[eagle] Re: Revised Module Suggestion

Dick Jansson-rr rjansson at cfl.rr.com
Mon Oct 15 10:40:21 PDT 2007

Thank you for your comments.
The truly "high power" modules, such as transmitters with power dissipating
devices of many Watts will still need to be constructed using the E05 01
(1100 alloy aluminum) Heat Sink, which will in turn be bolted directly to
the spaceframe panel (Y side). 
The intermediate power modules, such as the URx with the small internal heat
sinks could be special variation of the shown milled baseplate. With such a
provision, however, you would not have the latitude to shift the location of
the heat sink as you have done already without changing the mechanical
dimensions of the machined part, thus calling for another special part. It
also raises issues of the precision of locating such heat sink pads. I like
the concept, however, if it can be worked out, but it will require some
pretty good dimensional control.
Your concept of having the entire PCB rest on a thermal plate raises many
issues. Among them is the fact that you very often have traces on the bottom
side, placing them on the aluminum would not be very "cool" electrically
speaking. This issue also raises other issues, such as flatness. There are
Juan's flatness and then there is Dick's thermal flatness and these two
worlds are probably very much different. Any kind of protrusion, traces or
solder lumps or vias will take such a surface quickly out of the class of
being a thermal surface into a Juan surface. Any non-purely-flat-surface
becomes non-acceptable as a thermal surface. That is why the heat sinks for
the URx are only in contact with specific copper-faced areas on the bottom
of the URx PCB.
Your thermal gap filler material offers also only limited help, even if it
is space-worthy. Space worthiness is a matter of the outgassing
characteristics of the material, and most of these kind of materials are not
space acceptable. The thermal characteristics of such materials may also be
pretty limited, when I compare them to some of the power-density heat flow
Compared to locally clamping a PCB to a bottom heat sink, your bridge idea
does not conduct much heat. We can very quickly get into a discussion of the
conductance values of such heat sinking methods and in those discussions the
bridge method runs out of steam rapidly. You certainly would not need top as
well as the bottom heat sinking of a locale on a PCB.
Thanks for the discussions.
Dick Jansson, KD1K
 <mailto:kd1k at amsat.org> kd1k at amsat.org 
 <mailto:kd1k at arrl.net> kd1k at arrl.net 
From: wa6htp at gmail.com [mailto:wa6htp at gmail.com] On Behalf Of Juan Rivera
Sent: Monday, 15 October, 2007 15.38
To: Dick Jansson-rr
Cc: Bob Davis; AMSAT Eagle
Subject: Re: [eagle] Revised Module Suggestion
That looks nice!  It appears to solve the issue of getting that front panel
at exactly 90 degrees to the baseplate and also increases the stiffness of
the baseplate.  Increasing the useful front panel space also eases the
problem of working around the CAN-Do PCB with all of the necessary I/O
Would it be possible to customize the baseplate for the few modules that
draw high power?  It would be nice to machine the baseplate and heat sinks
as one chunk of metal instead of the existing method of having several
individual heat sink pieces.  I would like to see the PCB laying flat on top
of the baseplate with milled cutouts to accommodate any devices attached to
the bottom side.  In a perfect world there would be no components on the
bottom and the PCB would make contact with the baseplate across the entire
surface.  Another possibility that might be worth considering would be the
ability to include "U" shaped heat sinks that would bridge over the top side
of hot components and attach to the baseplate through holes cut into the PCB
on either side of the component.  Thermal gap filler could allow room for
CTE mismatches so that the device isn't crushed. 
73,   Juan - WA6HTP

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://amsat.org/mailman/private/eagle/attachments/20071015/a0d0bd30/attachment.html

More information about the Eagle mailing list