Slashdot Mirror
Ask Slashdot: How Would You Suggest Making Rugged, Weather-Resistant ARM Systems?
New submitter pecosdave writes: I need suggestions for commercially-made ARM systems that will work in temperature ranges from -35F to 140F (-37C to 60C) for an engineering project. These things are going to be in metal boxes on the side of Texas Highways. The existing Intel systems we're using in other areas are all fan-less, but I'm not going to rule out systems with fans. Considering the extremes of Texas temperatures I'm actually contemplating putting fans on top of our fan-less systems anyways. Almost everything I can find pre-made with ARM is a bare board, or something not nearly as temperature tolerant as some Intel systems I can find. The very nature of an ARM processor should be more tolerant simply because they produce less heat, but I can't seem to find any manufacturers exploiting that fact. Slashdot reader pecosdave added more details in a comment: "It's more closely related to speed cameras, but it's not a speed camera. It's for a toll road, and its main job is to take pictures of a sign at about 10 FPS, though less is probably fine, with a time-stamp so if someone runs the toll we have a separate picture of the current price. If there's a problem with the sign it shows up as well. They just want something local to store it I guess in case the fiber link goes down. We're going to run it rather low-res too to keep the CPU and storage overhead low. I figure 640x480@10FPS is reasonable, but that's not set in stone."
There are plenty of industrial manufacturers that will get you a custom chip in an IP65 enclosure. My company does these things all the time, but it's a bit more complicated than just putting something in a box. You have to spec the comms, peripherals, serviceability, lifetime, support, software, updates, ...
Fans on top of fan-less design is a bit weird partially because you're going to sacrifice your IP rating and your fan shorting out or seizing up could bring the power and thus the system down. Also a fan is for moving hot air out and cooler air in. An ARM chip in most data collection circumstances won't get as hot as an asphalt highway in summer so you're just exchanging hot air for hot(ter) air, what's the point of a fan?
Custom electronics and digital signage for your business: www.evcircuits.com
A lot of ARM systems (even raspberry PI) will live life fine at 60C. You will need a nice passive heatsink inside your box (if it is large) or thermally couple the device to the box wall and put a passive heatsink on the outside. You'll need an active heater of some kind in order to operate reliably below 0C, but that is easy. Paint your box white too... that will help keep it cool in the sun. Good luck! Also post to an appropriate reddit rather than slashdot!
(this is offended to the end of comments you post, 120 chars)
you can do it very easy with a RPI3 +. Raspicam. I run mine 24/7 [the RPI] connected to my 3d printer in this case with an IR raspi camera module and it streams 24/7 running repetier server for the printing part and python3 for the webserver and camera stream.
. Because of low heat you can just use ABS to print a nearly air tight box to put it in and 5v power supply and bob's your uncle.
You might be able to find a module intended for automotive use. The operating temperature range would be similar to that.
This is my industry. I've worked with super ruggedized military grade ARM boards. I would go fanless, and for your application you shouldn't need much power. Products end up as metal boxes with fans if they even need that.
You've gotta find vendors which make automotive and military/aerospace grade chips, so the usual Samsung and TI OMAP chips won't always work. NXP is bigger into this space with IMX.6 chips and similar. As much as tons of developers dislike Freescale/NXP, they have a huge presence in military and embedded products for this reason.
Says the guy who clearly never needs to use a road.
I just thought about double sealing it in a small container. The first layer filled with a compressed gas and the second filled with oil. Then have radiator tubes going through it and around, put a tiny 5 inch fan on the bottom so the breeze from the cars helps spin the fan, possible mount a heat sink on top...
I'm just imagining this thing built like a car radiator or computer heatsink, with the breeze off the road that should keep the fan going even with no electric, the pressurized gas + oil double seal should keep the components safe from the elements, fumes, bugs and debris and be a good indicator if the housing has been compromised (maybe have a pressure sensor inside to monitor that too).
Judging from the number signs riddled with bullet holes that I have seen in Texas, you may want to make it bulletproof.
Please take your anti government shit rant back to 4chan where it belongs.
Search for DIY and off the shelf Raspberry Pi trail cameras.
https://petapixel.com/2018/02/06/motion-detecting-wildlife-camera-made-raspberry-pi/
https://peaknature.co.uk/blog/how-to-build-a-raspberry-pi-zero-trail-camera--part-1-what-you-need
https://www.raspberrypi.org/blog/solar-powered-nature-camera/
https://thepihut.com/products/naturebytes-wildlife-camera-kit?variant=28137973841
1. Give them a sun shading roof that will not be an inch from the main case.
2. Use the pole as a big cooling tower, with fins on the outside, maybe a heat pipe system inside. The main box could have fins too.
3. Use a very long pole with lots of fins buried deep underground where the temperature does not change much.
4. Some ARM SoC designs can work at up to 110ÂC or.more, as do certain boards.
4. Please screw *everything* into place. Make sure *all* connectors and chips and heat-generating components have a hunk of metal pushing them down that is part of the case. That will prevent them from coming lose and lead off heat.
5. Evacuate ALL the air from the case after sealing it shut, so no condensation can happen. The metal will lead away heat. No air needed. No holes needed. No fans needed.
7. Maybe paint it white (for the entire solar spectrum).
If that is not enough, just add more shade, increase the albedo, add more fins and a deeper pole. Amd better vacuum sealing.
Thermal management for high temperature is easily managed. You're very concerned about the high temperature when that is the easy one to deal with. Any ARM chip thermally bonded to the side of the case / large heatsink which is somehow externally accessible and allows minimal airflow should be able to handle the upper side of your temperature requirement.
Your problem will be at the other end. Sustained electrical performance at -37C is something many components are not rated for. Even in industry there are problems sourcing instruments that handle that kind of temperature. We had many industrial systems fail during the last polar vortex including an industrial PLC mounted in external cabinets.
Your problem will be you somehow need to design a heating system which isn't defeated by your cooling system. That's an especially tough task if you're talking about passive systems for cooling.
Who the fuck said this was a government road? Could be a private road.
Sounds like you want an industrial PC. IPCs have a number of properties that will be useful in you envity (e.g. wide operating temperature range, dust resistant, etc.). According to the Wikipedia article they are mostly Intel systems, but you might be able to find some ARM ones if you look around.
... against the people for which the state even gets to exist at all!
"I was just doing my job." never worked in any trial. If you are mentally healthy, you are aware of what you are doing. And you can say no. So if you did't, but said yes, *you* chose to support evil. Not just your superior.
You absolutely do not want any fans. Fans circulate air, which will only draw in dust, moisture, bugs and other foreign materials that will exponentially increase the rate of failure on your hardware. Remind yourself that most ARM manufacturers are tailored to one market: mobile devices. You will likely find the price of hardened components (the term you should be looking for) to be quite more expensive than off-the-shelf devices like a Raspberry Pi. The operating range for most of those common devices is well outside your requirements, not to mention the durability factor (dust, vibration, etc). For such a simple project, a dedicated construction or outdoor security camera with integrated storage and networking will probably tick all of your boxes. You won't find much gain in reinventing the wheel here.
Well, here is not "Ask Slashdot to do your job" either, so all of you can fuck right off.
Funny rant from that guy, about roads provided _by_ the very government he's pissed at. Unbelievable unaware.
How many people have to run this toll to justify this backup devices existence? If you figure baseline $600 minimum for hardware that would be 120 people at $5 a toll. so 10 a month for a year, and this is a backup. Is that reasonable?
Also check with Advantech for hardware. They appreciate the difference between Intel and Arm exactly for the temperature differences.
I'm more worried about the -30C temperature. Worked for an unnamed ARM manufacturer for a while and we had some problems of our PLLs failing to lock below -25C. An environmental chamber is your friend. I wonder if a peltier unit with a heatsink would help. You could run it in reverse if it gets cold.
You mean *outside*. :D :)
Like I said in my other comment: Lots of fins on the outside, aluminium case metal protrusions directly pushing down on hot components, heat pipe from them through the pole and as deep underground as you can, plus lots of fins on the buried pole end. Don't forget to evacuate the now unnecessary air to prevent condensation.
You actually think it will need to be -35 to 140? It's hot in Texas but not that hot. And it most certainly does not get much below freezing.
I'd suggest you review some design standards first before post here.
Rethink your upper temp environment for that metal box in the sun, it's probably 100 deg. F off. Now if you bury the CPU 6' underground, you start off with a much more stable, cooler environment. Plus it'll be bullet-proof, and when the monster truck with a 6"x8" strapped to the front bumper hits it, the extra anchor depth will do serious damage to the vehicle, and not much to your buried electronics.
The plastic, like all plastics, will turn brittle incredibly quickly under that hot outdoor sun. If it doesn't get soft outright.
It is also horrible at leading away heat, and will not survive a single e.g. baseball bat.
Basically, your kind of people with that kind of mindset are the cancer that entered our industry, and literally why we can't have nice, lasting things anymore.
Well, as an engineer who designs outdoor electronics (with ARM CPUs) you should probably stop asking for freebie advice on Slashdot and actually hire a design house to design it for you.
Next up by BeauHD on Ask Slashdot: A kid needs help with his homework assignment of writing a sort routine.
Or... like it says in the summary, they're not taking pictures of cars. They're taking pictures of the toll road sign.
I assume it's for disputes, where the defendant tries to claim the electronic sign wasn't working.
As for the actual question being asked, seems like they're trying to find a solution that fits their preconceived beliefs that ARM based systems must be more power efficient that x86, so therefore must do better in hot environments.
You'll find that ARM systems are gear towards mobile applications. The chips produced generally max die temps around 80 degrees C. You don't want things hotter than that sitting in a passively cooled thing in someones hand.
Intel CPU's on the other hand, come out in flavours that are capable of working at 105C or higher.
The record high in Texas was 120 F in Seymour, TX in 1936 and -23 F in Tulia, TX in 1899. The range is considerably less along MO-PAC.
With enough power available at each site, anything is possible.
Have gnu, will travel.
I've used Technologic Systems' hardware for decades: http://embeddedarm.com/ Quite a selection to choose from. Fanless, -40C to +85C. Should be able to do what you need, and they keep making their products for many years, so you'll be able to get replacement parts in the future.
Funny rant from that guy, about roads provided _by_ the very government he's pissed at. Unbelievable unaware.
It happens all the time. For instance, some inmates complain about the prisons provided _by_ the very government they're pissed at.
Nobody should be allowed to complain about the government. They should just be grateful for whatever they get.
You will need something that survives way above 60C if this thing will be kept in a metal box in Texas, in the summer...
Make use of the lower ground temperature. You'll also see less extremes in temperature. The deeper the better. Put it in a concrete pipe and insulate the lid.
I'm pro-government, and I probably hate toll roads even more than the other guy! LOL
But the more serious answer to the question is, if it is an "engineering project" then your engineer should be telling us, you shouldn't be asking slashdot.
Are you going to need cooling? Yes, you're going to need cooling. Assuming that you have money, since you don't have an engineer you must still have the money, so you should buy a refrigerated water cooling system, and you're done asking about that part.
Since you're in Texas, you definitely want a TI-based ARM SOC, like from Octavio.
Or just buy a BeagleBone and put AC in the case, and done. That uses the Octavio.
It is all about the case, not the computer.
Why ARM? I get the impression that you think it will cope better with the heat by generating less of it, not that it will do better dealing with the expected workload. That impression is reinforced by the fact that you seem more concerned about the heat than the cold.
Solving the correct problem is step one in engineering. Choose a ruggedized industrial solution that has been designed to deal with those temperature ranges. A proven design will easily operate at 50 Celsius, which is hotter than the hottest temperature ever recorded in Texas, although some airflow (and shade!) is helpful when the temperatures climb that high. Similarly, the lowest temperature ever recorded in Texas is -30C, so why the hell are you looking for -37C?
You don't get any big-dick points for over-engineering.
Advantech was just the first industrial vendor that spring to mind. There are others, easily found with the power of google. https://www.advantech.com/succ...|pd=industrial%20communication
This isn't true. Check out Texas Instruments. For example, all their Sitara processors are available up to 105C.
http://www.ti.com/processors/s...
I didn't even look at other lines, I just clicked the first product lines that would do the stuff in the question, eg push 32 bits around moderately fast.
I design outdoor electronics for power utility automation where the operating environment and reliability requirements are a lot more stringent then they are for some county mountie cash grab camera. ARM MCUs in the -55C to 105C extended rage are not hard to find. Prepackaged SBC's in IP67 or even IP68 housings are a common military COTS and industrial control requirement so there is whole group of manufactures making these.
Your average consumer or maker/hobby targeted boards will not get you very far if you want it run more than a year or two out of doors. I don't know about the lone start state but Arizona has seen 50C, albeit briefly. When you start adding worst case solar loading and self heating to that and 85C comes at you faster than you think, even after accounting for cooling from ambient air flow (wind).
It was just recently announced they will begin doing tolls on Highway 75 in Dallas County.
PGBT is under construction to add a lane. They just opened an express lane from Dallas to Denton a year or two ago. 635 has massive express lane underneath... Each of those highways still gets stupid traffic... More lanes, more tolls, going after toll offenders is not a long term solution.
https://www.dallasnews.com/news/collin-county/2019/03/21/highway-75s-hov-lanes-will-soon-free-collin-county-1-catch
Dallas can be hot...
https://www.cbsnews.com/news/texas-city-makes-heat-record-100-days-of-100s/
"Wichita Falls has become the first Texas city to have 100 days of triple-digit temperatures in one year, the same day the Dallas-Fort Worth area joined the state's long list of cities with a record number of 100-degree days in 2011."
Thanks for getting this back on topic.
TI and others make units designed for automotive use which are speced for those kinds of temperatures. They are expected to work under the hood of a car, in Texas.
Just Google automotive micros or socs as needed, or call any manufacturer.
Milspec parts are similar temperatures plus a higher reliability rating.
Or, submitter could ask Zach, who works in signaling, which would then text Ray. :)
I work for a major process control equipment company.
You should look to folks in that industry and the designs that are used. We use ARM micros. Don't be surprised if you get above 140F in Texas in a closed metal enclosure in the sun.
1. Find all the parts that make a computer.
2. Find the code to make all the parts that make a computer work with ARM.
3. Test the ARM OS and ARM ready computer parts.
4. Find a factory that makes the needed temperature tolerant enclosure.
Heat will need to be moved. A fan might become more of a problem over time/the ability to cool.
Look at people in the USA who make trail cameras/remote camera https://en.wikipedia.org/wiki/... .
Why? They know of all weather US conditions.
Software and hardware works.
Add power and backup networking as needed.
Talk to them about heat long term.
Made in the USA. Jobs for the USA.
Domestic spying is now "Benign Information Gathering"
I would like to know what cooling system you are using in your metal boxes to keep them within the -37 - 60 deg C temperature range at the side of Texas highways. Most industrial grade components are rated for up to 85 deg C, but if that metal box is going to be in the sun, I'd opt for components that can handle 105 deg C if I were you.
Try the Jetson TX2i from Nvidia. It is the "industrial" version of there TX2 development board. Off the shelf, it has a temperature range of -40 to 85C, a quad core ARM A57, and a GPU that will easily meet your requirement of 10 FPS. I've got the TX2 processing 2K video, so 640x480 at 10FPS shouldn't be a problem...
https://developer.nvidia.com/embedded/buy/jetson-tx2i
Many of them are available in industrial temperature range (-40..+85C).
White does not keep things cooler than black. White acts like an insulator, black like a heat conductor. So if you paint it white, the inside will be cooler when the outside temp is hotter. But if the electronics heat up the inside to hotter than the outside temp, the white will actually inhibit heat flow out, thwarting your efforts to cool the interior.
For best temperature regulation (easiest to alter the interior temperature), paint it black, and shade it from direct sunlight to inhibit external heat flowing inside during the day.
White works for things like building roofs because there's relatively little heat generation inside. The biggest heat factor is the outside. So a white roof keeps your house cool in the day (prevents exterior heat from going in), warm at night (prevents interior heat from escaping out). But for most electronics, the biggest heat load is generated inside, and your goal is to maximize heat flow out. If your SoC is at 60C, it's highly unlikely the exterior temperature will be hotter (in the shade). So even during the day in the middle of a heat wave, you still want to maximize heat flow from inside to outside. And that means painting it black.
Wow, Texas has the same temperature profile as equatorial Mars?
Table-ized A.I.
You'll find that ARM systems are gear towards mobile applications.
I'm sure these systems will be quite mobile once people find out there are free PCs sitting around on TX roadsides ...
It must have been something you assimilated. . . .
Put the electronics in a Pelican case. Use a motorcycle camera on the outside. Use waterproof, circular mil-spec connectors for the camera and power supply to the case. Done.
Browsing at +1 - no ACs, I ignore their posts. So refreshing!
Seriously, these are cheap systems. Fit them out and then embed them in thermal conducting, non-electrical conducting epoxy. 10 years ago, I would embed small systems in it and it works great.
I prefer the "u" in honour as it seems to be missing these days.
At least I'm pretty sure they do everything you want. They're called "dash cams". If they can survive the interior of a hot car, they can survive this.
(T>t && O(n)--) == sqrt(666)
There are companies who already make this sort of hardened equipment.
But there's more than just temperature variations to consider... you need them to be able to operate in a range of humidity conditions, and survive intrusion attempts by insects, rodents, birds, etc.
There's only two things from Texas,
Steers and Queers.
Not sure you're going to have to go to these lengths, but one design I've seen used heat pipes going from the hottest chips up to a container of wax.
The idea is that the wax would solidify during the night and gradually melt during the day, keeping the electronics cool enough in the process.
It's quite a hassle, but in my opinion putting a fan on is a mistake. Lots of luck to you.
Selection of the wax is potentially problematic. Common or garden variety paraffin wax seems to melt over a wide temperature range. You want wax with a narrow range for melting, at a temperature where it'll keep your chips cool enough. (85C??)
Many IP cameras already have a micro SD card slot and can record video to the SD card in addition to streaming it offsite. A quick search on Amazon found one that is IP67 rated and has temperature ratings from -40c to +60c for $86, Dahua IPC-HDBW4431R-ZS. Iâ(TM)ve used Dahua cameras before and their optics and image sensors are great but their network security is lousy. Keep them on an isolated VLAN and donâ(TM)t let them connect outbound to the Internet.
Cool! (or Hot!, I suppose)
So is that typical, or is TI an exception?
I know I still love my TI-85 calculator - I must have a first-year 1992 model as I think of the timing. That sucker has taken almost 30 years of abuse and is still going strong. I just have to remember to replace the AAs every couple of years when I cant crank the screen darkness high enough any more.
Anyway - I recall them being quite proud of the durability of the thing in their marketing, and could easily see them being an outlier in the "durable-products" category.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
The solution is actually pretty obvious. Insulation.
If it needs extreme temperatures (let's say outside in Alaska and outside in California/Las Vegas/Texas) the most reasonable thing would be to apply some heatsinks to the CPU, RAM, etc.
The main concern would be the colder side, as the cold side might actually crack the chips or solder if the board expands, by which cause you might solve by routing any heat the board generates with a fan that circulates the air slowly, while reversing it to blow heat away if the board gets too warm. This requires some design considerations. If it's absolutely required to heat the board, you might actually put a peltier on the heatsinks to trigger one way or the other. This of course will burn more energy.
It's not even "your engineer should be telling you", it's "go look up industrial PCs and cases and decide which one fits your needs". This shouldn't even be on Slashdot, it's standard commercial gear you can get from any number of vendors, you just need to find whatever matches your requirements.
...and survive intrusion attempts by insects, rodents, birds, etc.
...and Texans. Might wanna make it bullet-resistant too. (I am not joking.)
I will be happy to bid on the project if you send me a request for quotation. Please be advised that such design effort is likely to cost quite a bit of gold pressed latinum.
Solution: Don't be a criminal piece of shit.
Don't speed. Don't commit felonies. Don't do stupid shit.
It's really, really fucking simple.
That is typical for quality suppliers. If your chips come from a random small fab in Shenzen, probably not.
Across most integrated circuits that are available from multiple vendors, you'll find high quality ones from TI, including automotive and milspec parts.
For ARM in particular, high quality is their thing top to bottom. At the bottom you have MSP432 which is a really cheap microcontroller, but with a better ADC and lower power usage than similar units from other manufacturers; for a few cents more, of course.
They don't make the cheapest or fastest parts, but they make the higher quality version of a broad range of standard parts.
The MSP432-EXP dev board is under $20 and comes with a decent hardware JTAG emulator/programmer that easily replaces an "entry level" one costing a few hundred bucks.
And they're OSS-friendly. Even though they push their own embedded IDE that uses a proprietary compiler, they also support GCC, and most of their stuff works will a regular open toolchain. I found a bug in the boot code for the GCC version of one of their firmware demos, and their support engineers found and provided a work-around in about 4 hours. And I'm just an independent developer!
All of those nice things said, they've rejected 100% of my sample requests. lol
Many manufacturers make ARM-based industrial-grade systems, e. g. the VIA AMOS 820 or the VIA AMOS 3003.
https://www.viatech.com/en/sys...
https://www.viatech.com/en/sys...
Is this gadget is attached to a post, driven into the ground?
Get down a metre or so, and the ground temperature shold be pretty stable. Put the gadget inside the pole, insulate the pole, and use it as a heat sink. For air circulation, if you need it, have extra black tube above the gadget. That will heat up, and the pole will 'draw' like a chimney. This may not solve all your problems, but it may get a few degrees.
According to my experience in road control systems and standalone embedded devices:
1. using a fan will provide you dust&dirt in huge quantities.
2. If you want to protect your device from heat, provide a reflecting shell-in-a-shell with thermal breaking envelope that assure you a permanent shadow.
3. About the heat generated by your system, reduce to 0 the unneeded processes and fine-tune your cpuspeed to suit your needs.
4. Displace the power supplly if possible is also a good idea.
5. In extreme scenarios, peltier cells can be explored.
Regards from spanish desserts
The US just does not have one... at least fitting the definition of a democratic government, by the people, for the people.
The US has an oligarchy of neocon-fascist corporations. And their employees aka lobbyists aka politicians are all you get to "vote" for.
Btw, democracy and the "free market" are mutually exclusive opposites, and I choose the former. Because the "free" in the latter means "the freedom from the freedoms of others". As in: human/worker/environmental rights. And the freedom to not uphold basic human social decency (like not letting somebody die or starve or sleep on the street) towards your fellow citizens. (I see it as my civic duty, and part of why I get to be proud of my country in the first place.)
Have you looked into something like a esp32? It generally has a better thermal range compared to ARM and I have seen cameras for it.
So it can claim efficient Atoms, it hid all the power wasting components in the accomanying chip. Which wad bigger and needed more cooling than the actual CPU.
All things actually included, ARM is still an order of magnitude more efficient.
Oh, and there are companies specializing on ARM systems for extreme/industrial envirobments.
Agree, all the data is out there. Seems to me someone thought of using an rPi or other hobby board and then realized it wouldn't handle the temperature and promptly asked on slashdot, as for having an "engineer" on board, I suspect rather strongly he IS the "engineer". So Texans need not worry about running any toll booths any time soon.
On a side note to the original poster, the reason fanless is chosen is that fans have a really short mean time to failure (depending on conditions, fan quality etc.) relying on them for cooling will require active monitoring of said fans / temperatures because if the fan fucks out the hardware is close behind. That means higher maintenance costs etc. If that was not enough, if the ambient temperature is higher than the rated temperature of the silicon it doesn't matter how many fans you pack in, it's going to die.
For shits and giggles I googled it for you BAE Systems RAD750 3U cPCI Single Board Computer
If it's good enough for Mars, it's should be good enough for Texas. Erm, it might require you to actually learn a programming language instead of cutting and pasting from stack overflow though. Good Luck!
There are three kinds of falsehood: the first is a 'fib,' the second is a downright lie, and the third is statistics.
A private road with a private toll booth. Sounds really plausible...
There are three kinds of falsehood: the first is a 'fib,' the second is a downright lie, and the third is statistics.
The raspberry pi 3 is thermally challenged at room temperature.
"Having the thermal throttle kicking in and not letting you have 200fps on whatever thing you're trying to emulate in RetroPi"
!=
"I won't be able to take a shot every now and then".
Also, the poster want to just take a shot every now or then of a street sign.
A Raspberry Pi Zero (lower consumption, cheaper) would to the job okay.
It's SoC is qualified for -40 to +85 (source), though you'll have to check if the other chip used (wireless functionality of Zero W for remoting, the chip used in the camera you plug into the MIPI connctor, the uSD card, whatever power solution) can sustain the necessary temperatures too.
Some parts (e.g.: uSD) should be available in industrial or even "consumer, high endurance" (cf. some Transcend parts) variant.
And all are dead cheap, meaning that in case of failure, it's simpler replace the box with a spare and then see later on if parts can be salvaged from the dead one.
Also, Raspberry Pi Foundation has guaranteed that parts will be kept in production for quite some time.
So if 5 years down the line, the project runs out of "spares", it would be trivial to source additionnal parts.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
For shits and giggles I googled it for you BAE Systems RAD750 3U cPCI Single Board Computer
Dude, that's a bit mean :-).
Check products from Olimex.com. They have done the legwork of sourcing all components in industrial ranges. For example, Allwinner A20 boards are available in industrial temperature ranges.
Other than that, seal everything from moisture. You don't necessarily need fans, but make sure there is ventilation and install the devices in shade to avoid direct sunlight.
Cheap, durable, tamper resistant and good heat propagation properties.
... you can fuck yourself.
You need the CPU units buried, with only sensors exposed.
I have a Zero W (with a stick-on SoC heatsink) in a box in my back yard that runs fine when the temperature in the box gets in the 50-60C range. Guess I could try it in a freezer to see how it does with cold.
"When information is power, privacy is freedom" - Jah-Wren Ryel
Send your requirements to me, I'll design and manufacture these for you.
...thick sleeves :-)
And theft resistant.
Could you have a Beowulf cluster of toll sign monitors?
Aah, change is good. -- Rafiki
Yeah, but it ain't easy. -- Simba
I've used NetBurner stuff before and have been satisfied. I don't know whether it's powerful enough for your needs, but it looks like they have a new ARM-based board now:
https://www.netburner.com/prod...
In the past, we just seal up the housings against the elements, using appropriate vent holes if need be. I agree with other posters, if this field is new to you, you may want to consult local engineers who do this sort of thing as there are gotchas and maybe even laws to consider.
Ambient temperature in the shade can reach above 100F easy.
A three foot deep hole in some of the areas we've already got three feet deep holes is a water well.
Even though this isn't a TXDOT project TXDOT is still involved and wouldn't approve something like that. They're pretty strict about enclosures.
The preceding post was not a Slashvertisement.
My company makes an arm device that is packaged, hardened, and certified for long term outdoor use in Texas. It's widely deployed at drilling sites. See here for details https://www.detechtion.com/hub... We are located in Houston and happy to talk in person if you're interested...
Just plunge your board in a solid resin:
https://www.abchimie.com/en/resins-potting/
I have seen such layers of protection in an old IOT style gas meter in France, the whole PCB was molded in some kind of transparent resin.
Check out the BBB Extended Temperature range ->> -40C to +85C
Plus a full Linux system for your development and deployment. I have used several for industrial apps with a custom daughter board with serial port connectivity.
This might help
https://www.sciencemag.org/news/2018/09/cooling-paint-drops-temperature-any-surface
We use imx6 industrial parts that are rated -40 to +95 C - that should be enough? Is the problem that you want something off the shelf? (We design our own solutions)
Is there anything that can go down past -40 (C or F, doesn't matter at -40)? Ie outside in saskatchewan? That's what I want to know. Cold killed my last Toughbook, and has taken out at least 2 laptops since.
GENERATION 26: The first time you see this, copy it into your sig on any forum and add 1 to the generation.
try eevblog.com forums instead
About 10 years ago I helped build a RPU for an RWIS systems that was installed across the state of Nebraska. We used custom SBC boards built by EMAC Inc. ( http://www.emacinc.com ) that were built with automotive spec tolerances (-45C to fucking hot). These were built for snow and ice control, so I mostly cared about the cold side of things.
We used a version of the IPAC9302 board with extended temperature range parts that were custom manufactured for our project. http://emacinc.com/products/pc_compatible_sbcs/iPac-9302 We attached those to a expansion board we designed and built in house. The complete system connected to a cellular modem to handle incoming SNMP requests as specified for NTCIP weather stations. For onsite monitoring we had a mix of RS422 serial ports and ADC ports that we used to collect environmental data and control a serial camera interface.
We also built sensors that we built for installation in the road surface in traffic areas. For these we used automotive grade SiLabs 8051 microcontrollers on custom boards. We encapsulated the sensors in high durability epoxy. If you do any kind of encapsulation, do not rely on ANY sort of mechanical interconnects---EVERYTHING MUST BE SOLDERED. Our early prototypes used snap in daughter boards, but we found that thermal expansion effects in the casting process as well as normal operation were opening an closing connections intermittently.
Be sure to look into current NTCIP standards for the sort of devices you are building.
Take a look at Phytec. They make ARM system on modules rated for -40 C to +85C operating temps. SoMs are small but theyâ(TM)re dev boards are quite large (but include lots of connectivity options).
Do not build when you can buy.
I have used Samsung XCover3 phone for about 6 years everyday, during heavy sporting, wet conditions, submerged into beer mugs countless times for demo, kicked around on the floor of pubs... and worked without a minute hiccup.
All of its electric components are filled up with plastic, so no moisture can get in.
Finally I replaced it by an xcover4, just because the USB recharge connector worn out.
Even in Texhoma, Tx, there ain't no -35F in Texas.
https://slashdot.org/comments....
Carry on then, you're doing great.
if the ambient temperature is higher than the rated temperature of the silicon it doesn't matter how many fans you pack in, it's going to die.
The silicon part of the chip won't have problems below 150C; other parts of the package will fail first unless it is a ceramic package or the temperature increase was caused by over-current.
There are 0 toll roads on the planet Venus. Zero (0). Where else would have outside free air temperatures that high, and also have enough gravity for a toll road to make sense?
That said, if the temperature inside the box without a fan would exceed them temperature limits of the [IC package] that does not imply that you can't use fans. When you turn the device on, it is not yet at the thermal steady-state maximum, and fans can reduce the rate of temperature increase. Assuming that you're on planet Earth and the box is mounted in regular outdoor desert air, the ambient air outside the box will always be well below the thermal limits of the package. Therefore, fans will always help. Generally for this type of thing you would wave a "fanless" CPU, and a multiple redundant case fans.
You need to measure the thermal resistance from the semiconductor junction all the way to ambient. Junction to IC case will be in the datasheet; as will the values for the heat sink. Then you calculate heat sink to ambient inside case at min and max air flow, and then inside case to outside. Now you should know exactly how many watts of power your CPU can use within the given temperature envelope.
If you do those calculations you find that if you don't properly ventilate the case, the ambient air temperature will climb to the failure point with even moderate CPU usage, but that low power systems do fine even in extreme environments without fans. What you want is a system that will survive without fans, but then to use case fans anyways because thermal stress reduces the lifespan of the parts. But water cooling is better because water pumps are more reliable than fans.
You could offload some of the heat with indirect shielding. It would be a step up from the 'painting it white' suggestion, you stop the direct heat from reaching the unit by putting (open) box around it. With an air gap (or vacuum) you then insulate the internals and can use some semi-active cooling like fans or a Peltier plate on the bottom for example.
A little extra engineering would get you a pretty reliable solution into some rather extreme environments.
Sorry I don't have specific off the shelf examples though
There were lots of good suggestions here, but few were actual pre-built in chassis systems - like the Intel ones I mentioned. This came from an online buddy, I found a few in-vehicle systems that were alright, but this one stands out. Not the exact form-factor I was looking for, but it will do.
https://www.rugged.com/a177-tw...
The preceding post was not a Slashvertisement.
you have said it right man kudos! Recruitment News Portal N-Power Jobs