HDS Systems        
Results 1 to 24 of 24

Thread: Review: Ultra-thin Raspberry Pi aluminum case assembly, mod & thermal test

  1. #1

    Mpr Review: Ultra-thin Raspberry Pi aluminum case assembly, mod & thermal test

    Note: The room temperature is 23.0C+, we are in summer after all!

    The
    raspi (here: Raspberry Pi 3 Model B 2015 ) is probably the #1 best-selling single-board-computer (SBC) ever. While more modern, higher performance, easier-to-use competitors exist (ASUS Tinker Board, LattePanda, a. o.), they also cost more and draw more energy from the USB power source, which could be a critical factor for the application.

    Thanks to raspi's immense popularity several commercial aluminum cases are/were available on the market for the various raspi versions, apart from the non-commercial aluminum cases built by DIY engineers. Today i am presenting the slimmest most compact commercial alu case, its assembly, along with a non-invasive mod plus a set of thermal tests. Feel free to use my test results as reference and compare them with your own test results of this or other raspi cases.

    Most raspi (or SBC) owners have an adequate inexpensive 4US$ case solution yet, are happy with what they've got or diy made, or don't want/need a nicer case. I am fully aware that the problem with expensive cases such as this (14-28US$ shipped, depending on source!) is that, as soon as a newer raspi version is released, one cannot recycle the case for the newer electronics. So the main motivation why i am posting today is: i planned a mod, did the mod, am extremely pleased with the result, and wanted to share it, because knowledge is power and sharing is fun!

    pro's
    • no thermal paste, thermal grease, or thermal glue needed. 1.0mm thermal pad suffices.
    • non-invasive mod re case or pcb not necessary
    • extremely effective heat-sinking, all 3 chips are cooled at the same time
    • few simple tools needed, very easy installation
    • the most compact case on the market, with also elegant design
    • leaves full access to all internal ports
    • absolutely amazing build quality, fit and finish
    • fully CNC-machined parts, thick walls, no flex, absolutely rigid
    • the two LED's are clearly visible through the back
    • easily accessible micro SD card

    con's
    • expensive purchase, depending on source
    • black anodization looks ugly, i'd prefer white color
    • dust could enter through the top interface ports
    • generic, no-name, does not come with bottom label or any branding
    • i had to trim the copper shim as part of the mod, was work. i don't like work
    • WiFi signal remains strong enough? (signal strength tests pending)

    Summary:
    Never mind the mod, this case brings joy and satisfaction to the raspi hobby! The construction, design, build quality, accurate machining, heavy wall thickness, effective cooling, slim form factor makes this case a wonderful choice for people wanting a commercial case made out of metal. The price may be steep compared to other commercial cases but imho the product is worth every cent. And you're not overpaying a bit, since the Chinese end consumers don't get it cheaper through Taobao. Fully approved and highly recommended by the squirrel!


    Part I. The assembly with the mod.


    Each of the 40+ pics is to highlight 1 certain aspect which i will be commenting upon. The pics are in chrono/logical order, as i went on with the assembly project. Now without further ado, let's get to the details, may the story time begin!

    Originally the inside of the case heatsinks chip1 and chip3 only and does not contact chip2. After having taken careful measurements of all and everything, my calculations told me that i would need a copper shim of exactly 1.50mm thickness plus a 1.0mm thermal pad to bridge the air gap between chip2 and case (inner wall). With this knowledge i ordered a pack of 20x20x1.50mm heatsink copper shims. Fast shipping, fast delivery, the shims arrived as expected, with smooth flat surfaces. I was curious about how the surface would polish up:


    How does one trim 20x20mm to 10x10mm? In the end i wasted 3 shims with my attempts. I tried hard to cut the shim with various Stanley knives. On very thin shims such cutting would work, but on 1.5mm thick shims it failed! I was able to cut through the thickness with my Victorinox and Leatherman pliers wire cutters, however this method bent the shims irregularly. The slightest bend and one had to toss the part:


    How about a vise and a metal saw? The vise would have crushed one end of the part, leaving the other end stress-free. I was too lazy to go after this failsafe method and decided to try some sawing action with the file of my Spirit X multitool. Took ages lol but we were getting somewhere:


    The bottom case (bottom half of the enclosure) is to cool/heatsink chip3. Sophisticated machining, flawless anodization, thick material thickness, loving it:


    For all intents and purposes we'd still need a thermal pad, if we weren't using thermal paste/grease. The manufacturer provided 1.0mm thick thermal pads, perfectly fitting. The pads are not really sticky but as far as i can tell they do have some sort of adhesive on their surface:


    I did a quick check to see if the case was touching the pad. The 'cooler3' did leave a mark on the pad, cool:


    The top case (top half of the enclosure) does not have a 'cooler2' for chip2. One could regard my thick shim as heatsink, cooler, or bridge, feel free to call it as you wish:


    Since there are electronic components located in the immediate vicinity of chip2, the shim should not be larger than chip2. Size comparison:


    So there's chip2. But how would you stack the shim and the pad, in which order? Chip+shim+pad or chip+pad+shim? Anywho, do you have thermal paste at hand? Keep in mind that the shim is a free-floating part in the stack, unless you fix it to either the case or the chip. Fix how? Best would be thermal glue. I decided to simply use tacky glue and attach the shim directly to the chip:


    Why? Because perfect positioning of the shim was crucial. By moving and pressing the shim gently onto the chip i minimized the glue layer and made sure that no air bubble was trapped in between. Some glue spilled over, no big deal:


    No big deal, because the spilled glue contracts after the volatiles have evaporated. It was also time to cut some thermal pad in the size of 10x10mm:


    A test seating. Nice. The rear part is made out of some transparent black plastic material which reminds me of acrylic glass/Plexiglass. Both faces are covered/protected with a pseudo-adhesive paper:


    The part was cut professionally, probably with laser-cutting technology. It fits accurately into the CNC-drill machined groove. Due to the geometries minimal wiggle room in some direction would be natural:


    The finish of the plastic faces is uber smooth. Mirror finish:


    I came up with the idea to create a snug fit by adding a layer of adhesive film on top of the mirror finish. The clear adhesive film could also protect against scratches. Usually this kind of film is used in gemani to protect book covers against dust, humidity, dirt, wear and tear. It does not stick aggressively to surfaces and does not leave glue residues. I also use it as screen protector on some displays:


    It is not necessary to cover both faces of the plastic part with the adhesive film, just the outside face. But the point is to wrap the film around the edges so that this would create a snug fit in the precisely machined groove:


    Sticking the film to the upper edge ensures that now all 3 directions, x-, y-, and z-directions, are padded by the film, which would leave no possible wiggle room after the complete assembly:


    Yes, i can feel it, the fitment is now snug. The film will prevent any kind of rattling sound, even if there will still wiggle room left. But there is isn't, hehe:


    Looks legit so far, what do you think? A pic before the addition of thermal pads:


    A pic after the addition of thermal pads. All thermal pads used in the assembly have the same thickness, 1.0mm:


    What does the inside of the upper case half look like? Looks quite elaborate, if you ask me. As you can see, there is only a 'cooler1' for chip1. Other raspi commercial aluminum cases do have a 'cooler2' for chip2, which is why i got inspired to add a cooler in this case for chip2. Anyhow, here are 2 shots in succession:



    Finally i put the upper case half on top of the lower case half w/ its seated PCB, flipped the entire case, screwed in the screws with the provided Allen key together. The bottom side of the case has recesses for the four rubber feet and an optional label, what a nice touch:


    Wow, this was interesting. After tightening the four screws, i noticed that the two halves didn't match up perfectly at the seam. No perfect fit, that is alignment, by maybe half a millimeter. So i loosened all four screws, corrected the alignment by that half a millimeter, and then retightened the screws. Voila, this did the trick. Now i got the perfect fit, really beautiful, you cannot feel the seam anymore:


    The plastic part had a mirror finish. Unfortunately the adhesive film ruins that look. Also note the thickness of the plastic and the aluminum, so sweet:


    The following is optional, placing a label sticker in the recess. But i really wanted to have a label in there! I thought that the logo from the raspi retail packaging was a suitable pick. So i cut it out, roughly with a Stanley knife:


    As you could tell from the color, the material is made out of some cardboard paper, good quality stuff. Since i didn't have thin double-sided adhesive tape in the household, i thought that i could stick the paper to white duct tape with my paper glue. Duct tape sticks reasonably well on anodized aluminum and could be removed without leaving residues. And while we're at it, why not protect the paper label side with clear packaging tape? I do have lots of tape in the household:


    So i stuck the white duct tape temporarily on non-sticky paper, sourced from adhesive film remains, put some glue on it, then the logo paper, then the clear packaging type, thereby creating a diy laminate consisting of 5 layers. I felt like a pro! Then i placed this laminate under a table foot with an eraser for even weight distribution and left it there for drying overnight:


    The next morning i cut the label out as best as i could. Poor job. I failed at the dimensions and also the corners. If i had a second try, i'd do a much better job. Can't be satisfied with it but since i don't possess another retail box, i must live it, be a man and *uck it up:


    Argh, the glue did not stick at all to the duct tape surface! What a fail. To not waste time any further, i decided to glue the label on the anodized aluminum. Not a great decision because this would mean a permanent solution basically:


    Here again, you can see that the really smooth finish of the paper looks ruined after adding a layer of clear packaging tape. The rubber feet fit nicely into the circular recesses. They are of the clear type and have tiny bubbles inside. Cheaply manufactured i would say:


    No need to trash the retail packaging bag. I closed the cutout hole and harhar taped the entire thing with clear packaging tape. Why? Because i am going to use it as carry bag, storage bag, and even operation bag. May no dust ever come inside my raspi in headless operation mode:


    Oh! You still don't know what the product looks like from above? Finally the first shot showing what the assembled product looks like. I really dig it:


    The slits in the aluminum are perfectly aligned with the internal ports. Really all ports are accessible, which is extraordinary for an aluminum case, just compare with other raspi aluminum cases on the market:


    Since i don't make use of those three ports, i sealed them with electrician's tape. Protects against entry of my dandruff:


    Only after the assembly of the case can one insert the micro SD card. It is easy to insert the card and pull it out through its convenient accessibility:


    Yet another shot of the rear, now with the card inserted. I believe that the two nubs on the plastic have a function, you name it:


    Aha yes sir, the black plastic is transparent after all! The two LED's are clearly visible. The red LED means that the raspi is connected to power, the green LED signals (write? read?) activity on the micro SD card. One must not disconnect the raspi from power when the green LED is on because such human interaction could corrupt the SD card:


    Part II. The thermal tests.

    I conducted a total of 5 tests. In all tests i waited until the raspi had reached a stationary state (steady state) in its environmental setting at no load (CPU&GPU = 0%). Basically you power up the raspi and do nothing further. The raspi will operate under no load, idling, but still consume energy to keep the operating system (Linux) alive, thus its internal temperature will be higher than normal room temperature. Then, at time t=0, i put the raspi under full load (CPU&GPU = 100%) for 15-27min and recorded the internal temperature, as reported by the raspi OS, every minute. When the hot temperature seemed to have reached its own steady state, i cut the load back to zero (CPU&GPU = 0%) and continued to record the internal temperature at various intervals, until the temperature was back at its load-free steady state level.

    The only difference between the 5 tests was their external setup. As this raspi case proved to be an extremely effective heatsink, conducting the heat of the 3 chips effectively and fast to the outside of the entire aluminum body, any variation of the external setup could be made visible in the temperature measurements and affected the result.

    Test 1 : raspi suspended vertically in the air and pointing with its rear to an upwards blowing fan:


    Test 2 : same as test 1, but with fan removed:


    Test 3 : raspi lying flat on a horizontal wood desktop. That's the standard setup, what the case is designed for:


    Test 4 : raspi sitting seitkant vertical on a horizontal wood desktop:


    Test 5 : raspi sitting hochkant vertical on a horizontal styrofoam piece:


    A picture is worth a thousand words, so i won't bore you with the detailed description or explanation or interpretation of the temperature graphs. To me, they all make sense in comparison, no real surprises here:

    Note that raspi's temperature measurement resolution is around +0.55C, even though its one decimal could suggest a resolution of +0.1C, if you didn't know any better. For your reference, here is a table of explicit numerical data extracted from the above chart:

    Test
    final C at 0% load
    time to heat up
    final C at 100% load
    time to cool off delta C
    test1 35.2 C 4 min 43.5 C 4 min 8.3 C
    test2 41.8 C 8 min 48.5 C 9 min 6.7 C
    test3 44.1 C 25 min 60.8 C 50 min 16.7 C
    test4 39.4 C 18 min 55.8 C 40 min 16.4 C
    test5 40.9 C 17 min 57.3 C 30 min 16.4 C

    Last but not least a size comparison of the assembled raspi. The 92.0x61.0mm dimensions equal very much the footprint of a credit card (86x54mm):


    The weight is 140g with USB microphone and shutdown button:


    Hope you enjoyed the story time. Let me know if you have questions regarding this assembly & test project. Imho it was a real success, even though i messed up the label sticker lol.

    And if you are a raspi owner too, feel free to do perform the 100%-load test too (for 15-27min) and share the reported temperature. I am quite curious how this raspi case's thermal performance compares to other raspi aluminum cases with passive cooling.

    tags: < box, case, casing, chassis, cover, enclosure, frame, half, holder, housing, shell >
    Last edited by kreisl; 06-18-2017 at 01:38 PM.

  2. #2

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    A quick follow-up. Based on the findings in the OP i decided to build a diy stand to support the raspi in the vertical position, "sitting seitkant vertical" as i called it in test4. In the meantime i did wifi (or wlan) signal strength tests, for example "#5", and learned that near the wifi router the signal strength is very strong, i am getting "Link Quality=59/70 Signal level=-51 dBm" about half a meter away from the router. At 3m distance Link Quality was at 41/70, still pretty good. Since i operate the raspi headless, i.e. without USB (or bluetooth) keyboard, without USB (or bluetooth) mouse, and without PC screen monitor, it makes sense to me to leave the raspi powered & unattended near the wifi router in the corner of the room. If no dog or squirrel kicks balls or balloons in the room, i wouldn't need a stand or holder for the raspi vertical position.
    I wanted a quick good solution, not actually build something out of wood, plastic, or titanium . Also no cardboard. Styrofoam scraps came into my mind and i was hoping that i could carve something, a stand, out of that material. Note that gluing styrofoam is a challenge, a pita, and i never met any RL person who glues styrofoam parts to build something from that material:


    I sat in front of the two styrofoam scraps brainstorming for two hours. Thinking hard how i could make any use of that material, how i could possibly build or carve a mini stand for the raspi. And also minimizing the footprint of the stand, otherwise it would defeat the purpose of the exercise, of the diy project result. It'd be easy to design and build a big secure stand for the vertical position raspi, if space and materials (and tools) are of no concern. No, my design criteria were making a quick good solution with minimal efforts and, if possible, without using glue.

    Well, after the two hours i landed the following design idea. A successful mod of the styrofoam scraps, as it turned out. I could have cut the material anywhere but i went for the corner. Feel free to guess why i preferred cutting a corner to cutting a side:


    The thickness of the styrofoam elevates/lifts the raspi off the ground. And cutting manually a straight(!) hole through that thickness was challenging, let alone knowing the exact position of the hole. With much care, taking measures and cutting precisely, i succeeded with the first try. I had cut out the hole at the correct position and the hole was straight, not askew. Well done kreisl:


    It is pretty clear why the hole has to be as straight as possible, isn't it? Just think about what you're seeing in the next pic. Here it is:


    A quick check of fitment before proceeding. So far so good:


    Done!


    Yep, the stand holds it securely. Great job:



    The stand itself sits stably on flat surfaces, no concerns here:


    Now it has its place in the corner, below a bookshelf or what. Since all ports are sealed with electrician's tape, no dust could enter. And it makes it easier to dust off the gadget once in a while:


    I can reconfirm that in this setup the raspi reports a temperature of 39-40C, idling. In the retail paper bag, the steady-state temperature idling was 45C which was also okay but at 100% full load it went up to 69C due to heat accumulation in the bag. In geman winter i'll be happy to operate my headless raspi in the paper bag, no problem. But for now, summer, i'll leave the raspi out (of the bag), operating in vertical position, held securely in the syrofoam stand.

    No raspi owner shall feel inspired to build a stand too, because the vertical position is not the natural setup of a raspi. But we learned from the OP that the orientation of the case does have a measurable thermal effect on the system.

    So.

    Whatever.


  3. #3
    Flashaholic* TinderBox (UK)'s Avatar
    Join Date
    Jan 2006
    Location
    England, United Kingdom
    Posts
    3,464

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Nice case, Great review

    I went though 3 or 4 cases before i found one i liked for my Pi3B, Though it`s only plastic not aluminium and i just use standard heatsinks though.

    Using the case as a heatsink is a great idea, but making sure you have a good contact between the chips and the case is essential, especially the soc, it supposedly throttles when the temp gets above 80c

    I tried using fans, but the small fans are just too noisy the high pitch whine gets on my nerves, Some owners use a large slow fan as they are much quieter but they are too large to fit to the standard size case.

    I only really use my Pi as a SPMC/Kodi player.

    John.
    "Build a man a fire, and he'll be warm for a day. Set a man on fire, and he'll be warm for the rest of his life." - Terry Pratchett.

  4. #4
    Flashaholic* maukka's Avatar
    Join Date
    Dec 2015
    Location
    Finland
    Posts
    642

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Interesting case and project. How are the temps compared to an enclosed plastic case or free air?

  5. #5
    Flashaholic* TinderBox (UK)'s Avatar
    Join Date
    Jan 2006
    Location
    England, United Kingdom
    Posts
    3,464

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    I read this test a year or so ago, That tested what makes the most difference to the soc temprature, heatsink or fan or open cover or a mixture of the three.

    https://www.element14.com/community/...-initial-tests

    John.
    "Build a man a fire, and he'll be warm for a day. Set a man on fire, and he'll be warm for the rest of his life." - Terry Pratchett.

  6. #6
    Flashaholic* maukka's Avatar
    Join Date
    Dec 2015
    Location
    Finland
    Posts
    642

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    So only a fan made a significant (more than a couple of degrees) difference in a long sustained load.

  7. #7

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Quote Originally Posted by TinderBox (UK) View Post
    I only really use my Pi as a SPMC/Kodi player.
    What's "the soc", you mean "chip1"?

    This highly effective/conductive case sinks the heat well but the heat still needs to leave from there. To the environment. A cold room or air movement from a window or fan would immensely help with that, the heat convection. Yeah my case is fully closed, sealed with electrician's tape, no dust can enter. No fan needed, especially in the vertical orientation. The passive cooling works really well, the entire case feels warm fast, like a zebra light

    I am still new to raspi, running jessie 2.4.0 with PIXEL 1.2.

    Even though i don't make often or regularly use of my raspi, my main uses of it so far are, all headless:
    1) 24\7 logging machine for MC3000 and UT61E thru DataExplorer Raspbian version
    2) downloading big files at slow rates, e.g. overnight when PC is shutdown
    3) portable/mobile Mathematica (well, you need a USB powerbank if you're on the road)
    4) streaming music from my online music subscription
    5) running Alexa 24\7, too

    I hope that in future more Win10 single-board-computers hit the market as hard and with long-lasting impact as this one hit wonder raspi did with Linux distros. If i really needed a SBC based on Windows OS these days, i'd buy the Lattepanda in the blink of an eye. Lattepanda is still new, the company, the product, and they might need several iterations (like raspi) for perfect maturity and maybe they won't survive in the long process. So many young and small tech/gadget companies have come, and then gone *RIP*, especially the ones who started off from a kickstarter or indiegogo campaign! I'll check back on Lattepanda in 3.0 years
    In the meantime it is fantastic to have an uber large super active and probably never dying user community around the raspi
    Last edited by kreisl; 06-28-2017 at 03:18 AM.

  8. #8

    Default Review: Raspberry Pi aluminum case assembly, mod &amp; thermal test

    I am confused about the need for such a case at all?
    I have several Pi and have yet to ever have a thermal issue no matter what I do with them?

    I like this Pimoroni Case the best of the ones that I have tried as there is great access and open air flow by design.

    The Pi-Top Ceed in the pics background also does the job that way and is the slickest KODI setup I have tried so far but I just can not stop trying either!

    I need to get my Tinkerbox going with KODI and see how it does overall?

    I love your work and write-up but I just don't understand the need?

  9. #9

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Quote Originally Posted by TinderBox (UK) View Post
    I read this test a year or so ago, That tested what makes the most difference to the soc temprature, heatsink or fan or open cover or a mixture of the three.

    https://www.element14.com/community/...-initial-tests
    Great blog entry, thanks John! I studied the entry and, for our comparison purposes, redid their test after discovering and fixing 2 mishaps in that blog post:
    - the stress command needs to be installed, it is not available on a freshly installed raspi
    - the shell script file needs to be made executable after its creation

    So the full sequence of Raspbian commands, apart from reboots, be:
    sudo apt-get update
    sudo apt-get upgrade
    sudo rpi-update
    (this command is NOT recommended!)
    sudo apt-get dist-upgrade

    (a reboot of the raspi is required)
    sudo apt-get update
    sudo apt-get install stress
    sudo nano temptest.sh

    (copy paste the temptest code, then exit&save)
    sudo chmod +x temptest.sh
    sudo -s
    (this command is not needed afaik!)
    ./temptest.sh > test1.csv

    This worked like a charm! As soon as the test starts, the CPU load jumps up to "100%" and stays there for exactly 550s, then drops back to "0%" where the logging continues for another 50s. I repeated the test three times and got same results. The following 5 pics are screenshots from my xiaomi phone.

    2 pics in phone horizontal orientation. the pics were taken near the end of test1:
    ,

    and 3 pics in phone vertical orientation. i started test2 at 11:43p.m., resting temperature was 40~41C:


    shortly before test2 ended, i took a screenshot to capture the temperature as shown in the system tray:


    a few seconds later test2 ended, you can see how the load dropped to 1%:


    the CSV-file was easily found on the raspi system, i transferred it to my Win-machine where i imported it to Excel and in no time got the Excel graph. For a better direct 1:1 comparison of their original graph versus my Excel graph i had to scale the pixels of my pic in the x- and y-directions for an exact match. Since i don't know how to do a superposition of two PNG-images in Photoshop, i am offering an animated GIF instead:


    In my 3 identical tests (raspi was in room corner on stand) the reported temperature didn't surpass 54.8C during the duration of the (admittedly short) test. 10min stress test was not long enough for my raspi to reach steady-state final max temperature!

    Compared with the blog entry, my raspi case setup would have corresponded to
    Test#---Fan---Cover---Heatsink---Graph Colour:
    7--------0-----x-------x---------Violet------.
    From the animated GIF you can see that the violet graph eventually reached 80C and that all other graphs ran hotter than my case during the ~550s test runtime, no matter if their case was extra-cooled by a fan or not! Even without a fan my alu case ftw beats the performances of their 8 test configurations I guess i made my point with this post .

    It just means that my passive cooling setup does a great job indeed at keeping the temperature below 56C (recall test4 from the OP). We're also learning that the stress command appears to be sufficient for thermal tests, it is a useful Linux command to make thermal engineers' lab life easier.
    Last edited by kreisl; 06-24-2017 at 07:13 AM.

  10. #10

    Default Re: Review: Raspberry Pi aluminum case assembly, mod &amp; thermal test

    Quote Originally Posted by AVService View Post
    I am confused about the need for such a case at all?
    I have several Pi and have yet to ever have a thermal issue no matter what I do with them?

    I like this Pimoroni Case the best of the ones that I have tried as there is great access and open air flow by design.
    Very nice case you got there, thanks for sharing! Hah, the "great access and open air flow by design", i do believe that the top cut-out is for the installation of an aluminum cooler (heat sink), similar to a PC CPU cooler.

    No thermal issues? That's true, the raspi is self-protected against overheating, reduces CPU clocking speed so that temperature doesn't get much higher than 80C. I just don't want to let my raspi reach that limit/limitation. Your case setup corresponds to the brown or orange graph, i.e. raspi without heatsink without fan. If you run the stress command, as shown above, you should get a graph similar to the brown/orange one; your raspi hitting the 80C-mark after 6min.

    There is no need for such a 14-28US$ case (I paid 14US$ fwiw). From vendor's stock numbers i could tell that my alu case is not a popular model, not a best-seller. Maybe shoppers don't dig the looks. Most raspi owners prefer the inexpensive plastic ABS cases which look very nice, i must admit.
    But lemme compare the situation to flashlights. You can buy plastic $ flashlight or aluminum $$$ flashlight. What shall it be?

    Also note my signature
    ~ bitterness about poor quality remains long after sweetness of low price is forgotten ~

  11. #11

    Default Re: Review: Raspberry Pi aluminum case assembly, mod &amp; thermal test

    Quote Originally Posted by kreisl View Post
    Very nice case you got there, thanks for sharing! Hah, the "great access and open air flow by design", i do believe that the top cut-out is for the installation of an aluminum cooler (heat sink), similar to a PC CPU cooler.

    No thermal issues? That's true, the raspi is self-protected against overheating, reduces CPU clocking speed so that temperature doesn't get much higher than 80C. I just don't want to let my raspi reach that limit/limitation. Your case setup corresponds to the brown or orange graph, i.e. raspi without heatsink without fan. If you run the stress command, as shown above, you should get a graph similar to the brown/orange one; your raspi hitting the 80C-mark after 6min.

    There is no need for such a 14-28US$ case (I paid 14US$ fwiw). From vendor's stock numbers i could tell that my alu case is not a popular model, not a best-seller. Maybe shoppers don't dig the looks. Most raspi owners prefer the inexpensive plastic ABS cases which look very nice, i must admit.
    But lemme compare the situation to flashlights. You can buy plastic $ flashlight or aluminum $$$ flashlight. What shall it be?

    Also note my signature

    My entire life is about adding value through quality so your last comment assumes a little too much about my comments I think?

    Also to compare palstic and aluminum is not a valid stereotype entirely but you were making some sense to me up to that point.

    I agree the case is sweet and that is enough for me,I just wondered if there is a practical or technical consideration involved related to thermal issues with a Pi that I am not aware of.

    I am putting together a project using a bunch of Pi in an outdoor installation and am pondering the level of thermal and weather protection really required for my uses here vs. the practical application of the design so this is a pretty timely post to stumble across in any case.

    Thanks for taking the time to create it.

  12. #12

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Quote Originally Posted by kreisl View Post
    We're also learning that the stress command appears to be sufficient for thermal tests, it is a useful Linux command to make thermal engineers' lab life easier.
    Since the element14 shell script plus the stress command are an improvement to my measurement setup, i redid all tests and also incorporated much longer resting times between the tests. This time I chose an equal test runtime of 36.5min[sic] (i had entered 40min but the stress command finished ~8% earlier than that) for all tests. Test4* was the corner/stand setup, which was similar to the original test4 setup from the OP iyrc. Moreover, this time for test4, the raspi was "sitting seitkant vertical" not on the wood table desktop but on an indoor marble windowsill. Wood (and air) acts as an electrical/thermal/ insulator, a barrier, whereas stones/marble/concrete/etc is great at heatsinking (or storing heat). Making direct contact with the windowsill instead of the table was very effective, the temperature barely reached 50.0C during the test runtime:



    Looking at the above course of graphs, the 36.5min still seemed too short for test3 and test4* to reach final steady-state max temperature omg. So i decided to compose code for an extended test which was to graph 5min of 0%-load (Vorlaufphase) plus 60min of 100%-load (Hauptlaufphase) followed by 60min of 0%-load (Nachlaufphase):
    sudo nano temptestx.sh
    (copy paste the extended temptest code, then exit&save)
    sudo chmod +x temptestx.sh
    ./temptestx.sh > test4x.dat
    ./temptestx.sh > test3x.dat



    So for the extended test i repeated test4* and test3. Yet this time for test3, i let the raspi be lying flat on the indoor horizontal marble windowsill. The raspi's four small rubber feet prevented contact between the alu case and the windowsill. By how much would it make a difference, whether one places the raspi with its four feet on a table or on a windowsill? Let's see:



    Now i finally have got an answer: If i leave my raspi in my preferred corner on its styrofoam stand for hours at 100% load, the temperature will hardly scratch the 60C-mark and, starting from the idling state, it takes 60min to reach that mark, aha. Good to know i guess. And my curiosity didn't kill the raspi. Also good.

    Thanks for reading. Fully automatic testing of the thermal performance and doing so headlessly is reeeeeally phun! 😍
    Last edited by kreisl; 06-19-2017 at 01:41 AM.

  13. #13
    Flashaholic* TinderBox (UK)'s Avatar
    Join Date
    Jan 2006
    Location
    England, United Kingdom
    Posts
    3,464

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Have you guy`s checked to see if the IO chip get really hot when you have a few usb devices connected to it, this was on the prevous Pi i dont know the Pi3B is affected, I fitted an heatsink to the IO chip on mine anyway.

    John.
    "Build a man a fire, and he'll be warm for a day. Set a man on fire, and he'll be warm for the rest of his life." - Terry Pratchett.

  14. #14

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    I tried to research but couldn't figure out which one the IO chip is. I call them chip1 (big chip), chip2 (where i installed the copper shim), and chip3 (chip on bottom side). Which USB devices connected to your Pi made your chip really hot? Just connecting or doing what with the usb devices? My USB scanner and USB printer are WinXP compatible only and i have no idea how to mount my old external (Win) HDD.

    In any case it is normal that chips get hot when in real use

  15. #15
    Flashaholic* TinderBox (UK)'s Avatar
    Join Date
    Jan 2006
    Location
    England, United Kingdom
    Posts
    3,464

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Quote Originally Posted by kreisl View Post
    I tried to research but couldn't figure out which one the IO chip is. I call them chip1 (big chip), chip2 (where i installed the copper shim), and chip3 (chip on bottom side). Which USB devices connected to your Pi made your chip really hot? Just connecting or doing what with the usb devices? My USB scanner and USB printer are WinXP compatible only and i have no idea how to mount my old external (Win) HDD.

    In any case it is normal that chips get hot when in real use

    It`s the usb and ethernet controller, the big chip next to the usb sockets, on the pi3B the current avalible was increased so it most likely does not get as hot as the previous generation, but how hot it gets depends on what you have plugged in.

    Just if you can run a Pi3B in a sealed case, and your are stressing the cpu/gpu to 100% you have also to take into account any heat from the IO chip, If you have nothing plugged into it then you wil be fine.


    John.
    "Build a man a fire, and he'll be warm for a day. Set a man on fire, and he'll be warm for the rest of his life." - Terry Pratchett.

  16. #16

    Default Re: Review: Raspberry Pi aluminum case assembly, mod &amp; thermal test

    Quote Originally Posted by AVService View Post
    bunch of Pi in an outdoor installation and am pondering the level of thermal and weather protection really required for my uses here vs. the practical application of the design
    certainly i wouldn't know, i am newer to raspi than anybody else . just thinking, outdoor .. comes with weather and humidity. and electronics corrode with humidity. plastic material deteriorates with sunshine. no doubt that one must keep the PCB as dry and dust-free as possible but i wouldn't know how, outdoors. even the conditions in my garage could corrode some electronic components, i once saw my Ultrafire WF-188 rusting away in my PC room for no reason, very shocking, which i will never forget.

    thermal protection, how about an extreme heatsink from old PC parts?



    After the 10min, the youtuber gets 49.4C reported, which is higher than in my test4. His test result is of little value because he tested for 10min total only, not for the full duration until steady-state final max temperature was reached, and because the temperature reading of the raspi fluctuates a lot, as can be seen from my graphs. Even the 11s-floating average produces a shaky graph because of those wild-ish second-by-second fluctuations.

    This thread has shown that a thermal test should run imho for at least 60min. Then it is fair to compare resulting numbers from various thermal setups. In all fairness, he did several youtubes on raspi thermal cooling, demonstrating various setups with and without fan. Instead of the stress command, he uses the sysbench command. On a freshly installed raspi, this command needs to be installed first:
    sudo apt-get update
    sudo apt-get install sysbench
    sudo nano ec10test.sh

    (copy paste the test code from the explainingcomputers video, then exit&save)
    sudo chmod +x ec10test.sh
    ./ec10test.sh

    Using the script from that video and my test3 marble setup, i am getting 53.7C as momentary reading. A meaningless value, as explained before:


    Too bad, one cannot compare his 6 data points with my 6 data points or anybody else's 6 data points produced from his script. 10 or 20min are not enough time in a passive cooling environment.

    My alu case could keep the raspi cooler than his "extreme passive cooling" heatsink solution, which would be to be seen after 60min of thermal test runtime. Funny, after all he decided to discard the DIY cooling solutions and in the end went for a commercial aluminum case, the popular KODI case (FLIRC case) which does not perform too well regarding thermals but offers a "very nice soft touch plastics up on top that's a really nice feel, you'll be"
    Last edited by kreisl; 06-18-2017 at 01:10 PM.

  17. #17

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Quote Originally Posted by kreisl View Post
    Moreover, this time for test4, the raspi was "sitting seitkant vertical" not on the wood table desktop but on an indoor marble windowsill. Wood (and air) acts as an electrical/thermal/ insulator, a barrier, whereas stones/marble/concrete/etc is great at heatsinking (or storing heat). Making direct contact with the windowsill instead of the table was very effective, the temperature barely reached 50.0C during the test runtime
    Quote Originally Posted by kreisl View Post
    After the 10min, the youtuber gets 49.4C reported, which is higher than in my test4.
    Here a photo of the setup of the above quoted test4:



    And today in the morning i ran the EC10 test (Explainingcomputers 10min test) and got the following result:



    Raspi reports 46.7C after the 10mins! This beautifully proves the point(s) from my previous posts. And with this result, some readers may(!) conclude that my alu case is superior to the "extreme passive cooling" heatsink solution from that video. Technically it would be imo wrong to draw such a (or any!) conclusion from the EC10 test results, but it could still be true, a fact: my raspi case could be superior with regard to thermals! However the EC10 test (5x2min) cannot prove it nor his EC20 test (10x2min). 10min or 20min are not long enough, one needs 60min for a fair comparison of passsive cooling solutions.

    Note that 34.9C was the steady-state temperature after idling seitkant vertical on the windowsill for the whole night and morning.

    Anyway, there you have it: 46.7C beats 49.4C, if that's what you wanted to hear! q.e.d.
    Last edited by kreisl; 06-19-2017 at 05:34 AM.

  18. #18

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Today i reached another milestone at modding my new raspi alu case. Before i composed the OP, my raspi had lost WiFi (btw in gemani we never use the word "wifi" but the word "wlan" which also exists in English and means the same) connection and wlan connectivity, seemingly for no reason. Well, as i later found out, the reason was the very use of this alu case, at a distance of over 10 feet from the wlan router, which made the wlan signal reception unstable/weak/unreliable inside the alu case. Clearly, i cannot recommend using this case at distances >10ft away from the router or repeater.

    Once the raspi loses its wifi connection for whatever reason, you cannot control or operate the raspi headlessly anymore. For example, i lost the VNC connection, my VNC client app on my phone refused to connect to the VNC server application on the raspi. Without the VNC connection i didn't know wtf happened to my raspi or in which operational state the raspi was. So the only way to find anything out was by carrying a PC monitor to the raspi and connect both devices thru HDMI cable and also connect USB keyboard and USB mouse. What a hassle!

    As mentioned earlier, the origin of all problems was the loss of the wlan connection but i hadn't realized it by this point. The raspi was still 10ft away and, knowing nothing about linux or configuration of wifi on linux, i tried to "fix" the linux wifi configuration manually by clicking on the wifi symbol in the system tray and playing with the GUI wifi settings there. Unfortunately by doing so, i totally misconfigured the wifi config on the raspi beyond good and evil, and even looking up instructions on the www (on my Windows PC) couldn't help it. By that time i learned that one should backup the linux config files (especially the wifi config files) after a fresh linux installation and that one must not fix a "running" system! Yes, running! The loss of the wlan connection could have been easily fixed by a simple reboot of the raspi and then placing the alu case next to the router; it was so not necessary that i had tempered with the wlan settings on the raspi!

    I had taken photos for the OP beforehand, so i went on and composed and finished the OP. For performing the thermal tests, one didn't need a wlan connection after all, fortunately. But it was clear to me that i had to reinstall the entire OS (NOOBS Jessie v2.4.0 Raspbian with PIXEL v1.2) after publication of the OP, which i did.

    To cut the whole story short: it happened a couple of times that my headless raspi lost, not for no reason!, the wifi connection and then did not automatically reconnect to the wifi router to get back online on its own. When such an incident happens, the user must either carry a PC monitor to the raspi (what a hassle no thanks!) or carry the raspi, without damaging it!, to a PC monitor.

    Now how do you carry a live headless raspi to a distant PC monitor/station?

    Wrong answer: You simply cut the power of the headless raspi, unplug the USB cable, then carry the device.
    Right answer: You perform a safe shutdown of the raspi OS, cut the power of the headless raspi, unplug the USB cable, then carry the device.

    The point is: Cutting the power of a raspi while the raspi is performing a WRITE operation on the microSD card can corrupt the card physically and/or the data (or the data integrity) on the card. Raspi users all over the world have experienced data loss/card loss in such instances. Basically, it is NOT SAFE to simply unplug the raspi from power randomly. Normally one would shutdown the raspi by invoking the shutdown command from the Raspbian GUI ("Raspberry" > "Shutdown" > "Shutdown options:" > "Shutdown"), but how do you perform a safe shutdown if the raspi is headless and lost its wifi connection?

    Under such circumstances I can think of 2 safe shutdown methods:

    1. you carry a USB keyboard to the raspi, plug it in and press the following key sequence on the keyboard: UP/DOWN, UP/DOWN (to wake up a sleeping raspi), ESC, ESC, ESC (to get out of active menus or dialogs), WINDOWS (to invoke the Raspberry Start Menu), 1x UP (to navigate to "Shutdown"), ENTER (to invoke the "Shutdown options:" window), ENTER (to 'click' on the preselected "Shutdown"-button). With some luck, since you're not seeing anything on a screen like a blind nut, the raspi will be shutting down and the green LED will stop showing activity. That's the point when you can safely cut the power and unplug the USB cable!
    2. you use my hardware key which invokes the raspbian safe shutdown procedure automatically. The raspi will be shutting down and the green LED will stop showing activity. That's the point when you can safely cut the power and unplug the USB cable!

    Option 2 sounds intriguing and convenient? Guess why and what this entire post is about

    The mod, Part I. Creating the hardware key.


    We're going to need a 2-pin (female) connector. I had white 2-pin connectors and a black 4-pin connector lying around in the house. I liked the black one but it was a 4-pin connector. I didn't like the white one because one side was not fully insulated with white plastic, the electrical contacts were exposed on that side. Not good:


    The white one has the same distance between the pin receptacles and therefore works okay for our purpose. If you don't have a black connector, you could use the white one instead, after insulating the exposed contacts on the one side. Probably the white connectors were designed for thinner pins:


    By the way, instead of my hardware key one could also use the Xiaomi key (the so-called Mi Key), but it has a resistance of 20milliOhms and is not as reliable as my simple key:


    Stripping the insulation was a breeze, no tools needed. Just catch some insulation with your fingernails and pull. The insulation will rip but not the copper inside:


    First i twisted the copper, bend the two ends outward, then hooked them with each other:


    Then i twisted the two ends together. I am genius, this is a huge diy project, that's why this post is so long:


    Now it was time to cut the black connector in half. A 4-pin or 3-pin connector is too wide, it has to be the width of a 2-pin connector. I used a Stanley knife to cut the plastic, which was an easy task:


    I held the connector with my Victorinox pliers while performing the exact cut:


    I did some shaving at the cut lines. While doing so, i cut into my thumb. Ouch. I ended up with a 1-pin connector and the needed 2-pin connector. 4 = 1 + 2, oic hh:


    Then i carved ROW3 on the connector as a reminder:


    Then i painted the chars white. You would know how i did that. A quick size comparison with the Mi Key:


    I taped the copper with red electrician's tape and cut the shape of a flag. A Red Flag! Oh my god i love my work!! So much, that i started thinking about finding a way of attaching my creation to my key chain:


    But then i had a better idea! Why not cover the copper with heat shrink tubing? I mean, that's the original purpose of heat shrink tubing, isn't it? To cover/insulate the connection of 2 wires:


    The diameter of the tube was too big and it began to catch fire. Better than no tubing i guess:


    Again with the red electrician's tape! Red is easy to see and it reminds me of 'warning' or 'caution'. Ingenious choice for color-coding my hardware key invention:


    Since i, by now, know how/why i lost the VNC client-server connection, i also know how to prevent it from happening again. This means, i will always maintain a working VNC connectivity and my raspi will never lose its WiFi connection. This also means that i don't need the hardware key anymore: if i want to shutdown the raspi safely for any reason, i can simply do so through the VNC connectivity from a remote location.

    That's why it is time now to think how/where to store my invention securely, at a place where i would not forget about it. Makes perfect sense to me, putting the key in a small clear bag together with the Mi Key and other raspi-related stuff, and then bag them in the Mi Key retail paper bag, finally in the raspi paper bag:



    And at last you can put the paper bag back into the raspi retail sales box. That's a very neat arrangement of your things and you won't lose any of the small parts easily. Back in the old days i never kept the retail boxes of the stuff i got/bought. But since i got into flashlights, i keep all the Fenix JETBeam Sunwayman etc retail boxes, they are imho the most reasonable way to keep the stuff tidy and complete. I wanna see the guys who really trashed the raspi retail box and the paper bag. I find the stuff highly useful for storing the small parts, accessories, and, above all, for storing away the raspi once you're done with it.


    The mod, Part II. Preparing the raspi OS to respond to the hardware key.

    You can do the following on a headless raspi, no problem. For example, from within your VNC client app on your phone, while sitting on the j*hn. In your raspi OS, open up a web browser and surf to this post. Copy the following line to clipboard:
    Now open up a terminal window and use the Edit>Paste function to paste the code into your terminal window. Press ENTER. When raspi OS boots, it will autostart the script, so that the raspi is prepared to respond to the hardware key all the time, even when the raspi has been shutdown (such that the red LED is still permanently on). Since the safe shutdown script is running in the background all the time, it consumes some energy: freshly booted, the raspi consumes 1.92~1.95W this way. For comparison, when the Amazon Alexa AI script is launched in addition, the raspi consumes 2.05~2.08W; this is Alexa's price for being ready all the time, listening to you non-stop until you say the wake word "Alexa, ".


    The mod, Part III. Making use of the hardware key.

    When the raspi has booted and is running, touch the third row (counted from the raspi rear) with the hardware key for a short moment, say for <0.5s . Do NOT let the key stick in ROW3! The short contact of the key with ROW3 will initiate the shutdown sequence on the raspi OS and from then on the OS shuts down automatically. On a headless raspi, you can see much green LED activity, then this LED stays off:


    Powered from a USB wall outlet, the shutdown raspi will go into a standby mode with the red LED continuing to stay on; standby power consumption is ~0.86W. Similarly to a modern Samsung PC monitor, when you shut down/turn off the monitor through the Samsung controls, the monitor continues to draw power from the mains and is in fact in standby mode, ready to be woken up by the user.

    Powered from a USB powerbank, the shutdown raspi will first go into that standby mode. Since 0.86W translates to a really low current draw, the powerbank typically shuts down powerbanking operation. At that point the raspi's red LED will turn off and the raspi can be woken up only through a power cycle.

    When the shutdown raspi is in standby mode (=red LED constantly on), you can wake it up through my hardware key. Simply let the key touch ROW3 for a short moment, say for <0.5s. Do NOT let the key stick in ROW3! The short contact of the key with ROW3 will initiate the boot sequence of the raspi OS and from then now the OS boots up automatically.

    In other words, you can use my hardware key to safely shutdown the (headless or not headless) raspi AND/OR, instead of a power cycling, to boot a shutdown raspi, waking it up from standby mode.

    And last but not least, maybe this is the best thing about my hardware key, you don't need it. The mod is so noninvasive and primitive, you could simply use your car key instead. My key's task is to short the pins of ROW3. It is super cool to have a dedicated hardware key such as mine, so beautiful, but one doesn't need such a fancy kresil approved hardware key built during an entire summer afternoon for the task. Reminds me of bloatware.

    Feel free to subscribe to this thread. Maybe thread title should become "Story Time: Slowly Setting Up My First Raspi"

    EDIT: Today (2017-06-30) i managedto cause my raspi (or raspi OS) to freeze up on me! The Alexa AI software was running, as always, and i was streaming my online subscription music through Chromium web browser, also as always. Then i opened another browser tab, surfed to my Gmail inbox, opened a TXT email, and then i lost the VNC connection on my phone, the music stopped playing, and it became clear to me that something on the raspi had crashed: either it was the WiFi connectivity (for whatever or no reason) or it was the entire system which had crashed on me. Since the raspi was headless, i couldn't know what was going on! I looked at the 2 LED's: the green LED was permanently on (yikes!) and the red LED had intermittent offs. ((If the red LED is not permanently on, what does this mean? We should look it up in the raspi operation manual!))
    First question which popped into my mind: Could i use my shutdown key to shutdown the system? I tried, it didn't work.
    Could i use my 'blind keyboard method'? I tried, it didn't work.
    Since i ran out of safe shutdown options, i carried my HDMI PC monitor to the raspi (What a hassle!) and connected the two to finally see what was going on. And? Well, the screen was frozen, and the frozen pic showed me a disconnected WLAN, a disconnected VNC, the time of freezing, and a frozen yellow flash symbol in the upper right corner.
    The second question was: Why did my raspi crash and freeze? And was it Jessie software which froze or was it raspi hardware which froze? I can only guess the latter, since the yellow flash symbol indicates nearing critical low power input from my USB 5V/1A power supply. I don't know for sure if low input power was the reason for the system crash. The third question then would be: Why did it have a problem with the power supply, since it didn't have such a problem before, when i was using several browser tabs in the same session situation?
    Since the system was completely unresponsive (USB keyboard, mouse, my shutdown key, etc), there was no safe shutdown method possible and i had to pull the power plug, while the green LED was permanently on, argh!
    From this anecdocte we are learning that a safe shutdown is not always possible, for example when the raspi (software or hardware) has frozen up on you and has stopped being responsive. In such a case, you're left with pulling the power plug, hoping that this won't damage the microSD card or the data on it.


    EDIT2: There is one simple and quite natural way to lose the VNC connection on a headless raspi. On your phone, in your VNC Viewer app, go to your raspi, Raspberry > Preferences > Raspberry Pi Configuration > Interfaces: > VNC: > Disable > OK . As soon as you click on <OK>, you lose the VNC connection lol, and you cannot enable VNC back again, because you cannot see what's going on on the raspi! So you'd do the safe shutdown with the hardware key (or your car key), carry the raspi to a 'head station' (hdmi monitor, USB mouse) nearby and reactivate VNC. The lesson from this incident: do not disable VNC haha!
    Last edited by kreisl; 07-01-2017 at 11:12 AM.

  19. #19

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Story time 14:59 continues, today i am posting about the installation of DEX on raspi.

    Alt1. The most basic way is to open a terminal in the GUI and then enter 6 command lines, one after the other:
    cd /opt
    sudo tar -xzvf ~/Downloads/dataexplorer-3.3.0-bin_RaspbianLinux_ARM_32.tar.gz

    clear
    dir
    cd DataExplorer
    ./DataExplorer

    This command sequence has assumed that you manually downloaded the *.TAR.GZ-file in the default system folder /Downloads/ through the Chromium web browser. However this crass short sequence may not work on a freshly installed raspi OS right off the bat, some OS preparation may be missing. Alternatively, one makes sure that all preparatory commands are included:

    Alt2. So our recommended way to install DEX on raspi is by using Ravel's more elaborate recipe, at the time of writing its v0.0001b version, in executable script form. For this, open a terminal in the GUI:
    sudo apt-get update
    sudo apt-get upgrade
    sudo nano installdex.sh
    (copy paste his entire blue text including the full download URL's, then exit&save)
    sudo chmod +x installdex.sh
    ./installdex.sh

    Done! This sequence downloads all required files automatically and starts DEX during the installation procedure. You should click away all DEX pop-up windows, exit DEX, connect the powered charger with raspi, and launch again DEX, select your DEX-supported device (MC3000) from the menu, etc. People with prior experience with DEX know what to do from here. Note that the v0.0001b recipe installs DEX3.2.9. When the DEX3.3.0(final) is released, the recipe should be updated accordingly.

    As always, you can do the above headlessly, e.g. the installation, operation and control of DEX on raspi through your smartphone via the VNC Viewer app. (There are other ways to access and control your raspi Desktop remotely, of course.)

    Today i followed Alt2 to install 3.2.9 and i did some mc3k logging (click to enlarge!):



    Then i grabbed 3.3.0(beta) and followed Alt1 to install it. The 3.3.0 did not uninstall or overwrite the 3.2.9; i have 2 separate DEX installations on my raspi now. The 3.3.0(beta) is interesting because it has debut support for Uni-Trend Uni-T Unity UT61E digital logging multimeter, hence also on the raspi.

    Seeing is believing. I had got the following raspi photo, this is how it should be, perfect:


    I fired up my 3.3.0(beta) and got this instead:



    Then i realized that i had not connected the DMM to the raspi yet haha!

    The UT61E comes with an original serial data cable (RS232, female plug) packaged in the retail box, great reliable cable. (Never mind the optional original USB data cable, i don't find it reliable.) But the raspi does not have a ready-to-go RS232 port! Of course, one must connect the UT61E through its original serial data cable to the raspi "somehow". There are 2 straight-forward commercial solutions: either using a FDTI-chipset based USB2Serial adapter or using a working(!) RS232 shield (Pi Hat). At this point i cannot recommend using USB2Serial adapters based on the much much cheaper CH340/341-chipset (or cousins like the HL340) because they are not straightforwardly supported by the latest Raspbian builds/kernels; my neighbor guesses that one could get CH340-based hardware or adapters to work on raspi but there's no guarantee/instructions/support/documentation/proof, so you've been herewith expressly warned. Same problem holds true for USB2serial adapters based on the ubiquitous inexpensive Prolific PL2303, a chipset rather common and supported on old Windows computers but more often than not a pita on Mac or Linux systems, feel free to google raspi forums. In contrast, FDTI chipsets are supported directly by the Raspbian kernel, so you're good to go, no need to manually install Linux ARM drivers from a CD or floppy disk. If you wanna save some bucks and buy a USB2Serial adapter based on CH340/341/HL340/PL2303 for your raspi, be my neighbor's guest, but do not ask me once you run into the slightest problem with it.

    Apparently the cost of a nice quality FDTI usb2serial adapter is similar to the price of a Pi Hat RS232 shield, ~13EUR shipped. In general, a usb2serial adapter would be more flexible, you could use it on any computer, PC, Mac, Linux computer in order to convert a USB port into a real RS232 port. Your choice. I chose the shield, yeah why not.

    Story time will continue with photos/screenshots/review of the serial board, once i got it. It is the only missing part to get my raspi log the multimeter for endless hours!

    (The serial board is being restocked on Banggood afaik )
    EDIT: it is in stock now, i've ordered it. Looking forward to receiving and writing about it!! Since it has a RS232 female plug too, i also ordered a RS232 male-to-male gender changer from BG, 0.69US$ apiece lmao.
    Last edited by kreisl; 09-07-2017 at 07:40 PM.

  20. #20

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Quote Originally Posted by kreisl View Post
    EDIT: it is in stock now, i've ordered it. Looking forward to receiving and writing about it!! Since it has a RS232 female plug too, i also ordered a RS232 male-to-male gender changer from BG, 0.69US$ apiece lmao.
    Today a share of the beautiful-looking products (John 29:85).

    I got the stuff from this post almost 2 months ago in 2 BG packages:


    The so-called "shield", also called 'serial shield/UART shield/serial board', in form of a Pi Hat, came well-packaged, well-protected. Bubble-wrap on the inside, sealed plastic bag, plus generous external foam padding. The so-called "gender changer" was shipped without any padding for protection:


    Let's get the stuff out for detailed inspection:


    Nice photo:


    I see:


    How to convert a female port to a male:


    The fitment looks nice:


    Now comes the fun part:


    Will it fit? Hmmm:


    Why yes it does:


    The maker should have glued a plastic spacer on the board:


    Let's plug and play:


    Since there was no spacer or holder, i used my white holder as support which had the perfect height:


    I read somewhere that it was a plug'n play product, nice!!
    () tl;dr to cut the whole story short, after 2 months of trying to figure out how to integrate the Pi Hat into my raspi system and failing to make any progress, i have given up. My raspi does not recognize the GPIO Serial Port Expansion Board RS232 For Raspberry Pi in any way. Maybe the shield or the gender changer was broken? So i got another set, with the same negative result.


    At least the green LED on the board lights up as soon as the raspi is connected to power. Looks so attractive haha.

    I am still a total newbie regarding raspi and stuff, especially regarding Linux per se. So chances are that i've been doing something wrong. Then again, there are no instructions, no documentation, nutting, not even other customers, users, or owners with a hey. And i won't accept "help" like suggestions, hints, tips, guesswork, ideas, instructions, etc, until some shield owner other than me reports and proves(!) positive results; in fact, i heard someone, who had examined the traces and component population on the PCB of the shield, basically the schematics if you will!, claim that this Chinese product 2015 is a total fake product and could not even work in theory! So until then i will regard this post as dead end and i am moving on, see my next post below, hooray.
    Last edited by kreisl; 09-02-2017 at 08:36 PM.

  21. #21

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Here, in my 3rd post regarding raspi & UT61E, i am finally showing how it's done (Matthew 32:55)!

    The other day Raspbian Jessie ("Raspbian 8") got upgraded to Raspbian Stretch ("Raspbian 9") and the recommended procedure is to install the OS from scratch on a blank formatted microSD card, which i did. After some unimportant personal customization of the raspi desktop ("PIXEL") preferences/settings/configuration, such as enabling SSH and VNC Interfaces, the only thing i installed as a must-have was the Hardware Key script from post #18 Part II.

    Then i edited the DEX installation text file, because DEX3.3.0 final version was released in the meantime, and used it to install DEX3.3.0. After the installation, i connected the (turned on) UT61E to the raspi through my expensive USB2Serial adapter; much to my astonishment, Raspbian 9 recognized the plugging without a reboot, and so did DEX! After choosing UT61E from the DEX device selection dialog, everything was good to go and logging UT61E on the raspi has been working flawlessly since.

    If you are interested in photos/screenshots which i took along the way, here we go:

    The (famous) German trading company Logilink has been offering various USB2Serial adapters in the past but only recently decided to release an ultimate hassle-free adapter to replace all previous products in that product category. That ultimate adapter is much more expensive but guaranteed hassle-free because it uses the FTDI chipset series (FT231X, FT232RL).

    I found an online retailer who offers the product at 9.45EUR shipped, so i pulled the trigger-happy. See what i got for the money, front face of the packaging says "Designed in GERMANY" lol yeah whatever:


    The definite manufacturer's model designation is "AU0034" and the stuff was "Manufactured in P.R.C." there we go:


    I had looked hard on Aliexpress to find a FTDI-chip USB2Serial adapter as low-priced as possible but they were all between 7-8US$/pc, looked awfully cheap build quality, and with no support whatsoever. I am absolutely pleased with the Logilink build quality and the wealth of package contents:


    Note that "Chip FTDI" is the most essential and decisive characteristic of this USB2Serial adapter! You're on your own if you buy a USB2Serial adapter with some other chipset like CH340/341/HL340/PL2303 for your raspi:


    The only thing cheap about the Logilink product are the label stickers on the adapter. They hardly stick on the rubber-plastic surface and they don't even fit properly into the recesses. Shame on the geman designers:


    Here a photo with the complete contents, nice huh. The printed instruction manual is rather detailed, 8 pages (per language) with lots of screenshots. Both the adapter and the included USB extension cable come with clear plastic protection caps on the USB A plug. The product has a manufacturer's website, a product webpage (with Downloads, Technical Support phone/email), all included in the 9.45 so to speak:


    The pressed CD-ROM is labeled "2015 Logilink", which must be the initial release date of the product as "AU0034 v1.0" because the CD content is much newer than that, March 2017 wow:


    All right, this is what the adapter looks like plugged into the USB extension cable. Clear/transparent stuff is easily lost, so i put some red tape on the caps for improved visibility hh. The caps fit nicely on other USB A male plugs, for example on my USB mini microfone (for the raspi/alexa application):


    Do we need the USB extension cable? Why yes because we don't want to strain the raspi's USB female port with unnecessary mechanical load/momentum. Because of the product thinness, the USB male plug rides several millimeters lower than the USB female port. It would need some mechanical support for matching heights:


    Instead, we make use of the included USB extension cable, no problem. Also see how massive the UT61E RS232 plug is. All fitments are tight and perfect, and we don't need the opposing integrated pin screws wtf:


    Why would you want to leave out the USB extension cable? When you buy a Chinese USB2Serial adapter from Aliexpress or Ebay, a free USB extension cable might not be included, beware. Because of all the mechanical load/momentum on the USB port, it is imho really not a good idea not to connect the adapter with a USB extension cable to the raspi:


    We don't need to power the raspi through the mc3k USB power output port, i know i know. Never mind huh. But i like this configuration with the very short USB power cable. Note that recent UT61E product packaging comes with the (formerly optional!) UNI-T USB data cable instead of the depicted UNI-T RS232 data cable. Don't wait for DEX to support the UNI-T USB data cable but go buy the UNI-T RS232 data cable plus the FTDI USB2Serial adapter, if logging UT61E on raspi is very urgent and important for you, that's my advice. I will assume that most UT61E owners in the world do own the RS232 data cable already, because that's what UNI-T factory had been shipping for most of the time since the UT61E initial release to market 7+ years ago. By the way, i once bought and tested the UNI-T USB data cable and it produced data sets with missing data points; from that experience i don't rely on the USB cable anymore but prefer the original old RS232 data cable. There you have it:


    In this thread we already covered how to install DEX on a freshly setup raspi (Jessie), didn't we? I used procedure Alt2 on my freshly installed raspi (Stretch). I edited Ravel's original v0.0001b recipe (which was valid for DEX3.2.9) this way, now being valid for installing DEX3.3.0:


    Other than that, nothing has changed, the Linux commands are the same:


    As you (might) know from Ravel's recipe, during the DEX installation procedure the program does get launched and you are required to click away/ignore a couple of windows/dialogs and eventually exit the program. By that time you have not connected your mc3k or the UT61E to the raspi.

    So you've exited DEX. Basically, DEX has been installed by now. But you still see a black window (terminal) on your desktop which is active. That's because Ravel programmed the installation script this way, very cleverly. Because, when you now plug a powered(?) device into the raspi's USB port, that terminal window will show live, return some feedback, that you have plugged a recognized device! This is fantastic! Usually it is a pita to have Linux recognize new hardware or to even find working third-party drivers (for the different Linux processor architecture: i386, x64, ARM, MIPS), but since we chose a USB2Serial adapter with FTDI chipset, the adapter is recognized by the raspi ARM kernel instantly without the need of installing a driver externally from the Logilink CD-ROM. Genuine Plug'n Play, if you will:


    Now you can start DEX manually. When you choose UT61E from the DEX device selection dialog, you'll see that DEX has recognized the existence and availability of your raspi's serial port, too:


    If you're familiar with DEX or DEX+UT61E from your other computer systems (WinXP/7/10, MacOS, Desktop Linux distro), then you'll know how to play. Click on 'Start gathering' and it's your time to play! As we can see, DEX logs the UT61E at the natural 2.0Hz sampling rate and with up to 3 decimals:


    Since the UT61E is a so-called "22000 Count" DMM, there are (few) ranges where the DMM can display up to 4 decimals, for example "2.1357 V"; in these instances DEX would log only the first 3 decimals (2.135V). Apart from that minor limitation, i prefer logging UT61E through DEX software to other capable UT61E logging software (ABACOM REALVIEW 3.0, UltraDMM, UT61E Interface Program v4.01, a. o.). For example i like that the graphics tab shows all data points, not just a floating window of fixed time frame; my UT61E with unplugged probes and in the "mV"-setting shows some interesting pattern, maybe it is registering a ghost in the room:


    With this post i hopefully managed to show you in sufficient detail what you need and need to do in order to setup your raspi for logging your UT61E through the original UNI-T RS232 cable.

    All kudos go to the programming team of DEX, free open-source multi-platform software Made In Germany

    EDIT: FYI this thread can also be reached under the easy-to-remember URL http://tinyurl.com/mc3k-raspi
    Last edited by kreisl; 09-22-2017 at 12:43 AM.

  22. #22

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    My raspi is getting older and older (the latest model is 3 B+ 2018) and continues to run 24/7, doing nothing but streaming music subscription whenever i want to listen to music. I never put strain or anything on the MicroUSB socket/plug (5V power), but today for no reason, when i dusted off the corner where my raspi was sitting for months, i noticed how the plug ..etc.. (let's save the words and see for yourself).

    I once bought the Geekworm 3A adapter, which is advertised as a raspi power adapter with on/off switch, from GB or BG. It's also available on ebay for the same price and HKJ reviewed this kind of adapter before. I can't recommend this particular item. Maybe the electronics are good for 3A and the cable is nicely thick for this purpose, but the power switch and the MicroUSB connectivity (poor plug, poor socket, high contact resistance, loose connection) kill the theoretical max. performance. I really hate MicroUSB connectivity. This is the 4th product in my household where the connectivity completely failed after a short time, so either the plug or the socket had to be replaced. And let's not forget the hundreds of mc3k users who've been struggling with getting a good PC Link connection with their MicroUSB cable.

    Today it's the plug which failed. Well, it still works, i.e. delivers power to the raspi. But when i saw that the plug developed loose parts, i cut everything off and disassembled the plug and the switch, here we go:


    A close up of the loose metal part:


    What's the point of today's post?

    I felt like sharing, that's all. I regret having bought the Geekworm 3A adapter back then. If i could dial back in time, i would ignore the Geekworm adapter and would have bought the official raspi power adapter. I think it got a positive review by HKJ iirc. It's only a 2A or 2.5A adapter but that's sufficient for raspi.

    Of course, i will not throw the Geekworm adapter away. I am throwing away the switch and the plug.

  23. #23

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    Does anyone know whether this case will fit the Raspberry Pi 3 B+ (the 1.4 GHz model, not the 3 B 1.2 GHz)? Thanks.

  24. #24
    Flashaholic* vadimax's Avatar
    Join Date
    Dec 2015
    Location
    Vilnius, Lithuania
    Posts
    2,068

    Default Re: Review: Raspberry Pi aluminum case assembly, mod & thermal test

    I am not the most clever person in the world for sure. Several days ago I have discovered a SBC universe for myself. And I started from ordering an ASUS Tinker Board S + Anidees Aluminum Case. The last one looks like a piece of art:



    But immediately I ask myself: will it allow to dissipate the generated heat (this board consumes 1.5 more current under full load)? Of course, I will ground memory chips on the case body with copper sinks, but what to do with the upper side chips? Small cable slots under the cover look insufficient to me to produce enough convection for cooling...
    Last edited by vadimax; 09-16-2018 at 03:03 AM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •