Current Server Time: 13:38 (Central Europe)

#apertus IRC Channel Logs

2013/06/24

Timezone: UTC


22:11
s3bastian
left the channel
10:37
se6astian
joined the channel
10:47
se6astian
left the channel
10:48
se6astian
joined the channel
12:53
Bertl
morning everyone!
12:55
se6astian
hello!
12:55
se6astian
today I also woke up at 11 AM ;)
12:55
se6astian
the luxury of the paramedic night
13:09
Bertl
hehe
13:10
Bertl
I did some HDMI testing tonight, and it seems that the available code is quite picky about the attached devices
13:10
Bertl
i.e. I think we need to improve that if we plan to stick with the ADV7511 in long term (for whatever reason)
13:11
se6astian
what did you try to attach so far?
13:11
Bertl
all the monitors I could fine here :)
13:11
Bertl
*find
13:11
se6astian
can the HDMI output mode be configured in the PL/PS 50HZ/60Hz, etc?
13:12
se6astian
and what monitors worked/didnt work?
13:12
Bertl
depends on what interface is used, the PL code from analog, which interfaces with the linux driver basically creates a 24bit framebuffer
13:13
Bertl
this can be modified within certain parameters from the kernel/userspace
13:14
Bertl
the funny part is, that my LG monitor, which works fine with devices like the raspberry pi, does not show any output via HDMI on the zedboard, but definitely gets detected and also keeps showing a black screen (i.e. doesn't turn off)
13:15
Bertl
my Dell monitor OTOH shows the output just fine, as does my old HP
13:15
Bertl
my Phillips doesn't get detected and doesn't show anything
13:16
Bertl
my suspicion is the EDID/DDC stuff (IIC) which might be incomplete or broken
13:16
se6astian
I see
13:16
Bertl
I plan to do some PL only projects to test the HDMI stuff at low level, we'll see what I can make work there
13:19
Bertl
btw, I recompiled a better/newer busybox, if anybody is interested, just drop me a note
13:19
Bertl
ah, and I managed to boot u-boot via TFTP from inside u-boot :)
13:20
Bertl
(which is nice if you want to test a new u-boot version first)
13:21
Bertl
I also added nfs support to all three kernels (digilent, xilinx and analog) ...
13:32
se6astian
nice!
13:32
se6astian
so many good things and so much progress in such a short time
13:33
se6astian
I feel like nothing can stop us ;)
13:37
se6astian
a pcb routing question
13:38
se6astian
our lvds lanes have max 300mhz
13:38
se6astian
so 3.3ns period
13:38
se6astian
Stuart said that 1ns = 6" of trace length
13:38
se6astian
our traces are just 2-3" long
13:38
se6astian
does that mean we can ignore any length differences of the lvds trace lines?
13:40
Bertl
depends, any length difference will result in a timing difference, 3" (if the 6" = 1ns is correct) means 500ps delay
13:40
Bertl
those might not be critical, but I'd definitely keep the differences small
13:40
Bertl
especially between lvds pairs
13:42
se6astian
you mean like positive/negative lanes of same LVDS pair right?
13:42
se6astian
since they are all next to each other on PCB that should already be that way
13:42
Bertl
altera has an IMHO nice document about LVDS routing: www.altera.com/literature/wp/wp_lvdsboard.pdf
13:42
se6astian
just some pairs are closer to FMC than others
13:44
Bertl
how much do they differ?
13:47
se6astian
I am expresspcb has no option to meassure trace lengths
13:47
se6astian
I am afraid expresspcb has no option to meassure trace lengths
13:47
Bertl
http://www.ti.com/lit/an/slla014a/slla014a.pdf (this is from ti)
13:48
Bertl
well, you can estimate them by adding up the line lengths
13:48
se6astian
thanks, good reads
13:48
Bertl
or simply counting the grid squares (as a rough estimate)
13:48
se6astian
will try to estimate
13:50
Bertl
of course, the closer the lengths (in general and to each other) the better the signal
13:58
se6astian
stuart seems to have cared about routing well here already
13:58
se6astian
all traces LVDS traces are P/N grouped next to each other and have a length of 1.1-1.4"
13:59
Bertl
that should be more than fine
14:00
se6astian
great
15:46
se6astian
just checked: FMC LPC connector is around 25$
15:46
se6astian
http://avnetexpress.avnet.com/store/em/EMController?term=ASP-134604-01&x=0&y=0&action=products&langId=-1&storeId=500201&catalogId=500201&hbxSType=&N=0&filterButton=true
15:55
se6astian
about the Zedboard
15:55
Bertl
cool, btw, you mentioned an FMC cable or flexible extension .. any idea what that would cost (e.g. for testing)?
15:55
se6astian
can you choose in PL what each of the pins from the FMC connector should be routed to?
15:55
Bertl
yes
15:56
Bertl
well, no, you cannot reroute the board
15:56
Bertl
but you can choose that in the design
15:56
se6astian
I am verifing the PCB atm
15:56
se6astian
do I need to care about which LVDS goes to which pin on the FMC connector?
15:57
Bertl
there are 34 differential pairs routed to the FMC
15:58
Bertl
2x 16 LA Bus pairs in two different banks IIRC
15:58
Bertl
plus two CC pairs
16:00
Bertl
the CC pairs are 'special' in such way that they are clock capable
16:01
Bertl
I just looked them up in the HW_UG, those are 0,17 and 1,18
16:04
se6astian
sent enquiry to Samtec
16:26
se6astian
0,17 and 1,18 are "row, coloumn" ?
16:26
Bertl
no, the actual pin numbers
16:26
Bertl
i.e. 0 and 17 is the first pair
16:26
Bertl
1 and 18 the second pair
16:28
se6astian
do you have a datasheet at hand that shows how pins are numbered through?
16:29
Bertl
the zedboard schematics should have that info
16:30
Bertl
but let me check what I can find
16:31
dmj_nova
do we know what format of the sensor's output?
16:32
se6astian
http://zedboard.com/misc/files/ZedBoard_RevC.1_Schematic_preliminary.pdf page one has the FMC schematics indeed
16:32
se6astian
but I also need the numbering scheme for the footpring pads
16:32
se6astian
hi david, what kind of "format" do you refer to?
16:34
dmj_nova
the signalling protocol it uses...The format of the data it's sending. I know it uses a number of LVDS channels, but what does it send over those channels?
16:36
se6astian
dmj_nova did I send you the datasheet yet?
16:37
dmj_nova
not to my knowledge
16:37
se6astian
ok, one sec
16:38
se6astian
sent
16:38
se6astian
page 20
16:38
se6astian
point 4.3 pixel read out
16:40
Bertl
regarding FMC, it seems that the pinout is part of the specification which can be bought for ~50USD :)
16:41
Bertl
but I'd check with the zedboard gerbers
16:42
se6astian
damn, well stuart routed the FMC pins already so he must have gotten the info from somewhere
16:42
se6astian
I just want to double check
16:52
dmj_nova
We should definitely make a virtual sensor module
16:54
Bertl
yes, I also suggested to build a sensor replacement, which connects to the FMC (or later instead of the sensor) for development and testing
16:54
Bertl
i.e. a simple FPGA, maybe with HDMI input or so
16:55
dmj_nova
FMC?
16:56
Bertl
the connector on the zedboard, which will be the interface to the CMV12000 (in the prototype)
16:56
dmj_nova
ah, yeah
16:57
dmj_nova
for now, we can probably get away with a virtual sensor on the fpga
16:57
Bertl
sure, and the code could be used lateron
17:13
se6astian
I just noticed that the OUTCTR_N
17:13
se6astian
LVDS negative control channel output
17:13
se6astian
LVDS positive control channel output
17:14
se6astian
lines are a lot longer than the normal LVDS lines
17:15
se6astian
it provides: LVDS control line with frame and line information
17:15
se6astian
so its probably not a big deal right?
17:20
Bertl
does anybody know how to deal with:
17:20
Bertl
WARNING:PhysDesignRules:2500 - This design does not have a PS7 block. Instantiate the PS7 block in order to ensure proper fabric tie-offs and correct operation of the processing_system7.
17:21
Bertl
i.e. how do I instantiate the PS7 block to make the tools happy?
17:21
Bertl
(note: this is a PL only design, so I don't need/want the PS to run anything)
17:25
se6astian
no idea sorry ;)
18:11
se6astian
hurray found a schematic error in the design!
20:09
dmj_nova
se6astian: So it occurred to me that we could give the user a lot more flexibility with their image if they can manually adjust the three sloping zones
20:24
se6astian
sure
20:24
Bertl
I think we should expose all possible settings to the user, maybe in different user levels (basic, advanced, expert)
20:27
dmj_nova
yes, we don't want to make things too complicated
20:28
dmj_nova
In fact I would actually recommend finding really good defaults and not directly exposing many of the actually adjustable things to the user unless they choose to enable it
20:30
dmj_nova
in the case of adjusting the sloping zones, it should allow much more detail/flexibility for grading where it's needed most.
20:31
dmj_nova
for instance more bits on the face in shadow while letting the sky have detail
20:33
dmj_nova
left the channel
20:33
dmj_nova
joined the channel
21:11
se6astian
time for bed, will be at the university shooting graduation ceremonies tomorrow ;)
21:11
Bertl
good night and have fun then!
21:11
se6astian
"shooting" in the sense of "filming", for our non filmmakers here ;)
21:12
Bertl
awww *G*
21:12
se6astian
which leads me to one of my favourite quotes from my ex boss who has been a camera man for 40 years
21:12
se6astian
he was in jerusalem at the david star hotel
21:13
se6astian
where high ranking US diplomats were currently staying
21:13
se6astian
he was there to shoot parts for a documentary film
21:13
se6astian
and needed a pan of the city scape on camera
21:13
se6astian
so he went to the receptionist
21:13
se6astian
and asked her
21:13
se6astian
"excuse me, can I shoot from the roof"?
21:14
se6astian
good night :)
21:14
se6astian
left the channel
21:16
dmj_nova
hehe
21:23
Bertl
somehow the 'ex boss' and 'has been' worries me how the story ended ...