| <Khlut> | hi | 00:00 |
|---|---|---|
| <andrey> | lindes, it is getting. Image direction (in 1/10-th of a degree), and hacked dest. longitude/dest latitude for pitch/roll - there is no special Exif fields for that | 00:02 |
| <lindes> | is there any way to put it in as an exif comment or somesuch, so that at least it's recorded? | 00:03 |
| <lindes> | or even as a side file or something. I suppose I could poll it around the same time as I get the picture, but if things are moving a lot, that may have a big difference | 00:03 |
| <andrey> | camogm has an option to record kml files, same as here http://community.elphel.com/pictures/GEO_TAG2/ | 00:03 |
| <lindes> | oh, cool | 00:04 |
| <lindes> | will it put that data in there, then? | 00:04 |
| <lindes> | sure looks like it | 00:04 |
| <andrey> | ccam.php?description=aa%20bb%20cc | 00:04 |
| <lindes> | neat | 00:04 |
| <andrey> | I think it is set up with 64 bytes | 00:05 |
| <andrey> | it should not be really difficult to add custom fields to Exif. You will need several things: | 00:07 |
| <lindes> | ImageDescription = "aa bb cc?" | 00:07 |
| <lindes> | ? | 00:07 |
| <andrey> | yes | 00:07 |
| <andrey> | modify elphel.cvs.sourceforge.net/elphel/elphel353-7.1/apps/web/imgsrv/Exif_template.xml?view=markup | 00:07 |
| <andrey> | it is a template fro Exif fields processed by http://elphel.cvs.sourceforge.net/elphel/elphel353-7.1/apps/web/imgsrv/exif.php?view=markup | 00:08 |
| <lindes> | heh, <!--Compass fileds - hack: using dest lat/long as no roll/pitch fields are available in Exif--> :-) | 00:09 |
| <andrey> | template defines fixed/variable fields, and later works like format in printf, so only the actual data is stored for each frame, combined with temlate when requested | 00:09 |
| <andrey> | There is a device driver http://elphel.cvs.sourceforge.net/elphel/elphel353-7.1/os/linux-2.6-tag--devboard-R2_10-4/arch/cris/arch-v32/drivers/elphel/exif353.c?view=markup that allows you to write data to the variable fields, they all go to "page 0". While serving frame interrupts, driver (in cc353.c) copies page 0 to one of the page 1...2047, incrementing number and saving the index with the image in the circbuf. When you read Exif file (raw d | 00:14 |
| <andrey> | ata - /exif.php?exif=0) the template is combined with binary data | 00:14 |
| <andrey> | I did not want to use undefined fields to make sure other software would not remove them. And as you see, Flickr even shows it | 00:17 |
| <andrey> | http://elphel.cvs.sourceforge.net/elphel/elphel353-7.1/os/linux-2.6-tag--devboard-R2_10-4/include/asm-cris/elphel/exifa.h?view=markup - another Exif-related file, visible for applications (like this one - http://elphel.cvs.sourceforge.net/elphel/elphel353-7.1/apps/php-5.2.1/ext/elphel/elphel_php.c?view=markup) | 00:20 |
| <_polto_> | Andrey, Gimp support RAW with ufraw & lcms (just apt-get install ufraw lcms). resolution (2592+4)x(1936+4) what is the offset ? | 00:33 |
| <andrey> | 0 | 00:44 |
| <_polto_> | hmm. | 00:50 |
| <_polto_> | and is the image type : RGB or indexed, planar RGB, RGB-alpha ? | 00:52 |
| <andrey> | no, monochrome, 2 bytes/pixel | 00:56 |
| <_polto_> | i have not that chose | 00:58 |
| <_polto_> | no monochrome | 00:58 |
| <_polto_> | This is where we should go instead LW/LT (i's tomorrow) http://www.libregraphicsmeeting.org/2008/index.php?lang=en&action=projects | 00:59 |
| *ChanServ gives channel operator status to _polto_ | 03:28 | |
| <beug> | hi all | 11:20 |
| <_polto_> | hi | 11:21 |
| *ChanServ gives channel operator status to _polto_ | 13:55 | |
| <_polto_> | t'as fini ? | 13:58 |
| <_polto_> | ;) | 13:58 |
| <lindes> | hello. anyone still here? | 20:33 |
| <lindes> | I have some bad news, but I'm hoping it's not tragic news, and looking for help diagnosing... | 20:33 |
| <lindes> | at a minimum, one of the bridge rectifiers in our camera is toast. the "magic blue smoke" has been released. :-( | 20:35 |
| <lindes> | it's difficult for me to tell what else may or may not be fried | 20:53 |
| <lindes> | there's a small amount of visual damage elsewhere, but I don't know if it's just a side-effect of the rectifier going, or another failure | 20:53 |
| <siraf_> | is David Lindes here? | 20:58 |
| <lindes> | hey | 20:59 |
| <lindes> | yes | 20:59 |
| <lindes> | this is he | 21:00 |
| <siraf_> | have you spoken to anyone else about your issues yet? | 21:04 |
| <lindes> | only here within our group | 21:05 |
| <siraf_> | ok | 21:05 |
| <siraf_> | can you tell me which board it is? | 21:07 |
| <lindes> | yeah, it's... one sec | 21:07 |
| <siraf_> | also, please email joan@elphel.com with that info and the following | 21:08 |
| <elphel> | unfortunately it was the only board with 12VDC and tested with 10369. We have some 48V, but none is tested with 10369 (so connector may be not working) | 21:09 |
| <lindes> | it's the main board, I think... not the 10369, but the... I'm having trouble finding the appropriate label. but the one that the camera itself actually connects to, and the ethernet jack and such | 21:10 |
| <siraf_> | and send joan the address and a description of the problem, we may be able overnight the replacement | 21:10 |
| <lindes> | we can run on 3.3v instead. | 21:10 |
| <lindes> | by overnight, that'd be arriving tomorrow? | 21:10 |
| <siraf_> | saturday | 21:10 |
| <elphel> | I understand it is 10353 - the only with 12V installed and the only tested to work with 10369 left (others are in China, Russia, Ukraine) | 21:11 |
| <elphel> | With 3.3 I'm afraid, both boards will be fried, not only one. And 10369 are too precious now until we got mor assembled - those I hand-built myself | 21:12 |
| <elphel> | Faris, we do not have any replacements | 21:12 |
| <lindes> | one sec... | 21:13 |
| <elphel> | Can you id it happen thet rectifies got burnt? And what else did burn? | 21:13 |
| <Joan> | siraf_ Hello | 21:15 |
| *ChanServ gives channel operator status to _polto_ | 21:15 | |
| <siraf_> | elphel: just relying what joan told me | 21:15 |
| <Joan> | What did I say? :-) | 21:15 |
| <elphel> | Faris, that was the only with 12V and we do not have any 12V DC-DC left, need to reorder | 21:16 |
| <lindes> | ok, so... for sure, the bridge rectifier that's next to the header pins is toasted. there's some burn damage near by... which I'm unsure if it's just from that one component, or other components had damage as well | 21:16 |
| <Joan> | elphel, don't we have some extra at the office? | 21:16 |
| <elphel> | ?? | 21:17 |
| <_polto_> | Joan, no more. | 21:17 |
| <lindes> | the FKC08-12S33... | 21:17 |
| <Joan> | not cameras, just 12V components | 21:17 |
| <elphel> | We had only one 12V, but that was in old body, w/o mounting holes. We asked Deon to insatll the last 12V DCDC we had | 21:17 |
| <lindes> | mostly looks in tact, but has some possible damage. hard for me to tell visually. I haven't yet probed anything, not knowing what to probe | 21:18 |
| <lindes> | elphel: ok, we won't try supplying 3.3 to it. :-) | 21:19 |
| <elphel> | But wjhat exactly happened? I've never seen something like that during normal camera operation the current is easily handled by the rectifies | 21:19 |
| <lindes> | (sorry, I'm a little behind... distracting environment at the moment.) | 21:19 |
| <_polto_> | how did it toasted ? do you know for sure why and how ? (to avoid to do it twice) | 21:20 |
| <lindes> | Joan: if you're able to overnight something, that would be fantastic. I'll have to figure out the logistics of where it would go and such though | 21:20 |
| <lindes> | ok, so here's the story of what happened: | 21:20 |
| <lindes> | I had gotten things working using a 12V "wall-wart" transformer | 21:21 |
| <lindes> | but wanted to test things in a more mobile configuration | 21:21 |
| <lindes> | so, I wired up a pack of 8 AA recharchables (NiMH) as a power supply | 21:22 |
| <lindes> | at the beginning of the day, I think those were reading 10.9ish volts or so | 21:22 |
| <lindes> | they currently read 9.98, though of course it may have been different at the time of the problem | 21:22 |
| <lindes> | anyway, things were working fine for a while... | 21:22 |
| <lindes> | my script to capture a still once per second was running | 21:23 |
| <lindes> | and I was ftp'ing images off periodically to make sure things were working | 21:23 |
| <lindes> | and all seemed happy | 21:23 |
| <lindes> | I had that running for maybe an hour or so, including for parts of time where I just sort of let it sit beside me, without ethernet connected and without doing anything with it | 21:23 |
| <lindes> | I wanted to get a feel for how quickly the CF card would fill up, among other things | 21:24 |
| <lindes> | after a while, though, it was warm enough that I decided to turn it off | 21:24 |
| <lindes> | had it off for maybe 30-45 minutes | 21:24 |
| <lindes> | and then plugged it back in again for when I was going to go on a bus ride | 21:25 |
| <lindes> | (I had a GPS plugged in, etc.) | 21:25 |
| <lindes> | I again connected to it, and all seemed fine for a while | 21:25 |
| <lindes> | then I disconnected and just let it run | 21:25 |
| <lindes> | pointing it around, etc. | 21:25 |
| <lindes> | at some point, I suddenly noticed that the batteries were hot | 21:26 |
| <lindes> | previously, they had been cool even when the camera unit was hot | 21:26 |
| <lindes> | I quickly disconnected the batteries | 21:27 |
| <elphel> | Maybe the voltage gradually went too low, current increased... I did not use 12 V much myself | 21:27 |
| <lindes> | but I was out and about downtown, so I couldn't really do anything but put it in my bag | 21:28 |
| <elphel> | Can you send high-res images of the damage? | 21:28 |
| <lindes> | and later, when I opened it up, I could smell (even from in my bag, actually) the smoke, and opened things up and saw the damage... | 21:29 |
| <lindes> | probably, yes. | 21:29 |
| <lindes> | shall I start collecting those images now? | 21:29 |
| <Joan> | elphel, are you at the office? | 21:30 |
| <elphel> | Most likely DCDC was the first to fail (trying to handkle low voltage) and ieee802.3af chip was bypassed to use 12V (it should disconnect) | 21:30 |
| <elphel> | Joan, no - I'll be there later | 21:30 |
| <Joan> | ok | 21:31 |
| <elphel> | And burning DCDC could fry other parts | 21:31 |
| <elphel> | It is rather hot even during normal operation | 21:31 |
| <lindes> | i'll be away from irc for a bit... need to move where there's better light | 21:37 |
| <lindes> | back... will work on uploading presently. we've got a quick meeting | 21:59 |
| <lindes> | actually, working on images in the background | 22:06 |
| <lindes> | urgh, but I shot a bunch of them in raw only. :-( don't have software to convert where I am. A few JPEGs coming soon, though. | 22:07 |
| <lindes> | (and I can take much higher quality images later, too) | 22:08 |
| <_polto_> | lindes, when is the launch ? | 22:09 |
| <elphel> | Most likely it will not be easy to fix and we do not have spares to get it to launch. Mistake was to let it run on batteries unattended the DCDC seems to have no overcurrent/thermal protection, and PoE chip is disabled. I do not have quick solution, sorry. | 22:10 |
| <lindes> | http://www.flickr.com/gp/96686711@N00/5XS1co | 22:11 |
| <lindes> | IMG_0768 is I think maybe the best | 22:11 |
| <lindes> | launch is Saturday morning, sometime after 9am | 22:12 |
| <lindes> | I can get you more images in a bit | 22:13 |
| <lindes> | "unattended" was in my hands at the time of the problem, FYI. ;-) but yeah, I probably should have been paying closer attention to it | 22:14 |
| <lindes> | (of course, in flight, we'll be nowhere near it.) | 22:14 |
| <lindes> | based on the above images, are there particular angles you'd like me to try to get? | 22:18 |
| <lindes> | you still around? | 22:22 |
| <_polto_> | we are moving to the office. | 22:28 |
| <lindes> | ok | 22:29 |
| <lindes> | someone else has the software I needed to decode raw files, so more images coming soon | 22:45 |
| <andrey> | I am in the office, yes. We have couple 9-36 (4:1) converters, but they are lower power (rated 5W vs 8W) so I do not think it will work with 10369, compass, GPS. And I'm not sure the PCB itself survived. | 23:00 |
| *ChanServ gives channel operator status to _polto_ | 23:00 | |
| <_polto_> | hi again | 23:01 |
| <lindes> | howdy | 23:01 |
| <lindes> | are there continuity checks I could do or anything? | 23:02 |
| <lindes> | to get a better feel of the state of things? | 23:02 |
| <lindes> | also, more images are uploading as I speak | 23:03 |
| <andrey> | I also have concerns about the 10369 that was very close to the fire. And there is a crystal that had wrong package on PCB so it was hanging over the board - really close to that damn diode bridge | 23:03 |
| <lindes> | i have photos of the 10369 on their way | 23:04 |
| <lindes> | side-question: I'm uploading all these images as "private", on the premise that you might not want folks to see them... but then, you operate in a very open philosophy as far as I can tell, so... let me know if you'd like me to make these images public or not. I'm happy to do either. | 23:05 |
| <andrey> | What I would try - just clean up the mess and try to apply 48V bypassing the bridge and controlling current and see if it will power up with reasonable current 10353 alone - 3W total, so ~0.25A@12V | 23:06 |
| <lindes> | you mean 62.5mA at 48V, or 250 at 12V? | 23:07 |
| <andrey> | If that will not work - you'll have to remove DC-DC | 23:07 |
| <andrey> | about that, yes (10353 alone) | 23:08 |
| <andrey> | current depends on the camera state, before FPGA kicks up and sensor is running it is about half of that | 23:08 |
| <lindes> | where would I supply that power? through the POE still, or somewhere else on the board? | 23:09 |
| <andrey> | 3W I measured at 38V, maybe at 12V DC-DC is less efficient (plus more losses on bridges) | 23:09 |
| <lindes> | ok, well we have a variable power supply here, so I can supply whatever for testing purposes. | 23:10 |
| <andrey> | directly to the DC-DC (through some current limiter) - there is no fuse now ( bridge played it's role) | 23:10 |
| <andrey> | There is a PDF with circuit diagram and layout | 23:11 |
| <lindes> | which is the DC-DC? the big silver box opposite the bridge? | 23:11 |
| <lindes> | ahh, yes, I have that... let me get it open | 23:11 |
| <lindes> | 6 more photos up | 23:12 |
| <lindes> | more coming | 23:12 |
| <andrey> | 2 pins on the 4-pin molex header (closer to back panel) are DC-DC input, 2 others (closer to sensor - output). So apply controlled 12V to the 2 that are closer to the back panel (with network and USB), polarity - "+" closer to the edge of the PCB (on both input and output). So you can apply 12V, check current and 3.3 on the other 2 pins | 23:16 |
| <andrey> | but first you need to carefully remove burn parts and clean the board | 23:18 |
| <lindes> | ok, so... hmm, I want to make sure I'm doing things correctly, and I'm concerned that my hardware hacking skill level combined with IRC as communications may be difficult to manage for getting this right... | 23:21 |
| <andrey> | No secret of burnt boards - you may as well post them on the wiki. That nice photos can help others to avoid similar damage - when using battery power some external cut-off switch is needed, DC-DC itself seems tot to handle it. | 23:21 |
| <lindes> | cool, I'll probably blog about it later, then, and make at least some of the photos available | 23:22 |
| <andrey> | Can you get somebody's personal help? Somebody who can handle hardware | 23:22 |
| <lindes> | one sec... I'm talking with someone about it. | 23:23 |
| <andrey> | first step - to clean up the mess - you need to know what you are doing | 23:23 |
| <lindes> | ok, so I have been given some alcohol swabs and some pointy tweazers (sorry, I don't know proper terminology for that), so I can swab the board with that... | 23:26 |
| <andrey> | most components around are not essential - DC4 is bypassed (it is PoE chip), capacitor, TVS | 23:26 |
| <lindes> | sorry, alcohol (isopropyl) wipes | 23:26 |
| <andrey> | No, before that you need a soldering iron in the experienced hands | 23:26 |
| <lindes> | ok. to remove the bridge? | 23:27 |
| <andrey> | completely remove (desolder) bridge, wipe, probe/scratch to remove all carbon, wipe again ... | 23:27 |
| <lindes> | 6 more photos up | 23:28 |
| <andrey> | Can you find a hardware guy around? | 23:28 |
| <lindes> | well, there are some, it's just we're in crunch time, so it's difficult to get much time from one. I can try, though. | 23:29 |
| <lindes> | oh, I think I know someone that could help | 23:29 |
| <andrey> | You may help him/her with a printed portion of the PDFs (part of the circuit diagram around the DC-DC) and maybe appropriate portions of PCB layers | 23:33 |
| <andrey> | PCB layout is later in the PDF where circuit diagram is on the first page | 23:34 |
| <lindes> | 3 more photos up. I've got someone to help me | 23:34 |
| <lindes> | looking for where I put the pdf | 23:34 |
| <andrey> | On a wiki, 10353 page - there it is | 23:35 |
| <andrey> | and be very careful with the 10369 while cleaning up the carbon. | 23:36 |
| <_polto_> | http://wiki.elphel.com/images/2/27/10353e.pdf | 23:36 |
| <lindes> | ok, he's using low melting point solder and flux to get things off. he seems much more experienced than I. :-) | 23:36 |
| <lindes> | thanks, _polto_ | 23:36 |
| <lindes> | are we going to be connecting the 10369 right away? | 23:38 |
| <andrey> | So none of the components around are essential, the most needed is 4-pin power connector fro the 10369, and only the 2 pins that are in better shape (3.3/gnd) | 23:38 |
| <lindes> | i thought you were saying just the 10353 | 23:38 |
| <andrey> | definitely not. | 23:38 |
| <andrey> | No 10369, even no sensor - only 10353 first | 23:39 |
| <lindes> | ok | 23:39 |
| <lindes> | he thinks we may have lost the top layer of the PCB | 23:39 |
| <lindes> | some traces are lifted, and we may have some bridging | 23:39 |
| <andrey> | Not many things in that corner - PDF should help you. | 23:39 |
| <andrey> | You do not need traces going to the 2 connector pins that are 12 V... bridge is only required by the ieee802.3af (to work in any polarity in the line), TVS - surge protection... | 23:41 |
| <lindes> | ok, well the trace that goes through the one pin to items on both sides is gone | 23:47 |
| <lindes> | on what you're calling layer 4 on the pdf | 23:47 |
| <lindes> | he's going to just jumper that | 23:48 |
| <lindes> | connecting to Z1, if I'm reading this right | 23:50 |
| <lindes> | ok, that's done | 23:52 |
| <lindes> | ohhh... our power supply only goes up to about 16V | 23:53 |
| <andrey> | Why do you need 16 ? 12 is nominal | 23:53 |
| <lindes> | I thought you had talked about doing 48 | 23:54 |
| <lindes> | we can do 12 | 23:54 |
| <lindes> | we need to know where it should connect, though | 23:54 |
| <andrey> | just Just look carefully on the circuit diagram and layout. You need to provide 12V to the DC-DC -it has several pins in parallel. The source - pins on RJ-45. DC4 is disabled (solder jumper) DC4.4 is connected to DC4.5 | 23:56 |
| <andrey> | basically you need to restore the traces, but use straight connections instead of the bridge (correct polarity, of course) | 23:57 |
Generated by modified irclog2html.py 2.4 by Marius Gedminas - find it at mg.pov.lt! This version is modified by Andrey Filippov for Elphel