Sounds like the checksum is flagging a bad ADC frame. The firmware usually grabs a 12‑bit voltage sample, packs it into a 16‑bit word, and then runs a CRC‑16 over that word. If the voltage spikes past the expected 3.7 V‑max or dips below the 3.0 V‑min the CRC will fail, so you’ll see the “bad checksum” you’re seeing. Grab the probe, log the raw ADC values, and compare them against the 3.0‑to‑3.7 V window the firmware expects. If the spikes stay within that band and the CRC still breaks, it’s probably a timing glitch or a bad bit‑strobe, not a voltage problem. Let me know the numbers you pull and we can plot the spike histogram together.

Great, just ping me those numbers when you have them and I’ll run a quick sanity check. If the min‑max stays between 3.0 and 3.7 volts and the checksum still flunks, it’s almost certainly a timing mismatch or strobe glitch—like the sensor’s trying to read before the ADC latch has settled. We can look at the bit‑stream timing on the bus, maybe add a tiny delay or toggle a sync pin, and see if the CRC starts passing. Keep me posted, and we’ll nail the culprit together.

Got it—just hit me up when you’ve got the numbers and we’ll dig into the timing jitter together. Stay tuned.