ChanServ changed the topic of #glasgow to: glasgow interface explorer · code https://github.com/GlasgowEmbedded/glasgow · logs https://freenode.irclog.whitequark.org/glasgow · discord https://1bitsquared.com/pages/chat · production https://www.crowdsupply.com/1bitsquared/glasgow · CrowdSupply campaign is LIVE!
aquijoule_ has joined #glasgow
aquijoule__ has quit [Ping timeout: 272 seconds]
GNUmoon2 has joined #glasgow
G33KatWork has quit [Ping timeout: 256 seconds]
G33KatWork has joined #glasgow
mwk has quit [Ping timeout: 272 seconds]
ali_as has quit [Remote host closed the connection]
ali_as has joined #glasgow
ali_as has quit [Client Quit]
aquijoule__ has joined #glasgow
aquijoule_ has quit [Ping timeout: 246 seconds]
FFY00 has quit [Remote host closed the connection]
mwk has joined #glasgow
FFY00 has joined #glasgow
FFY00 has quit [Remote host closed the connection]
FFY00 has joined #glasgow
FFY00 has quit [Remote host closed the connection]
FFY00 has joined #glasgow
FFY00 has quit [Remote host closed the connection]
FFY00 has joined #glasgow
modwizcode has quit [Quit: Later]
electronic_eel has quit [Ping timeout: 246 seconds]
electronic_eel has joined #glasgow
aquijoule_ has joined #glasgow
aquijoule__ has quit [Ping timeout: 264 seconds]
PyroPeter_ has joined #glasgow
PyroPeter has quit [Ping timeout: 272 seconds]
PyroPeter_ is now known as PyroPeter
GNUmoon2 has quit [Remote host closed the connection]
GNUmoon2 has joined #glasgow
_whitelogger has joined #glasgow
GNUmoon2 has quit [Ping timeout: 268 seconds]
bvernoux has joined #glasgow
GNUmoon2 has joined #glasgow
PyroPeter has quit [Ping timeout: 256 seconds]
G33KatWork has quit [Ping timeout: 256 seconds]
G33KatWork has joined #glasgow
PyroPeter has joined #glasgow
lxdr has quit [Ping timeout: 240 seconds]
minicom has quit [Ping timeout: 240 seconds]
pie_ has quit [Remote host closed the connection]
miek has quit [Ping timeout: 240 seconds]
minicom7 has joined #glasgow
miek has joined #glasgow
pie_ has joined #glasgow
lxdr has joined #glasgow
ali_as has joined #glasgow
thaytan has quit [Ping timeout: 264 seconds]
thaytan has joined #glasgow
egg|anbo|egg has joined #glasgow
Saphire has joined #glasgow
<Saphire> Mmh, kinda looking at the dates for stuff and don't think I can get this in crowdfunding stage. How would one get it after that, and how long to wait until it's available? Crowdsupply has some preorder thing but kinda confused if/how that's applicable?
minicom7 is now known as minicom
<electronic_eel> Saphire: once the Crowdsupply "campaign" phase ends, you can pre-order on Crowdsupply.
<electronic_eel> when you buy during the campaign-phase, the planned shipment date is end of may. the pre ordered ones will be delivered afterwards
<electronic_eel> so basically the earlier you order, the earlier you get your glasgow. but it will still take a few months
<Saphire> So if I'm like a week late, it's still going to be roughly may?
<Saphire> (Well, at least for shipping OUT, yay international shipping)
egg|anbo|egg has quit [Remote host closed the connection]
<whitequark> it will be manufactured in batches, so depending on which batch your order will go into, the shipping time will vary
<whitequark> only esden can clarify further than this, i think
<Saphire> Mhh. Just really don't want to miss it and end up having to wait not until May but like September..
<Saphire> ...will the price increase or is this same as it will be?
<d1b2> <Attie> I think that's another question for @esden
<tnt> One thing for sure is the earlier your order the sooner (or equal) you'll get it so if you know you want one, why wait ...
<whitequark> not everyone can spend $130 on a whim
<whitequark> at the start of 2020 i could barely afford food much less neat tech toys
<tnt> Fair enough, but from the above, it didn't look to me as a "if I order" but rather a "when I order it".
<d1b2> <Attie> still sound hey well be a cash flow concern...
<d1b2> <Attie> *still could very well...
<whitequark> yeah i'd assume they want to save for it or something
<d1b2> <icb> I think pre-order price can be higher than crowdfunding price, but doesn't have to be. iCEBreaker was $69 in crowdfunding, and is still $69 even with it in-stock, though did have free US shipping in the crowdfunding phase that is $8 now. Worldwide shipping has been $16 the entire time
<whitequark> sometimes there's an early bird discount
<whitequark> which... i think was gone in like 20 minutes
<tnt> There was. But they're all gone.
feldim2425_ has joined #glasgow
feldim2425 has quit [Ping timeout: 244 seconds]
feldim2425_ is now known as feldim2425
stefandeml has joined #glasgow
modwizcode has joined #glasgow
<modwizcode> Committed and bought myself a glasgow :)
<whitequark> nice! :D
<modwizcode> Yeah I tweeted this but essentially what sold me is that marcan was using it as a UART for the linux dev and I realized how useful being able to program the voltage each bank is at :p
stefandeml has quit [Quit: Ping timeout (120 seconds)]
<Saphire> tnt: I literally have like, 250$ total available for spending for at least a week or two ^^'
bvernoux has quit [Quit: Leaving]
<cyborg_ar> Well hopefully the larger than expected volume can get some price breaks that allow the crowdfunded price to remain after the campaign
<cyborg_ar> There is a difference between a couple hundred boards and a thousand
<d1b2> <DX-MON> given what Esden said last night about having to order things and wait for the lead times.. the campaign is firmly in the realm of "you're getting full-scale run price breaks" I think
<cyborg_ar> 130 is a pretty good price for what glasgow is
<cyborg_ar> Hopefully some of it goes to support software dev
<modwizcode> Is there a reason the project didn't go with something like an ECP5? I feel like I've seen similarly priced boards with that and it's got much greater capacity i think
<fridtjof[m]> iirc that's planned for "next gen", aka revD?
<d1b2> <TiltMeSenpai> I think it's revE
<modwizcode> Hmm maybe I should wait
<modwizcode> whitequark mentioned a revE but I thought she was being theoretical :p
<d1b2> <daveshah> Glasgow - certainly rev A and B - predates the ECP5 tooling being stable enough
<modwizcode> Ahhh
<d1b2> <daveshah> I think rev E is a good way away still
<d1b2> <daveshah> and likely to be several times the price too
<modwizcode> That was my only concern was that maybe I should just get an ECP5 board and stuff.
<d1b2> <TiltMeSenpai> I remember E for ECP5 for some reason
<gruetzkopf> yup
<modwizcode> I mean I've seen super cheap ones
<modwizcode> but I don't understand the cost structure well enough
<agg> ecp5s are very fun but a lot of what glasgow gives you is the rest of the hardware, especially the frontend with pull up/down, variable drive voltage, a lot of input protection
<agg> and of course the USB interface and software stack, none of which will work with a generic ecp5 board either
<fridtjof[m]> yeah, couldn't find ecp5 on the revD minispec just now. revE then, my bad
<agg> though e.g. the colorlite ecp5 boards are _so_ cheap, maybe get a glasgow _and_ one of those :p
<electronic_eel> also revE will not have the ports with the level shifters, but SYZYGY ports
<agg> you could use the glasgow as a jtag programmer for the ecp5 board...
<d1b2> <TiltMeSenpai> "E for ECP5" is a decent and convenient way to remember it then lol
<modwizcode> Right the USB stack and configurability is why I wanted a glasgow anyway
<gruetzkopf> (with a revC/D frontent as a module)
<modwizcode> I was just curious about reasoning
<modwizcode> I was thinking about also getting one of those super cheap obsolete bitcoin Zynq boards, I think there's no open toolchain for that though.
<d1b2> <TiltMeSenpai> if you want USB with an FPGA, i've had pretty good success messing around with USB stacks using LUNA
<modwizcode> I have decided that in general avoiding direct interactions with USB is in my best interest lol
<modwizcode> I think the most I want to touch is playing with the built-in stuff on an NXP arm chip based board and the stuff in an STM32 (if I can ever get it working). I appreciate greatly the glasgow software stack taking the USB out of the equation.
egg|anbo|egg has joined #glasgow
<modwizcode> Incidentally I saw another interface board project on crowdsupply which had a comparison table with glasgow and like the only negative they dinged it on was "gateware" which I thought was the most amusing nonsense.
<d1b2> <TiltMeSenpai> Tigard?
<modwizcode> Might have been
<modwizcode> I basically went through crowdsupply yesterday opening a tab for every fpga and interface board type thing, I know that was one of them I looked at
<electronic_eel> yeah, the comparison chart on the tigard page is a bit strange re glasgow...
<modwizcode> tbh it seems to solve a different problem and it's cost is totally different as a result, I just found it amusing that it was framed that way.
<modwizcode> They could have just been like "glasgow=expensive and flexible", "our board=cheaper and does the stuff you probably need most"
<modwizcode> I say expensive I just mean "more expensive"
<d1b2> <TiltMeSenpai> I think Tigard was developed for Joe Fitz's classes, the idea is more plug and play without host side software
<modwizcode> Yeah that's a fair distinction too! It just was written up super strange lol
<d1b2> <TiltMeSenpai> similar to how you can plug in a bus pirate and go poke an spi flash chip immediately
<modwizcode> Some of these boards are very much like "is this the thing I want to get at this price or is it kind of legacy", like I feel like a bus priate just never sounds like the right choice
uberushaximus has quit [Ping timeout: 256 seconds]
<whitequark> there is definitely a place for a board like glasgow without the host software
<whitequark> in fact, STARSHIPRAIDER would not rebuild bitstreams on demand
<whitequark> so really it's glasgow that's a major outlier here, nothing else works remotely like it
<whitequark> (because you couldn't really make anything like it before FOSS PNR)
<modwizcode> yeah I haven't thought much about the trick of on-demand building of the applets.
<whitequark> so if you don't that you can do things like write linux drivers and stuff
<whitequark> tnt's icepick would have worked that way
<whitequark> it... makes total sense, it's just not a board *i* wanted to make
<gruetzkopf> i don't remember STARSHIPRAIDERs intended IO topology anymore
egg|anbo|egg has quit [Remote host closed the connection]
<whitequark> it would have a pin mux on the FPGA
<modwizcode> I should learn basic PCB design and try to design an addon board that lets you fully do USB-PD
<electronic_eel> modwizcode: isn't that more of a software problem than a hardware/pcb design problem?
<electronic_eel> i mean there is https://github.com/ReclaimerLabs/USB-PD-Breakout and wiring the i2c from it to glasgow is easy
<modwizcode> I think the issue was that most existing boards don't break out all the pins you need to do the serial stuff that the M1 macks need
<modwizcode> *macs lmao
<electronic_eel> how do they do serial? is it just plain uart over an alt mode or something else?
<modwizcode> uh I'm like 99% sure it's multiplexed over not alt pins and that's the issue
<whitequark> modwizcode: you should definitely talk to marcan
<whitequark> wait you know marcan.
<whitequark> nevermind
<whitequark> as for PD, take a look at the chromium twinkie board
<modwizcode> I mean I don't think I know him that well lmao
<modwizcode> I'll take a look at that board
<whitequark> that has a fully discrete PD frontend
<whitequark> i... don't entirely understand whether it's good, yet
<modwizcode> I think I just saw you complaining about it and that reminded me it might be decent project to try
<modwizcode> I was looking at crowd supply and like hoping that one of the various pd boards would do what I want.
<whitequark> the idea would be to make slight modifications to twinkie until it stops being bad
<whitequark> my idea, anyway
<whitequark> it's possible i'll take an even closer look at it and just give up on that design
<modwizcode> ... my irc client doesn't let me select text wtf
<whitequark> FUSB302 is good but for a *debugger* i think it is actually pretty nice to have a fully discrete frontend that just has no PD parts at all
<modwizcode> oh it automatically copies if I select stuff okay... that's a weird way to handle that on windows.
<whitequark> given the pervasiveness and insidiousness of PD bugs
<modwizcode> wait you meant like it doesn't even use a chip
<modwizcode> I thought you meant like independent of a supporting board
<whitequark> yeah it uses comparators and ADCs and stuff
<modwizcode> Interesting
<whitequark> and a bunch of resistors to drive it
<modwizcode> god I hate google's repository system
<whitequark> it is infuriating
<whitequark> really, everything about google is infuriating if you are not a googler
<modwizcode> I can't find anything. Also they seem to link to have there be no main branch and all the code is in side branches
<whitequark> "the recommended way to build chromium is to become a google employee but some alternative workarounds are available if that's not practical" was it
<modwizcode> hahah I think I remember that
<modwizcode> I am assuming they have a nifty tool that works if you are a googler that makes it all super nice. But I do not have that
<whitequark> honestly i could probably interview with google just to tell them that i don't *really* want to work there but i want to build chromium and android more easily
<modwizcode> LOL I wouldn't doubt they'd be like "okay"
<modwizcode> I did read your tweet but I don't remember it having much specifics, besides the host tool being annoying was there something wrong with the twinkie board?
<whitequark> it was a six layer board full of WLCSPs and BGAs (i think they used BGAs for everything they could get in BGA/WLCSP) that would cost a fortune to manufacture and assemble
<modwizcode> ohhh
<modwizcode> Wait... but why
<whitequark> literally i do not think they could have made it any harder to manufacture
<whitequark> well
<whitequark> i think they asked the engineer who did the chromebook to also do this and they just kinda did it the same way
<modwizcode> I looked at like their overview thing and I was like oh that's so nice and simple. I have been mislead by my laziness
<whitequark> it makes sense if you're making like a cellphone. it'd make total sense to do that
<whitequark> if you're making a tool for other people it is completely nuts
<modwizcode> I mean if I was designing a board I feel like I would try to optimize for the kinds that have fast turnaround which I feel like I don't remember 6-layer boards being.
<whitequark> i'm quite sure google can get fast turnarounds on any kind of PCB known to humankind
<whitequark> (there are definitely express services for 6-layer boards. you just can't afford them)
<modwizcode> I had the opposite opinion. They can probably afford to just wait and not care too.
<modwizcode> But yeah they can probably throw money at it I guess. For all I know they have in-house board prototyping
<modwizcode> uhh wtf is this repo I just went up from the firmware dir and why is there so much other crap with this board
<whitequark> ok look
<whitequark> this is... google thing
<whitequark> do not think about google thing. that only leads to suffering
<modwizcode> I will try to avoid thinking about it as long as I can determine where the damn pcb files are
<whitequark> there's a release archive on the webpage
<modwizcode> I just assumed it was the repo tared up
<whitequark> i haven't really looked at the repo because their repo gives me heartburn
<modwizcode> I wonder how they even make their repos like that. Like it seems really hard to do.
<d1b2> <TiltMeSenpai> Repo (the tool) is evil, can confirm
<modwizcode> Maybe that's the reason
<whitequark> modwizcode: this is google thing. do not think about google thing
<modwizcode> That tool might make it very easy to do this
<modwizcode> hah
<d1b2> <TiltMeSenpai> I suspect some of their code bases would make git sad if it was all 1 repo
<whitequark> TiltMeSenpai: well microsoft has a monorepo
<d1b2> <TiltMeSenpai> so they developed Repo to do Git things but on git repos instead of files
<modwizcode> I think they probably encountered an early git scaling bug that probably is not relevant as much today
<whitequark> idk, i find repo reasonably ok in android, as much as anything pertaining to android might be described as "ok"
<whitequark> modwizcode: the other thing repo does is helps with submodule madness
<whitequark> they use a lot of external vendored code
<modwizcode> Also they probably do that thing where they don't link repos they just copy the files in. Like my boss that would rather copy and paste instead of learn branches
<whitequark> and using submodules at their scale would just... not really be viable
<whitequark> you don't want the entire thing to be stored in a merkle tree when it's six hundred repositories
<modwizcode> I mean doesn't repo just kinda replicate submodule in a weird way
<whitequark> repo is late bound submodules
<d1b2> <TiltMeSenpai> I think "helps" might be a strong word but yes, using submodules probably wouldn't be fun
<whitequark> git-submodule is eagerly bound submodules
<whitequark> s/submodules/subrepositories/
<modwizcode> hmm
<whitequark> what repo does is fundamentally reasonable and also what most people expect submodules to do, possibly
<modwizcode> I might not understand submodule as much as I thought.
<modwizcode> wait wtf this board is really simple
<modwizcode> how did they even manage to make this a 6 layer board
<whitequark> by not caring about making it anything less
<whitequark> same as with all the WLCSPs, and the ridiculous flashing procedure
<modwizcode> I mean you have to understand, I decided to avoid learning PCB design unless I have to. I have mostly succeeded so far
<whitequark> this is a level of self restraint i find incomprehensible
<whitequark> i don't know everything. i only know what i know. which should approach everything
<modwizcode> You have to understand it's more imposed by looking up PCB costs going "oh I can't afford that at all" and not having any reasonable starter projects that would not be totally doomed to failure (I don't think my first PCB board project should involved DDR3 but my ideas all do...)
<whitequark> oh.
<whitequark> ok yeah not being to afford PCBs sucks.
<modwizcode> Now the costs have gone done and level of spending money has gone up.
<whitequark> i've been there
<modwizcode> So the logic no longer applies
<modwizcode> Well I think trying to design a version of this that's no a 6-layer board would be worthwhile (assuming you haven't already done that)
<whitequark> already done
<modwizcode> Figured
<whitequark> i have the prototype on my desk
<whitequark> and this is what i was swearing about
<modwizcode> I thought you might have just gotten a 6-layer fabed
<whitequark> ha, no, i work on community open hardware, not the weird google flavor that's effectively useless to anyone
<modwizcode> I could probably build this on a breadboard if I had type-c breakouts actually
<modwizcode> If I figured out how to fix my stm32 board so usb worked
<whitequark> yay yak shave!
<modwizcode> I still don't know what that means I've been too lazy to google
<whitequark> wait, wrong link, sorry
<whitequark> modwizcode: https://seths.blog/2005/03/dont_shave_that/ here's the etymology
<modwizcode> ah
<whitequark> for example, today i was going to do PD debugger bringup, and instead i've started to reverse-engineer an ethernet media converter
<whitequark> the only thing that stopped me is the realization that i left all of my IC clips and jumper wires and revC+ glasgows elsewhere
<modwizcode> heh
<modwizcode> Apparently I have class now but this class (like others) decided for some godforsaken reason to switch to teams (others have used zoom), (which is significantly worse than the built-in video software in blackboard) and in this case I have to be invited in separately. So I cannot go to class. Awesome.
<whitequark> wonderful
<modwizcode> Maybe I'll fix my stm board instead of class
<modwizcode> My parents got me a hot air + soldering station thingy that might make resoldering the USB port on it doable
<whitequark> niiice
<whitequark> for more complex rework or PCBs with ground planes i find that preheat significantly improves the chances of the board actually leaving the rework in an intact and functional state
<modwizcode> My theory is it wasn't the typical resistor issue (well it might be I think I knocked another resistor off while trying to do the fix) but actually port without good connection to the datalines. I never tested that though.
<whitequark> there are some combinations of connectors and PCB layouts that make preheat effectively mandatory
<modwizcode> I've only tried using the hot air thing once. I successfully desoldered a crystal from an old PCI modem card I had sitting around
<modwizcode> but like I don't think I did a good job, I'm not certain it would still work.
<whitequark> mostly, I'm telling you this so that you'll look into it if you get frustrated
<whitequark> I sure hope someone told me this
<modwizcode> Trying to put it back was hellish bc I couldn't get the holes to stay open.
<modwizcode> I will keep that in mind.
<modwizcode> Preheating it how though
<d1b2> <xabean> I forget who, some hardware hacker I follow on twitter removes components with blowtorch hot-air?
<whitequark> um.
<d1b2> <DX-MON> I need to look into preheaters and such.. would probably make working on things like MXKeyboard's hardware way easier
<d1b2> <xabean> was that you whitequark?
<whitequark> yes. yes i do that
<whitequark> i put them back with a blowtorch too
<whitequark> i mean, i *normally* use a hot air gun, but, i can do with a blowtorch.
<whitequark> i think i've never done a BGA with a blowtorch. i should try that sometime
<d1b2> <xabean> yep, this is why I follow folks on twitter. Epic badass hacks.
<modwizcode> If you do bga with a blowtorch I wanna see a video :p
<whitequark> the hardest part will be keeping the flux from getting on fire, i suspect
<russss> plumber's flux will surely do it, that's designed to be used with a blowtorch. Might dissolve other things as well, I suppose.
<whitequark> that's brazing, not soldering
<russss> it's soldering for water pipes, at least here in the UK. The solder is similar melting point to electronics solder
<whitequark> oh i see
<whitequark> i wasn't familiar with that specific process, sorry
<russss> aircon refrigerant pipes use brazing for some reason, maybe less porous to refrigerant gases
<Ultrasauce> the main difference is plumbing can have a few hundred um of the exterior surface turned into various copper salts without being destroyed
<whitequark> i haven't really met soldered water pipes yet, it's either threaded connections or, lately, welded plastic
<modwizcode> The plastic water pipes make me uncomfortable and I do not remember what I read that resulted in me feeling that way
<russss> push-fit plastic is becoming more common here
<whitequark> modwizcode: something related to plasticizers leeching out?
<modwizcode> probably yes
<modwizcode> I'm not sure why I felt like that was any better/worse than metals leaking out.
<whitequark> i think PP/PEX pipes don't use plasticizers but i might be wrong
<russss> I think some of the early plastic water pipes had a tendency to degrade pretty quickly too, but you can get stuff now which is guaranteed for 50 years.
* sorear dimly recalls inorganic flux being a thing
<d1b2> <xabean> I once had to replace the main cut-off valve for the water feed to the house I inherited from my Dad. I got it all soldered up, turned the water back on at the street and discovered a pinhole leak. :(
<whitequark> sorear: sure but it's hard enough to solder BGAs with an appropriate flux
<whitequark> really it's just a question of better blowtorch technique
<whitequark> not that hard
ali-as has joined #glasgow
lxdr has quit [Ping timeout: 256 seconds]
Stormwind_mobile has quit [Ping timeout: 256 seconds]
ali_as has quit [Remote host closed the connection]
lxdr has joined #glasgow
Stormwind_mobile has joined #glasgow
<_whitenotifier> [GlasgowEmbedded/glasgow] electroniceel pushed 1 commit to wip-revC3-hardware [+0/-0/±1] https://git.io/JtmMZ
<_whitenotifier> [GlasgowEmbedded/glasgow] electroniceel b0f7bb1 - revC3: add pin 1 marker to silkscreen for LVDS header
Stormwind_mobile has quit [Ping timeout: 256 seconds]
GNUmoon2 has quit [Ping timeout: 268 seconds]
<d1b2> <rwhitby> BTW, I have a couple of twinkies (got them direct from Google engineers at one of the first USB-PD interop workshops in Cambridge after I pointed out enough bugs in their implementation). And I've also implemented PD controller firmware. Happy to help with any open source implementation.
<modwizcode> so I did not fix my board :p maybe after this class. Got distracted by fixing my stupid aarch64 kernel project
<modwizcode> oh that's uh helpful
<d1b2> <rwhitby> I've also been talking with marcan about this topic.
<d1b2> <rwhitby> (and using my own proprietary board to fuzz the VDMs)
<_whitenotifier> [GlasgowEmbedded/glasgow] electroniceel pushed 1 commit to wip-revC3-hardware [+0/-0/±2] https://git.io/JtmS3
<_whitenotifier> [GlasgowEmbedded/glasgow] electroniceel 4cc578e - revC3: add an 4.7µF capacitor to VIO_AUX for improved decoupling
<modwizcode> Btw the reason I mentioned making a pcb was that I was thinking that it'd be pretty nice to write the interface as a glasgow applet so all you'd need is the minimal interface to the PD "bus"
<d1b2> <rwhitby> Right, he's interesting in sending VDMs and not having to shave the PD raw messaging yak. Hence the fusb302 on Glasgow idea that I suggested to him since he was already doing fusb302 on arduino.
<d1b2> <rwhitby> He only did arduino cause that was the code library he found for the fusb302, so another preshaved yak there.
<modwizcode> Yeah that's one usecase. But what's the minimum interface logic assuming that Glasgow takes over the role of the fusb302
<modwizcode> There's a pretty decent library for that I think for arduino so it suits his use just fine but it's not as nice for a generic PD peripheral
<d1b2> <rwhitby> AFE. Everything else can be done with enough gateware and code. But timing for some things is tight at the PHY level.
<d1b2> <rwhitby> I'm reviewing the fusb302 right now to see if there is anything Twinkie can do that it can't.
<d1b2> <rwhitby> And have some fusb302 and fusb302b on order.
<_whitenotifier> [glasgow] electroniceel commented on issue #255: Schematics & Layout TODO for revC3 - https://git.io/JtmS1
<d1b2> <rwhitby> Big thing going for it is that support for it is already in linux.
<modwizcode> Support for it how?
<d1b2> <rwhitby> But I've not written tcpm for it before. I've done NXP (my team wrote the code in ROM for that chip) , Cypress, analogix, etron and some TI, but not on semi.
<d1b2> <rwhitby> Support as a Linux port controller. Not as debugger though.
<electronic_eel> rwhitby: i had a glance through the fusb302b datasheet and it looks like it can only do PD 2.0 and not PD 3.0. the fusb307b seems to be capable of pd 3.0
<d1b2> <rwhitby> Yes, but not easily available?
<d1b2> <rwhitby> Pd3 of course is desirable.
<d1b2> <rwhitby> My current board does PD3 using NXP PHY.
<modwizcode> What does PD3 bring again?
<electronic_eel> PD3 means you can request arbitrary voltages and currents, not just preselected profiles
<d1b2> <rwhitby> Collision prevention, PPS, in the future extended to 50V/5A.
<d1b2> <rwhitby> All the long extended message, authentication and firmware update messaging.
<d1b2> <rwhitby> For what marcan wants, PD2 is perfectly fine.
<electronic_eel> farnell seems to stock som fusb307b, but availability seems to be limited for now. probably quite new
<d1b2> <rwhitby> https://gist.github.com/rwhitby/b98254e84ae402e87a25c6b981303511 is what I've been doing with macan, using my current board
<d1b2> <rwhitby> fusb307b still only does chunked extended messages, so still not as capable as the ptn5110.
<whitequark> this is a good reason to go with twinkie's design for a debugger
<d1b2> <rwhitby> (but otherwise good for pd3, and maybe you can do unchunked at the next level up using raw transmi)
<d1b2> <rwhitby> as long as you have someone to write the software levels for PHY, protocol, policy layers.
<d1b2> <rwhitby> (or are willing to work with the google code)
<electronic_eel> the ptn5110 should work well for a debugger, wouldn't it? so you don't have to mess with comparators and similar yourself
<whitequark> the google code can probably be a starting point
<d1b2> <rwhitby> ah yeah, I forgot twinkie was just part of the EC repo. that was indeed a pain to rebuild and flash.
<whitequark> electronic_eel: oh cool, i didn't realize those exist
<whitequark> that seems like it could replace half of the stuff on twinkie
<whitequark> without fundamentally changing what it des
<whitequark> *does
<electronic_eel> me neither until rwhitby just mentioned it
<whitequark> not entirely clear if there is an point
<whitequark> since twinkie doesn't really implement a source or sink so it has no need for actual power management features
<d1b2> <rwhitby> NXP didn't market it well. It's been around for a number of years, and I use it in all my customer engagements, but NXP has pretty much EOL'd it.
<whitequark> we're back to < whitequark> this is a good reason to go with twinkie's design for a debugger
<whitequark> see, this is basically why i bothered with twinkie in first place
<whitequark> it's... flawed... in most ways in which it does anything... but it's still better than being at the mercy of vendors for something as odd as pd
<d1b2> <rwhitby> it's the same old make vs buy decision
<electronic_eel> rwhitby: do you have the EOL info from some sources internal to NXP? bc they still list it as active on their website. just some package variants are eol
<whitequark> what i'm still not sure about is whether the twinkie AFE has any serious issues
<whitequark> because if not, we should obviously just use it, it's not really complex or expensive
<d1b2> <rwhitby> for example, ptn5110 is generic enough that it can actually be made to send PD messages tapped on the side of an existing host->active_cable->device link and inject any type of message (including the SOP*-Debug messages apple uses).
<d1b2> <rwhitby> The chip is not officially EOL, and I'd say they have enough inventory for many years of hobbyist use.
rwhitby has joined #glasgow
<electronic_eel> whitequark: how did you plan to do the usb pd protocol? a huge chunk of fsms in gateware or to do it with boneless code?
<whitequark> electronic_eel: nfi, twinkie has *something* that works to *some* extent
<whitequark> i mean, initially i hoped i could just use google code
<whitequark> then it turned out that google code is fucking horrible and only barely functions at all
<whitequark> i wasn't surprised that their hardware was nigh-unusable as OSHW and that was basically ok, i knew it was open-sourced as a matter of convenience
<whitequark> but i did not expect the frontend to be too broken to actually use it
<whitequark> well, firmware and software.
<d1b2> <rwhitby> Total Phase has a product that is basically a Twinkie inside and some nice (but closed source) software.
egg|anbo|egg has joined #glasgow
<_whitenotifier> [glasgow] smunaut commented on issue #255: Schematics & Layout TODO for revC3 - https://git.io/JtmQ3
GNUmoon2 has joined #glasgow
<rwhitby> Twinkie was designed before there were any commercial PHYs available for PD.
<rwhitby> https://www.nxp.com/products/interfaces/usb-interfaces/usb-type-c-true-plugn-play/usb-type-c-and-power-delivery-evaluation-kits:USB-TYPE-C-SHIELD are the PTN5110 dev kits. The MCUXpresso SDK USB-PD stack is behind an NXP account EULA (which I think just says you have it use it with NXP chips).
<_whitenotifier> [glasgow] electroniceel commented on issue #255: Schematics & Layout TODO for revC3 - https://git.io/Jtm74
egg|anbo|egg has quit [Remote host closed the connection]
<rwhitby> https://www.onsemi.com/pub/Collateral/EVBUM2559-D.PDF has all the interesting USB-C pins as test points.
<modwizcode> that's pretty neat
<d1b2> <rwhitby> key question for whether you want a sniffer or a participant. FUSB307 is no good as a sniffer, as it will GoodCRC anything it's enabled to receive.
<d1b2> <rwhitby> Twinkie is designed to be a sniffer, with some additional injection capability.
<d1b2> <rwhitby> FUSB302 seems to have a mode where you can turn off or delay the GoodCRC, so that needs to be investigated to be able to use it as a general purpose sniffer.
egg|anbo|egg has joined #glasgow
egg|anbo|egg_ has quit [Ping timeout: 246 seconds]