I am doing #electronics, #embedded programming, #python scripting, hardware security and recently some sewing.
Public Key
npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 Profile Code
nprofile1qqsvyattu6cjz6nl34ksl82gxk0hn4s07h4005swvlj9ajxddvg6hdqpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsgkaj3e
Show more details
Published at
2026-09-02T13:52:05Z Event JSON
{
"id": "020167656d0ca1ee82243a638abfcdb2adde784897f4edfb2c5314653a2c6936" ,
"pubkey": "c2756be6b1216a7f8d6d0f9d48359f79d60ff5eaf7d20e67e45ec8cd6b11abb4" ,
"created_at": 1788357125 ,
"kind": 0 ,
"tags": [
[
"proxy",
"https://chaos.social/users/jaseg",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.ditto.pub"
]
],
"content": "{\"name\":\"dr.-ing. jaseg\",\"about\":\"I am doing #electronics, #embedded programming, #python scripting, hardware security and recently some sewing.\\n\\nPhD (Dr.-Ing.) in EE/CS currently building open source HSMs with prototype fund and nlnet\\n\\nEmail: whatever you like at my domain. I've got a catch-all alias.\\n\\nPronouns: er [DE] / they [EN]\",\"picture\":\"https://assets.chaos.social/accounts/avatars/109/319/034/414/087/546/original/b30e5f683d0e025f.png\",\"banner\":\"https://assets.chaos.social/accounts/headers/109/319/034/414/087/546/original/68fd0fbd075a6ce6.jpg\",\"nip05\":\"[email protected] \",\"fields\":[[\"Website\",\"https://jaseg.de/\"],[\"github\",\"https://github.com/jaseg\"],[\"gitlab\",\"https://gitlab.com/neinseg\"],[\"Accept-Language\",\"en;q=0.9, de;q=0.9, jp;q=0.5\"]]}" ,
"sig": "11aee523274ce8fea7ce9ca78f5cd37a96e462e6fc4fcadd20a31f64af209e33408a69a1d988c1987c0fac2615c9e8c078f8f2230bb0ea8b9f042f2b0497264f"
}
Last Notes npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg war grad im supermarkt wo bleibt das weihnachtsgebäck es ist schon august npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg AFAICT the technical reason for this is that the new flat is in a relatively new building, and the previous tenants didn't have fiber internet, so in the provider's databases this flat is listed as "doesn't have a fiber uplink yet" and they aren't able to move over an existing contract to it. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg When I started working with this arduino uno q board a few weeks ago I was thinking “ it’s kinda stupid but it’s cheap”, but over time I’ve come to like it a lot. If you ignore arduino’s bloated software tooling, it’s just a tiny ARM linux SoC board with a chonky bonus microcontroller that has nicely accessible IOs. Even though it has increased in price by 40% since, it’s still a good deal if you compare it with a pi. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg lol I just read one of the stranger copes I’ve seen so far in an academic paper: They built a crypto hw accelerator and tested it in a chonky FPGA. Since their accelerator ended up being slower than a pure software implementation, instead of comparing actual performance numbers (throughput or time), for their headline numbers they compared “cycle count per op”, ignoring that a CPU clock cycle isn’t the same as a FPGA clock cycle and that the CPU had more than 10x the clock speed 🤦 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg last night, a friend and I recycled some old PCBs left over from another project into notebook covers! We cut the pages and stitched the spine by hand. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…cz2l Ansonsten ist normale Baumwolle bei normaler Luftfeuchtigkeit antistatisch. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…cz2l Es gibt für Textilien Sprays, die nach dem Waschen aufgetragen werden und die das Textil antistatisch machen. Die halten halt nur bis zur nächsten Wäsche, und es ist nicht sinnvoll möglich, bei einem so behandelten Textil zu garantieren, dass es wirklich gleichmäßig antistatisch ist und dass die Wirkung auch bei trockener Umgebungsluft anhält. Aber wenn's keine kritische Anwendung ist, ist das vielleicht egal. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…tv4h the fact that you got two DSLAMs after less than 2 hours and a dozen boosts says a lot about this corner of the fediverse 🤣 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg Has anyone ever noticed how it's really emailGPT rather than chatGPT with how verbose these things are? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…kcq6 hopefully not, but if we can find a 90sqm 3 room flat for that price that’d be alright npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…43fw I wrote a paper a while ago where we were considering a scenario of using a datacenter as a controllable electronic load for some experiments. We figured it would be pretty hard to actually break anything, even at much larger scale. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…3tdq what’s your target supply voltage and your cost target? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…3tdq I’d be very careful about reducing R3 without doing proper simulations. I’d do an analysis of what cutoff frequency the R3/C filter gives you at the output node and make sure that’s at least somewhat matched to your opamp bandwidth. These days you can get a unity gain stable opamp with 10-50MHz of GBP for not that much. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…3tdq I’d replace TL072 with something less than half a century old, like from the past decade or so. That’s going to give you better noise and bandwidth. Maybe you can even find one that can give you whatever current you need, those BC547 aren’t good for much anyway and some modern CMOS opamps can handle dozens of milliamps just fine. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…3tdq as is you have a weird high pass that’s also a capacitive divider depending on cap matching. If you want a low pass response on that node, I think you have to pick a reference node for the other end of that 100n cap. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…3tdq I’d drop the second 100n on the voltage divider. It forms a capacitive divider and gives high frequency ripple on the supply rail a low-impedance path to the divider midpoint. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg chat am i doing this right npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg My wife and I are looking for a flat in Berlin. If you know of anything, please dm me or email me at contact (at) (my nick) dot de. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…hrzl it just wastes energy since the secondary inductor will only output the AC component anyway npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…hrzl some microcontrollers have built in modules for encoding data for infrared protocols using on/off keying. maybe you can repurpose one of them? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…hrzl ideally, you want a data signal without a DC component if you want to put the data signal on the coil directly without modulation. One easy way to achieve that is to use 8b/10b encoding. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…hrzl you need to impedance match your driver to the coil’s impedance at the target frequency. Common CMOS outputs usually have in the order of 30-50 ohms of output impedance. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…hrzl I’ve tried this before, and settled on capacitive coupling as a better way to get high data rates. With air core coils, the limiting factor is self resonance due to distributed inter-winding parasitic capacitance npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I have zero need for a mechanical typewriter but I found this one for free on the street so I couldn’t really walk past 😅 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg translation in alt text npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…zk9d @nprofile…5ve4 I happen to have this ~100 year old cookbook on Alsace cuisine from my grandma who was living on the french border. Here’s what that has to say (translation in next toot) npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…vnlz @nprofile…g8vl @nprofile…gvk2 every household should have a thermal camera to check on drying laundry and a laser welder to properly reseal partially used cans of beans and such. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…g8vl @nprofile…vnlz @nprofile…gvk2 a similar cost reduction also happened with laser engravers and laser welders. I think those now are at the point where they start saving on stuff you might actually want. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…vnlz @nprofile…gvk2 That thing is tiny! At 23kg total it can’t contain a lot of lead though, and I’d imagine you’d still need a decently sized space to to get at least some r^2 “shielding”. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg have you ever seen anyone refer to resetting a watchdog timer as “petting” it npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I’m giving away a small drill press (lightly used) npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I’m giving away a small vacuum cleaner npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I’m giving away a tubular lock pick (new, unused) npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I’m giving away five new in box Verifone card payment terminals (and maybe the guts of a sixth if you’re interested) npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I’m giving away: A bunch of industrial load cells, different types ranged 5kg to 100kg, 8 total. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I dont like the term “digital sovereignty” because of the pronounced nationalist and authoritarian connotations that it carries, regardless of what the actual intentions of its particular users are. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6380 the error detection I think just monitors the raw noise source for long series of ones or zeros. The data ready signals when a full output word is ready, since the CPU could request data faster than the noise source can provide it. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…pfl7 that’s so pretty! One thing that I love about this kind of multicolor plastic is that every piece is unique, and that making them look good is a craft that takes actual skill. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg lol wtf, STM32G4 has a HW RNG peripheral that ST apparently designed themselves, but for some reason they didn´t think of the idea of letting you enable interrupts only for errors, not for errors + data ready. All it would have taken is two more interrupt mask bits in the config register they have anyway 🤦 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg Tomorrow (Friday) 15:10 at #gpn24 I’ll give a talk on Tamper-Sensing Meshes in the Wild. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…d5pv If you’re interested, let’s have a call to talk about this stuff. I’ve just started a project in which I’ll further develop this stuff as open source hardware throughout the next year. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…d5pv Sure, here’s some pics of the board and of one test coupon, and below are links to the paper and the Kicad files and source code. In the paper there’s plots of some time-domain waveforms, and in the git is the raw data of a few thousand total. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…d5pv tbh that could work. In npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…d5pv The limiting factor in my case ended up being slew rate. I suspect that using fast RF schottkies for both the sampling gate as I did, and to generate an edge to use as stimulus you may be able to get down to dozens of ps rise time with a small budget. The higher you go, de-embedding the pulse shape gets more difficult though. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…d5pv The microcontroller I used there had some pretty neat self-calibrating delay line block at some of its timer IOs that lets you set duty cycles with ~200 ps precision. ultimately, that ended up being enough for me but I also tried out controlling delay by adjusting the supply voltage of a 74LVC gate using a 12 bit DAC an fed I got somewhere in the dozens of ps of precision like that. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg The kitchen’s food stocks are looking interesting. I wonder what I can cook from nothing but two large onions, eight apples and ten bulbs of garlic 🤔 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…d5pv neat! In a paper last year I showed a similar idea. I got up to several GSp/s equivalent sampling rate with an STM32 and a DisplayPort redriver IC for about 10€ total: https://chaos.social/@jaseg/115411504259693224 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg i made a thing npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…yf6g @nprofile…tdn8 I use such laundry disinfectant regularly and I can recommend it. The unscented type I have makes the wet laundry smell a bit weird, but the smell entirely dissipates when it dries. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg at the day job, a supplier of highly advanced and very expensive bits and pieces is desperately confused what we’re trying to buy from them so I’m left to having to email an mspaint drawing of the very expensive, highly advanced bit we need 🤪 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…khrh if you don’t have tattoos, I highly recommend a visit to an onsen or sento if you haven’t done that yet. Some of my favorites in the tokyo area are maenohara onsen saya no yudokoro and jomon natural onsen shiraku no yu. If you have tattoos, azabu kokubisui onsen takenoyu allows tattoed guests and is also a pretty unique experience. For these, bring your own small towels and soap. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6wcx words mean things. An audit is a paid thing because it takes a lot of time. Unpaid research isn’t the same because it generally won’t cover nearly the same attack surface due to time constraints. A key characteristic of an audit is that within its scope, you systematically look for issues to make it less likely that you miss something. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…5lsr this isn’t about “refining” something. calling a scientific paper an audit is an incorrect statement. calling your e2ee audited when it was not is also an incorrect statement. these marketing statements need to be removed not refined. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…5lsr weird mistake. Why do you list that paper as an “audit” on the page you linked? It was not an audit. Also the actual audits you list there lack a direct link to the report in at least one case, and they do not support your assertion that your e2ee implementation as such was audited as they only cover small components of your system. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg delta.chat advertises that they provide “🔒 Audited end-to-end encryption safe against network and server attacks”, but if you click through it turns out that supposed audit: npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…vc5j I think the scenario of this experiment was that they already had a large number of boards deployed that were lacking “proper” hardware capabilities for this kind of fine-grained adjustment and they evaluated retrofitting that capability to these exising boards in software. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…vc5j here you go: https://chaos.social/@jaseg/116567746676154596 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg Stopping the system fans to heat up the system to shift clock phase by a few femtoseconds is one of the less hinged ideas I’ve seen come out of CERN lol. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…k7h5 I wish you good luck. I hope they properly routed at least some of the serdes. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…k7h5 somehow it feels wrong to see a capable chip like that constrained by a board design like that. I guess this was some sort of cryptocurrency miner where they needed basically none of the IO? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…ynvr I’m using fastmail too and I found their spam filter to be generally ok but sometimes oddly insensitive to certain particular types of spam. Usually, what gets through is spam with a valid origin server in a reputable location. In these cases, I usually escalate straight to their hoster’s abuse contact and train my a client-side filter on the mail. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I’m really not a fan of linters, and espec of code auto formatters. In a time when AI usage already homogenizes most types of online communication, I really don’t see the value in removing the tiny specks of personal style I express in the way I align arguments across code lines, or in my positioning of comments. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…g2l8 if you want to message me if you need a review on the board or your thesis. I just finished my PhD in EE and I have experience doing such reviews. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…g2l8 what input amps are you using for the EEG? Do you do RLD or some other analog common-mode mitigation? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…g2l8 heh I was using the ADS1204 back when that was new. The design is up on github, but mind that this was a loooong time ago and I was still new to PCB design back then: npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…g2l8 That’s cool! A simple EEG was my first larger project when I started out doing electronics. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I have a pretty robust BGA layout extractor working locally now. You give it a datasheet with a large BGA, and it spits out a YAML file compatible with KiCad's footprint generator code that includes the size of the layout along with any missing balls and cutouts. I might polish that up and package it as a re-usable tool at some point. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I just found out that the KiCad library BGA footprint generator is limited to grids no larger than 80 x 80. I guess that limit will have to go for me to enter AMD/Xilinx VSVA6865 🤣 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…5ve4 frankfurt airport station has a rewe that might be open npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…thd8 do you need to run the whole OS on the target architecture, or only a single binary? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I think it should be the law that every laptop that has more than one Type C port capable of charging the device must have at least one of them on either side of the device. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…vnlz thank you for that suggestion, I actually needed that. I just started using uv a few months ago after I heard people say good things about it and I’m just really annoyed it’s getting fucked up by capitalism now. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg The 500 page ENTSOE final report on last year’s blackout in Spain just dropped for some light saturday morning breakfast reading: npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg fedi folks, who wants one? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…thd8 for some reason I read CADgirl at first. I think I spent a bit too much time in engineering meetings today 🫠 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…thd8 does the catgirl benchmark, is the catgirl benchmarked, or both? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…nhxu which country/area/timezone are you looking for work in? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…8760 @nprofile…8vnr This one says it’s for automotive headlights. There, this sort of topology makes sense because it allows you to steer a certain amount of total power between 12 LEDs easily without needing 12 separate drivers. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I've just had an AI-generated pull request end up in one of my repos, whereupon without my consent github's fucking AI tool immediately sprung on to it and started doing a "code review", spamming the PR with dozens of unsolicited comments. So now I not only have a hard-to-review AI PR, but also a dozen comments by a different AI to wade through to make sense of it all. I might just close both if they can't convince me they actually tested this. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…mv7s @nprofile…yery I’m pretty sure I recently saw that being sold as a product for military purposes npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I just noticed that I’ve pushed almost 500 commits to gerbonara 😮 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg gitlab has like three functions I ever want to use, "repo", "merge request" and "issue". Why they are hidden among three dozen useless features I'd never use under multiple levels of headings in a sidebar is beyond me. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg #gitlab is such a UX/UI shitshow. I was looking where the fuck they hid the issue tracker when re-designing their clusterfuck of an UI this time around. Turns out, issues are now "work items" because clearly the term "issue", used by every single git hosting platform ever, was just too obvious and easy to understand. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg der chinesische online-retailer joybuy bootstrapped gerade in deutschland sein geschäft und überbrückt die zeit bis sie ihr eigenes logistiknetzwerk aufgebaut haben mit dhl express. ich habe eine bestellung über 140€ aufgegeben, die versandkostenfrei in 6(!) einzelnen paketen zugestellt wurde. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg Darmstadt-Wixhausen ist kein attraktiver Ortsname npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6cfx I think their main input 12 V to 5 V/40A regulator that doesn’t have a heat sink at all is going to act as a fuse before we’d get an opportunity to blow up the fancy fpga 😅 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6cfx there is a table in the datasheet. I think their theoretical max power dissipation if you fill the entire thing with ring oscillators or something can get pretty whacky, but at a reasonable utilization I don’t think you’d get all the way up there. I haven’t worked with parts quite that large though. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6cfx i think any aluminium extrusion heatsink cut to size would be fine. A CPU cooler is definitely overkill. These chips don’t dissipate massive amounts of heat, and they dissipate it across a fairly large area so the heatsink isn’t that critical. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6cfx I have seen worse in practice. A vendor sold us a dev kit that had the tiniest heatsink for the FPGA, which didn’t actually cover the whole chip, and whose intended means of attachment was a (non sticky!) silicone pad. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6cfx at full throttle that FPGA might need a heat sink npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…6cfx what are you going to do with them? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg I find it annoying that this way. LLMs shape some particularities of how I write even though I consciously choose not to use them to write for me. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg Recent LLM iterations have picked up the habit of overusing the “it’s not X it’s Y” contrasting sentence pattern so much that I find myself avoiding using it in my own writing even in places where it would be appropriate stylistically or help with readability. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg note that they count the missing balls in the middle, but not the missing balls in the corners #electronics npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg Today in semiconductor vendors having slightly too much fun packaging their silicon, AMD/Xilinx names a package "VSVA1365" that does not, in fact, have 1365 balls. At least they helpfully include a note in the datasheet 🙃 npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…9jea that microscope mount on the monitor arm looks neat! How susceptible is it to vibration? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…9jea thank you, that’s very helpful :) npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…9jea can you starte manufacturer and model? npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg @nprofile…uptt If messages were encrypted, I think it would be really important that there is a very clear distinction between encrypted and unencrypted posts. Using the same part of the UI for both encrypted and unencrypted messages with the only distinction being a hard to understand setting behind a button I think invites confusion as to what the precise security guarantees are. npub1cf6khe43y948lrtdp7w5sdvl08tqla027lfquelytmyv66c34w6qgh67h8 dr.-ing. jaseg the symbol’s outline is wayyy too thick, the VCC and GND pins are in the wrong places, the pins are grouped wrong, and the gate I/Os aren’t aligned properly. I can’t imagine how they came up with this.
