Andrew Zonenberg

I'm looking for something that I absolutely could DIY but am pretty sure exists as a commercial product.

• Large format red/black/white (yellow optional) eink display, 7 inch or so
• Ethernet or wifi
• Local push based update (i.e. I can make a POST request or something directly to the device to dump an image onto it without needing to go through a cloud server or be on a VLAN with internet access. Timer-based pull update (device downloads an image at regular intervals) is also acceptable but less desirable
• Good documentation

Anybody have suggestions? I see random eink displays all over the internet but most either provide no obvious docs on how to update them, or want to go through some kind of cloud management system.

Tempted to just buy a panel and bolt a stm32 or something on it

@revk@toot.me.uk @penguin42@mastodon.org.uk @dragonarchitect@rubber.social I really at least want red. If you make one with a B&W + red panel that I can update via LAN without going through any cloud services, I'll gladly buy at least one and probably a few off you

If there's nothing export controlled or ITAR restricted on your Christmas list... ... why are you so boring?

Don't get stuck high and dry with a noncompliant swamp. To ensure proper interoperability, always ask your suppliers if they ensure compliance with ISO 414141 "Standard for Swamps, Wetlands, Marshes, and Bogs". Better known simply as "the bog standard"

Oh and mouser, if you're using customer part numbers (i have some overstock bags that a customer bought me years ago that use part numbers in their database system), does not include the MPN anywhere in the barcode. So those are basically useless and will probably have to get relabeled with a house label that actually includes the MPN. Modern digikey bags will use P, 1P, and 30P tags so you get MPN, customer part number, and digikey part number unambiguously. But older digikey bags and other distributors don't do that.

@samhenri.gold or a case of club mate?

So the plan is to extract information (preferably as much as possible from automated scanning, this is still very much a manual PoC) from the original label, add any missing information like "who was the distributor", remove stuff like invoice number that I don't care about, and print a new label. The boxes at the right side of the label are intended to be a consumption tracker. Scribble out extra boxes at seal time to show how many parts were in it originally, then as you take stuff out cross out more boxes to show how many remain. This only works for small quantities of ICs, but passives are usually in clear bags where you can estimate quantity just by looking at how much tape there is.

I may tweak the exact layout and set of fields, and add my own logo to the free space, at some point. The fundamental idea is, when you buy parts from a distributor they come with a label that has various information in a distributor-specific format. They expect you to scan a whole order from one supplier into your system and track it there, then use an entire reel or something at once. That's not how I work. I have parts in my inventory that are many years old because I bought a thousand or a whole reel and am working my way through slowly. If the part is moisture sensitive, I'll cut open the bag, take a few parts out, use them, then heat seal it back shut.

@NatalyaD@disabled.social a bunch of foss projects, especially gnome/gtk, have been doing this more lately (unilaterally removing a feature or changing a behavior people like, declaring it to be the One True Way, and ignoring the clamoring for a setting to keep the previous behavior). I don't like it at all.

Mocking up a new ngscopeclient UI feature I'm only half kidding, I think a "wtf is this signal" auto-ID / auto-decode feature might be interesting to try to build at some point. But it won't look like this.

@whitequark@social.treehouse.systems @petrillic@hachyderm.io here kitty kitty

And there's the problem: each time I cut the bag it gets smaller, and eventually there's no more bag left. If I buy 25 STM32s on cut tape Digikey won't have left enough spare bag for me to cut off the seal and re-seal it 25 times. Right now, I either label-printer or sharpie the part number on a new ESD bag, then hand write the number of parts left. As I consume stuff I'll cross out the previous count and write a new count. The bags get pretty cluttered after a while.

@ryanc@infosec.exchange [ cw very big spoiler, probably at least 1m^2 ]

Kid is having school pictures soon and there's various options for prints etc, but a ton of options are just prints with no digital file or copyright assignment (although there are options that do include that). Why would anyone do that? If i pay someone to create an original artistic work to my specifications for my exclusive use, be it a photo or a drawing or a song, I expect it to be a "work made for hire" and own the copyright. Commissioning a piece you can't duplicate seems absurd.

Things you don't notice until you actually get the instrumentation to study it: some of my smaller IT gear has a terrible power factor. The average power factor of my main IT rack lower PDU (running most of the servers and switches) is an excellent 0.99 with about 1 kW of load. The upper PDU on the main rack is averaging 0.86 with about 350W load, with the main offenders being the FS 1G edge switch (14W @ 0.58 PF) and the PoE injector (48W @ 0.57 PF). But the PDU on the development/UPS rack is a different story. It's averaging 0.76 with 133W / 175 VA load. The Xilinx AC701 is last place with 9W of active power and a whopping 0.42 PF. The USB PD dongle for [redacted board] is only slightly better with 6W at an 0.49 PF, then the powered USB hub is pulling 7W at 0.59 PF, and the Siglent SPD3303X-E is pulling 25W at an 0.65 PF. The NUC is pulling 7W at an 0.75 PF and the only reason my average isn't a lot worse is the old 24-port Cisco switch pulling almost 80W at an 0.96 PF - probably the only thing on the rack that actually has deliberate PFC.

@ErinRose@chaos.social polycule 2fa token?

For now everything is going in the lab fridge, I have a lot of work to do to get ready for upcoming conference travel and won't have time to design/build the test board until probably early October. But in a month or so I should be ready to have a little "wine tasting" party with all of these.

I also will probably A/B test electropolished JLC stencils and non-EP oshstencils, as well as perhaps different paste printing setups, and different size/shaped apertures. Lots of variables to experiment with and I will not be bruteforcing all combinations, I'll start with my current setup and change only one variable at a time and see how things go. This will generate a lot of data (ideally I would get high-res microscope photos of every board after pasting, placement, and reflow but that is likely to be a lot of work and I'm more likely to do low-res of everything then only high-res post reflow or something, and perhaps full res of a few selected ones at intermediate process steps) so it'll take time to crunch.

Digikey's format has changed over time, some of my bags from a few years ago have less fields than the new ones. And before 2018 when they started using this format it was a 1d barcode with only some kind of internal numeric ID.

@gewt@social.treehouse.systems if you don't want them to be, just multiply by sqrt(-1) and you'll get a 90 degree rotation about the complex plane. Boom, it's not real anymore and can't hurt you :p

@erin@sleepy.computer boop the led snoot

@dianea@lgbtqia.space I have had one defective BGA (the STM32H7 with two open pins last weekend) in about ten years of using TS391, and it was a fairly old syringe of it being used on a not-fully-qualified oven (and easily fixed with more flux and a second reflow in my home oven). What I'm more interested is things like tombstoning and mid chip solder balling and bridging as well as stencil release.

@craigjb@fosstodon.org see here, much lower resolution than displayport but it's the same unified UX

@jpm@aus.social yes but like if i can pay $20 and own the rights and make as many prints as i want cheap, why would i ever pay $50+ for a bunch of hard copies i can't duplicate

@whitequark@social.treehouse.systems so i might be a werido but I don't care about reproducibility from a hash equality perspective i care from a functional perspective. If you have e.g. an ELF date stamp change or something, and you have tooling that demonstrates it's the *only* thing that changed, i consider that sufficiently reproducible. The challenge is mostly that such a fuzzy equality checker is a lot more work than just doing a sha256sum on the msi file

@xabean@infosec.exchange @gsuberland@chaos.social The format is documented / standardized but it's a key-value array and e.g. Digikey today, Digikey in 2022, and Mouser today populate different sets of fields

@revk@toot.me.uk @penguin42@mastodon.org.uk @dragonarchitect@rubber.social Can you push images via MQTT or does it take some kind of textual/JSON description then render the image locally or what? I do have a local MQTT already for aggregating sensor data so it would be straightforward to push whatever format to it

Thermography is a nice idea but the oven window is some kind of glass that isn't IR transparent (good, since we don't want to be radiatively cooling the board too much) so there's no easy way to see it

@jenesuispersonne@piaille.fr It looks like you have to either write custom firmware or use their "ai powered" dashboard tool? I just want a HTTP endpoint I can POST a bitmap to

Alternatively, any other good ideas on how to measure non uniformity of a reflow oven without taping like a hundred thermocouples to a dummy board?

@jpm@aus.social it's like how digikey fedex shipping is cheaper to do 2nd day air than 3 day saver

@craigjb@fosstodon.org let me know if you have any other fixes to suggest. Other than enhanced framing which I've been too busy to touch but if you send a PR I'll merge it

@0h00000000@ioc.exchange which is why it's not an *easy* option

I'm now able to parse every digikey, mouser, or in house datamatrix barcode I've tested it against.

There's a few caveats:

• Manufacturer name (1V tag) is not present in most or all digikey barcodes. You can probably look it up in their API, or in some but not all cases there's a numeric vendor ID encoded in the digikey part number prefix

• Pre 2018 digikey bags use a 1d barcode containing only a numeric ID that i don't know the meaning of

• Avnet, newark, and farnell are not tested. I either didn't have parts from them to check, or the barcodes were damaged and unscannable

• LCSC uses a completely different format (JSON in QR, not ECIA format 06 in datamatrix). This looks straightforward to parse but will be a different code path.

@gnarf@hachyderm.io last time i ordered stencils they didnt support coating and everyone who did it was $$$. It's been on my list of things to try once somebody offered it for a reasonable price

@emily@fedi.uni.horse instructions unclear now wearing Taiyo Yuden nails

People have told me thermally stable fluxes aren't as good and I want to get some data to see if that is in fact the case *on my process*. Plan is to make a stress test board with a lot of different fine pitch stuff and hundreds of passives, then run a bunch of boards with each paste at various places in the oven then collect data and see what happens. Ideally on both my home oven and the kibbyshop, although doing the home oven will be more difficult because I would either need to tweezer-assemble the boards or bring pasted PnP'd boards home and reflow them without knocking anything loose during the drive home.

@NatalyaD@disabled.social intentionally removed a year or so ago. A lot of people are upset about the change but so far upstream hasn't reverted it or added back a preference. See https://github.com/mastodon/mastodon/issues/33799

@xabean@infosec.exchange @gsuberland@chaos.social for example mouser populates key 1V manufacturer/vendor name, while digikey does not. Digikey today populates 9D, date code, a few years ago did not. Digikey until 2017 was the wild west with none of this info in the barcode whatsoever

@fullywoolly@mastodon.social yeah i know right? Most of the heavier weight stuff does. The overall PF of the entire lab measured at the UPS output (excluding lighting and HVAC which are not on UPS) is a very respectable 0.97. All of the servers, larger switches, and big test equipment seem to have active PF or that number would be a lot worse.

@GarnetKoebel@hachyderm.io I still do a lot of hand assembly or hand placement of low-volume parts we dont want to burn a feeder on. For the most part I don't track passives at all, other than "looks like I'm low I should get more". For ICs I try to keep exact counts.

All of these are ChipQuik branded pastes with no-clean ROL0 flux. So that's a variable I'm not touching. At top is my current default, TS391SNL10: SAC305, type 4 mesh, thermally stable for a year without refrigeration. Next is NC191SNL, also SAC305 type 4 but with a "smooth flow" flux that is supposed to print better. Next is SMD291SNL, a fairly standard refrigerated SAC305 paste, in both type 4 and type 5 mesh (I want to see if the type 5 works better for WLCSPs and 0201s and other fine pitch stuff). Finally, CQ100Ge which is the only different alloy in the ensemble - Sn100c or something very close to it. This is also type 4.

@0h00000000@ioc.exchange yeah a thermal viewport is an option but I'd have to cut the oven open to install it

@a@pdx.social @whitequark@social.treehouse.systems @CyReVolt@mastodon.social Hmm well almost all of those signals look like power and the surrounding support parts all look power too (large capacitors etc). The flat flex cable to the left doesn't go to the display? Is it possible one of those 20 pins is like analog NTSC video?