Here are a couple of plots:
M215IG manufactured in week 50 of 2016 (Part no. 800-00363-r02) running image 520-00044-r1-v01.23.00:

M215IG manufactured in week 45 of 2015 (Part no. 800-00351-r02) running image 520-00044-r1-v01.22.00:

These inverters sit side-by-side in the array since the newer one was provided by Enphase as a replacement for a failed inverter from the same lot as the older one. Both are being read by the same Envoy-S and both are reporting data every five minutes. The difference seems to be that the newer one appears to report TWO data points at (nearly) each timepoint, thus creating the stairstep look. Note that the Envoy-S recently updated the firmware in ALL the older inverters in the array and none of them were updated to the same level as the newer ones.
Somewhere along the line, Enphase has moved to fully-digital controllers for their inverters and the newer graph looks similar to (but not exactly like) a discrete-amplitude plot. I have no idea if this is related to what I am seeing, or not.
What I don't understand is why Enphase would change to this new format, particularly at a time when they are changing the recording intervals for many customers from five minutes to 15 minutes. IMO, the older, smoother format would not be nearly as objectionable to most people, even if 15-minute intervals were used.
Anyway, it's just a curiosity. It's certainly not critical, but I'm just wondering if others are seeing the same thing in their results. I'll be installing some of the M250-72s next week and expect to see the same jagged curves as I see with the 2016 inverters.