Dick,
I know this is
an unpleasant subject and I apologize, but I don’t think the existing enclosure
is going to satisfy the real-world requirements for SMT PCB flex. I’m basing
this on several things. The first is a paper written by AVX (see below), one
of the largest surface mount capacitor manufacturers in the world. They recommend
a maximum PCB flex specification based on bend radius to prevent component
cracking. It works out to 0.0084” in any 1-inch segment. They also caution
against placing any components near mounting screw holes as they are areas of especially
high stress. As you know, this board has ceramic chip capacitors (the most brittle
components on the board) extremely close to the heat sink mounting hardware.
This imposes even more need for absolute rigidity and flatness across the entire
mounting surface. Given the current expectation to cold soak this Receiver
down to -60C it places almost impossible demands on the PCB, the components,
and the enclosure. Coefficient of thermal expansion mismatches between these
items is aggravated by this huge ∆T the receiver is expected to survive. I
have yet to find anyone who has experience with a payload that was expected to
survive repeated excursions down to -60C.
Next is my
experience with this prototype enclosure. Yesterday I disassembled the PCB and
heat sinks and took a few measurements. Here’s what I found:
|
Heat Sink No. |
Minimum Dim. |
Maximum Dim. |
|
1 |
0.183” |
0.186” |
|
2 |
0.181” |
0.185” |
|
3 |
0.185” |
0.189” |
As you can see,
the tolerances on the heat sink thickness are pretty bad. I’m not able to
measure the standoffs because of the material extending out the opposite side
of the base plate. I checked the parts list for this chassis and those
standoffs are not listed so I don’t know what the manufacturer’s tolerance is
on height.
I also found
that seven out of the ten PEM standoffs are loose and can be rotated with your
fingers. This raises a concern over grounding. But the main problem is that
the enclosure is flexible and warped. I think there are several reasons for
this. The sheet metal base might have been slightly bowed to start with and
then the large amount of work that is done to it (34 PEM standoffs and captive
nuts) might have aggravated the warp. And of course milling holes in the base
to gain access to the connectors might have contributed to this. I also notice
that the base becomes more warped after the front connector panel is attached.
I inspected this closely and found that some of the PEM captive nuts that are
applied from the bottom are not completely flat on the top surface. When the
connector plate is drawn down to the base plate the base plate is flexed over
those protruding PEM nuts causing more warp. This changes the dynamics of the
enclosure. Attaching the lid and finally mounting the enclosure to the
satellite frame will all cause the PCB mounting surface to shift, tilt, and
bend in unpredictable ways.
And lastly, we
may need to split the enclosure into two compartments with the CAN-Do module
and other switching power supply components in the front and analog circuitry
in the back. This may be necessary to resolve radiated EMI issues.
So there are my
concerns and findings. To me they all point to the need for a milled
enclosure. Such an enclosure would have a flat plate on top as a cover. A
milled enclosure would allow the power supply regulators to be mounted on the
inside surface of one of the walls, greatly improving and simplifying heat
dissipation. Most importantly, all of the heat sink and PCB mounting surfaces
would be part of the same block of aluminum as the enclosure itself and would
be very rigid and flat.
Here’s a link
to the AVX paper I mentioned à http://www.avxcorp.com/docs/techinfo/smzero.pdf
(see page 4)
73,
Juan – WA6HTP
From: Dick Jansson-rr
[mailto:[email protected]]
Sent: Saturday, July 28, 2007 5:26
AM
To: [email protected];
'John B. Stephensen'
Cc: 'David Smith'; 'Dave Black
(Work)'; 'Dave Black (Home)';
Subject: RE: [eagle] Receiver Spec
vs. ATP, a few Suggestions and a Question or Two
Juan:
From a practical aspect, the hanging of
electronics on the outside of an E05 20 module is rather implacable as there is
no place to put them. The thin metal cover is not a place as the heat would not
be spread out for dissipation, and the cover would not be “divorceable” from
the rest of the module. You would not want to place them on the front connector
bracket as you are already scrambling for space for connectors. Those are the
only areas that are open to the outside world.
We already have solutions for the
mounting of these electronics on the inside of the module, where they belong,
so I do not see where you otherwise plan to place these power conditioning
electronics?
Dick Jansson, KD1K
From:
[email protected] [mailto:[email protected]] On Behalf Of Juan Rivera
Sent: Saturday, 28 July, 2007
05.23
To: 'John B. Stephensen'
Cc: David Smith; Dave Black
(Work); Dave Black (Home);
Subject: [eagle] Receiver Spec vs.
ATP, a few Suggestions and a Question or Two
John,
I took a few minutes to look over your new specs and compare them
against the Acceptance Test Procedure. I've got a number of tests in the
ATP for which there are no specs:
·
Image rejection
·
Internally generated spurs
·
Local oscillator leakage
·
Input and Output VSWR
Items that need to be addressed that aren’t in either document:
·
EMI conducted susceptibility
·
EMI radiated susceptibility
·
Short and long-term frequency
stability
I also take issue with the -60C minimum power-down temperature. I
think this is unrealistic just from a CTE mismatch perspective. The
reliability of anything subjected to that wide a temperature spread is going to
suffer. A way must be found to raise that temperature.
I have a few thoughts... The CAN-Do switching step-down converter
is only supplying 11 milliamps. If we take a slight efficiency hit we
could just go to a simple linear regulator and completely eliminate the
radiated and conducted EMI emission problem from CAN-do. That eases the
EMI filtering and shielding requirements for every single payload. That
seems like a good trade-off to me.
Rather than worry about trying to conduct heat through a PCB why not
just go to externally mounted regulators for the CAN-Do and the Receiver right
on the case itself. We’re not that pressed for space. That eases
the heat sink and the associated thermal gap filler issues. EMI will
still be an issue, but only for external sources instead of one that is inside
the enclosure itself. That strikes me as a huge bonus. Why not
design a single-sided PCB with the regulators hanging over the edge and the
whole assembly, PCB and regulators, mounted directly to the enclosure.
Stick it in with the CAN-Do module in a separate cell. Filter all the
signal and power through the common bulkhead.
I’m also curious as to why you are specifying two different types of
SMA connectors.
73,
Juan – WA6HTP