Current Server Time: 03:05 (Central Europe)

#apertus IRC Channel Logs

2015/12/14

Timezone: UTC


00:08
hozer
left the channel
00:13
hozer
joined the channel
00:57
Bertl
off to bed now ... have a good one everyone!
00:57
Bertl
changed nick to: Bertl_zZ
02:38
troy_s
joined the channel
05:35
Bertl_zZ
changed nick to: Bertl
05:35
Bertl
morning folks!
05:37
cbohnens
changed nick to: cbohnens|away
06:20
intracube
joined the channel
06:26
Bertl
off for now ... bbl
06:26
Bertl
changed nick to: Bertl_oO
06:51
intracube
left the channel
06:57
Guest78145
left the channel
07:21
se6astian|away
changed nick to: se6astian
07:27
se6astian
changed nick to: se6astian|away
07:30
tezburma
joined the channel
07:39
intracube
joined the channel
08:43
se6astian|away
changed nick to: se6astian
09:08
John_K
hey all
09:08
Bertl_oO
hey John_K!
09:08
John_K
I'm getting ready to assemble a few Beta boards this week
09:09
John_K
running into some issues figuring out what components to order though, none of the resistors on the Main Board have values, etc....
09:09
Bertl_oO
yeah, that's because not all of them are fixed yet
09:10
Bertl_oO
let me know which value you are missing, I should be able to tell you
09:10
John_K
all R's on Main Board
09:10
Bertl_oO
the pullups around the i2c mux are 4k7
09:10
Bertl_oO
the pullups for the pic are 10k
09:11
John_K
also curious which height specifier was used for the JX1, 1,3,4 seem to be options (61083-10*400LF)
09:11
Bertl_oO
and the termination resistors are currently unpopulated
09:11
Bertl_oO
for the LVDS pairs you probably want 100R (those are the 0201 ones)
09:12
Bertl_oO
anything missing?
09:12
John_K
matching what you said to R #'s
09:13
John_K
there are some 0402's between the PCIE connectors
09:13
John_K
guessing those are the i2c mux
09:13
Bertl_oO
yep, precisely
09:14
John_K
is pull down for PIC also 10k?
09:14
Bertl_oO
that would be the termination
09:15
Bertl_oO
i.e. probably 50R later, but for now leave unpopulated
09:17
John_K
what about resistors near T1/T2?
09:17
John_K
and resistor for the buzzer
09:17
Bertl_oO
ah, yes, we didn't fix them yet, but the usual protection should do
09:18
Bertl_oO
for the buzzer we used 15R for now
09:20
Bertl_oO
but I have to admit, we haven't tested the buzzer yet :)
09:20
John_K
hehehe that's fine, I'm considering omitting it
09:21
John_K
but may include it just to play with later on
09:21
Bertl_oO
yeah, that's what we did
09:22
John_K
is the 2-pin header for an external speaker?
09:22
John_K
or status LED or something
09:23
John_K
it's the other MOSFET circuit besides the speaker
09:30
Bertl_oO
that one is for fan control
09:30
John_K
aha
09:30
Bertl_oO
i.e. to control a fan usually attached to the microzed FPGA heatsink
09:31
John_K
that makes sense
09:31
John_K
does the uZed burn that much power with the current design? or just being forward looking?
09:35
John_K
ok one last question for the main board, what height connectors were used for PB-SW/PB-SE the DF12 components and uZed mating connectors (61083-10*400LF)
09:35
cbohnens|away
changed nick to: cbohnens
09:38
Bertl_oO
some kind of cooling is required, otherwise the FPGA will overheat
09:39
Bertl_oO
the DF12 are only available up to 5mm and the 61083 are available in 4-8mm IIRC
09:39
Bertl_oO
so we did choose the 5mm variant, but 4mm should be possible as well
09:39
Bertl_oO
(5mm stacking height)
09:40
John_K
ok, 5mm variant for each?
09:40
Bertl_oO
yes, they have to have the same height
09:40
Bertl_oO
(DF12 and 61083 that is)
09:40
John_K
makes sense
09:41
Bertl_oO
I'm currently in a train to munich, and the connection is a little flakey
09:42
John_K
it's ok - it meshes well with me filling out these BOMs for order :)
09:42
Bertl_oO
where are you ordering?
09:43
Bertl_oO
I probably asked that last time, but I don't remember :)
09:43
John_K
I'm in California, so trying Digikey for most
09:43
John_K
I've got the CMV12000 and BGA sockets already
09:43
Bertl_oO
ah, okay, please be so kind and add the part numbers to the wiki or send them to se6astian
09:44
John_K
of course, I've been going off the Google Docs and modifying a copy for now
09:44
John_K
is there any harm in fully populating the power board?
09:46
John_K
or should I just DNP components as shown on the v0.18 photo on the wiki? (for me v0.22 board)
09:54
Bertl_oO
no harm in adding additional regulators, they will consume a little more power but that's it
10:01
John_K
what values are the Trim Resistors for the regulators?
10:01
John_K
10k? 100k?
10:04
matray1
joined the channel
10:14
se6astian
100k
10:14
se6astian
for best range 18k to gnd
10:14
se6astian
10k trim
10:15
se6astian
output 43k
10:15
se6astian
which gives a range of 1.389V to 3.667V
10:15
se6astian
greetings from herbert :)
10:15
se6astian
his laptop keeps disconnecting from the train wifi
10:15
se6astian
we are getting close to salzburg
10:15
John_K
I figured as much
10:15
John_K
nice
10:16
John_K
I'm slightly confused as to your first four lines
10:16
John_K
the trim pots should all be 100k?
10:17
tezburma
left the channel
10:19
Bertl_oO
we figured that the voltage range for most sensors will be between 1.5 and 3.5V
10:19
John_K
makes sense
10:19
Bertl_oO
so the most generic option is to have 18k/100k/43k
10:19
Bertl_oO
for the CMV12k (if you consider that sensor only) you can also go for
10:20
John_K
I'm going to be playing with both CMV12k and CMV20k only
10:20
Bertl_oO
20k/100k/62k
10:20
Bertl_oO
which will give you 1.681V - 3.731V
10:20
John_K
for NW/NN/NE respectively?
10:22
John_K
(might be an idea to drive these with a DAC from a small uC in the future instead of using trim pots)
10:24
Bertl_oO
check with the datasheet, the internal reference is 0.41V IIRC so you can do the math to figure out the ranges
10:24
Bertl_oO
in the future we will completely rework the power board and add a management unit
10:25
John_K
yeah that makes sense
10:26
Bertl_oO
especially as the sensor frontend is planned to be "plug and play" i.e. the power board adjusts the voltages for the sensor accordingly
10:26
Bertl_oO
(this is for example required when rotating or swapping out the sensor)
10:28
John_K
absolutely. I started making an interface board for the CMV20K but am going to have to change the size of the sensor board as the sensor covers the current mounting holes
10:30
tezburma
joined the channel
10:31
Bertl_oO
shouldn't be a problem (the mounting holes)
10:32
Bertl_oO
i.e. there should be enough space below the sensor for a screw head, and the sensor itself will use different holes to attach it to a lens mount
10:32
John_K
was considering eliminating the dummy interface board and fabbing a larger PCB that matches with the main board
10:33
Bertl_oO
that would be possible but would also break the modularity
10:33
John_K
probably, yeah. I need to figure out what kind of a socket I can use for the CMV20k as well
10:33
John_K
true
10:34
John_K
What other types of things have been imagined to replace the dummy board?
10:34
Bertl_oO
we investigated the socket 1 for the cmv20k
10:35
Bertl_oO
the dummy board will be replaced by a smart FPGA board to connect all the 80 LVDS lanes available on the sensor interface
10:35
John_K
the purpose made socket (by that company I can't remember)?
10:36
Bertl_oO
nah, the intel zif socket, but I'm sure andon is selling a socket for the cmv20k as well
10:36
John_K
ah
10:36
Bertl_oO
but the cmv20k should not be very problematic because of the large pitch
10:40
Bertl_oO
i.e. I'm sure there are several socket solutions available
10:40
John_K
right
10:42
se6astian
changed nick to: se6astian|away
10:47
se6astian|away
changed nick to: se6astian
10:47
se6astian
changed nick to: se6astian|away
11:07
se6astian|away
changed nick to: se6astian
12:13
cbohnens
changed nick to: cbohnens|away
13:16
Bertl_oO
off for a nap ... bbl
13:16
Bertl_oO
changed nick to: Bertl_zZ
13:57
matray1
left the channel
15:20
danieeeel
left the channel
15:22
LordVan
joined the channel
15:28
danieeeel
joined the channel
16:54
irieger
Hey folks. Has anyone seen some strange data coming from the sensor when taking the same image/exposure another time? I have taken a shot 2 hours ago which had a max raw value of 4092 so used the 12 bit up to the top. Now I have the same setup, same exposure of the lights (checked double with light-meter), same aperture, same sensor registers for everything and the image maxes out at code value 3199
16:56
se6astian
hi irieger
16:56
irieger
drives me nuts currently. I'm just trying to tune the sensor output to have a bit more dynamic range ...
16:56
se6astian
very strange
16:56
se6astian
did you reboot the camera already?
16:57
irieger
was out in between. But checked the sensor register and except of the temp sensor of the sensor everything is the same. And temp register was only 1 bit less and in a second capture 1 bit more so should make much off a difference
17:00
alexML
irieger: only the max value is different, or also the median?
17:00
se6astian
changed nick to: se6astian|away
17:00
irieger
haven't checked that.
17:00
se6astian|away
changed nick to: se6astian
17:00
irieger
wait a second.
17:01
irieger
median for the img with max 4092 is 1302, for the other it is 1277
17:02
alexML
median ratio is 1.02
17:03
alexML
so, the exposure appears to have changed by 0.03 stops
17:04
alexML
the peak is likely a specular highlight (just a guess, since I didn't see the raw files)
17:07
irieger
damn. you are right. Should go to bed. This stuff makes me tired ;-)
17:08
se6astian
:)
17:09
irieger
It's quite difficult with all the stuff interacting when changing just one register having to tune everything else and just see what improved or got worse
17:10
irieger
Could someone take apart a Blackmagic camera and record the SPI when they set the sensor up? ;-)
17:23
se6astian
I thought you have a BM 4k? :D
17:24
irieger
Jep. But I neither have the will to destroy it (the case seems not to be made to be opened), nor do I have the equipment and knowledge to meassure the serial interface ;-)
17:26
irieger
alexML: Btw. it's more some wrongly coming pixel value than a specular highlight. I'm currently overexposing a exposure chart a few stops
17:30
alexML
ah, quite likely, since this sensor doesn't clip sharply to white
17:32
Bertl_zZ
changed nick to: Bertl_oO
17:59
se6astian
changed nick to: se6astian|away
18:58
alexML
some progress on correcting pattern noise: http://www.magiclantern.fm/forum/index.php?topic=11787.msg158766#msg158766
18:59
arpu_
left the channel
18:59
arpu
joined the channel
19:01
danieeeel
what is the ISO on those? seems to be pretty noisy given that the illumination feels like enough
19:03
alexML
no idea, they are from Ursa Mini, and there's nothing obvious in the exif
19:06
alexML
read noise seems to be about 3 units
19:09
alexML
in the original exif (16-bit DNG), white level is 27580, so in the 12-bit DNGs, it's 1724
19:09
alexML
so, about 9 stops of DR
19:10
slikdigit
joined the channel
19:11
alexML
probably somewhere around iso 1000, from my old calculations
19:11
tezburma
left the channel
19:44
mgielda|away
left the channel
19:44
se6astian|away
left the channel
19:44
philippej
left the channel
19:44
philippej|away
joined the channel
19:44
se6astian|away
joined the channel
19:44
philippej|away
changed nick to: philippej
19:44
se6astian|away
changed nick to: se6astian
19:44
mgielda|away
joined the channel
19:44
karol|away
left the channel
19:45
karol|away
joined the channel
19:45
Bertl_oO
left the channel
19:45
Bertl_oO
joined the channel
20:08
arpu
left the channel
20:09
arpu
joined the channel
21:25
Bertl_oO
off to bed now ... have a good one everyone!
21:25
Bertl_oO
changed nick to: Bertl_zZ
21:33
LordVan
left the channel
21:33
se6astian
off to bed too
21:33
se6astian
good night :)
21:39
se6astian
changed nick to: se6astian|away
22:32
aombk
left the channel
22:34
aombk
joined the channel
22:34
troy_s
alexML: What technique are you using to find the matrix off of those passports?
22:47
alexML
troy_s: matching with an image from a Nikon camera
22:48
alexML
(not quite exact, but better than random :P )
22:48
troy_s
?
22:48
troy_s
Just using curves or?
22:48
alexML
I've matched the hue from the square boxes
22:49
alexML
https://github.com/apertus-open-source-cinema/misc-tools-utilities/blob/master/color-calibration/match_hues.m
22:59
danieeeel
left the channel