Current Server Time: 06:03 (Central Europe)

#apertus IRC Channel Logs

2014/01/21

Timezone: UTC


23:15
gwelkind
left the channel
23:16
carsonau
left the channel
00:13
gcolburn
left the channel
01:31
gabriel__
joined the channel
01:32
gabriel__
changed nick to: Guest6616
01:37
Guest6616
left the channel
01:54
Guest6616
joined the channel
02:26
carsonau
joined the channel
03:14
Guest6616
@FergusL I think he's referring to switching barriers
03:14
Guest6616
http://en.wikipedia.org/wiki/Switching_barriers
03:15
Guest6616
Oops, this is gwelkind, let me log in derp derp derp
03:15
Guest6616
changed nick to: gwelkind
03:17
gwelkind
there we go, yeah, sorry I was trying to solder a PIC board, barely looking at my computer
03:32
Bertl
'trying' doesn't sound promising :)
03:53
gwelkind
hahah, yeah, it went really poorly
03:54
gwelkind
I soldered on the LEDs and every capacitor backwards
03:54
gwelkind
Every polar capacitor anyways
03:54
carsonau
did you manage to get any magic blue smoke? :)
03:55
gwelkind
Dunno, I couldn't see a thing through the waves of vaporous tin smog
03:55
gwelkind
My lungs feel armor plated :P
03:55
gwelkind
I assumed that cathode was the positive end, because cations are positively charged!
04:03
Bertl
hmm, if you see tin smog, your temperatures are way to high :)
04:04
Bertl
and simply don't use polarized capacitors unless you really need to (for typical microcontrollers, you usually don't)
04:05
Bertl
carsonau: hey, IIRC you wanted to talk to me recently?
04:05
carsonau
yeh sebastian told me to talk to you about hardware stuff and in what ways i can contribute to the project
04:06
Bertl
great! so what is your main interest and where would you like to help?
04:06
Bertl
(and most importantly, what would you like to talk about :)
04:07
carsonau
i think i'll be most valuable in schematics / circuit design
04:07
Bertl
okay, I read that you have some experience there, did you get a chance to have a look at the FE design yet?
04:07
carsonau
what's the FE design?
04:08
carsonau
I've had a look at the existing EagleCAD files for the prototype board
04:08
Bertl
the sensor frontend we are using in the alpha prototype, yes
04:08
Bertl
so what's your comment on that so far? any obvious issues, potential problems, improvements?
04:09
carsonau
you designed that one right?
04:09
Bertl
correct
04:10
Bertl
but please, do not hold back
04:10
carsonau
i'm surprised you didn't need to length match all the LVDS pairs... and purely relying on the FPGA adjustable IODELAYS
04:11
Bertl
interesting that you mention that ...
04:11
Bertl
do you know the LVDS frequency?
04:12
carsonau
was it 300 mbits/sec for the CMV12000?
04:12
carsonau
i think there was a new version of the CMV12000 recently
04:12
carsonau
that could do higher?
04:12
Bertl
correct, we are at 300MHz LVDS clock
04:12
gwelkind
left the channel
04:12
carsonau
for my project we used the CMV2000 and i recall it could go up to 480mbits/sec
04:12
Bertl
the more recent versions can do 600MHz
04:12
carsonau
wow!!!
04:12
carsonau
32 LVDS channels?
04:12
carsonau
or more?
04:13
carsonau
64?
04:13
Bertl
yep, do you have an idea how long 300MHz are wavelength wise?
04:13
Bertl
i.e. 32 channels atm, 64 max
04:13
carsonau
300 mhz........around 3 ns period
04:13
carsonau
45cm?
04:13
carsonau
:P
04:13
carsonau
rough estimate
04:13
Bertl
yeah, it's about half the length in copper
04:14
Bertl
so, what matching between different pairs would you suggest for good timing?
04:14
Bertl
10% of the wavelength?
04:16
carsonau
10% sounds reasonable
04:16
carsonau
but the more the better (from a timing perspective) right? the more margin you have without need to use IODELAYS for tuning
04:17
Bertl
in theory yes, although anything below 1% is probably useless, especially as there are package differences and similar
04:17
Bertl
http://vserver.13thfloor.at/Stuff/AXIOM/LVDS_netlen.txt
04:17
Bertl
here are the pair lengths in mm (on the frontend)
04:17
carsonau
ahh right
04:18
carsonau
you can account for the package differences as well! :)
04:18
carsonau
(a lot of work :P)
04:18
Bertl
so we roughly have 24mm max and 8mm min
04:18
Bertl
what does that give with roughly 250mm wavelength?
04:19
carsonau
i see
04:19
Bertl
so while I could have trimmed them to match more closely, the gain would be questionable
04:20
Bertl
but an interesting detail is that there is a really weird delay I attribute to the FPGA
04:20
carsonau
what is it?
04:21
Bertl
i.e. all LVDS channels are adjusted with the configurable INPUT delay elements
04:22
Bertl
so I'd suspect a pattern which closely matches the total wire lengths
04:22
Bertl
s/suspect/expect/
04:22
Bertl
sorry, late here
04:23
Bertl
http://vserver.13thfloor.at/Stuff/AXIOM/cmv_train2.out
04:23
Bertl
but that's what I get instead
04:24
Bertl
note that the LVDS channels are connected in a completely different way and that the configured delays do not match with the wire lengths either
04:25
Bertl
so I presume the synthesis tools add a systematic delay which I haven't figured out yet
04:26
carsonau
do you know if the zedboard
04:27
carsonau
has its pairs matched?
04:27
carsonau
or is your LVDS_netlen.txt
04:27
Bertl
the FMC specification requires good matching
04:27
carsonau
ahh right
04:27
Bertl
and I checked with the net length on the zedboard as well
04:27
carsonau
what about the bond wire flight times in the FPGA package?
04:27
Bertl
I also looked at the beforementioned package differences, although they are a magnitude smaller
04:28
carsonau
ahh right
04:28
Bertl
so I presume it is an artificial delay the tools add for whatever reason
04:28
Bertl
I have to write that together in a meaninfull way and ask our contact at xilinx about that
04:29
carsonau
that is strange
04:29
Bertl
but if you find that interesting, you might want to help putting together the various lengths from all involved parts
04:29
Bertl
i.e. the FE lengths, the zedboard lengths and the package lengths
04:30
Bertl
another interesting interpretation might be that the sensor produces those delays
04:30
carsonau
could be
04:30
Bertl
i.e. that the LVDS channels are not in sync as the documentation suggests
04:31
carsonau
sebastian told me a little bit about the non-prototype axiom
04:31
carsonau
is there a page on the wiki where i could read more about your plans for that?
04:32
Bertl
probably the best source of information for the beta prototype (the next step) are the IRC logs
04:32
carsonau
haha
04:32
Bertl
basically we figured that something like the Zynq comes handy
04:33
Bertl
we also figured that we probably want multi gigabit tranceivers
04:33
Bertl
which limits it to the 7030 and the newest member the 7015
04:34
carsonau
have you decided whether or not to use KiCAD for the beta?
04:34
Bertl
we want to have between 0.5 and 1GB of memory to handle the data, we would like to see some of it on the FPGA side
04:34
Bertl
we definitely would like to go with kikad if possible, really depends on who does the design
04:35
Bertl
(so you are running in open doors here if you suggest to use kikad :)
04:35
carsonau
sorry i'm not familiar with the expression "running in open doors" :P
04:36
carsonau
what does it mean?
04:36
Bertl
yeah, that is probably a bad translation, sec
04:37
Bertl
"to kick at an open door"
04:38
Bertl
does that make more sense to you? I.e. you do not need to convince me of kicad
04:39
Bertl
the reason I chose eagle was that I've been using it for quite some time, and I wanted to complete the frontend quickly ... and I tested beforehand that I can convert most of the eagle layout and schematic to kicad later
04:39
carsonau
oh ok
04:40
Bertl
a second argument pro eagle was the fact that OSHpark processes eagle files directly
04:40
Bertl
i.e. there is no need to break it down to gerbers
04:40
carsonau
i am presuming you're going to do the beta prototype PCB's as well
04:42
Bertl
I'm not sure, if there is a chance that we find somebody (person, team, partner) which is interested and willing to create the next prototype step with us, we are open in this regard
04:42
Bertl
of course, open source software is always preferred, but 'open' document formats are good enough
04:43
Bertl
for example, it's a real PITA to convert anything from Altium
04:43
carsonau
haha
04:43
Bertl
but since eagle switched to XML style documents, it is quite simple/easy to work with them
04:44
Bertl
I'm hoping that the CERN involvement in kicad will speed up development significantly
04:44
carsonau
what improvements would you like in KiCAD at the moment?
04:45
Bertl
last time I checked, I had some troubles working with the router (not autorouter) and mostly visual issues
04:45
Bertl
i.e. displeasing presentation, jerky scrolling/zooming, etc
04:46
Bertl
(I was able to fix a few things with patches though :)
04:47
Bertl
a nice eagle 'feature' are ULPs, and I haven't figure how/if kicad allows for something like that
04:47
Bertl
also back/forward annotation was not working as expected
04:47
Bertl
but that was last summer, so a lot might have been improved already
04:49
Bertl
looking at the CERN page I see python scripting, which sounds good :)
04:52
Bertl
btw, have you seen the P&S router in action? that's one of the most interesting features for me atm
04:53
carsonau
haven't used that in kicad
04:53
carsonau
in fact haven't used kicad much at all
04:53
carsonau
i installed it two years ago
04:53
carsonau
got fed up with using it
04:53
carsonau
then crawled back to non-open source tools
04:53
carsonau
haha
04:54
Bertl
I see, so what do you use atm?
04:54
Bertl
http://www.youtube.com/watch?v=zxHDAHpR5Ls (P&S)
04:55
carsonau
i mostly 90% use altium
04:55
carsonau
10% eagle
04:56
Bertl
so, do you know how to properly export a schematic/layout in altium so that it can be used e.g. in kicad or eagle?
04:56
carsonau
never had the need to...so i do not know :(
04:58
Bertl
I see ... what did you pay for your altium package?
04:59
carsonau
during university, was in a solar car team, and Altium sponsored us licences...
04:59
carsonau
and now I use Altium at work...
04:59
carsonau
so I've never had to pay personally for Altium
04:59
Bertl
so the company you're working for pays that for you
05:00
carsonau
the company has the Altium licence
05:00
carsonau
so all the work projects are done in Altium
05:00
carsonau
I use Eagle for my personal projects
05:00
Bertl
I see
05:00
Bertl
anyway, back to the prototype ... anything else but the wire lengths?
05:01
carsonau
nope :)
05:04
carsonau
all looks good
05:06
carsonau
did you design the elphel cameras?
05:07
Bertl
nope, not involved at all
05:09
carsonau
when do you plan for the beta prototype PCB design to start happening?
05:10
Bertl
as soon as we finish the crowd funding
05:11
carsonau
what are you working on at the moment?
05:12
Bertl
noise compensation and FPGA image processing
05:12
carsonau
is the current focus to get good images and videos from the alpha prototype for the purposes of crowdfunding?
05:12
Bertl
correct
05:13
Bertl
and of course, we want to get an idea if the chosen path is the right one
05:48
cksa
joined the channel
05:49
carsonau
left the channel
06:01
Bertl
off to bed now ... have a good one everyone!
06:01
cksa
goodnight!
06:20
carsonau
joined the channel
06:22
cksa
left the channel
06:46
jucar
left the channel
06:51
jucar
joined the channel
08:09
philippej
joined the channel
08:09
jucar
left the channel
08:48
jucar
joined the channel
09:09
shinokubo
joined the channel
09:27
carsonau
left the channel
10:07
FergusL
Hi there
10:13
philippej
Hi FergusL
10:35
shinokubo
left the channel
10:41
shinokubo
joined the channel
11:08
aombk
joined the channel
11:08
aombk
left the channel
11:08
aombk
joined the channel
12:24
shinokubo
left the channel
13:01
aombk
left the channel
13:10
shinokubo
joined the channel
13:36
ThatCantBe
left the channel
14:50
gcolburn
joined the channel
15:01
gcolburn
left the channel
15:26
Bertl
morning everyone!
15:30
philippej
morning sir !
15:51
philippej
left the channel
15:59
se6astian
joined the channel
16:00
se6astian
good evening
16:09
mars_
left the channel
16:09
mars_
joined the channel
16:28
shinokubo
left the channel
16:51
opusprod
joined the channel
16:52
opusprod
I everyone
16:54
Bertl
welcome opusprod!
16:55
gwelkind
joined the channel
18:11
gwelkind
left the channel
18:22
gcolburn
joined the channel
18:37
shinokubo
joined the channel
18:37
shinokubo
left the channel
18:39
shinokubo
joined the channel
18:49
philippej
joined the channel
18:56
opusprod
left the channel
20:02
jucar
left the channel
20:12
gwelkind
joined the channel
20:16
jucar
joined the channel
20:30
shinokubo
left the channel
20:37
gwelkind
left the channel
20:51
shinokubo
joined the channel
21:18
philippej
left the channel
21:48
se6astian
good night
21:48
se6astian
left the channel