On Jun 3, 2008, at 2:12 PM, Alexander Thom wrote:

> John Doty wrote:
>> On Jun 3, 2008, at 10:11 AM, der Mouse wrote:
>>
>>>>> Console based processing would have to work as usual, perhaps
>>>>> throwing a stderr / stdout warning that there might be out-dated
>>>>> footprints for certain files.
>>>> Unacceptable.  In a large project, such warnings are lost in the
>>>> spew.
>>> Then surely the right thing to do is clean up the spew!
>>
>> Yeah, except nobody ever really achieves that with make.
>>
>>>   Even if it
>>> means a grep -v to throws out the commands reported by make...or  
>>> using
>>> @ on the commands you don't want to see, if you can hack the  
>>> Makefile.
>>>
>>>> Some customers imagine hermetic packages improve reliability,
>>> And, don't they...for some environments?
>>
>> Maybe if you're operating your circuits in high pressure steam. But
>> in space? Nope. There's a widespread fervent *faith* that hermetic
>> packages are more reliable, but actual experience with operation of
>> electronics in space tell a different tale.
>
> Is it possible for you to elaborate on that? Can you point me to any
> reference material on the subject?

There is none that I know of. But there is also none that relates the  
pseudo-reliability that aerospace R&QA "experts" deal in to actual  
reliability in the real world. Test data doesn't count, because you  
can get any answer you wish by choosing the tests.

If you restrict design to hermetic parts you generally wind up  
engaging in a large number of poor engineering practices.

Good engineering chooses technology to match the goals and  
requirements of the project. The aerospace approach bypasses this,  
choosing technology by arbitrary criteria that are often unrelated to  
any requirement.

A well chosen part will have tight specifications, allowing the  
engineer to develop margins in the design to improve reliability.  
Available hermetic parts generally have relatively sloppy  
specifications, so margins suffer.

An inexpensive part that meets tight specs can only be made by a  
manufacturer that has its processes under tight, effective control.  
It is reasonable to expect that this should correlate with  
reliability. Sloppy specs and high prices are not an encouraging sign  
here, but that's what you get with hermetic parts.

Heat is a major cause of part unreliability. Available hermetic parts  
are generally less efficient and therefore run hotter than the best  
available parts.

A major hazard in a space launch is shock and vibration. Hermetic  
parts are more susceptible to this than plastic encapsulated parts.  
They are also heavier, which means they stress the boards they are  
mounted on.

One crazy idea is to build your prototypes with commercial parts and  
then switch to so-called "high reliability" parts for flight. Huh? In  
my experience, this almost always creates problems that, if you're  
lucky, you'll detect in testing. The "high reliability" parts don't  
behave predictably in real systems. Should that not make a thinking  
engineer suspicious?

Given all of these problems, and given the complete lack of evidence  
that "high reliability" parts are really more reliable than parts  
procured from Radio Shack, what should a rational person conclude  
when confronted with the truth that the only kind of "reliability"  
that actually matters is whether *system* meets its *specific*  
*requirements*?

How can anyone believe that torquing system design away from an  
implementation with low stresses and large margins improves  
reliability? Blind faith and ignorance...

I know of two missions that flew mixtures of commercial parts and  
aerospace parts. On ALEXIS, the hermetic, rad-hard aerospace memories  
failed, while thousands of commercial parts lasted for a decade  
without failure. The story on HETE-2 was similar: the only electronic  
subsystem to fail in six years of operation was a GPS receiver build  
to aerospace standards, while most of the spacecraft and its  
instruments used ordinary commercial parts. In that case, the GPS  
receiver was a power hog and ran very hot: it seems likely to me that  
shortened its life, and I have no doubt that its inefficiency was due  
to a poor choice of parts constrained by inappropriate standards.

> At my place of work we are severely
> constrained by IC technology selection criteria as specified by a  
> *true
> believer* in the ways of hermetic sealing.

The burden of proof should be on him, not you. He's the one asking  
for you to abandon good requirements-based engineering practice.

>
>>
>>> /~\ The ASCII                               der Mouse
>>> \ / Ribbon Campaign
>>>  X  Against HTML           [EMAIL PROTECTED]
>>> / \ Email!       7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B
>>>
>>>
>>> _______________________________________________
>>> geda-dev mailing list
>>> [email protected]
>>> http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev
>>
>> John Doty              Noqsi Aerospace, Ltd.
>> http://www.noqsi.com/
>> [EMAIL PROTECTED]
>>
>
>> ______________________________________________
>> geda-dev mailing list
>> [email protected]
>> http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev
>>
>
> Best regards
>
> Sandy Thom
>
>
> _______________________________________________
> geda-dev mailing list
> [email protected]
> http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
[EMAIL PROTECTED]




_______________________________________________
geda-dev mailing list
[email protected]
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

Reply via email to