Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: [EFM] Extended Temperature Optics


For semiconductor devices the "reference" specification has always been
junction temperature, Tj. After all, this is the temperature that the Si
experiences, and is what affects the MTBF.

The junction-to-case thermal resistance for the various packages used for
the device, THETAjc, is also specified.  A system designer can then combine
this with the power dissipated in the device, the ambient temperature, and
the THETAca (case-to-ambient) of the particular system hardware
configuration, to predict junction temperature, and verify it is in the
desired range.  Things get interesting when you add in derating factors in
order to get the MTBF higher.

Some semiconductor manufacturers may specify a case or ambient temperature
range, under certain assumed conditions, as a rough guideline to the
designer.  But these are only intended to be approximate values for
non-critical applications.  So it's not surprising that different
manufacturers will report these differently.  But they all should specify a
Tj range and a THETAjc as the basic fundamental specifications for the
device and package in question.

There should be Telcordia specifications for the temperature a line card can
expect to experience in an RT, as well as in a C.O.  Our committee members
that supply DSL equipment should be familiar with these, I would think.

Barry O'Mahony
Intel Labs
Hillsboro, OR, USA
tel: +1 (503) 264-8579

-----Original Message-----
From: FEffenberger@xxxxxxxxxxxxxxxxx [mailto:FEffenberger@xxxxxxxxxxxxxxxxx]
Sent: Wednesday, January 23, 2002 1:55 PM
To: stds-802-3-efm@xxxxxxxxxxxxxxxxxx
Subject: [EFM] Extended Temperature Optics

(forgive the wide distribution, but I'm sort of trawling for info...) 

I've done a preliminary search on the origin of the fabled "-40 to +85 C"
temperature range. 
This is what I've found.

Various semiconductor companies talk about these ranges.  Most often, they
are specified 
as ambient temperature (Ta).  In cases where serious power is being
dissipated, a de-rating 
curve or equation is supplied, and in some cases, certain air-flow rates are
suggested.  However, 
there is a lot of inconsistency in the industry.  For example, Allegro says
the following: 
Operating Temperature Range.
C =Commercial (0 C to +70 C)
S =Standard (-20 C to +85 C)
E =Extended automotive/industrial (-40 C to +85 C)
G =Extended industrial (-40 C to +105 C)
J =Extended industrial (-40 C to +115 C)
K =Industrial/military (-40 C to +125 C)
L =Automotive/military (-40 C to +150 C)
X =Special (i.e.,wafer probe at +25 C only)

While National Semiconductor says:
In general, the temperature ranges are defined as follows:
Military: -55 to +125'C
Industrial -40 to +85'C (some variation, check datasheet)
Commercial 0 to +70'C
Automotive -40 to +125'C (some variation, check datasheet)
Note that in this case, even the specification indicates that it is not

I found two standards that talk about temperature:

There is the Mil Spec: MIL-STD-883E 1012.1
This talks about either Case or Mounting surface temperature, and NOT
(In fact, they give a diagram showing how you should build a liquid cooled
heat sink to 
clamp the case temperature to the tested value.)  This document specifies a
wide range 
of temperatures, including -55 to +85 and -55 to +125.  It seems the
military is 
concerned about attacking cold places (Siberia?) as well as hot.  

There is also JEDEC A105-B "Power and Temperature Cycling"
This talks about testing such that "the entire mass of each device under
test to reach 
the specified temperature extremes".  This would seem to be a 'case'
And, it requires the power to be turned on and off during the test (kind of
like a soak test).  
However, this test is considered destructive and is for device
Specifies temperature ranges of -40 to +85 and -40 to +125 C.  

That is what I've found so far.  I can't say that this is very definitive,
and I looks to 
me like the electronics industry has avoided this issue.  I invite anybody
who has 
any other information to please contribute it to the group.  

I'll save the rest of my comments for the optics exploder. 
Frank Effenberger