Heating an enclosure to -20°C

[Rigby]

People,

I have a project where I need to deploy a part that can only go down to -20°C and remain within spec.  I expect that the ambient temperature will go far below this limit.

I have planned to use a pwm pin on a micro, driving a MOSFET and powering a heat-sunk LED or power resistor and effectively creating a PID loop to keep the temp above -20°C.

Is this how this kind of thing should be done or is there a better way?

The enclosures will likely be watertight and airtight, if that is important.

Thanks!

[Rigby]

The enclosure will be maybe 1 cubic foot in volume, at the maximum, and will probably be uninsulated.

[timb]

You should really insulate the enclosure if you can. Your best bet would be to use a series of 1 to 3W power resistors of low ohms, mounted to the BOTTOM of the PCB. You could also use zig-zagging traces of sufficient width all over the bottom or an inner layer. Use a calculator to figure out the resistance per in/cm for your particular board's copper weight.

One resistor or LED is not going to do the job, especially uninsulated.


Sent from my Tablet

[wraper]

If the device is well insulated, there might be no need in heating at all if schematic consumes enough power itself. As of PWM, I think it is not the best choice as will create additional interference. Better to just switch on/off slowly (seconds). Depending on insulation, >2-3W of total power consumption might be completely enough to keep it above -20^oC. LED is not a good choice, much worse than resistors.

[Skimask]

Bring it over tonight!!!

[IanB]

The enclosure will be maybe 1 cubic foot in volume, at the maximum, and will probably be uninsulated.

Insulation, insulation, insulation. You need insulation before you go anywhere else.

Your enclosure will lose heat to the surroundings at a rate proportional to the temperature difference between the inside and the outside. If your enclosure is not insulated the constant of proportionality will be large and the rate of heat loss will be correspondingly large. Unless you have a mains power supply and a good powerful heater, you are likely going to be in trouble.

Of course, if your enclosure also has to work in the tropics at temperatures of 40°C then things are more complicated. But you you haven't mentioned anything about that.

[m12lrpv]

I asked a similar heating question here a little while back and was told to insulate so I did and to be honest it has made a huge difference and it doesn't mess with your circuit.

I also put in a number of switched power resistors to generate heat and that is giving me a lot of trouble due to interference with my signal processing so I gave up and switched it off.

[Rigby]

I think some part reselection is in order.

[tonyarkles]

What kind of environment will it be working in? Is this just outside in the cold? Or like someone mentioned, in a cryogenic environment?

One nice place to look for high/low temperature parts is Automotive-rated parts. They're often spec'd down to -40.

[Rigby]

It will be outside in Missouri.  So -40°C is unlikely, and -20°C is guaranteed.  Conversely, 40°C is likely in summer.

A better idea is to find a part that will brave -20°C without issue.  Insulation will not be easy.

[T3sl4co1l]

You may also have better luck heating critical components rather than the whole thing; an uninsulated enclosure more or less guarantees you're never going to get all points above -20C without a lot of work.

And, how much cold does it need to withstand?  -21C ambient is a lot different from, say, a satellite in space.

Tim

[Rigby]

Wolfram Alpha shows the coldest temp at one of the locations in 2013-2014 to be -32°C. 

So, I'm going to switch from the Spark.io Core to the Electric Imp, if research confirms my recollection that it can go to -40°C.  Otherwise I'll need to either insulate the existing enclosures or find new ones.

It is the TI CC3000 on the Spark that is only rated at -20°C.

[LabSpokane]

What's your power budget look like?  Do you have mains available or is this solar/battery?

Heated/fan cooled enclosures are off the shelf items from places like Lcom, Tessco and the like. If this is a onesy twosey deal that will be cheaper than your time on a commercial project.

Also, if you have access to an environmental chamber, try to test your gear. You might be surprised at the low end temps it will be reliable at.

[Rigby]

Two in illinois have mains available at the install sites.

Six in are going to be in Missouri, and will be solar powered only.

[Rigby]

Only the IMP-003 part is rated at -40°C and fortunately the data sheet includes schematics that even my stupid ass can follow and design a board for.

The part isn't available, though.

Going to reach out to a contact I have at Electric Imp and find out about stock delivery dates.  See how that goes.

[SeanB]

So insulate to allow internal heat to keep it alive, and add a small fan and a pair of fine mesh vents at the bottom, with one going to the top of the case via an insulated tube so heat can be moved out by forced flow, and put in a thermal switch to cool it on those hot days when sun power is plentiful and you would dump it in any case. The vents at the base mean you will not get convection, and the fine mesh keeps the bugs out. With the fan not running the air will not lose heat too much to the outside without the fan running.

[Dave Turner]

I'd be concerned about moisture penetration if the case is vented and the internal temperature cannot be kept above the condensation point.

[LabSpokane]

In Missouri, the case has to be vented.  Tightly sealed enclosures in the SE US can become ugly science experiments in short order. 

[Dave Turner]

LabSpokane - why's that?

[SL4P]

The assumption here is the circuit doesn't generate much heat in itself, and will be in a modest ambient environment at other times.

If the enclosure is ever in direct sunlight and no means of cooling on warmer days - you may have two problems to solve ... ?

[LabSpokane]

LabSpokane - why's that?

Humidity.  The midwest is both hot and very muggy (at least to me) in the summer.   And forgive me for calling it the SE, I had Mississippi stuck in my head for some reason. Missouri will have the same problem to a slightly lesser extent.  We used to have enclosures turn into their own miniature swamps.  Maybe GoreTex vents will be enough, but this is where you really need an environmental chamber.

For the cold test, I'd buy some dry ice and stuff the device into a cooler and see if it croaks at -20C.  I've run embedded PCs well below -20C and they ran fine.  Granted, there was no RF crystal involved.  I think a lot of these temp range specs are picked off boilerplate and really aren't actual device failure points (plus a safety margin).  I'd also consider scoping the crystal during the test and see what the thermal drift is really like.  One goofball kind of solution may be to let the receiver be at the same environmental temperatures, ensuring that both crystals are temperature-biased similarly.

[Rigby]

The other endpoints will be ambient temp, what ever that is on that day, and have no active heating or cooling other than the waste heat generated by the electronics inside.

I imagine that the electronics are derated for whatever reasons, and I should test.  I also imagine that I'm going to be asked to deploy uninsulated, airtight, with dessicants, and hope that there aren't too many bitterly cold days ahead.

We'll see how it goes.

I want to use some Electric Imp IMP-003 chips and design a board to do what I need and I'm not sure I have the chops.  The part doesnt look like it's going to be available until late February.  I know our design group would charge a minimum of five figures for the design service, though we do have our own PCB manufacturing facilities and whole design and assembly operations, so that would be a whole package, from-spec-to-delivered-working-boards price.

I'll update when I know more.

[SL4P]

...We'll see how it goes....

Kudos - a true engineer, not afraid to tell 'how it really is' sometimes!

[timb]

The CC3000 will work past -20. I've seen it used in a bunch of outdoor sensor arrays in north-eastern Canada. I've got a bunch of contacts with the Spark team if you want some data. (I helped man the Spark booth at Maker Faire last year.)


Sent from my Tablet

[Dave Turner]

Depending on the environment when the enclosure was sealed and the quality of the enclosure & penetrations humidity should not be a problem. Consider a hard disk.