@gsuberland lets say we have 20W TDP on VCC, i think around 14V normally, so lets say 1.5A. for EMC, currently i have in my notes: EMC Directive 2014/30/EU (EN 55032:2015/A11:2020, EN 55035:2017/A11:2020). case is fully aluminum and GND connects to case, if relevant.
@gsuberland oh ok. can i ask you to elaborate a bit more on when a CM filter would make sense, and by ground bounce, do you mean like different GND potentials on different sides of the boards?
i still have to clean up the silkscreen due to some changes but otherwise i'm pretty ok with the layout. compromises: hdmi traces are not coplanar all the way to the end, but i never ever had issues with hdmi SI. focus is on usb3 and pcie, and power. and ethernet is still on an inner layer, but it worked fine before and with correct stackup it should be fine still.
@gsuberland good feedback, thank you! i believe the SoM does not need decoupling on the carrier as it has it's own decoupling at the input. i'll look at the caps in more detail, thanks!
for TVS on those, my thinking is that i'll have TVS on the "port boards" with the actual outside connectors. there could be some static events from handling the internal cables of course. but i wasn't sure if it's a good idea to have TVS on every board's internal connections as well.
surprise! we're finally bringing the MNT RCORE processor module (with RK3588 and 16/32GB RAM) to @crowdsupply, which is a great alternative to buying from our own shop–especially if you're in the US and don't want to deal with the ongoing tariff drama. it's for classic MNT Reforms as well as for Pocket Reform, where it enables a new, third M.2 slot for modern Wi-Fi and BT. Check it out: https://www.crowdsupply.com/mnt-research/mnt-reform-rcore-v2-with-rk3588
BTW for some reason KiCAD is faster on my RK3588 Pocket Reform at home (where i did this in the last days) than on the i9-9900 at work (at least since I installed an intel GPU there when my radeon gave up)
getting even there-er! last polishing and checks tomorrow (incl. diff pair deskew) and then we can hopefully very soon build the 4th revision of this thing.
now available: a green LED version of the MNT x Kolibri HALO-90 Electronic Earrings (open hardware, sound reactive). also after the link: a new video demo by @holo_memory demonstrating all modes and colors of the earrings (warning: some strobing lights, but video doesn't auto play).
@woody@socketwench this is confusing to me because you say that the Framework is an open source laptop, and i don't understand how it is supposedly more open than the MNT laptops?
about RISC-V: risc-v processors (the actual chip) might be marketed as "open" in a to me incomprehensible metric, but they currently are not. the only difference to ARM is that you don't have to pay a license fee as a chip maker for the instruction set. there is more in a SoC than the CPU: what about GPU etc?
watching the "living with apple lisa" video for breakfast, i'm astonished that lisadraw had vector editing and copy paste of graphics into lisawrite. how was this technically achieved at the time (early 80s?) in terms of object model/file formats and programming language?
GPUs were a big deal but i feel like we're still in the 90s mentality of them being mostly trade secrets, and open general purpose GPUs are a thing that's missing. it's weird that they're so much harder to use than CPUs. i wonder what the future will hold here
thinking about my computer use over the last 35-ish years, it's fun how the basic "office apps" are already so old and mostly conceptually unchanged (except in small details). for me the biggest, most important milestones/improvements in that time were things that improved stability and interoperability (between software as well as network/internet): unicode, tcp/ip, html/text markup instead of binary formats, svg, jpeg, mp3/ogg, xml, json, memory protection, journaling file systems, ssh