<pie_>
would you put reconfigurable hardware on a space probe or is that just too much extra complexity for no reason
<pie_>
*reconfigurable routing
eduardo_ has joined ##openfpga
eduardo__ has quit [Ping timeout: 240 seconds]
pie_ has quit [Remote host closed the connection]
pie_ has joined ##openfpga
<awygle>
pie_: sorry, yes, time zones. Also, I would, I have, and importantly, other people do
<pie_>
ah.
<awygle>
Check out microsemi RT line
<pie_>
note i specifically ment to say reconfigurable routing
<pie_>
(for no specific reason)
<awygle>
There are also rad hard virtex 5s for like 10k apiece
<pie_>
i figured ;_;
<awygle>
I worked on a project (which may never have flown) which just ran every external pin into an FPGA on every board, and then out again to its destination, to allow routing around broken wires or cables.
<awygle>
If you want something cheap that's "pretty OK" for radiation, the igloo line is fine for LEO stuff
<awygle>
They don't rate it for space but it's the same process and architecture as a space rated line
<awygle>
Just make sure you turn on the TMR and the EDAC
pie_ has quit [Read error: Connection reset by peer]
pie_ has joined ##openfpga
<pie_>
(or just never gonna fly at all, but yknow :P)
<pie_>
<pie_> im just pipe dreaming about cubesats btw
<pie_>
<pie_> personal project -> never gonna fly with 10k$ rad-hard :
<pie_>
<pie_> :/ :P
<pie_>
awygle, so since youre around can you give any pointers for material?
<pie_>
awygle, also im vaguely rememebring you may have mentioned working for aerospace before but either its just dejavu or i forgo
<pie_>
*forgot
<pie_>
the whole idea came from me poking around wrt formal methods in software lately
<awygle>
Yep, I've worked mostly in cubesats until very recently
<awygle>
What kind of material?
<pie_>
oh
<pie_>
for fault tolerant hardware arch stuff
<pie_>
i mean what i asked about earlier
<awygle>
Oh yes yes
<pie_>
what kind of cubesat stuff did you do
<pie_>
(lol pros vs amateurs i always find a little funny)
<pie_>
(also awesome, i have a contact now at least xD)
<pie_>
software is easy wrt procurement at least :/
<awygle>
Let me do a little digging. NASA has a ton of resources, and there's an avionics workshop every year
<pie_>
yeah ive rediscovered the technical reports server a couple times
<awygle>
With all the talks and papers online
<awygle>
Wrt to books, get a copy of SMAD
<pie_>
im rediscovering a lot of "retro computing" stuff lately because a lot of programming language theory originates form then and computing in general seems to have a bit more vibrant since it was an emerging field, more weird shit lying around
<awygle>
I don't really have any other suggestions, but I can't imagine there's much difference between them. The fundamentals are all the same.
<pie_>
ive never been really into retro computer stuff because i dont like "artificially" constrained systems, but given the more difficult circumstances i imagine people had to be more clever
<pie_>
oh yeah i think i have some SMAD pdfs laying around, dont rememebr what it stands for :P
<awygle>
Space Mission Architecture and Design
<awygle>
The new version is called something else though
<awygle>
Space Mission Engineering
<pie_>
did you work for gov or private?
<pie_>
howd you get into cubesats?
<pie_>
(if you dont mind me asking :P)
<awygle>
My best "cheap cubesat" tip is to use FRAM or MRAM for non volatile storage
<pie_>
something thats kind of bothering me is that theres a lot of stuff about process and such, but i havent seen anything on like for example, is normal pcb manufacturing spaceworthy? how do you design actual circuits for something high reliability? etc
<pie_>
i havent seen any "space electrical engineering" cookbooks ;PP
<awygle>
So PCB manufacturing, there's IPC class 3A for space
<pie_>
s/high reliability/space/
<pie_>
which is actually the reason i started looking into space stuff at all, earthy stuff is built to shit standards :PP
<awygle>
It's basically the same as 3 but with additional inspection and out gassing requirements
<awygle>
Double (at least) your voltage tolerance, especially on power transistors, to give you best chance of surviving an ion hit
<awygle>
Can't use wet electrolytic caps, only tantalum and polymer (and ceramic obv)
<pie_>
yeah
<awygle>
Use things like progressive discovery and plan for "okay half the ship is broken, how can I still accomplish this thing"
<pie_>
thats the fun part xD
<pie_>
progressive discovery?
<awygle>
Plan for temperature. LEO is *cold*, usually
<awygle>
Batteries especially don't like cold
<pie_>
i was actually wondering about that, i never actually knew how space cooling actually works
<pie_>
i mean i know about black body radiation
<pie_>
i just always found it weird that it would be enough to actually cool something down (but obviously works givent hat everything doesnt cook in space....)
<pie_>
*given that
<pie_>
(?)
<awygle>
Progressive discovery means that you don't assume that you have a given capability. So instead of turning on your 10W transmitter you first check if your antenna deployed.
<pie_>
ah yeah of course, none of this EAFP junk ;PPP
<awygle>
Yeah space cooling mostly involves big radiator fins. Cubesats spend so much time in shadow they tend to be frozen. You need heaters more than radiators. Also you get like a 40C differential between the solar panels and the battery when charging.
<awygle>
(on bigger cubes especially, 3u or 6u)
<pie_>
man things must love the temp gradients and thermal cycle
<pie_>
pfff id be lucky with a 1u probably ;_;
<pie_>
i was about to say you could use that temperature gradient for power generation but then i realized the whole point was probably to heat up the batteries lol
<awygle>
To answer an earlier question I worked for Planetary Resources, as well as (briefly, as an intern) Andrews Space before Spaceflight Services existed
<pie_>
"Industry Asteroid mining"
<pie_>
wait wat
<awygle>
Yeah. If you're interested I can talk more about it in PM, I try not to discuss former employers in detail in public
* qu1j0t3
has job envy
<pie_>
awygle, thats fine
<awygle>
The thing about working in space is that it's incredibly interesting but also incredibly consuming. I had a really hard time leaving work at work, and I frequently didn't leave work until very late.
<pie_>
i might ask then at some point :D dont want to go into it a lot right now
<awygle>
pie_: anytime
<pie_>
^_^
<pie_>
try not to kick the bucket :PP
<pie_>
awygle, btw did you learn most of this stuff on the job or...?
<awygle>
Yeah mostly. Good mentors etc. Space people love to talk space, the hard part is getting them to stop (source:this conversation)
<qu1j0t3>
LOL
<awygle>
Also go to SmallSat if you can
<awygle>
Great conference to meet folks actually doing things
<pie_>
haha
<awygle>
I'm happy to put you in touch with anybody you might want to talk to, if I can. My contacts are mostly US-based but I know some people at GOM and could probably dig up someone at Clyde too
<pie_>
next smallsat is quite a ways off hehe (though thats probably better that way for the moment, maybe i can go :D)
<awygle>
No idea where you are of course lol
<pie_>
awygle, ok thanks, i really appreciate the thought :D
<pie_>
maybe something will come of this in a while lmao
<pie_>
im in hungary
<pie_>
for the forseeable future anyway
<pie_>
GOM and clyde?
<awygle>
GOMspace is Dutch, Clydespace is British
<awygle>
I also have some Lithuanian contacts, now I think about it, and at least one in Spain
<awygle>
Nothing in Hungary that I can name off the top of my head
<awygle>
If you're interested in large systems, the constellation post mortem is _fascinating_
Hootch has quit [Quit: Leaving]
digshadow has quit [Ping timeout: 260 seconds]
<awygle>
anybody have strong CI opinions? travis, jenkins, circle, etc?
digshadow has joined ##openfpga
<qu1j0t3>
i've used a few, but i can't really recommend anything without knowing more about what you're trying to do
<qu1j0t3>
lately we've been just using a lambda function + codebuild --> amazon ecs, but that won't work for everyone
<qu1j0t3>
i hated jenkins and teamcity is annoying as well, but maybe some people love them
<qu1j0t3>
we also used codeship for some clients but that costs money if you need more than free tier
<qu1j0t3>
(both codeship's non-docker and docker products, which are rather different)
<qu1j0t3>
we have one client doing on-premises deployment and codebuild/codedeploy works ok for that.
<awygle>
Trying to run a docker container on azure, which builds an IP core, runs some verification stuff, gives me a binary y/n and also some stuff about resource utilization
<awygle>
I'd also like it to be portable from gitlab to github
<qu1j0t3>
i guess it comes down to taste.. e.g. i'd prefer a hosted solution than deploying something like teamcity or jenkins myself
<awygle>
I tend to agree, but I'm concerned about the flexibility of hosted solutions. I have no real experience though so maybe that's unfounded
<qu1j0t3>
i haven't really had any trouble with flexibility ... codeship pro has a bit of a learning curve but you can do reasonably tricky things
<qu1j0t3>
their system is a bit like docker compose for builds
<qu1j0t3>
i've also had good experiences with their support
<awygle>
Hmm, hadn't even heard of codeship I'll check it out
<qu1j0t3>
not sure how it stacks up vs travis, circleci, etc
<rqou>
i selfhost jenkins
<rqou>
worksforme(TM)
<rqou>
whitequark doesn't like jenkins though
* qu1j0t3
hated using it and would hate hosting it even more
<awygle>
I would be more open to Jenkins if the literature didn't seem to scream JAVA quite so much
<pie_>
just think of sexy japanese ladies
<jn__>
pie_: not sure that works for everyone
* awygle
winces
<awygle>
That was a bad joke in several ways
GenTooMan has joined ##openfpga
nrossi has quit [Quit: Connection closed for inactivity]
pie_ has quit [Ping timeout: 248 seconds]
eduardo_ has quit [Ping timeout: 246 seconds]
eduardo_ has joined ##openfpga
pie_ has joined ##openfpga
SuperChickeNES has joined ##openfpga
dx- has joined ##openfpga
Marex_ has joined ##openfpga
dx- is now known as dxdx
dx has quit [Killed (karatkievich.freenode.net (Nickname regained by services))]
<pie_>
i bet redhat is microsoft is being an evil nsa puppeteer
<jn__>
(but less relevant; they speculate that the linux foundation is behind this)
<jn__>
the Conservancy should please survive though. coreboot tries to get money through it.
<pie_>
psyopsss
* jn__
closes the tab
<awygle>
is one of these the guy suing everybody because he wrote like ten lines of ALSA or something?
Hootch has joined ##openfpga
Hootch has quit [Remote host closed the connection]
<pie_>
though actually its more like the cia to do things like this no? :P
<pie_>
"Others' comments notwithstanding, whether this whole episode has its roots with the VMware suit *or* GPL enforcement "style", it has the potential to drive a large wedge in the Linux/FLOSS community." for roughly the nth time?