<hl>
yeah. apparently this somehow got signed off on without his campaign managers hearing about it, so they first heard about it when it was unveilled on TV... at which point the campaign managers apparently started literally screaming
<_whitenotifier-1>
[GlasgowEmbedded/Glasgow] whitequark pushed 1 commit to master [+0/-0/±1] https://git.io/fjLCv
<_whitenotifier-1>
[GlasgowEmbedded/Glasgow] whitequark 19d07e9 - fx2_crossbar: explain how and why it works.
<electronic_eel>
it has little "horns" going into the board, but is still smt
<esden>
reflowing of and pick and placing of those is a pain just fyi
<whitequark>
^
<electronic_eel>
I mean placeable by pick'n'place
<whitequark>
esden actually runs a pick&place
<esden>
electronic_eel: I have ap pnp here and I do that I know what it means
<electronic_eel>
what is the problem you had with similar ones?
<esden>
they sink a lot of heat and they like to snag when placing
<whitequark>
electronic_eel: fwiw the current plan is that the large IO connectors will go unsoldered as the only PTH parts
<whitequark>
unless specifically requested to be soldered when ordering
<esden>
you have to rework them a lot and they take a lot of solder paste
<whitequark>
like on other 1bitsquared boards
<esden>
and pth stuff is a pain in general
<whitequark>
the board *is* designed for wave soldering but it's just future proofing
<whitequark>
since you have to mask off a lot
<esden>
I think you guys added via around the pads of the usb connector IIRC ... that improves the mechanical connection ... I will have to check that
<electronic_eel>
soldering the THT stuff yourself is easy
<esden>
it is usually the lamination of the pads that fails not the conector itself
<esden>
electronic_eel: prototyping is easy ... making a product is hard... ask Elon ;)
<whitequark>
lol
<davidc__>
BTW, re DFM. I took a look at the layout in kicad.. there's a few pads/vias that look like they might have solder wicking issues with the soldermask margin specified in kicad
<esden>
it is funny how often I have this conversations with "experianced engineers" that never actually ran a pick and place machine or manufactuered a thing themselves. :D
<esden>
it is much more complicated than you think from the perspective of your chair in the office :D
<electronic_eel>
esden: I think the lamination stuff is true and vias will probably help - good idea
<whitequark>
davidc__: do you think you can file a bug?
<esden>
yes I was planning to do that for the icebreakers to improve the situation
<davidc__>
whitequark: can do. But I wanted to check if the margin is what you plan to actually get the boards made with
<gruetzkopf>
a great idea is to have those people populate like 10 boards of 200 SMD parts no larger than 0402 themselves
<davidc__>
whitequark: could also be a "davidc__ does not know how to use kicad issue"
<whitequark>
davidc__: it never hurts for us to take another look, even if it turns out there is no issue in the end
<whitequark>
DFM is hard
<esden>
I still have to go through a "thinking period" to decide how to place the via for the USB footprint so that they don't just wick all the solder away from the pad.
<esden>
I think a way to do that is to hide the via next to the pad under solder stop.
<whitequark>
tented via?
<whitequark>
yeah
<esden>
or cover the via with solder stop on the top. That last one is hard to enforce with the PCB manufacturers
<whitequark>
oh I see
<gruetzkopf>
why do they dislike via tenting
<esden>
via in pad tenting
<whitequark>
the ones that do via tenting can't get it to work reliably anyway
<whitequark>
at least in my experience
<davidc__>
whitequark: yeah, looks like its an error between keyboard and chair. Not sure what the contour I was looking at is, but its not edge of the mask
<davidc__>
gruetzkopf: air inside the via expands, and you get a little bubble
<davidc__>
gruetzkopf: in the hard soldermask
<davidc__>
gruetzkopf: then the parts don't sit flat
<davidc__>
in my experience, via tenting is OK as long as you only do it on one side
<davidc__>
but that may not be true with $YOUR_FAB
<esden>
anyways, I will try to squeeze some time to try a few possible ways of adding those via over the weekend. (fighting all the phone calls for my birthday today... :P )
<electronic_eel>
often depends on the drill size if via tenting works with the fab or not
<davidc__>
whitequark: I'll check over the weekend using some reasonable fab specs for wicking/solder stealing issues
<gruetzkopf>
i've been lucky with shenzhen jlc so far in that regard
<gruetzkopf>
but then again i haven't yet run anything above 150pcs
<esden>
whitequark: we could fill the via on the glasgow for via in pad...
<davidc__>
esden: isn't that $$$
<esden>
I am curious how much more that will cost but it can't be too much
<whitequark>
esden: ViP is absurdly expensive
<esden>
davidc__: pcbway keeps surprising me with their prices ... worth a test
<whitequark>
we considered it for BGA
<whitequark>
and rejected it soundly
<davidc__>
gruetzkopf: I've had bubbles at cheap fabs and high end ones. To their credit, the high end ones warned me first :)
<electronic_eel>
via in pad is really $$$, but just plugged vias might be reasonable
<whitequark>
i think plugged vias isn't a reasonable requirement for OSHW unless absolutely necessary
<whitequark>
a lot of people want to make their own revC boards
<esden>
maybe "Via filled with resin" will be enough whitequark?
<whitequark>
this would just add a hundred or two dollars to their prototype cost
<esden>
at least for the usb connector not bga
<esden>
obviously...
<esden>
ok let me try some numbers here and see what happens so I can satisfy my morbid curiosity :D
<electronic_eel>
whitequark: if you make your own few boards, soldering the usb socket by hand is no problem
<gruetzkopf>
usb sockets are *annoying*
<esden>
electronic_eel: yeah that is not a problem indeed
<whitequark>
yeah, and I don't want to have several revC variants that differ slightly in connectors
<esden>
you could swap the connector for your own boards that you make a few of no problem
<gruetzkopf>
im thinking of doing a usbc test board to check process reliability
<esden>
whitequark: also true... but that is not really an electrical difference so it "might" be ok? ... dunno
<whitequark>
esden: it's a maintenance problem
<whitequark>
suddenly we need to do 2x the work
<whitequark>
make sure the boards stay in sync, etc
<esden>
whitequark: fair enough ... but I would not keep that design in the repo
<whitequark>
if it's not in the repo it might as well not exist
<whitequark>
i mean
<whitequark>
if it's your personal change, that's fine
<esden>
that is what I meant, just a personal change not in the repo
<whitequark>
oh sure
<whitequark>
what i mean is
<whitequark>
the design in the repo should be *reasonable*, whether it's for a single board or a small run
<whitequark>
if you run your own assembly line of course you can make changes to fit your constraints and capabilities
<esden>
I do wish that kicad had a modular system already .. where you could have replacable modular sections of the board layout ... same with population options >_<
<whitequark>
yeahhh
<esden>
I really miss that functionality
<whitequark>
kicad still can't do panelization too, right?
<whitequark>
there's some horrible plugin
<esden>
I will definitely complain about that at kicon end of the month during my talk :P
<esden>
plugins don't count I am not maintaining an outdated plugin ... what is this blender?
<whitequark>
ha
<esden>
(I need to calm down my blood already starts boiling) :P
<esden>
the only decent plugin that I might be ok using is the html bom export tool thing.. it seems well done and well maintained as far as I can see.
<electronic_eel>
the html bom export is really cool
<esden>
electronic_eel: yeah no... but no ... using array feature in kicad despite being a hack is still better
<electronic_eel>
I recently bought a dictaphone usb footswitch to press the N (-ext) key when populating boards by hand
<esden>
none the less I can't wait to see a proper panelization features in kicad directly. Most of the underlying plumbing seems to be already there.
<whitequark>
glasgow wouldn't benefit from it exactly but something you might want to do when panelizing is add circuitry to test the entire panel
<esden>
electronic_eel: ohh that is nice! I should get a footswitch too!
<whitequark>
hmm this reminds me, i should write a driver for the ATECC chip
<esden>
whitequark: yeah I never done that and keep dreaming that I will add such testing stuff eventually. :D
<electronic_eel>
panelization in kicad itself would be nice - but needs to allow enough freedom (like with/without mousebites,...)
<whitequark>
esden: seems easy enough for jtag?
<whitequark>
the frequency will suck, but 10k is prob enough?
<whitequark>
of course none of the chips on glasgow *support* jtag
<electronic_eel>
whitequark: the ATECC is the i2c crypto ic I saw?
<whitequark>
yes
<electronic_eel>
whitequark: do you plan to use to make sure no one tampers with the serial numbers?
<whitequark>
the serial numbers are just informative
<whitequark>
the ATECC chip is purely for traceability
<whitequark>
it's essentially a replacement for trademarks that doesn't rely on the power of nation states to imprison you
<electronic_eel>
you mean that no one calls his clone Glasgow
<whitequark>
not exactly
<whitequark>
you're explicitly permitted that under the terms of the 0BSD license, derivative works are fine
<whitequark>
but each derivative work should be clearly marked
<electronic_eel>
so, no one should be able call his clone an "original Glasgow"
<whitequark>
so if someone files an issue like "glasgow killed my $10000 DUT" it's one thing if it was esden's glasgow and the other if it was from their neighbor's garage made with fake level shifters
<whitequark>
there's no "original Glasgow" per se
<whitequark>
i do not and will not manufacture them myself
<electronic_eel>
whitequark: you showed some die shots on twitter. Did you try that with this crypto ic?
<electronic_eel>
I'm curious if they have some visible hardware tamper protection
<whitequark>
electronic_eel: i don't have any spares
<electronic_eel>
like extra layers around their eeprom or the like
<whitequark>
but also it doesn't really matter
<whitequark>
for the cost of tampering with that IC you could make several thousand glasgows instead
<electronic_eel>
I know that it is not important for Glasgow
<whitequark>
there's no local retailer that stocks it so
<electronic_eel>
But I'm just curious if they really do add something or if it is just marketing blurb
<whitequark>
oh, hm, maybe there is... 15 days lead time
<whitequark>
dunno
bgamari_ has joined #glasgow
danilonc_ has joined #glasgow
bgamari has quit [Ping timeout: 255 seconds]
zignig has quit [Ping timeout: 255 seconds]
vmedea has quit [Ping timeout: 255 seconds]
Ekho has quit [Ping timeout: 255 seconds]
zignig has joined #glasgow
Jasjar has quit [Ping timeout: 246 seconds]
danilonc has quit [Ping timeout: 246 seconds]
Jasjar_ has joined #glasgow
Ekho has joined #glasgow
electronic_eel has quit [Quit: Konversation terminated!]
<_whitenotifier-1>
[GlasgowEmbedded/Glasgow] whitequark pushed 3 commits to master [+0/-0/±3] https://git.io/fjLPm
<_whitenotifier-1>
[GlasgowEmbedded/Glasgow] whitequark 79e038c - support.logging: dump_hex: show total bytes, not elided bytes.
<_whitenotifier-1>
[GlasgowEmbedded/Glasgow] whitequark 82b391b - applet.memory.onfi: explicitly flush IN FIFO.