Monday, 2019-03-11

*** tpb has joined #tomu00:00
*** xkapastel has joined #tomu00:03
*** lathiat has quit IRC00:17
*** lathiat has joined #tomu00:23
*** earthnative has joined #tomu00:27
*** xkapastel has quit IRC02:13
xobsSuccess!02:55
* xobs uploaded an image: image.png (225KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/EdBnjKljVHUUTCRCKZznkcdd >02:55
futarisIRCcloudxobs: Congrats. Saw your post on twitter.03:19
futarisIRCcloudxobs: On another similar note, are we using the Microsoft OS Descriptors for USB? Makes it a little easier for Windows 8 and 10 machines, for use with WinUSB.03:19
xobsfutarisIRCcloud: those are the BOS descriptors, correct?03:20
xobsOr possibly WCID.  They're there.03:21
futarisIRCcloudhttps://docs.microsoft.com/en-us/windows-hardware/drivers/usbcon/microsoft-defined-usb-descriptors03:21
tpbTitle: Microsoft OS Descriptors for USB Devices - Windows drivers | Microsoft Docs (at docs.microsoft.com)03:22
futarisIRCcloudhttps://docs.microsoft.com/en-us/windows-hardware/drivers/usbcon/automatic-installation-of-winusb03:22
tpbTitle: WinUSB Device - Windows drivers | Microsoft Docs (at docs.microsoft.com)03:22
xobshttps://github.com/xobs/foboot/blob/master/sw/src/usb-dev.c#L11003:22
futarisIRCcloudhttps://docs.microsoft.com/en-us/windows-hardware/drivers/usbcon/winusb-installation03:22
tpbTitle: foboot/usb-dev.c at master · xobs/foboot · GitHub (at github.com)03:22
tpbTitle: WinUSB (Winusb.sys) Installation - Windows drivers | Microsoft Docs (at docs.microsoft.com)03:22
xobsI probably need to figure out where to add the GUID.  I thought it was in there, but it might not be.03:24
*** Farscrap has quit IRC04:00
*** rohitksingh_work has joined #tomu04:48
*** AmosSam has left #tomu07:18
*** rohitksingh_work has quit IRC07:31
*** AmosSam has joined #tomu07:37
*** rohitksingh_work has joined #tomu08:00
*** AmosSam has left #tomu08:32
MadHackerxobs: Congrats on USB bringup. Good to see it's working well enough to identify itself!08:33
*** AmosSam has joined #tomu08:35
xobsMadHacker: thanks! it's so nice to be working.08:40
MadHackerThat switch from "does it even work?" to "can I make it better?" is always great. :)08:45
tntxobs: congrats, now just need to implement DFU and UF2 :)09:10
xobstnt: DFU is mostly there!  UF2 is... not so much.09:11
xobsUF2 will have to wait.  It needs more than EP0.09:12
xobsAlso, I may be a little bit distracted.09:14
* xobs uploaded an image: IMG_20190311_170648.jpg (1096KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/ZfmfkTVnbwImvsWRpenqqjNl >09:15
* xobs uploaded an image: IMG_20190311_170701.jpg (1003KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/PQJgeCVoawVIjoLXkXOweXzJ >09:15
xobsI just got a delivery.09:15
tntpretty :)09:16
tntWhen is the 'expected ship date' for fomu btw ?09:16
xobsOn the site it's stated as being the end of June.09:17
xobsNow I need to modify the plastic case, check to make sure these work, possibly replace the 48 MHz crystal with a 12 MHz one, and order the rest of them built.  And hopefully no changes to the PCB.09:18
tntlol the crystal and the flash are bigger than the fpga :)09:19
xobsYeah.  The crystal is taller than I thought it would be.09:24
*** xkapastel has joined #tomu09:28
MadHackerxobs: Are those circuit stickers as teste indicators under the 12345??09:38
MadHackertest even.09:38
xobsGood eye.09:39
MadHackerI presume the crystal height is still OK for space int eh port, though? It doesn't look excessive.09:39
xobsMadHacker: Yes, actually. The factory added them on their own.  They did a really good job with this jig.09:39
xobsI hope so.  It definitely doesn't fit in the current case without drilling.09:39
MadHackerIt's a neat jig.09:40
MadHackerProper linear slides, not just using the latch and the plate itself in slots on the sides, too.09:40
MadHackerSomeone wanted fancy precision alignment. ;)09:40
MadHackerActually, that's interesting in its own right. The one time I've needed that, I've managed with tapered pins on the top plate and alignment holes on the bottom instead.09:42
MadHackerBut that needs longer travel on the latch, so I can see why using proper slides might be nicer.09:43
xobsThese test points are super tiny, too.09:43
MadHacker'tis pretty, as these things go.09:43
MadHackerYeah, it's not an easy thing to manage alignment-wise.09:43
*** cdmatter has quit IRC09:44
*** cdmatter has joined #tomu09:45
*** futarisIRCcloud has quit IRC09:45
xobsThe photos lie slightly.  The component-side goes down.  I designed it so that all test points are on one side, to avoid needing to have test points on both sides.09:46
MadHackerHm, should have designed in some assymetry. You know someone's going to put the boards in upside down and wonder at the 100% failure rate. ;)09:46
xobsThere is some asymmetry, they just didn't take advantage of it.09:47
MadHacker(I jest, but I've had similar, same connector at opposite ends of a board, and people plugging things in the wrong one and considering it a test fail)09:47
tntxobs: I was scratching my head wondering if the probes were coming out of those white things or something :)09:48
MadHackerTBH the more annoying bit is that it's *not* the same connector - two different barrel plug diameters - but that way around fits enough to make people think they're doing it right.09:48
MadHackertnt: I presume they're just pipette tips protecting the pogo pins.09:48
MadHackerOr maybe centrifuge vials. Same shape.09:49
tntMadHacker: well if the component side goes down, aren't they just mechanical pressure to push the whole assembly down on pins coming from the bottom ?09:49
MadHackertnt: You don't want to do that if you can avoid it. Pins underneath means pins that are roughed up by whoever's trying to align the board.09:50
MadHackerPins above mean the pins aren't involved at all until the board is safely in place and you toggle the lever above.09:50
tntbut all the test points are on the component side ... the other side is USB connector.09:50
MadHackerI mean, OK, you can have enough travel to deal with that, but it's still only spring pressure not physical distance keeping the pins safe.09:50
MadHackertnt: The board is upside down; xobs was explaining above.09:51
MadHacker09:38 < xobs> The photos lie slightly.  The component-side goes down.09:51
MadHackerBut actually, hold on, you're right?09:51
tntMadHacker: exactly ... and on the picture, the component side is currently up ... so if the compoennt side , where the pads are, is facing down ...09:51
MadHackerxobs: Confused.09:51
MadHackertnt: I thought the pads were on the other side.09:52
MadHackerIn which case, why do the pins below?09:52
MadHackerBut OK, sensible otherwise.09:52
MadHackerI take it back, it's me that's confused - and I like the pins out the way, but yes, what you're saying all makes sense.09:52
MadHackerDoes that mean there's no test on the USB pads, actually?09:53
MadHackerI would have thought those were worth prodding.09:53
tntThey're probably routed to pads on the other side.09:55
MadHackerI clearly haven't had enough caffeine yet today, BTW.09:55
* xobs uploaded an image: VectorImage_2019-03-11_054757.jpg (278KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/feObkMauLSeWrXvwaEbtepka >09:55
xobsThe factory even labeled the wires.09:55
MadHackerNice. :)09:56
xobsYes, USB is routed to pads on the other side.09:56
MadHackerRight. Sorry, I got myself massively confused looking at that jig. I was expecting pins above somehow.09:56
xobsMe too, which is why the initial photo was done that way.09:56
tntis that two mains wire directly soldered on the prong of a usb PSU ?09:57
xobs#china09:57
MadHackertnt: Seems legit. They're even heatshrink-covered.09:57
xobsChinese factories really like having an IEC connector for all of their test jigs, so they just route it to a USB PSU on the inside to keep the standard connector on the outside.09:57
MadHackerIt's not the least safe thing I've ever seen. I'm nore concerned thate it's a Pi B and those can take a bit more juice than a generic USB PSU can handle without a bit of voltage falloff.09:58
MadHackerSulky pi is a thing.09:58
xobsIt's true.  But usually it's Good Enough.09:59
MadHackerThink the 3B+ is supposed to need 2.4A @5v peak; the issue is the 5V bit, which quickly becomes 4.5V on a cheapy charger.09:59
MadHackerYeah, as long as it's not under too much load it's absolutely fine.09:59
MadHackerThe blinky undervoltage warning on the HDMI output on current pis is nice.09:59
*** awe00 has joined #tomu10:25
*** awe00 has quit IRC10:30
*** AmosSam has left #tomu12:04
*** AmosSam has joined #tomu12:05
*** cdmatter has quit IRC12:07
*** cdmatter has joined #tomu12:08
*** awe00 has joined #tomu13:07
*** xkapastel has quit IRC13:18
*** rohitksingh_work has quit IRC13:31
*** rohitksingh has joined #tomu14:01
*** rohitksingh has quit IRC14:23
*** xkapastel has joined #tomu14:53
*** alexhw has quit IRC16:14
*** alexhw has joined #tomu17:21
*** rohitksingh has joined #tomu17:34
*** alexhw has quit IRC17:36
*** alexhw has joined #tomu17:38
*** awe00 has quit IRC17:39
*** rohitksingh has quit IRC20:07
mithrotnt: Which change did you make?21:15
tntmithro: ?21:15
mithro<xobs> It's a fantastic patch.  Mostly because it seems to fix the issue :)21:16
tntmithro: https://github.com/xobs/valentyusb/commit/3837e08eb590a8a12b9d5c5469aa0564a165368321:17
tpbTitle: rx: pipeline: experimental pipeline fix from @tnt · xobs/[email protected] · GitHub (at github.com)21:17
mithrotnt: You were able to make that pass timing?21:18
tntmithro: Thing is you can't just pass bits from 48M to 12M because the bit generated in the 48M domain are 'clock recovered' from the incoming stream. And the local 12M is != from the recovered 12M so they can drift, so for every local 12M clock cycle you could have either 0/1/2 bits from the recovered 12M.21:19
tntmithro: yeah, with appropriate yosys options it passes just fine.21:19
mithrotnt: Yes, the 12MHz clock inside the FPGA is different from the 12MHz clock on the computer21:19
tnthttps://github.com/xobs/foboot/commit/0c6e44478963cd67b096d0d27976c43bfefcd21621:19
tpbTitle: hw: foboot-bitstream: add -relut and friends to nextpnr · xobs/[email protected] · GitHub (at github.com)21:20
tntwhere is the fomu schematic for the final board ?22:10
tntIn the repo I only see the evt322:16
*** Farscrap has joined #tomu23:07
mithrotnt: I don't think xobs has published it yet?23:26
mithroxobs: ping?23:26
mithroopps, it is probably still a bit early for him...23:34
*** xkapastel has quit IRC23:53

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!