Tweets
Replying to @josecastillo
so the SAML22 has four sleep modes: Idle, Standby, Backup and Off. Off isn’t very useful; the only way back from that is to press the reset button, which is inside the watch case. I’ve been imagining mostly using Standby mode, where the clock or the buttons can wake up the watch.
(original)
whoa! while reading a microcontroller data sheet at 1am (AS ONE DOES) I realized that completely by accident, I may have designed the watch + accelerometer gadget to take advantage of an even lower power mode than I had hoped for. Notes, thread; 1/?
(original)
This thread reverse engineering the AirTag is riveting to follow; keep scrolling, things are happening! Also in case Twitter crop hides it, this is a dump of the AirTag’s SPI Flash chip. The first block contains an ASCII art STOP sign, which isn’t stopping anyone :) https://twitter.com/colinoflynn/status/1390486554586587139
(original)
Replying to @josecastillo
The thing of it is, the whole time, I was surrounded by people that supported me and nurtured my interests. That’s my origin story; if I do anything at all that works today, that’s the only reason why.
(original)
Replying to @josecastillo
really most of my origin story is things that didn’t work. In grade school borrowed manuals for software from my dad’s office. Tried to clone it in Turbo Pascal. Barely started; never finished. In high school I tried to write a GUI for the TI-83 in Z80 assembly. It never worked.
(original)
When I was 10 my family’s Windows 3.1 computer had a map of the world as a background. I opened it in Paintbrush and used the fill bucket to paint it rainbow colors. Accidentally hit save; then tried my best to fill it back in with the closest colors I could find. It didn’t work. https://twitter.com/RayRedacted/status/1387834538873827329
(original)
Replying to @josecastillo
I shared something like this before, but it was old Atmel demo code I’d hacked and contorted beyond recognition. It also wasn’t low power, and didn’t use interrupts for button presses. This code actually makes use of the nascent framework I’m writing for building watch apps :)
(original)
THE MARS WATCH LIVES! And the code is in decent enough shape that I can share. This is still just code to test and demonstrate what’s possible, but I think it’s pretty cool; with a bit of extra math, it could show local time and sol for each of our rovers. https://github.com/joeycastillo/Sensor-Watch/commit/7ff91a2f678a5eb1f6b985a257a8e59bde6a973f#diff-4043f8a5e3e4ef74f8f6100c676aaeb2dfd3c429506ca4f0ca0b698d5c51c756R70
(original)
Replying to @sneenyc
huh! That would explain a lot; the situation has become viscerally worse since installing Big Sur. Didn’t know there was a BIOS change. Thanks for the info!
(original)
Replying to @sneenyc
thanks! but I’m fairly sure it’s just atmel studio being a beast, and parallels working overtime to try to tame it. I just need to dust off my old windows machine; running this in emulation is not working.
(original)
Replying to @bigjoshlevine
“In general, if a Web page has more than three bits of entropy, different browsers will generate extravagantly unique mappings between the Web developer’s intentions and the schizophrenic beast palette that browsers use to paint the world.” https://scholar.harvard.edu/files/mickens/files/towashitallaway.pdf
(original)
RT @szeloof: TIL wood is… piezoelectric??? what?
https://sci-hub.st/10.1143/JPSJ.10.149(original)
Replying to @emilesnyder
Oh god this was so me last week. I guess technically I’m booting four operating systems, depending whether you count the broken Windows VM https://mobile.twitter.com/josecastillo/status/1387429599706034180
(original)
Replying to @josecastillo
Is it okay to give up on your dreams? In the long term, no; you gotta stick to that shit. But in the short term, absolutely yes: it’s getting dark, and you should probably eat some dinner.
(original)
Replying to @gennyble
Atmel Studio, which is very Windows-only.
(original)
Replying to @josecastillo
I HAVE DONE SO LITTLE I HAVE GONE BACKWARDS. HALP. I AM NOT GOOD WITH COMPUTER.
(original)
started off with the modest goal of hacking on the watch for a bit; my Mac spun up like a turbine and got so unresponsive I had to reboot with the power button. now parallels won’t start, and I’ve wasted an hour accomplishing nothing. I guess I broke my system, so less than that.
(original)
Replying to @kkress
I have, so cool! I also think it would be interesting to print Deep Dream-style feature visualizations onto cloth as an adversarial attack. Nothing to see here; I’m just a verdant hillside walking down the street :) https://microscope.openai.com/models/contrastive_16x/image_block_4_7_Add_6_0/191
(original)
Replying to @kkress
Looking forward to not getting a cold this fall; also looking forward to defeating pervasive surveillance tech with a fashionable cyberpunk bandana. masks are cool.
(original)
Replying to @enzinolombardi
Very soon, I hope; maybe as soon as the end of the month? I have parts on hand for a first run of 20, just working out one last bug in the bootloader, and need to build a programming and test jig.
(original)
Replying to @PantsSlide and @the_prepared
Some folks at the shop also plugged one of these analog accelerometers into it, which I thought looked very cool. You can see the voltage smoothly change as you rotate the sensor, and jump around sharply when you shake it :) https://www.adafruit.com/product/163
(original)
Replying to @Jaxter184
I could brainstorm some more but I’m just saying that the feed, in being both easy to implement and highly engaging, tends to suck all the oxygen out of the room. Personally I’d be interested in folks trying something new and more humane than yet another news feed social app.
(original)
Replying to @Jaxter184
The virtual party model: you and some friends are in a virtual space in real time; sound travels as it does IRL, so you can talk to each other or walk around and meet new people. It’s inherently time-limited (people don’t stay at a party forever) and interactions are ephemeral.
(original)
Replying to @Jaxter184
The personal site model: you and your friends have a page where you put things that matter to you. You can visit their page to see what they’re up to and maybe leave them a note, but neither the things nor the note live in a feed; you have to visit your friend’s page to see them.
(original)
Replying to @Jaxter184
The forum model: information is arranged by topic, not by time or by person. You start a conversation by posting a topic, topics bump to the top as people talk about them, and individual conversations are ordered chronologically (i.e. not in reverse like a feed).
(original)
Replying to @Jaxter184
I agree that feeds have a place; I guess I just lament the monoculture of lists, whether ordered by “hottest” or “latest.” Like, let’s pretend the feed didn’t exist; it was never invented. Let’s just play with some different interaction models for social media. 1/?
(original)
Replying to @gennyble
The 1st one, taking the most recent and/or engaging thing and shoving it near the top of a list. That’s not how human beings interact with each other. Consider the MySpace model: you have a page about you; people visit it to see what you’re up to. Less engaging, more human-sized?
(original)
Replying to @josecastillo
(this isn’t a criticism of those projects, incidentally, I’m just lamenting that a “news feed” somehow became table stakes for building a social media experience when there are probably better models out there we should be exploring.)
(original)
the News Feed was the worst design innovation to come from Facebook. Not because it’s a jumbled mess of endlessly disappearing content, but because it became the dominant model for all other social media. Even opensource alternatives like Mastodon & Diaspora adopt it whole-cloth. https://twitter.com/hackaday/status/1389563213641969675
(original)
I don’t have any particular reason to be playing with this today. I just think it’s neat. (The Feather is an Adalogger M0 that I had on my bench, running a few lines of CircuitPython to generate a sawtooth wave.) /cc @the_prepared
(original)
I have a knack for problem solving, and I think I’m a better-than-average coder. but the biggest asset I bring to a team is positivity.
(original)
Replying to @AndrewLeCody
Like, three days later he did condemn white supremacy, in a statement. But in that zoom call, his head of strategy had just said something racist. It’s hard to acknowledge that racism exists in the management team and condemn it, especially when you see yourself as not racist.
(original)
Replying to @AndrewLeCody
I think this situation had to do with self perception. When confronted with the possibility that you might be part of the problem, there’s an instinct to go on the defensive — “racism is bad, and I’m not bad” — instead of processing that information and acknowledging the reality.
(original)
I once had a toaster with a cancel button. it would eject the contents immediately, but the name always bothered me. life is change, time only moves forward, and some choices, you can’t take back.
there’s no canceling toast. https://twitter.com/ash2x3/status/1389348233420357633(original)
Replying to @femtoduino and @steveintransit
I stockpiled 110 SAM L22’s for the watch when Mouser, the last supplier with any stock, had 610 left. the remaining 500 hung around for a good month or so, but now… well, let’s just say we both made the right call.
(original)
Replying to @josecastillo
“As much as the conversation has been framed as ‘politics,’ it seems important to remember that the entire affair began when a third of the company volunteered to help the company become more diverse and equitable… [and] Fried and Hansson moved to shut the whole thing down.”
(original)
The employee called for the founders to denounce white supremacy. “That would be the bare minimum for me.”
“I’m not here to share my personal views on anything,” [Jason] Fried said.
It was in that exchange that several employees decided to quit Basecamp. https://www.platformer.news/p/-how-basecamp-blew-up
(original)
Replying to @adafruit
You can get one in any color you like, so long as it’s black. Or green!
(original)
Replying to @mattk
It’s not on my radar screen right now, but the idea is you can add additional parts on a small flexy board. With I2C, three analog inputs and two digital pins, you could easily add a thermistor or temperature sensor. Haven’t looked into a TCXO though. https://mobile.twitter.com/josecastillo/status/1360038133014949888
(original)
Replying to @josecastillo
now you might say “but joey, you just invented a less featureful version of the watch you started with.” And yes, that’s true! but this is a foundation. Want to log overnight temperatures, track your sleep habits or do TOTP authentication? having accurate time is the first step.
(original)
Replying to @josecastillo
posted teh codes. this version also adds a simple interrupt-based UI to set the time. the MODE button unlocks the time setting mode (and lights a green LED); LIGHT at the top left adds minutes; ALARM down by the seconds display resets the seconds to zero. https://github.com/joeycastillo/Sensor-Watch/blob/471bdc60c0d38459702888c722a10f2ed1a8379b/Smol%20Watch%20Project/My%20Project/main.c#L12
(original)
Replying to @josecastillo
also I somehow broke the thread but it continues here. https://twitter.com/josecastillo/status/1388958530120605696
(original)
Replying to @josecastillo
also I misspoke here, the TC was getting clocked 32 times faster, not 32,000. Still, less is more when more is no good.
(original)
Replying to @illdred
A colleague lent me his Atmel Power Debugger, it’s a neat little gadget purpose built for this situation. Has a configurable regulator on board, and two current sensors that can measure from nanoamperes up to 1 amp, I think. https://www.microchip.com/DevelopmentTools/ProductDetails/ATPOWERDEBUGGER
(original)
Replying to @josecastillo
I also figured out where I was screwing up my math calculating Martian time, so I’ll be able to add that bit of demo back to the repository later this week.
(original)
Replying to @josecastillo
Hit a wall at 100 µA but that’ll have to do for today. That’s at least 40 days of battery life, enough time to evaluate the accuracy of the crystal over the next month or so. And I got the seconds working, with a new tick interrupt from the low power RTC. Solid weekend progress.
(original)
Replying to @josecastillo
switched to the low power voltage regulator. (the SAML22 can use an LDO to generate its core voltage, or a buck converter with the addition of an external inductor. which I most definitely added.)
103 µA.(original)
Replying to @gennyble
the goal for today, that is. I’ll doubtless keep working on this as I hack on it further; the lower I can get the power consumption, the more useful the gadget becomes.
(original)
Replying to @gennyble
I was running some very old Atmel demo code (admittedly on a SAML22G16, which has less RAM and stuff) but that code was down at ~14µA, and 11 in sleep. I think the goal is <25µA, which would be about six months battery life; I’ll take 15 as a stretch goal. https://twitter.com/josecastillo/status/1362992767660662787
(original)
Replying to @josecastillo
OH DUH. Just read the instructions: “running code through a debugger will not yield accurate current measurements. This is because the target device’s debug module (OCD) requires a clock source which cannot be disabled while debugging.”
Down to 110 µA.(original)