SX at low temperatures

Kroum Kouzmanov

I am using SX52BD/PQ at 30 MHz external clock and power supply +5 V. It is programmed in HS3, with brown out level 4.2 V and DRT timeout 60 ms. When the ambient temperature decreases below (- 5) degr. C the processor does not start. The clock is running normally, the MCLR is tied to V supply and the V supply rising time is about 0.9 ms.
Please advise on the ways I could get the processor running at these low temperatures (down to -20 degr. C). Are there any special hardware/software requirements?
Thanks

Bean

Does the processor start if you pull MCLR low for a short period of time ?
If yes, you may need a reset circuit that holds MCLR low until the supply is stable.
Maybe nothing more than a cap to ground and a high value resistor to B+ on the MCLR line.
I would try different values, but maybe a 10uF and a 100K resistor ?
Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Check out· the "SX-Video Display Module"

www.sxvm.com
·

Jonb

Who knows what sub zero temperatures could do to components. Every component in the system could fail in mysterious ways. Make sure all your components are rated for -40C. Most Military spec. components can withstand such temperatures, but can occasionally consume additional power.

Post Edited (Jonb) : 3/7/2005 12:00:53 AM GMT

SteveW

Or, if you've got the power budget, throw away some power and heat the PCB when the temperature gets low. (I do this to keep LCDs working at low temps, without having to buy the whizzy low-temp ones)
It doesn't take a lot of power to raise a tiny PCB a few degrees, if it's insulated - say, in a block of expanded polystyrene.
Steve

allanlane5

SteveW -- what's the best (most efficient, easiest, lowest current) way to heat circuits for·low temperature use (-10 degrees Fahrenheit worst case)? I've seen people put in a 10 watt light-bulb in a box, just for the heating effects -- but that seems excessive for a battery powered circuit.

SteveW

If I'm just heating a PCB, then a scattering of (biggish) surface mount resistors does the job nicely. Getting heat into an LCD is harder, and I sometimes resort to an aluminum heatsink behind it, with a resistor rivetted to it. Whatever works in a design, really.
Unless your board is really low power, you might be able to do someting really crude, like power it up, wait 30 seconds for the silicon to warm up a bit, then hit reset again...(or hold reset for 30 seconds). You don't need the whole package, or even the whole die, to be up to temperature, just the active bits... Silicon doesn't conduct heat particularly well.
Note that, if you're playing games like this, the electrolytic capacitors will be out of spec, and damn near useless, and batteries will need looking after as well. All in all, if you can afford (or find) devices rated for your conditions, I'd recommend making your life easier... These hacks are, well, hacky [noparse]:)[/noparse]
Didgital silicon is generally pretty immune to low temperatures, it just gets faster, and slew rates get faster (more EMC, reflections, all the usual suspects). Voltage references will drift out of spec (hence power supplies) with cold, and capacitors will fail, but a digital circuit that cares about -10oC is likely hto have been cutting it very close in the first place. (LCDs, and to a lesser extent LEDs, are a different matter - the cold really slows LCDs down, and LEDs change colour [noparse]:)[/noparse]
If you're expecting to encounter sub-zero, you ought to consider hermetically sealing your box, to stop ice forming and doing damage.

Steve

Kroum Kouzmanov

Thank you, Bean.

I've got the processor running at -20C but I still have a problem: It takes up to 1.8 s to get out of reset and execute a simple pin-toggling loop. This delay depends on the temperature: it tends to increase sharply below -10C. I tried to manage with that situation but was unsuccessful and now I guess it may be due to one of the following:
- the internal rc_clk (fig. 12-1 in the Data Sheet) has a long start-up time at low temperatures (meanwhile that parameter is not listed in the specs, and also is that rc_clk for the DRT the same as the internal RC oscillator clock);
- I've missed something in the init program so the processor is unstable (going out of the reset and then resetting itself) until it heats up a little.

Any experience or suggestions?
Thanks