Looks like a probable formula change. Printer's fine.
I was quite concerned because the changes were so large. Did something suddenly happen to the printer?
So I investigated the gamuts of profiles made with before/after ink change. There was little difference at higher luminosities but at lower L*s the saturation on the +b* (with the a* held at 0) was higher with the new inks while with the -a* (with b* held at 0) the old inks had higher saturation. Other parts of the gamut were close, ie: orange, red, magenta and blue.
This led me to speculate the Cyan spectra differed and the new inks were cutting off earlier than the old one.
Next I compared the spectra for RGB (0,255,255) which produces pretty much pure Cyan.
The new ink exhibited both a lower peak amplitude but, more importantly a 3nm shift, with the peaks normalized, to shorter wavelengths along the region from 500nm to 540nm. This may not seem like much but it happens to be in one of the most sensitive areas of the visual response near the peak of the horseshoe CIEXY.
1. It accounts for the differences after I changed ink.
2. It's not possible for any change in the printer/driver/paper selection, etc. to produce this shift. It's a property of the Cyan (not Light Cyan) ink. The ink is dated 20180627 (manuf. I assume) and Exp 2010627
The largest dE 1976 occurred at Lab(20,-53, 8 ) v (16,-39, 5) which is about 15
Fortunately, this area is not very visually sensitive and the dE00 is much lower. But still!
Looks like I need to re-profile all the papers I use. No idea if it's just a bad batch but it is highly annoying.