I’m building a car power supply for my hand-held GPS. The voltage regulator I’m using is a LM317T, which, according to this (warning, PDF) spec sheet, has a temperature range of 0 to 125 C. Using it in the car, I can easily envision it going below 0C, in winter. Does the regulator just stop supplying power if that happens, or does something unpleasant (i.e. fried GPS) happen? I’m not worried about the high end of that range, since I am pretty sure that exceeds the GPS’s temp range, and it would fry anyway, regardless of what the regulator does. I’m only concerned about the low end of the regulator’s range. FTR, I’m using a LM317T, instead of a LM317AT (which goes down to -40 C) because I have one, so don’t have to buy something new.
Anyone know what happens when you go below the low temperature end?
It means the device has been qualified to operate within the published specifications over that temperature range. Outside that temperature range it may not meet some or all of the published specifications, or it may encounter a hard failure due to physical stresses.
Personally I wouldn’t sweat it, unless you’re planning on this for life-saving emergencies. I think the worst that will happen is the thermal stress will cause delamination in the package and crack the die or pop some bond wires.
Do it.
I can absolutely guarantee that you will have no problems.
The wider temperature range devices are tested at those temp extremes to make sure they meet the spec, while the consumer grade devices aren’t. But, they will work just fine at those moderate temperatures.
ETA: if you are really worried, stick it in your freezer and measure the voltage output.
There are in some cases modifications which need to be made to insure reliable start-up at sub-zero temperatures, especially with power supplies. The LM317T will probably work at the lower temperature range, but this can’t be guaranteed without extensive testing.
Like any voltage regulator, the LM317 has a built in voltage reference wth a non-zero TC. But I think you’ll find that it will be O.K. if you operate it below the lower temperature limit by 10 or 20 degrees.
You’ll need two resistors at a minimum. A better design would include bypass caps, protection diodes, and LC input filtering. You should also calculate the maximum power dissipated by the LM317 to see what (if any) heat sink you might need.
A lot of the graphs on the spec sheet show temperature on the x-axis so you can see which specs will be most affected by subzero soaks. A device that stops working at 0C/32F will be unusable in much of the country for several months a year.
I can’t help but wonder whether you can do better than an LM317. One thing that is certainly temperature dependent is the adjustment current, especially below 0. Since this current is usually supplied with resistors which themselves are temperature dependent (even worse with potentiometers), you may be in a situation with a large range of adjustment currents and thus, voltage outputs. This may be problematic in some applications. Unless your desired output voltage is unusual, I’d think you could find a suitable fixed regulator that will be more reliable.
In my experience, whenever I have stuck things in the freezer they’ve always worked well below their advertised spec. On the high end, when they overheat, things tend to fail much closer to spec. Also, once you lower the temperature to the point where the circuit does fail, they usually recover once they’ve had a chance to thaw out. Exceeding the high temp spec is much more likely to result in the device being permanently damaged.
That’s just my experience, and of course I’m sure the OP is aware that operating out of spec means you have to add a YMMV disclaimer and all of that.
Yes, I’m aware of the usual disclaimers for operating out of spec. I’ve built regulator based circuits before, I’ve just never had to concern myself with the low end of the temperature range before.
Even in this case, operating my hand-held GPS (for hikers and boaters type, not actually car type) in my car, I’m always having to change the batteries at least once on trips of any length, and at night, using it’s light really eats the batteries, so I had the idea of making a power supply that operates off car power. Car voltage levels are highly variable, so a simple resistor divider won’t do it, hence a regulator. Here’s a regulator in my junk box. What’s its specs? 0C low temp? Shouldn’t be a problem, you always turn the heat on when it’s even close to that cold. Oh, wait. Will it work when you first get in the car? Will it fry the GPS if you park to get dinner and the temperature drops? What if you leave it in there, when you park overnight?
Since I’ve never built anything that had to operate out of spec due to low temp, I have no experience to go by. Hence I asked here, to see if anyone here did. Thanks for the answers, all. Best suggestion award goes to beowulff, for the “put it in the freezer and measure to see what happens” idea.
The LM317AT and the LM317T are undoubtedly the same device, from a materials perspective.
What manufacturers commonly do is screen the devices differently at the test gate. So, package up a lot of parts, and screen them for LM317AT functionality. The passed parts are branded accordingly and the failed parts are screened again for LM317T and either branded or scrapped at that stage.
The test gate will use either a handler that does decrease the temperature, or they will have marginalized some timing, or power characteristics to mimic cold temperature testing.
In some cases thee manufacturer may have enough LM317AT parts in stock, and the next tested lot will only go through the LM317T test program, meaning that some parts may well be just as good as LM317AT parts, but just branded differently.
