19:18 <florolf> hi

19:19 <sorear> hello

19:22 <florolf> we've submitted glasgow revC1 to an EMS and they came back with some DFM problems, two of which might be relevant in the general case: J4 has pin 1 and 2 swapped compared to the datasheet and the footprint/placement of the usb socket are not optimal

19:22 <florolf> regarding the usb footprint, molex has published a pcn where they recommend a modified pad layout (https://media.digikey.com/pdf/PCNs/Molex/pcn506286.pdf). since the footprint is upstream, i guess i'll submit a PR there

19:23 <whitequark> sounds good

19:23 <florolf> the placement of the socket in the current board design is a little odd, even the current footprint has a "board edge" marker, but it is not being honored. is there a reason for that?

19:24 <whitequark> I'm not sure I understand what's odd

19:27 <florolf> i haven't checked this with the drawing yet in this case, but if the socket is placed too far in, the bottom flange will sit on the pcb instead of hanging over the edge. that could cause all kinds of problems during placement/reflow

19:27 <whitequark> there's a cutout in the pcb

19:27 <whitequark> with generous tolerance

19:29 <hellsenberg> florolf: that link seems to 404 for me

19:30 <hellsenberg> ah wait, nvm. trailing garbage in the filename

19:33 <florolf> whitequark: hmkay. that line represents the manufacturer's recommendation. after checking the drawing i agree that it's probably fine the way it is right now

19:36 <electronic_eel> J4 is the sync connector. Pin 1 of it is GND, pin 2 is sync

19:37 <florolf> electronic_eel: yeah, but it's angled, so it's not symmetrical. not a problem if you apply common sense, but the EMS (correctly) flagged it as questionable

19:37 <electronic_eel> there is "G" and "S" on the bottom silkscreen, which seems to match the pinout

19:38 <electronic_eel> angled? sorry, could you please explain it in more detail

19:38 <whitequark> florolf: there are two SKUs of these connectors I think, angled in a different way

19:38 <whitequark> we might have the wrong SKU in the BOM

19:39 <whitequark> the one I have has the latch on the top (far from the board), can you confirm that?

19:39 <florolf> let me check

19:39 <electronic_eel> whitequark: the glasgow greg davill made for me has the latch on top

19:40 <florolf> whitequark: yes

19:40 <electronic_eel> whitequark: where is the second of these connectors on the board? do you mean aux? that is vertical

19:41 <whitequark> two different SKUs I mean

19:41 <whitequark> that are very similar and so easy to mix, esden had the wrong one in stock

19:43 <florolf> whitequark: i'm looking at https://www.molex.com/pdm_docs/sd/022053021_sd.pdf, which is the one referenced in the bom

19:43 <electronic_eel> just checked: molex 22-05-3021 is the correct one, as it is in the schematics

19:43 <florolf> however, pin 1 is the leftmost one if you look at it from the front

19:44 <electronic_eel> ah, you mean molex says in their datasheet where pin 1 should be?

19:45 <florolf> exactly

19:45 <florolf> whitequark: what's the other SKU?

19:48 <florolf> i can't find anything from molex that looks similar, but has different chirality or something

19:48 <electronic_eel> the molex datasheet says "circuit one designation is used to define void location" - they don't say anything about the schematics. And our model with 2 pins doesn't have voids

19:48 <electronic_eel> but someone not looking at the silkscreen might indeed get confused by that

19:50 <whitequark> hm ok, so what do we fix here?

19:51 <electronic_eel> we could swap 1 and 2 in the schematics and in the layout, so that the layout stays the same

19:52 <florolf> fwiw, these are screenshots from their web tool: http://n621.de/fud/2019-09-02--glasgow

19:53 <florolf> when the connector is oriented correctly (2nd screenshot), pin 1 does not match (1st screenshot)

19:54 <florolf> i don't think this will ever cause any problems, because the correct placement is obvious, but i guess the correct fix would be to swap the 1 and 2 pads in the footprint and the pin assignment in the schematic

19:58 <electronic_eel> florolf: does your ems service allow to import & configure stuff like this all for yourself? or does an agent from them import and convert all the data for you and you just see the problems in their web tool?

20:00 <florolf> electronic_eel: thay's eurocircuits. they ingest kicad_pcb files and bom data as csv. you basically do all the P&P setup yourself (fixing rotations/offsets) and a human looks it over in the end

20:00 <florolf> fwiw, we'd happily buy glasgows from esden, but i think the timeline on that is still somewhat vague?

20:01 <electronic_eel> ah, ok. do they have a direct live online quote, before manual review? or do you get the quote only after manual review?

20:02 <florolf> electronic_eel: you get a quote right away i think

20:03 <electronic_eel> ah, nice. didn't know there were any ems with online quote. It is a nice tool for dfm optimization, just upload new variants until the price goes down

20:04 <florolf> it's not cheap (10 glasgows are ~150EUR each, without taxes), but not significantly more expensive than getting this done in china (at least from the services i checked)

20:05 <florolf> and the level of automation is pretty nice. in the happy path, you don't have to talk to other people at all

20:05 <ZirconiumX> An introvert's dream

20:05 <hellsenberg> :D

20:06 <electronic_eel> do they source the components for you in this scheme or do you that and send them a package?

20:07 <electronic_eel> and what is the regular lead time, like when you do not pay extra for express handling?

20:07 <florolf> electronic_eel: your choice, but in the cases where we sent them stuff, there was a seemingly random handling fee

20:07 <florolf> something like 15 work days?

20:07 <florolf> i'm note sure

20:08 <electronic_eel> wow, that is fast.