Searching \ for '[EE] Designing for low temperature' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/io/sensors.htm?key=temperature
Search entire site for: 'Designing for low temperature'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Designing for low temperature'
2011\10\14@184305 by Josh Koffman

face picon face
Hi all.

I had an idea for a project that will be subjected to some pretty low
operating temperatures (say down to -20C). I know there are some
tricks when using through hole components (ie leaving a bit of extra
lead to act as a flex point when things contract at different rates),
but I don't know much about using SMT in low temperatures. I would
guess that using as fat tracks as possible would be good to try to
avoid them getting broken. Would a metal clad PCB be an asset or
liability? I can think of arguments both ways. Any other tips? SOIC
would probably work size-wise (and have the largest leads), but 0603
resistors have zero flex.

Anyway, it's just an idea for the moment, but I'm curious.

Thanks!

Josh
-- A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
        -Douglas Adams

2011\10\17@040728 by Peter

picon face
Josh Koffman <joshybear <at> gmail.com> writes:
> I had an idea for a project that will be subjected to some pretty low
> operating temperatures (say down to -20C). I know there are some
> tricks when using through hole components (ie leaving a bit of extra
> lead to act as a flex point when things contract at different rates),
> but I don't know much about using SMT in low temperatures. I would
> guess that using as fat tracks as possible would be good to try to
> avoid them getting broken. Would a metal clad PCB be an asset or
> liability? I can think of arguments both ways. Any other tips? SOIC
> would probably work size-wise (and have the largest leads), but 0603
> resistors have zero flex.

Extended temperature range modules use Alumina boards for that reason. The
thermal expansion coefficient is the same as for all alumina and ceramic based
parts (SMD resistors and most caps). Your most important concern is likely
complete dry packaging including probably void elimination (potting in silicon
with ultrasonication might work well). Any void will collect condensate and that
will expand and contract with thermal cycling eventually causing whatever it is
under to separate or crack. So one uses conformal coating to create a surface
that is mostly convex on the outside. You also probably want to limit the size
of each module as much as possible. Take a good look at a ready made hybrid
circuit, it comes close to what you need. I am not an expert in the field but I
recently read the relevant information for a project. The easyest way to deal
with cold it so avoid it. A PTC and a few resistors will provide enough heat to
keep the board above the dew point and make most special measures above
avoidable. The simplest and oldest way to deal with this seems to be to provide
a small incandescent lightbulb in the case, undervolted with a resistor. Life
expectancy is >10,000 hours (over 12 years) for an undervolted bulb, but it will
likely still be bright enough to serve as 'Power' indicator.

-- Peter

2011\10\17@063359 by alan.b.pearce

face picon face
> Hi all.
>
> I had an idea for a project that will be subjected to some pretty low
> operating temperatures (say down to -20C). I know there are some
> tricks when using through hole components (ie leaving a bit of extra
> lead to act as a flex point when things contract at different rates),
> but I don't know much about using SMT in low temperatures. I would
> guess that using as fat tracks as possible would be good to try to
> avoid them getting broken. Would a metal clad PCB be an asset or
> liability? I can think of arguments both ways. Any other tips? SOIC
> would probably work size-wise (and have the largest leads), but 0603
> resistors have zero flex.
>
> Anyway, it's just an idea for the moment, but I'm curious.

For space applications we need to use laminate that has thermal expansion characteristics that match the components. The biggest hassle is leadless chip carriers (microchips QFN packages for example). For this purpose we use polyimide laminate commonly, or other more expensive options that even better match the thermal expansion characteristics.


-- Scanned by iCritical.

2011\10\17@105331 by RussellMc

face picon face
Re
> For space applications we need to use laminate that has thermal expansion characteristics that match the components. The biggest hassle is leadless chip carriers (microchips QFN packages for example). For this purpose we use polyimide laminate commonly, or other more expensive options that even better match the thermal expansion characteristics.

________

Dupont Kapton = Polyimide.

          http://www2.dupont.com/Kapton/en_US/

'Nice" 25 page mechanical properties document

  http://www2.dupont.com/Kapton/en_US/assets/downloads/pdf/summaryofprop.pdf

Data sheet - 6 pages

     http://www2.dupont.com/Kapton/en_US/assets/downloads/pdf/Gen_Specs.pdf

Trademark usage.
Technical content free, but gives a good feel of how you may and may
not use or imitate a trademark

         http://www2.dupont.com/Kapton/en_US/assets/downloads/pdf/Kapton_TrademarkGuidelines.pdf



         Russell McMahon

2011\10\19@233048 by K S
picon face
> expectancy is >10,000 hours (over 12 years) for an undervolted bulb, but it will

10000 hours = 416.6 days only.

(I just happened to do this calc. about an hour ago to determine a
capacitor lifetime hence picked up on it)

2011\10\20@031204 by Rupert Swarbrick

picon face
part 1 472 bytes content-type:text/plain; charset="utf-8" (decoded base64)

K S <spam_OUTkreshoTakeThisOuTspamgmail.com> writes:
>> expectancy is >10,000 hours (over 12 years) for an undervolted bulb, but it will
>
> 10000 hours = 416.6 days only.
>
> (I just happened to do this calc. about an hour ago to determine a
> capacitor lifetime hence picked up on it).

There may be a factor of ten in the calculation. If I've got it right,
12 years comes out as 105192 hours (ie ∼100,000). Maybe there was a
typo?

Rupert

part 2 181 bytes content-type:text/plain; name="ATT00001.txt"
(decoded base64)

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

More... (looser matching)
- Last day of these posts
- In 2011 , 2012 only
- Today
- New search...