I believe the vanilla behavior may be overriding the upgrade chances, because with 32% quality in an assembler I have:
Normal -> Uncommon 32%
Uncommon -> Rare 25.6%
This would suggest an outcome (for normal) of 68% normal, 23.8% uncommon, and 8.2% rare, if the exploding chance of uncommon -> rare gets the full 25.6%.
However, after performing a broad test (at rare cap), with 32% quality, I I turned up 68,120 normal, 29,387 uncommon, and 2,704 rare. This would suggest a distribution of about 68.0% normal, 29.3% uncommon, and 2.7% rare, which would suggest that the vanilla behavior of 10% exploding chance of upgrades has overridden the 25.6% chance to explode from uncommon to rare.
If this is intended, it should be written explicitly in the information section, because the tooltip REALLY suggests that the vanilla explode chance is being overridden. I would also suggest making the tooltip reflect the actual exploding upgrade chance to alleviate confusion.