1%, being calibrated back to an HP 5-1/2 digit systems voltmeter. ;-) )

The same thing comes up when I look at the ADC12 specs:

The TLV contains calibration factors for the ADC12's Gain and Offset and again, these are
specified in terms of parts in 32768 (~30 ppm). But Offset error is specified as +/-5.6 LSBs
(+/- 0.14%) max and Gain error is specified as +/- 2.5 LSBs ( +/- 0.06%) max. I know we're
starting to push up against instrumentation accuracy limits here, but even these might have
been calibrated to tighter limits.

It's okay if the Data Sheet specs really are the calibrated values, but I guess I'd like to see
the Data Sheet state that for both the ADC12 and VREF peripherals rather than simply staying
silent. And I'm asking, in part, because my Systems Engineer is pushing me on "Just how
accurate are these ADC channels?" so I really need to be able to tell him, "Yes, what the
Data Sheet says is the best we can do" or "No, using the TLV calibration factors, we can
do x-amount better".

Atlant

Atlant,
I see your point. I'll double check with our systems team. In fact, you may be right. I noticed in one of our FRAM parts, it mentions it was tested without TLV calibration.

Cameron:

Thanks! If the ADC can be more accurate than what's stated in the data sheet, you'll make our System Engineer (and our Controls Engineers who are the ultimate customers of the ADC data) very happy. ;-)