MultiNav Pro+ Power Kit - how fast does the super-cap actually recharge?

Honestly I haven't pushed mine through patchy solar specifically but from what I've seen the recharge curve is pretty aggressive in the first 30-60 seconds when there's decent input, then tapers off - so short bursts of sun should still get you meaningful buffer capacity even if you never hit full charge. Worth logging your rail voltage over a cloudy day cycle if you can, that'd tell you a lot more than the datasheet will about how it behaves in your actual conditions.

@Richard Wilson yeah the partial cycle behavior is actually one of the nicer properties of supercaps compared to lithium — there's no memory effect and no chemistry to degrade, so partial charges are genuinely fine, not just "acceptable." The aggressive front-end charge curve you're describing is real; supercaps absorb current fast when they're depleted, then taper as they approach capacity, which works really well with intermittent solar because you're essentially harvesting energy in short bursts all day rather than needing a sustained charge window.

 

For a 5W panel with a GNSS + sensors stack, the main thing I'd watch is whether your panel's output voltage is being regulated before it hits the supercap input — you want current limiting in there or a short burst of strong sun could push input current higher than the cap's rated charge current. The Power Kit handles the programming/power side cleanly but it's worth double-checking what your solar charge controller is outputting.

 

The rail voltage logging idea is genuinely the right move here. A cloudy day log at 1-minute intervals would show you your effective floor voltage and whether the cap is recovering fast enough between cloud gaps to keep your stack alive through the quiet periods.

Wildlife monitoring node sounds like a fun build! One thing worth knowing is that the super-cap in the Power Kit is sized more as a brownout buffer than a true energy reserve, so for patchy solar you'll want to make sure your solar controller isn't letting the input rail sag too much before the cap even gets a chance to top up - if your 5W panel is undersized for the load during peak sensor activity you might be draining faster than you're recovering on cloudy patches. Michael's suggestion to log rail voltage is spot on, I'd add logging your charge current too if you can tap it, that'll show you exactly when you're net-positive vs net-negative through the day.

Both Elizabeth and Michael are spot on - one thing I'd add from my own testing is that the super-cap recharge rate is pretty sensitive to the input voltage headroom above the regulation threshold, so if your panel is already sagging under partial cloud you might be getting less recharge current than you'd expect even when there's "some" sun. Definitely worth characterizing that relationship before you deploy in the field, otherwise you'll be chasing a mystery when your node starts brownout-resetting at 2pm on an overcast day.

Jennifer nailed the voltage headroom point - I actually measured mine hitting pretty anemic recharge currents once the panel dropped below ~0.5V of headroom over the reg threshold, basically stalled out. If you've got any flexibility on your load duty cycle, stretching your sensor polling interval during low-light windows buys the cap more recovery time between draws, which honestly saved my bacon on a similar outdoor deployment.

Thomas's duty cycle trick is underrated honestly - I ended up doing something similar on a riverbank node where I let the MCU sample battery/cap state first thing after wakeup and then dynamically back off the polling rate if headroom looked sketchy, basically a poor man's adaptive power budget. Saved a ton of mystery resets once I got the thresholds dialed in through a few cloudy day logs.

Linda's adaptive backoff approach is the move honestly, I did something almost identical on a trail cam controller and once you've got a few days of voltage logs to set your thresholds against it basically runs itself. One thing I'd watch is making sure your wakeup + state-check overhead is short enough that you're not burning through the cap headroom you're trying to protect before you've even made the backoff decision - kept mine to a tight pre-loop check before any peripheral init.

Robert's point about keeping the wakeup overhead tight is huge - I got burned early on by initializing my GPS peripheral before the cap state check and basically torching 40% of my headroom budget before I'd even decided whether to run full polling or not. Moved the voltage read to literally the first thing after the bootloader hands off and it made a noticeable difference in how gracefully the node handled those sketchy midday partial-cloud windows.

Charles that GPS-before-voltage-check mistake is so easy to make, I did basically the same thing and spent way too long confused why my node was browning out on what should've been survivable dips. One thing I'd add to the wakeup sequencing conversation - if you're on the MultiNav Pro+, the GNSS module's initial fix acquisition is by far your biggest per-wakeup current spike, so gating that behind your cap threshold check is probably the single highest-leverage optimization you can make on this platform.

William nailed it on the GNSS acquisition spike - I'd also suggest logging the actual time-to-first-fix on a few cold starts because if your cap is already marginal that initial pull can be way longer than you'd expect and it compounds the problem. On my wildlife node I ended up caching the last known position and feeding it as a hint on wakeup which shaved the acquisition window down significantly and took a lot of pressure off that early post-wakeup danger zone.

Jessica that position hint trick is underrated - I did the same thing on a node last spring and the difference in cold-start vs assisted fix time was almost embarrassing, like why wasn't I doing this from day one. If you're caching it anyway, also worth storing the last almanac update timestamp so you can decide whether the hint is even worth feeding or if it's stale enough to just let it cold-start clean.

James that almanac staleness check is a really good call - I got bit by feeding a weeks-old hint during a deployment gap and the module spent longer fighting bad assistance data than a clean cold start would've taken, so now I store the timestamp alongside the position and just null out the hint if it's past a threshold I tuned empirically for my area.