| 12:34 | quitte | I just received my hummingboard pro quad. it came with a heatsink attached. I didn't expect that. great. |
| 12:35 | quitte | however it seems to me the thermal transfer sheet is too wide and prevents proper contact of the heatsink to the die. |
| 12:35 | quitte | I'll mention that again on a workday |
| 12:36 | rabeeh | quitte ? |
| 12:36 | rabeeh | too wide? |
| 12:37 | quitte | rabeeh: yes. the q4 has that pcb like part and then the flip chip die |
| 12:37 | quitte | the transfer sheet covers the whole package |
| 12:37 | quitte | if you take of the heatsink and push it back you'll notice a tiny gap between sink and die |
| 12:37 | quitte | unless you apply a lot of pressure |
| 12:38 | rabeeh | don't apply lots of pressure; you will break it |
| 12:38 | rabeeh | most of the heat of the systems goes through the die to the heatsink |
| 12:38 | quitte | i know. i should have used grammar to indicate it ass an option |
| 12:40 | quitte | rabeeh: that's why I think I'll cut off strip?s of the sheet so that the sink makes better contact to the die. I thought I'll mention it. in case it is an oversight. if testts indicated that the melting of the sheet is enough to make proper contact all the better |
| 12:41 | quitte | pardon my typos |
| 12:41 | rabeeh | quitte: worst case scenario is 7.5 Watts |
| 12:41 | rabeeh | you can't melt anything with that; just make a scrambled egg :) |
| 12:42 | rabeeh | that will take some time also to cook :) |
| 12:42 | quitte | rabeeh: did you understand what I meant by pushing back the heatsink such that you can see the tiny gap between die and heatsink? |
| 12:42 | rabeeh | quitte: you should have 0.5mm 7W thermal pad between the die and the red heatsink |
| 12:43 | rabeeh | quitte: i think you are talking about CNC drilling a 10mmx10mm 0.5mm depth for the die to go INSIDE the heatsink |
| 12:43 | rabeeh | right? |
| 12:43 | quitte | no that already kind of exists. the heatsink has a lowered profile for the die |
| 12:44 | quitte | basically want to change the pad to go on the die only. so that the pcb doesn't push the die away from the heatsink |
| 12:45 | rabeeh | ok |
| 12:45 | rabeeh | so making the thermal pad smaller? |
| 12:45 | quitte | yes |
| 12:45 | rabeeh | is it the case? |
| 12:46 | quitte | there is a visible gap between the die and the heatsink if I align the heatsink so that the edge of the heatsink goes over the middle of the die |
| 12:46 | rabeeh | hmm... |
| 12:47 | quitte | the thermal pad is quite compressed at the outer edges of the mx6 package,too. where the heatsink does not make room to accomodate for the i.MX6 |
| 12:47 | rabeeh | i think the gaps that you are talking are not between the die and the heatsink |
| 12:47 | rabeeh | it's around the die and the heatsink which is fine |
| 12:48 | rabeeh | quitte: can u take pictures? |
| 12:48 | rabeeh | the idea is that the heatsink should never touch the die and the processor green package ! |
| 12:48 | quitte | sure but this might take some time. have to charge the camera somewhat |
| 12:49 | rabeeh | for reliability issues i worry that it might break the die |
| 12:49 | quitte | a picture is definately needed :) |
| 12:49 | rabeeh | ok. |
| 13:09 | quitte | rabeeh: http://imgur.com/L5r1Ib9 |
| 13:27 | quitte | http://imgur.com/RLeAwNe some clarification. even if I align the heatsink with the pads indentations as good as possible and apply slight pressure there is a gap above the die and no gap above the two stripes on the pcb |
| 13:28 | rabeeh | quitte: so it's pretty much marginal there |
| 13:28 | Artox | quitte: your heatsink is not attached by screws? |
| 13:29 | quitte | Artox: not anymore ;) |
| 13:30 | Artox | ah |
| 13:32 | quitte | rabeeh: I definately want to do something about it. I can either take of about 3/10mm of the bottom of the heatsink. or simply cut off stripes of the pad. |
| 13:33 | rabeeh | cut them then |
| 13:33 | rabeeh | i mean cut the strips |
| 13:34 | quitte | okay. i thought so,too. but it sounded a bit like the mechanical relief of those stripes was somewhat important. |
| 13:34 | quitte | thanks for taking the time |
| 13:35 | rabeeh | quitte: the issue i worry about is metal directly touching the pcb (either die, processor package) |
| 13:35 | rabeeh | they might work good for you; but in mass production it's a bad idea |
| 13:37 | rabeeh | quitte: do you have a caliber to measure with? |
| 13:37 | quitte | rabeeh: the pad shows the outline of the die quite clearly on both sides. so i guess as long as the heatsink never comes off it works out okay |
| 13:37 | rabeeh | can u measure the height of the thermal pad? |
| 13:37 | quitte | rabeeh: yes |
| 13:38 | quitte | the pad will compress too much |
| 13:38 | rabeeh | without compressing |
| 13:38 | rabeeh | 0.5mm? |
| 13:39 | rabeeh | i know a bit tough |
| 13:39 | rabeeh | i just want to make sure if it's 0.5mm or 1mm |
| 13:39 | quitte | precisely .5mm |
| 13:40 | rabeeh | ok |
| 13:40 | rabeeh | so it's the right thermal pad |
| 13:40 | quitte | as is the depth of the die indentation in the heatsink |
| 14:06 | michiwend | Hi, still having frame drops ~3 seconds playing videos with xbmc on archlinux and openelec |
| 14:07 | michiwend | I own a cybox-i 4pro |
| 14:07 | michiwend | It |
| 14:09 | michiwend | What is wrong with decoding videos? I thought cubox was made for exactly that use case |
| 14:12 | jnettlet | michiwend does it continue to drop frames or is it just a hiccup in the beginning? |
| 14:18 | michiwend | jnettlet: it continues |
| 14:21 | jnettlet | michiwend, what type of media and bitrate, and from what medium are you playing. If it is network wired or wireless? |
| 14:21 | michiwend | USB hard drive |
| 14:21 | michiwend | H264 encoded full HD videos |
| 14:21 | jnettlet | what bitrate? |
| 14:22 | jnettlet | although I have played 120Mbps h264 hbr with no problems at all |
| 14:22 | michiwend | I don't know... Have to look up |
| 14:23 | jnettlet | press o while playing the video |
| 14:23 | michiwend | But its with all my movies an TV shows on different bitrates |
| 14:23 | michiwend | I know |
| 14:24 | jnettlet | michiwend, I would recommend you try geexbox |
| 14:25 | michiwend | jnettlet: OK, will try... Had geexbox installed some time ago, same behaviour |
| 14:26 | michiwend | But than I miss a full featured Linux os with package manager... But, that's a minor drawback |
| 14:26 | michiwend | I'll give it a try later on |
| 14:27 | jnettlet | michiwend, the images based on the 3.14.14 kernel will be out soon enough and the performance improvement is noticable |
| 14:28 | michiwend | jnettlet: images for geexbox? |
| 14:28 | jnettlet | geexbox will probably be the first to release them I think |
| 14:28 | michiwend | Ah, sure... |
| 14:29 | jnettlet | it does have a packagemanager by the way opkg |
| 14:29 | michiwend | Hm OK... Its from openwrt |
| 17:08 | quitte_ | can I use the rectangular pad next to the jtag testpads for the GND ? |
| 17:29 | FoXMaN | i don't see any vpu messages in dmesg for imx_3.0.35_4.1.0 kernel, how can i check that vpu and firmware are actually loaded properly? |
| 17:50 | Artox | FoXMaN: I think firmware is only requested at runtime |
| 17:50 | Artox | e.g. when you try to use the vpu |
| 17:52 | FoXMaN | Artox: thanks. I'm trying to run xbmc from geexbox on the imx_3.0.35_4.1.0 kernel but it complains that it can't find opengl accelerated device |
| 17:52 | FoXMaN | i haven't seen any vpu related messages in dmesg so i assumed that vpu module wasn't properly loaded |
| 17:53 | FoXMaN | on the other hand it's impossible to compule mcx_vpu as module for that particular kernel |
| 17:54 | FoXMaN | also i couldn't find good information on what are the supporting libraries for video acceleration for the hummingboard hardware gpu |
| 17:56 | FoXMaN | an why am i playing with older kernel when geexbox is using newer one? because the most kernels in distros for cubox/hb don't support audio codec sgtl5k |
| 18:00 | Artox | FoXMaN: opengl sounds very fishy |
| 18:01 | Artox | you need a patched 3.0.35 when you want to use the same vivante opengl libraries as used with 3.10 and 3.14 kernels |
| 18:01 | Artox | about gpu, there is this libfslvpuwrap library |
| 18:01 | Artox | but developers recommend against using that directly |
| 18:03 | Artox | https://github.com/Freescale/meta-fsl-arm-extra/tree/master/recipes-kernel/linux/linux-boundary-3.0.35 |
| 18:03 | Artox | the 3 patches there |
| 18:08 | FoXMaN | Artox: thanks a lot, i'll have a look |
| 18:10 | Artox | FoXMaN: if you want a binary of 3.0.35 with those patches for quick testing I can provide one |
| 18:16 | FoXMaN | i see that 2/3 patches are already in https://github.com/SolidRun/linux-imx6/commits/imx_3.0.35_4.1.0 |
| 18:17 | FoXMaN | Artox: what is the base tree for your patched version? |
| 18:17 | FoXMaN | in general i'd prefer to stick to 3.14 but sgtl5k is not working there |
| 18:18 | Artox | exactly the one you linked FoXMaN |
| 18:18 | Artox | and if 2 are in |
| 18:18 | Artox | that still makes them 3 |
| 18:18 | FoXMaN | the patch is called linux-boundary (3.0.35): Add Vivante GPU 4.6.9p13 support |
| 18:18 | Artox | and its the important one |
| 18:18 | FoXMaN | what does the 4.6.9p13 relate to? |
| 18:18 | Artox | I have no idea |
| 18:19 | Artox | I think its even wrong |
| 18:19 | Artox | something about ther version of galcore kernel module |
| 18:19 | FoXMaN | ah you're right i remember seeing 4.6.9 in sources of galcore |
| 18:20 | FoXMaN | the base seems to have ENGR00264288-1 [GPU]Integrate 4.6.9p12 release kernel part code |
| 18:21 | FoXMaN | ok so the question is where does the original vivante/galcore driver come from |
| 18:21 | Artox | from Freescale |
| 18:21 | Artox | more or less |
| 18:21 | FoXMaN | with zilions of patches and repos it's sometimes hard to get idea where is the ported code being made |
| 18:21 | Artox | but I have no idea |
| 18:22 | Artox | it is no fun searching for either kernel- or userspace-parts of vivante driver |
| 18:48 | joelbaby | FoxMaN you are probably using the wrong version of geexbox. |
| 18:48 | joelbaby | you need to use the one herE: http://download.geexbox.org/snapshots/geexbox-xbmc-imx6-cuboxi/latest/binaries.cuboxi/ |
| 19:07 | FoXMaN | joelbaby: i did get some snapshot few days ago |
| 19:07 | FoXMaN | and everything is perfect for cubox-i |
| 19:07 | joelbaby | FoXMaN the stuff with 3.0.35 is very old |
| 19:07 | joelbaby | and probably before they put the hw acceleration in. |
| 19:08 | FoXMaN | but for hummingboard i really miss analog audio output support |
| 19:08 | joelbaby | but Rudi in forum said that hummingboard support is coming soon. |
| 19:08 | joelbaby | http://www.solid-run.com/community/topic1548.html#p11527 |
| 19:09 | FoXMaN | i'm impatient sometimes and i wanted to see how are the things organized from kernel point of view |
| 19:09 | FoXMaN | but i see a bit of mess and zilions of versions / patches and tons of unreliable information |
| 19:09 | joelbaby | so just compile a new kernel |
| 19:10 | FoXMaN | joelbaby: i tried 3.14 and it was good but no sgtl5k codec working |
| 19:10 | joelbaby | i put 3.14 into my personal geexbox shortly after it was released. |
| 19:10 | joelbaby | does it work on cubox-i? |
| 19:11 | FoXMaN | joelbaby: it worked on cubox and hummingboard but i needed to add some kernel opts for galcore to work |
| 19:12 | FoXMaN | joelbaby: i used this one https://github.com/linux4kix/linux-linaro-stable-mx6/tree/linux-linaro-lsk-v3.14-mx6 |
| 19:12 | FoXMaN | and then rabeeh_ told me to use 3.0.35 for proper sgtl5k support |
| 19:12 | FoXMaN | so i tried and found opengl problems but audio codec works perfect |
| 19:13 | FoXMaN | 3.14 is much faster |
| 19:13 | joelbaby | ok, solution: play the video on your cubox-i, and the audio on your hummingboard. you just need to start the movie on the two devices at the exact same time :) |
| 19:13 | FoXMaN | now i'll test 3.0.35 with patches recommended by Artox |
| 19:14 | FoXMaN | joelbaby: the only challenge is to keep them in sync |
| 19:14 | FoXMaN | i've ordered two devices: one for fun and other for xmbc fun |
| 19:14 | FoXMaN | i really needed ir-transmitter so cubox-i seemed perfect |
| 19:15 | FoXMaN | ;] but there's no support for tx yet |
| 19:15 | Chris-ECServices | what are your ideas for tx? |
| 19:16 | FoXMaN | Chris-ECServices: to control my tv and amp |
| 19:16 | FoXMaN | on/off basically |
| 19:17 | Chris-ECServices | but it's wrong direction? |
| 19:17 | Chris-ECServices | from view of the cubox |
| 19:17 | FoXMaN | Chris-ECServices: it depends on receiver |
| 19:18 | FoXMaN | led diode on tv and amp are very sensitive and can spot signal reflections |
| 19:18 | FoXMaN | somehow |
| 19:18 | Chris-ECServices | ;-) |
| 19:18 | Chris-ECServices | not as bad sensitive as cubox ir rx :) |
| 19:19 | FoXMaN | Chris-ECServices: that's the different story |
| 19:19 | FoXMaN | but true |
| 19:19 | FoXMaN | i don't understand one thing |
| 19:20 | FoXMaN | freescale is an enginieering company which makes their own ic |
| 19:20 | FoXMaN | socs and etc. |
| 19:20 | FoXMaN | designing SoC requires a lot of self-discipline and good organisation |
| 19:21 | FoXMaN | but their software releases are totaly disorganised for me |
| 19:22 | FoXMaN | like: what is the rule behind the BSP versioning |
| 19:22 | FoXMaN | ? |
| 19:23 | FoXMaN | or file naming |
| 19:23 | Chris-ECServices | just two different departments with to different kind of people :) |
| 19:23 | Chris-ECServices | hard guys and soft guys :D |
| 19:26 | joelbaby | Chris .. I'd like to 3D print a new box to fix the IR receiver |
| 19:27 | joelbaby | my friend put a massive USB IR receiver from his rpi onto the cubox, so he's happy with that |
| 19:27 | FoXMaN | joelbaby: i saw the model files for cubox case on the cubox-i forum |
| 19:27 | joelbaby | yeah, i asked for them :) |
| 19:27 | FoXMaN | ah it was you |
| 19:27 | Chris-ECServices | but what should a new case help? without case the ir is also bad :/ |
| 19:27 | joelbaby | also the hole they made for the spdif optical port is totally retarded |
| 19:27 | FoXMaN | i couldn't belive that solidrun just shared it without any problem |
| 19:28 | joelbaby | chris ... maybe a new red plastic bit on the front? |
| 19:28 | joelbaby | the light filter that lets the ir in, but not much light pollution |
| 19:28 | Chris-ECServices | but without case i dont have the plastic and it is as bad as before |
| 19:28 | Chris-ECServices | ah okay |
| 19:29 | Chris-ECServices | sorry :) |
| 19:29 | joelbaby | yes, but Ir usually doesn't work without that red thing in front. |
| 19:29 | Chris-ECServices | there should be some tape you can apply with this filter properties :) this would be a great addon :D |
| 19:30 | joelbaby | not sure that exists. |
| 19:32 | joelbaby | chris .. using a keyboard is easier than a remote. the ir receiver is annoying. i got one of those logitech k400r |
| 19:33 | joelbaby | i'm never quite sitting at the correct angle for the cubox to pick up the signals from the remote. |
| 19:33 | Chris-ECServices | dito i'm using the cheap microsoft wireless one :) |
| 19:33 | Chris-ECServices | sad but true... in cec i miss the options button :/ |
| 23:52 | matteofb | I'm running debian jessie on my cubox i4pro, but need iptables. I've been unsuccessful in updating the kernel (currently 3.0.35) to one compiled with iptables. Anyone willing to help me? |