Thanks for the replies:
I was ready to replace the pressure sensor (including 2 spares) with 
ones that come with their own weatherproof case and pigtail wires to 
eliminate the need for the epoxy encapsulation which is kind of minimal 
cross section, just enough  to encapsulate everything, with a small hole 
for atmospheric vent (it's a gauge pressure sensor) and a 1/4" brass 
close nipple for connection to pressure source. The sensor case is 
molded plastic, the coefficients of expansion may not be that different. 
But the unit returned to normal operation in what might expect time to 
evaporate any condensation. I don't discount the differential expansion, 
and even that it could of healed itself (temporarily). Following that 
line and replacing the sensors would not cure the problem of 
condensation. The easiest fix and least costly would be to pot the 
board. But what is interesting, it's only the one sensor that's an 
issue. The board is only 2.5" square, I would more think that it would 
be random between all the sensors, or just quit working. And that hasn't 
happened.

Not over enthusiastic about desiccant (silica gel), though it would 
basically work, and not that hard to do. Concern 5 years down the road, 
someone opens the aluminum box, the desiccant should be refreshed, and I 
don't think that's realistic.

Probably in the morning, I'll mix up a teaspoon of urethane and with an 
artist's brush paint the board with clear paint.

Danny Miller wrote:
> Mounting the pressure sensors in epoxy may be a problem.  With any temp 
> change or even the curing action of the epoxy it creates stress on the case.
>
> An absolute pressure sensor is a sealed reference chamber separated from 
> the outside port with a sealed diaphragm and very sensitive strain gauge 
> (gross oversimplification).  I suspect any mechnical stress which could 
> distort the case could put unexpected stress on the strain gauge which 
> would change the reading or even break something.
>
> Epoxy usually does not expand or contract much as it cures.  There is 
> always some thermal expansion coefficient but even if it is small for 
> your epoxy note that the stress comes from *differences* in expansion 
> coefficients.  If a PCB expands by a few mils and the encapsulating 
> agent does not, then there is stress between them.
>
> Danny
>
> Carl Denk wrote:
>   
>> I have a 18F1320 PIC with a temperature sensor and 2 pressure sensors 
>> mounted in a waterproof gasketed aluminum box. There are sealed 
>> penetrations for weatherproof connectors, except the power from a wall 
>> wart with the usual concentric connector on the bottom face of the box, 
>> and a LED and temperature sensor sticking through a tight fitting rubber 
>> grommets.The sensors are potted in epoxy coated with 3 part urethane 
>> paint. All this is mounted in an insulated enclosure with several backup 
>> heaters that maintains temperature above 40F (4.4C) minimum.
>>
>> This is twice now, that the one pressure sensor (Motorola Freescale 
>> MPX5700GP), after having the enclosure open for inspection for say 1/2 
>> hour in near freezing temperatures, that the pressure reading was what I 
>>   would think a maximum reading (110 PSI.). The sensor output is 0 - 5 
>> VDC, and appears to me that I'm getting a short from the 5VDC power 
>> supply for the sensor to the signal pin. The first time, I replaced the 
>> sensor immediately, but today, I waited a couple of hours, doing other 
>> chores, and when I checked, pressure reading was normal (30 PSI.). The 
>> board is thru the hole with solder mask double sided. After soldering, I 
>> took special attention to clean the board of flux. The board has no 
>> additional coating.
>>
>> I'm thinking it's condensation since the last time the aluminum box was 
>> open, was much warmer temperatures and might have been light rain. What 
>> would be appropriate actions- Coat most of the board with say a 
>> catalyzed urethane clear paint (have stock of lots of it), let the 
>> inside of the aluminum box open with cold low humidity ambient air (that 
>> won't help if I open later in the summer though)? or??
>>   
>>     
>
>   
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist