Last call!
📅 #pts26 CFP wraps up on March 31 (midnight, GMT+2)
🎤 Got something spicy to share about Security ✕ Open Source (or open formats & protocols)?
This is your moment to shine on stage this July ✨
🚀 Submit your talk now:
👉 cfp.pass-the-salt.org/pts2026/cfp
🙏RT = eternal gratitude
You more boldly left the erase button there. I decided to make people send the erase commands manually! (Originally this was for a talk about finding/erasing your data from old devices.. even then I didn't want to get complaints about it ..)
I appreciate the commitment to putting whatever SPI flash package they can buy down.
A still from happy gilmore with the text "Happy learned how to html/js/websocket - uh oh"
Me with LLM coding
On the embedded side example of bad choices: a simple one was the SPI sniffer tried to use bit-banging (for the sniffer, not control). There was almost no chance this would be reliable, but it would insist it should work for "reasonable" frequencies. Eventually got to PIO+FIFO+DMA based solution.
However as someone who's last major HTML project involved frames on Geocities, it's a bit of game-changer there, since it rapidly sped up development on the UI. This sort of "demo project" I felt like was a good use, since if things break it's not that bad.
If you didn't have the embedded experience it would make it hard to know when it's completely wrong, as it was equally confident about both code choices it would make (terrible and amazing). Especially around logic/timing I had to either do myself or hand-tune, as it would often flip things around.
It was also an experiment in LLM coding (which I plan on releasing a template of this project for), and a longer follow-up on that. But the tl;dr is it was like working with the most efficient but inexperienced intern or something. Sometimes terrible decisions were made with high confidence.
For an upcoming talk I built a stand-alone #hardwarehacking tool out of a R-Pi Pico that does Serial + #SPI Dump + #eMMC Dump with a bunch of stuff I wanted. It only needs your browser, and serves the webpage that is the interface from the micro. No internet, no install. github.com/colinoflynn/...
More #welding projects from the other day, circular saw wall storage
An eagle landing on a tree
Chanced an Eagle flying around the other day, some nice photo ops
Also have a lib for the binary protocol, which gives you more data including 3D speed information - in progress at github.com/colinoflynn/...
A circuit board soldered to the back of a 2" ceramic antenna
A screenshot of a GPS log saying "Sat 0 visible=11 used=8 Sat 1 visible=7 used=7 Sat 2 visible=7 used=2"
It's really absurd how good modern GNSS receivers are. Testing this cheap receiver module that I just roughly soldered to the back of the antenna (including an inductor in the path). It's about 1m from the window where it is on my desk, and it's still getting 8 GPS satellites & 7 BDS satellites.
A gmail "is:unread" search
A "confirm bulk action" checkbox
Annual mark all messages read to get back to inbox zero.
Not yet.... EU distributor in progress as well for some products, we'd been hoping tariff life would settle down (making Mouser easier again) but has yet to happen! The actual EU distributor will hopefully make things better long-term.
A text slide saying "turing complete things" and listing: C/C++ Arm Assembly Code PowerPoint Excel The x86 “mov” instruction
Towards fun topics in teaching computer architecture (this 5 min video on powerpoint as a computer is worth watching: www.youtube.com/watch?v=uNjx...)
A LCSC (electronics parts website) search result for "ESD clamp", showing a mechanical C-clamp
Yes LCSC, that is exactly the sort of ESD clamp device I want.
An embedded board with a power supply in the background set to 2.7V (many embedded systems run at 3.3V, so 2.7V is lower). A tool in the foreground has clips going to the embedded board.
Lazy #hardwarehacking SPI dumping trick: you can normally use a lower VCC - the processor gets held in reset by the brown-out/reset logic, but still in #SPI operational range. No need to find reset pin/test pad.
Browser based 2D field solver I've been writing for a while is now up. It calculates characteristic impedance and losses for many common and less common transmission lines.
A screenshot of a text file saying "Problems have been reported trying to import ARC E00 files into ArcView on Windows 95 computers. Related tables do not import into ArcView. It is not a problem on UNIX, Windows 3.1 or Windows NT platforms. If you experience this problem you can use our DXF/DBF product which contains the tables in dBase format. DBase files can be imported into ArcView. "
Just me downloading some data from NS government in 2026 where the readme.txt file contains helpful information about the superiority of Windows 3.1 platforms, really capturing the moment here.
Nothing like the great feeling that in 2026 a website still sends you your cleartext password in response to "forget my password".
Inside of a basic transmitter... the gimbels just use wire that flexes back & forth, nothing fancy! Again all the actual work done in a micro.
A specific RC receiver (AR630+) showing a price of $145 CAD
An internal photo of the receiver from FCC documentation
The annoying part of the integration is the clear high margins. e.g., the "receivers" are now not fancy things but 2.4GHz microcontrollers. There are 3rd part & other brands that are much cheaper, but these are also popular!
An RC plane in a tree
...and less successful.
A hobby zone AeroScout rc plane
Plane in question is an AeroScout trainer, here it was after a successful landing.
It's been ~20 years but decided to get back into RC, high integration levels make it easy easier (more fun IMO).
I do love that with USB-C we're back to checking carefully the ratings on the AC/DC power bricks, I guess at least you don't fry something plugging in the wrong one at least as it negotiates the higher power.
I thought maybe it was something with older USB PD versions, but even in older USB PD with just profiles, at 20V you can have a 60W device. I'm guessing it's a discrete USB-PD negotiation chip so they didn't actually interface it to the MCU. I'll have to take it apart to check.
A screenshot of a manual of a S100 charger, showing that if the input voltage on USB-C is 20V it assumes the power supply is 81-100W.
More USB-C fun, this product just uses negotiated input voltage as USB-C PD rating!? I've tested this as a 45W USB-C power supply that allows 20V will let me set a charging current that needs >60W, which then causes the system to reboot as the USB-C PD obviously isn't happy.
