Chip operating temperatures

tom cloud cloud at hagar.ph.utexas.edu
Tue Sep 24 20:47:05 GMT 1996


>At 12:15 PM 9/24/96 EDT, you wrote:
>
>>But what about temperatures below zero degrees? H

        [ snip ]

>about...  Most chips will work outside their temp range, but how well is
>another story.  Most analog parts simply wont meet published specs (but when
>do they?) like offset, slew rate, etc.  Most digital stuff still works, but

Precisely, but lets understand:  a digital part is just an analog circuit
in saturation (unless you're talking ECL, in which case it's analog stuff
that's NOT in saturation).  So, the analog parameters that shift cause the
degradation of the "digital" performance.  [BTW I have done analog and
digital design for a long time -- and I find that the parts most likely
to be on the ragged edge of specs are the digital ones -- esp. from
certain (smaller) mfgrs.]

>the timing parameters must be relaxed.  Every now and then you can find a
>part that wont work in the cold, but often it can be replaced (w/ another
>identical part).  Often the automative and comercial temp ranges are the
>same part, but one is tested and guaranteed to the more extreme temps.  The
>only real difference I can tell (from practical experience) is the
>packaging.  The Mil spec stuff is almost always ceramic which may take the
>temp cycling better, but the comercial and auto stuff appear to be in the
>same plastic package (hope it doesn't crack at low temp).

Typically, have three temp / performance ranges: commercial, industrial,
and military.  Price increases exponentially.  Military usually has best
specs, most tolerance for supply voltage and temp variation, etc.  Usually
has better plating on pins and is in ceramic header since epoxy softens
and melts at the upper end of the mil-spec temp range.  Also, can get
better moisture sealing with ceramic.  [One of the failure modes is
moisture drawn into a hot chip as it cools down -- most pronounced
with the plastic packages.]  If you'll look at the "storage"
specs you'll see that they can all be stored at high temps -- so,
it's the packaging and whether or not the device goes out of spec
that determines its rating.  At high temps, the carriers tend to migrate
across the junctions producing *permanently* degraded performance.  All
the cold will do is *temporarily* push the operating region out of
spec. -- make it not work right (i.e. you could be walking).

Bottom line: heat destroys; cold just makes it not work -- just like the
rest of the stuff in your car.

>  In a nutshell:  If you derate the comercial parts they often work well
>outside their spec temp range.  
>SMD  
>
>Remember that your interior can get VERY hot (when your away w/ the windows
>rolled up) so dont assume your EFI computer will never have to work at 140F+
>ambient if its in the passenger compartment (still want to be able to start
>your car, right? ;)

Be sure to have adequate heat sinking to the enclosure to get the heat
out of its interior.

>  Oh, and the way to keep from replacing the radiator because you forgot to
>add antifreeze is to use an air cooled engine;)  (Im a 911 owner)





More information about the Diy_efi mailing list