I’m working for a wind turbine manufacturer, and I want to set up a super-robust data acquisition system. Basically it needs to:
receive about 200 channels, single precision, 10-50 Hz through wired network
store data for some time (I need about 128 GB storage ideally)
process it into secondary (much smaller) data summaries
and then the data summaries will be transferred over the wind farm network.
Although it will be installed inside the turbine, conditions are still harsh:
Temperatures may range from about -20 to +70 C
condensation/humidity may be an issue
offshore use is also planned so salinity could be an issue.
Lifetime should be long, say 10 years, with no/very few manual interference. I want to have about 2000 units in the next year.
What do you think, will the BBB be an option for this task? Is airtight (and pressure resistant) casing available to avoid condensation/humidity/salinity issues? Is a micro SD card robust enough when inside this airtight casing?
If the BBB is too hobbyist for my purpose, do you know a more robust system that I could use?
I'm working for a wind turbine manufacturer, and I want to set up a super-robust data acquisition system. Basically it needs to:
Fun!
- receive about 200 channels, single precision, 10-50 Hz through wired network
- store data for some time (I need about 128 GB storage ideally)
- process it into secondary (much smaller) data summaries
- and then the data summaries will be transferred over the wind farm network.
The analogue I/O is all up to you. The data rates are not a problem, nor is the networking.
Achilles heel will be storage. YMMV according to your write frequencies/rates.
SD frames are (IMHO) dreadful for anything serious. And I'd be certain of SD media failure over that time frame.
Suitably housed USB/Network attached SSD an improvement but still very tricky over a decade or so.
Re comparative robustness - that depends. There are vendors selling AM335X SOMs (including RJ/USB) with better temp range specs.
Presumably these things are going to be properly housed even if they are inside the nacelle?
I guess the advantage here is that unit cost is not too much of an issue? How big are the turbines?
This will prove to be an interesting challenge for you.To handle your environments, you will likely need to enclose whatever board
you choose in an IP67 or Nema 4 sealed enclosure. I don’t think potting
or conformal coating will suffice if you have salt and/or potential for condensing
humidity. Serviceable contacts will eventually corrode (Power, Enet, uSD, etc…)
So when you enclosure your board you will have the potential to exceed the 70*C
inside the enclosure. Ambient plus the thermal load of the board.
So, unless you use some form of active cooling transfer - not sure that will help with
the condensing humidity or the saline. If you use thermo-electric (peltier) you damn
better make sure you get 100% sealing or you will condense on your internal heat(cool)
sink and eventually your enclosure will fill (been there done that). If you are not 100%
sealed, thermal cycling will pull moisture in - condense on the heat sink and fill the enclosure
Until something if not everything fails.
So you will likely need industrial (+85) if not automotive (+105) temp components - probably will need
testing to determine which will be needed.
Next, I haven’t heard of 128GB uSD cards being supported on BBB. I’m sure the community will chime
in on that one. So you will likely need a bank of FLASH memory. Or create a process that will fill what
you have/put on the board and periodically download/stream to a server or the like.
The IP67 or Nema 4 case will aid in longevity and reliability except in the regards of heat.
Use the appropriate connectors to get you network and power inside and all should be good.
There are likely other systems out there that can potentially handle this.
You’ll just have to do research.
Alternately CircuitCo or some other companies can do a custom version of the BBB that
can meet your needs.
I could even do this as I’ve done an industrial temp version of the BBB customized for my
company’s needs - just not sure if I’ll have time to do it. It could potentially be modified
to meet your memory needs. So as long as you don’t need functioning in temps above
+85*C a variation of it could work for you.
If you decide you want/need a custom board and are not going to do it yourself, please give
CircuitCo the 1st chance at this since they do so much to support this community.
Speaking of sealed case, and heating issues. Why not fill the area where the BBB will sit with non conductive oil ? No idea if this is being done “professionally” but it’s been done with PC motherboards before . . .
USomIQ SOM can have 128GB of SLC NAND flash, but it will be rather expensive due to high price of such NAND chips. However, with SLC you will not care about data lost because Micron SLC NAND stands for more than 100k of write cycles into a single cell
Отправлено с iPad
13 июля 2015 г., в 20:54, William Hermans <yyrkoon@gmail.com> написал(а):
About the enclosure: I think I will go for a non-sealed solution. I’m not sure if IP67 sealed solutions including cable seals are available, and it sounds a lot more bothersome. In case it starts leaking I will get humidity issues. And like you say, I might get really high temperatures as well. Our current computers are not sealed airtight (but they are weather-coated), and that seems to work fine so far (2000 units installed, although admittedly, not offshore yet). Since it will be inside a cabinet inside a closed tower at maybe 20 m above sea level, I hope the salinity is not going to let it corrode too much. Not sure though if all weather coating is comparable?
I am wondering about the serviceable contacts, like you say. Least I can do is use a naturally ventilating case with power and ethernet openings on the down side. Should I insist on a special material for these connectors, such as copper? How about the wires? The USB opening on the upside should be covered by the casing, I shouldn’t need that one anyway. And I think the uSD will be a weak point, I prefer to not use that one and get 128GB flash on the board instead, if someone can deliver that. Periodic transfer to the server of the primary data is not really an option for reasons of logistics, data management, and network availability.
Thanks again for the comments, I’m happy to hear any further tips!
Look into conformal coatings such as Miller Stephenson MS-460H or MG Chemicals. Sealing all the connectors and sockets will be a problem, but you can get a silicone paste used in sealing connectors such as in irrigation systems. Check out the 3M Scotchlok connectors which we are testing now in fresh- and salt- water aquatic environments. I like the suggestion of immersing the whole board in some nonconductive, preferably hydrophobic fluid, maybe something on this list?
Lightning may be an issue. There are various ESD protective devices. We have used them on Ethernet connections and had good survivability of our (now obsolete) systems used in weather stations where a competitor’s (without such precautions) got taken out repeatedly. So consider where any wires enter the enclosure, protect them from ESD and surges, and make your enclosure a Faraday cage.
We’ve have systems we designed in field for more than ten years and still working. Some are used on thruway systems (I’m not sure exactly how), some are in dairy plants, etc. We got back one system a while ago - it had been (accidentally filled with milk which over time curdled and rotted and became conductive and eventually 120V in the system (we voted against having any mains power in the enclosure but were overruled) arced through the milk pudding and killed everything. This system had red 7-seg LED displays which must have been observed to get cloudy as this situation developed but no one did anything until it died. It was a stinky charred mess when we opened it up. They asked if it could be repaired… but I digress.
Heat is the enemy, any liquid paste electrolyte caps or batteries will die. You want all solid caps. Redundancy is good. Look at the aging data for components and don’t push their operation to the edge of voltage or temp and design to the worst limits of aging, temperature, power supply, etc.
Over the years I’d say 90% of system failures we see are the power supplies (usually customer-supplied for one reason or another), so if they are solid you should be good. We typically design local on-board regulators and fuses (sometimes sacrificial PC board traces feeding a heavy zener diode) and have lost very few systems even when power supplies have failed. Then we feed a group of boards with a higher voltage (12V or 24V more recently) DC supply which is locally down-converted where used. The raw supply can fade or misbehave a lot and still not take out the local boards.
On a good day you get what you pay for, so use good quality components designed for the task and you’ll be happy later when your phone doesn’t ring with in-field problems.
Next, I haven’t heard of 128GB uSD cards being supported on BBB. I’m sure the community will chime in on that one. So you will likely need a bank of FLASH memory. Or create a process that will fill what you have/put on the board and periodically download/stream to a server or the like.
One user reported that an 128GB sdxc card did work with the beaglebone black. Also I believe Robert Nelson tested a 64GB sdxc card: reporting that the higher speeds of the sdxc interface were not possible( due to the sd card physical interface on the beaglebone black), but it was in fact usable. At lower speeds.
Wind turbines work on high places that os very nice for an wifi network that could use one central 3g for upload/download data. In this case 32gb will be good for you. You will get 2 industrial 3g for 10 or more wind turbines.