[time-nuts] Temp/Humidity control systems?

Poul-Henning Kamp phk at phk.freebsd.dk
Thu Oct 27 09:28:43 EDT 2016

In message <20161027131226.0d5a72e62b2c91f2e13b637e at kinali.ch>, Attila Kinali w

>> Stop!
>> Over insulating is a 100% sure-fire way to get unstable temperature inside,
>> because it amplifies the consequences of any change in power dissipation.
>> It is a classic mistake to build a 100mm insulated enclosure inside an
>> office-like enviroment and end up having less stable temperature on
>> the inside than the outside.
>No! Don't Stop! :-)
>That's why the next section I wrote described how to get rid of the
>excess heat in a controlled fashio. Of course, simply just insulating
>the room without giving it a lot of thermal mass or any form of control
>of the heat production/exchange will not stabilize the room.

Attila,  I did read that.

You are right in a certain set of circumstances, for instance in a
already climate-regulated office space.

But your model is a recipe for grief and disaster if you do it in
a typical basement, on about 2/3 of the inhabited area of the planet
where you cannot ignore the humidity.

In general in a basement, it is a better idea to set up a closed
space and manage the humidity inside it, than it is to move the
humidity in and out of it.

WRT to humidity, that's a very unlinear relationship, but you can
get far by remembering these five rules:

* Warm air holds more *absolute* humidity (ie: g/m³) than colder air.

* At the same temperture, wetter air is lighter than dryer air.

* The temperature is never the same.

* Condensation should be avoided, period.

* You cannot "feel" absolute humidity, always measure it.

And yes, go ahead and look up number two, it's true.  Dehumidifiers
belong on shelves under the ceiling, not on the floor.

Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

More information about the time-nuts mailing list