Notices by joey castillo (joeycastillo@mastodon.social), page 2

New battery test, day 700: 2.84 volts, as I extend the X-axis another hundred days.

I now have enough data to posit that 2.85 volts is the 70% mark for battery exhaustion: old firmware hit it on day 300 and died on day 425; Ben's watch hit it at day 440 and died at day 622. By that math, I predict 292 days remain, for a death date around November 8th. Which means this little watch that could might just make it to its THIRD @hackaday Supercon on a single CR2016 coin cell.

New battery test, day 675: 2.85 volts. Number hold steady! Tho I admit I've been on and off with this watch the last few weeks; I've been splitting my time between Sensor Watch and a stock Casio W800H, just to get a feel for a larger watch on my wrist. Anyway, onward to 2024!

New battery test, day 650: 2.86 volts. Number go down! But even when number go down, watch can still function just fine, down to 1.62 volts (which realistically we won’t get to until the battery has but minutes left). If I had to guess, I’d guess I have 30% of my CR2016’s capacity remaining, which means we should easily make it to the two year mark, and I should be able to wear it all semester and well into the summer. If I’m being bold? I think it‘ll tick from February 2022 to September 2024. 🤞🏽

@derek I have one that I’m pretty happy with, but alas I’ve been so busy with teaching this semester that I haven’t had a chance to put it in the repo. I’ll try my best to make some time to uploaded it this weekend or next.

working on my week three lecture slides, on transducers and sensors, and once again I'm going hard. Just finished a photoshop job on a classic illustration; Thanks, Phototransistor Man!

no matter what safeguards you try to build to help internet users, the modern web is enough of a cesspool that some asshole is going to come up with something like this.

My sister got me circuit board socks for Christmas and I’m relieved to discover that they are, in fact, CE marked.

@josecastillo Battery test, day 380: 2.59 volts. To reiterate the parameters of the test: this watch (red line on the graph) is running an old firmware from 2021 that lacks some low power optimizations I did early this year. I also started a new watch (green line) in February of this year that I've been wearing daily since; it's still sitting pretty at 3.08 volts here on day 275.

I'm going to post a Sensor Watch battery update later today, and this post is to connect it to the old battery test thread. If you're new to this: in late 2021, I put a fresh coin cell into Sensor Watch, my board swap for the Casio F-91W wristwatch. This month that watch hit one year on that same battery, and in this thread I'm tracking how long the battery will last. You can click the “Replying to @josecastillo” link repeatedly to get back to the beginning of the test. https://twitter.joeycastillo.com/josecastillo/status/1590059216798388224/

@josecastillo Battery test, day 410: 2.195 volts. (Mostly 2.19, but it did hit 2.20 once). No more big voltage drop when I press the mode button, which signals to me that its steady state current consumption is higher now. That and the steepening of the red line 😬. This microcontroller operates down to 1.62 volts, which gives me optimism that this watch with the old firmware will still be with me in 2023, if only for a few days. (New firmware, green line: 3.07 volts at day 305.)

@josecastillo Battery test, day 400: 2.4 volts. It’s low enough now that the extra power of beeping the buzzer on mode change causes a voltage drop to ~2.3, but it recovers to 2.4 over about ten seconds. New firmware is back up to 3.08 volts here on day 295! Both lines are red this time, but it’s the one on top.

@josecastillo Forgot to post this yesterday! Battery test, day 390: 2.51 volts. I think we're finally near the end with the old firmware; I expect it'll die within a week or two. There's an app note that once the LCD voltage is more than 700 mV above the system voltage, current consumption increases, and I think I can see that on the red line: it's dropping faster now that we've crossed the threshold. New, more efficient firmware though: still going strong! 3.07 volts at day 285.

Napkin math suggests that if we accept the rated capacity of 100 mAh and divide that by the 10,198 hours of runtime we achieved, we’d get an average current consumption of 9.806 microamperes — which tracks astonishingly close to the hypothesis I posted at the outset, fourteen months ago.

BATTERY TEST, DAY 425: THE END. I measured 1.8 volts two days ago, which officially marks the end of the graph’s Y-axis. Today I woke it up for this test and watched it sag from 1.7 to 1.4 volts as I changed modes, the LCD dimming slightly in between each button press. Then I caught the exact moment this test ended: 11:46 AM central time, or 10,198 hours continuously running the old firmware. 🫡

Battery test, day 420 (and 315): 1.97 and 3.07 volts. Might be the last time these two get photographed together. I also brought them both into the backcountry, where they hung on a branch for one last night of temperature logging. It’s wild that running on this same coin cell, that gray watch logged temperatures in Big Bend in December of 2021. Even with the old firmware, this watch ran continuously for all of 2022, and saw three calendar years. And at just shy of 2 volts, it’s still going…

NEW battery test, day 330: 3.07 volts. While the 2021 watch is dead, the stainless steel watch running the February 2022 firmware keeps running strong. In just over one month’s time it’ll hit the one year mark, and still today it sits at nominal voltage. Bold prediction: I’ll still have this very watch, powered by this same coin cell, on my wrist come January 2024.

NEW battery test, day 350: 3.06 volts. The one year mark approaches! Also: I caught up with my friend Ben on day 348; he's been wearing a green watch that started its life on the same day. One difference: he left the hourly chime on, and his battery is at 2.94 volts over this same period. THIS IS A GREAT DATA POINT! It tells me that year-long battery life still holds with the chime turned on, and it's still more efficient than the old firmware (which was at 2.8 volts at this point in the test).

Also just to give a sense of perspective: the graph I keep posting is the manufacturer's data sheet scaled by a factor of ten. In an apples-to-apples comparison, you can see where the blue line, the manufacturer's simulation, dies at 1000 hours; how the Sensor Watch with November 2021 firmware outlasts it to run for over a year; and how the February 2022 firmware outlasts even that test. With improved crystal and cap selection, the new hardware revision could outlast even this one. Wild stuff!

New battery test, day 365: 3.08 volts. One year on the wrist and we’re still at nominal voltage! This signals strongly, to me, that we aren’t yet at the halfway point of this test, and that as configured and shipped to backers, Sensor Watch may in fact have TWO-year battery life. Anyway I’m going to science about it for the rest of 2023; we have fucked around with the firmware, and now we find out what happens. Isn't that what the scientific method is all about?

Quick followup on day 354: it's been at 3.07 volts ever since, and I realized: in the days leading up to day 350, I'd been wearing a different watch to test some firmware changes. As a result, the steel watch had been sitting at room temperature for about a week — not warmed up to 90° F by my wrist. Lower temperature, lower voltage? Anyway it'll be a few weeks to months before I'm sure, but I think we're still at the top of the discharge curve; I don't think it's started curving just yet.