After two ADC damages in a well-filtered MK60DN512VLQ10 (4N22D) we are trying to chase the root cause. Possible candidates:
1) VADIN cannot tolerate VrefH
The last K60P144M100SF2V2 (6/2013) data sheet still requires a differential input of no more than 31/32 of VrefH on VADIN.
This was required together with e3863 for the older mask 4N30D:
Could an input voltage of VrefH degrade/damage the ADC in 16-bit differential mode and not only give a conversion error ?
Having seen no conversion errors on the 4N22D we used this mode for the full 3.3V (ADCx-DPx = VrefH = VSSA , ADCx-DMx = Ground) input.
Please see forum post [https://community.freescale.com/message/593170?et=watches.email.thread#593170] for damage description and circuit schematic.
2) Analog inputs not to be grounded
It has been a requirement for many years to ground open analog (only) inputs. Lately, for instance with forum post (https://community.freescale.com/message/572445#572445) and AN4747, it was suggested to leave them open. Having grounded all ADCx-DMx analog inputs we wonder whether this could be a damage cause?
3) Excessive Vref filtering
In our circuit we tried to provide good filtering for Vref & VSSA and might have overdone as forum member Chip Weller suspects.
While we definitely want to follow the advice, we are unsure about the best ADC solution.
We tried to follow the TWR-K60N512 schematic here (see Analog Supply TWR-K60N512.JPG). Now we see just a 0.1uF filter on the FRDM64 board (see Analog Supply FRDM-K64.JPG).
Question: Is the K64 ADC so much better that it doesn't need the extensive TWR-K60 filtering or has this TWR filtering caused unwanted side effects ? Would it make sense to switch to a K64 then?
We have stopped assembly of a 200 unit first series lot to hopefully sort out the damage seen in the pre-series units. Quick help would be very much appreciated.