Freescale KM1 and KM3 '16 bit ADC' performance?

[splin]

Has anyone used the Freescale Cortex-M0+ 'metering' MCUs which headline a 16-bit ADC? It's hardly 16-bit in reality, being a 12 bit with a bit of hardware oversampling.

I've been offered some KM1s (MKM14Z64) for free, but is it even worth bothering spending the time evaluating the ADCs?  I'd like to get 14+ bit noise free bits  with reasonable repeatability (yeah I know the INL won't match,  but it may be possible to improve that that with calibration). The specs are somewhat vague so I wouldn't expect much. I can't make any sense of the ENOB values in Table 28 - what are the last two entries supposed to be?

I wouldn't be surprised, for example, if the only way to achieve the data sheet noise specs is to operate the ADC with virtually everything else turned off.

So anyone think I'd be wasting my time and money trying these out? I'd have to get some PCBs made at the minimum.

Freescale published some comprehensive application notes for there use in power meters etc.  but none properly addresses the accuracy over time and temperature - they only say that 'we built and calibrated a meter and achieved these accuracies'.

[Dielectric]

You probably want to look at the sigma-delta ADC in the AFE instead.  Their best-case SNR spec works out to about a 15-bit ENOB.

[splin]

You probably want to look at the sigma-delta ADC in the AFE instead.  Their best-case SNR spec works out to about a 15-bit ENOB.

Good advice, except that no-one's offered me any free AFEs! Unless you're offering some? 

It doesn't look like any viewers here know much the ADCs specifically, but any more general experience with the Freescale MK series Cortex MCUs? Any bad experiences with poor documentation,  bugs,  dev-tools etc?

[Dielectric]

Maybe I was looking at a different datasheet, but the one I found had an AFE peripheral with the delsig ADC in it.  I've got no dog in this hunt, I was just curious how they'd spec a 16-bit SAR ADC since they typically sort of suck.

Just LOL if you're basing an architecture choice on getting a couple of freebies anyway.

[splin]

Maybe I was looking at a different datasheet, but the one I found had an AFE peripheral with the delsig ADC in it.  I've got no dog in this hunt, I was just curious how they'd spec a 16-bit SAR ADC since they typically sort of suck.

Doh,  for some reason I thought you were talking about one of the dedicated AFE chips (TI,  Microchip) etc such as those for metering applications with 'reasonably well' (at least compared to the MKM14Z64's delsig ADCs)!

The AFE may prove useful but it's too slow for my needs hence my focus on the so-called '16 bit SAR',  which actually appears to be a 12 bit ADC with some averaging hardware. Clearly its performance will totally suck compared to a real 16 bit SAR.

I remember watching a video years ago, by,  I think, a Rensas Engineer/Marketing support bod about the specifications and performance pitfalls of integrated ADCs.  He explicitly ridiculded the idea of a 16 bit ADC saying that even 12 bits was a huge challenge given the noisy environment on an MCU mixing with high-speed digital logic.  I got the impression he was referring to the Freescale 16 bit SAR ADC given that was the only one advertised at the time that I was aware of. Unfortunately I can't find it now probably due to the usual Internet bit-rot.

Just LOL if you're basing an architecture choice on getting a couple of freebies anyway.

It's a lot more than a couple, but even so it obviously wouldn't justify their use on cost grounds compared to the development time. Part of the motivation though is to have a bit of fun with a new MCU family - even if it is obsolete.