https://scruffycatstudios.com/categories/engineering/Engineering - Category - Scruffy Cat StudiosHugo -- gohugo.ioenThis work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.Fri, 10 Apr 2026 12:00:00 +0000https://scruffycatstudios.com/posts/heathcliff-issue-212-postmortem/Fri, 10 Apr 2026 12:00:00 +0000Juan Segoviahttps://scruffycatstudios.com/posts/heathcliff-issue-212-postmortem/Short version: we investigated a faint audio pop in a specific filter sweep scenario, tried a narrowly scoped DSP change, found it caused audible regressions on constrained targets, and concluded the right follow-up is an architectural design spike — not a quick patch. Why this matters (TL;DR for busy readers) We did the investigation the right way: reproduce first, measure, hypothesize, change, and measure again. When a partial fix regresses real users (or MCU targets), it’s better to roll back, document what we learned, and plan a proper architecture change.