[xray]

On the TV there was just a thing about fighting terrorists and
particularly nuclear devices.

One of the things in it was that a team had liquid nitrogen. They said
they didn't have info about what it was for, but sources said that it
was cold enough to freeze electronic devices.

So, is there any truth in that? I could see that batteries at that temp
might stop, but does really cold affect silicon circuits. Everything I
have heard makes me think silicon likes cold if it doesn't break wire
connections or such.

No biggie... just curious on your thoughts.

 

[Mikko Kiviranta]

One of the things in it was that a team had liquid nitrogen. They said
they didn't have info about what it was for, but sources said that it
was cold enough to freeze electronic devices.

So, is there any truth in that? I could see that batteries at that temp
might stop, but does really cold affect silicon circuits. Everything I
have heard makes me think silicon likes cold if it doesn't break wire
connections or such.

The silicon devices which rely on thermal excitation of charge
carriers to the conduction band tend to stop functioning at
low temperatures, including depletion-type MOS- and J-FETs. Even
though enhancement-type FETs keep functioning there may be too weakly
doped regions in the current path (like ohmic contacts) that
they stop functioning also. Typically the freeze-out temperature
is lower than 77K, though.

A complete circuit, including an integrated one, is typically driven
off its design margins (biases, offset voltages, timing...) when cooled
to 77K. Even though every active device in the circuit may keep
functioning, they no longer co-operate as designed.

Regards,
Mikko

 

[Deefoo]

On the TV there was just a thing about fighting terrorists and
particularly nuclear devices.

One of the things in it was that a team had liquid nitrogen. They said
they didn't have info about what it was for, but sources said that it
was cold enough to freeze electronic devices.

So, is there any truth in that? I could see that batteries at that temp
might stop, but does really cold affect silicon circuits. Everything I
have heard makes me think silicon likes cold if it doesn't break wire
connections or such.

No biggie... just curious on your thoughts.

Most silicon devices will stop working (reliably) at -40 degrees Celsius or
below, many wont even go that far. Storage temperatures should often not be
lower than -65 degrees Celsius. So yes, temperature is an issue.

--DF

 

[John Woodgate]

I read in sci.electronics.design that xray <[email protected]>

I could see that batteries at that temp might stop, but does really
cold affect silicon circuits. Everything I have heard makes me think
silicon likes cold if it doesn't break wire connections or such.

LN2 temperature would upset the working of many electronic devices. But
if one were deigning a device to go bang, and one knew that LN2 might be
used to disable it, it would be quite easy to arrange that the LN2 made
it go bang immediately.

 

[Jim Thompson]

[snip]

Most silicon devices will stop working (reliably) at -40 degrees Celsius or
below, many wont even go that far. Storage temperatures should often not be
lower than -65 degrees Celsius. So yes, temperature is an issue.

--DF

Nonsense!

It is standard operating procedure to design MIL devices for -55°C
operation.

And storage temperature is really of function of the packaging, not
the silicon.

...Jim Thompson

 

[Tim Wescott]

[snip]

Most silicon devices will stop working (reliably) at -40 degrees Celsius or
below, many wont even go that far. Storage temperatures should often not be
lower than -65 degrees Celsius. So yes, temperature is an issue.

--DF

Nonsense!

It is standard operating procedure to design MIL devices for -55°C
operation.

And storage temperature is really of function of the packaging, not
the silicon.

...Jim Thompson

Nonsense as stated, but practically sensible. While individual chips
may be designed for -55C operation (and I wouldn't count on it for
devices targeted at consumer devices) electronic _assemblies_ often fail
at cold. Digital circuits fail because CMOS tends to speed up at cold,
and each device speeds up differently. As a result a circuit which has
perfectly safe timing at 0 degrees or even -40 may fail at -55. Analog
circuits fail because everything changes; they usually try to operate
but fall out of tolerance in one way or another.

 

[Tim Wescott]

On the TV there was just a thing about fighting terrorists and
particularly nuclear devices.

One of the things in it was that a team had liquid nitrogen. They said
they didn't have info about what it was for, but sources said that it
was cold enough to freeze electronic devices.

So, is there any truth in that? I could see that batteries at that temp
might stop, but does really cold affect silicon circuits. Everything I
have heard makes me think silicon likes cold if it doesn't break wire
connections or such.

No biggie... just curious on your thoughts.

My comments on Jim Thompson's post notwithstanding there are devices
which are routinely designed to operate at 77K.

77K is a _very_ popular temperature for cooled infrared detectors, the
modern versions of which involve a focal plane of exotic semiconductor
diodes (like InSb, HgCdTe or GaAs with multiple quantum well diodes)
laminated to a fairly ordinary read-out circuit made of silicon. The
whole assembly is built into a dewar and cooled to 77K where the
electronics perk along quite happily.

The IR detecting diodes have such low bandgaps that at room temperature
they tend to go ohmic. The read-out electronics may or may not operate
at room temperature -- some are rated for room temperature, some have to
be cryogenic before you are allowed to turn them on. At any rate the
image tends to be really crappy unless you're quite close to the correct
operating temperature.

So you can design a chip, or even a system, that will operate from room
temperature down to 77K -- but you can't expect a consumer device to be
happy much below 0C, and even military devices must be carefully
designed to make it down to -55C without problems.

 

[Jim Thompson]

[snip]

Most silicon devices will stop working (reliably) at -40 degrees Celsius or
below, many wont even go that far. Storage temperatures should often not be
lower than -65 degrees Celsius. So yes, temperature is an issue.

--DF

Nonsense!

It is standard operating procedure to design MIL devices for -55°C
operation.

And storage temperature is really of function of the packaging, not
the silicon.

...Jim Thompson

Nonsense as stated, but practically sensible. While individual chips
may be designed for -55C operation (and I wouldn't count on it for
devices targeted at consumer devices) electronic _assemblies_ often fail
at cold. Digital circuits fail because CMOS tends to speed up at cold,
and each device speeds up differently. As a result a circuit which has
perfectly safe timing at 0 degrees or even -40 may fail at -55. Analog
circuits fail because everything changes; they usually try to operate
but fall out of tolerance in one way or another.

"...a circuit which has perfectly safe timing at 0 degrees or even -40
may fail at -55" sounds like bad design to me ;-)

Of course any temperature below ~70°F is too damn cold... it's cold
this morning in Phoenix, but it'll be 100°F by noon ;-)

...Jim Thompson

 

[Pat Ford]

[snip]

Most silicon devices will stop working (reliably) at -40 degrees Celsius or
below, many wont even go that far. Storage temperatures should often not be
lower than -65 degrees Celsius. So yes, temperature is an issue.

--DF

Nonsense!

It is standard operating procedure to design MIL devices for -55°C
operation.

And storage temperature is really of function of the packaging, not
the silicon.

...Jim Thompson

Nonsense as stated, but practically sensible. While individual chips
may be designed for -55C operation (and I wouldn't count on it for
devices targeted at consumer devices) electronic _assemblies_ often fail
at cold. Digital circuits fail because CMOS tends to speed up at cold,
and each device speeds up differently. As a result a circuit which has
perfectly safe timing at 0 degrees or even -40 may fail at -55. Analog
circuits fail because everything changes; they usually try to operate
but fall out of tolerance in one way or another.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

I wonder if the goal is more on the chemical side? does the detonator go
stable when that cold?
Pat

 

[John Larkin]

My comments on Jim Thompson's post notwithstanding there are devices
which are routinely designed to operate at 77K.

77K is a _very_ popular temperature for cooled infrared detectors, the
modern versions of which involve a focal plane of exotic semiconductor
diodes (like InSb, HgCdTe or GaAs with multiple quantum well diodes)
laminated to a fairly ordinary read-out circuit made of silicon. The
whole assembly is built into a dewar and cooled to 77K where the
electronics perk along quite happily.

The IR detecting diodes have such low bandgaps that at room temperature
they tend to go ohmic. The read-out electronics may or may not operate
at room temperature -- some are rated for room temperature, some have to
be cryogenic before you are allowed to turn them on. At any rate the
image tends to be really crappy unless you're quite close to the correct
operating temperature.

So you can design a chip, or even a system, that will operate from room
temperature down to 77K -- but you can't expect a consumer device to be
happy much below 0C, and even military devices must be carefully
designed to make it down to -55C without problems.

We routinely blast our prototype circuits with freeze spray,
especially FPGAs and other fast logic that may have timing margin
issues. That gets them down to about -55C, and we've never seen a part
quit working down there. We heat them, too, to see what margins are...
all logic families I know of slow down when it gets hot. We usually go
to 120C or timing failure, whichever comes first.

I guess some linear parts may crash cold... bias networks or whatever.
Jim may know.

I'm eager to get a real, big, LN2-boosted enviro chamber for serious
system testing, but currently we don't have space for it.

John

 

[Keith Williams]

We routinely blast our prototype circuits with freeze spray,
especially FPGAs and other fast logic that may have timing margin
issues. That gets them down to about -55C, and we've never seen a part
quit working down there. We heat them, too, to see what margins are...
all logic families I know of slow down when it gets hot. We usually go
to 120C or timing failure, whichever comes first.

Bipolar logic speeds up when it gets hot.

 

[Jim Thompson]

On Wed, 12 Oct 2005 09:50:21 -0700, John Larkin

[snip]

I guess some linear parts may crash cold... bias networks or whatever.
Jim may know. [snip]

John

None designed by me ;-)

I've been doing accurate temperature compensation since the early
'60's, so much so that my nickname at Motorola was "vbe Thompson" ;-)

...Jim Thompson

 

[Jim Thompson]

Bipolar logic speeds up when it gets hot.

[snip]

Beta certainly increases, but speed changes with temperature will
depend on logic type... properly designed PECL is pretty flat.

...Jim Thompson

 

[Keith Williams]

Bipolar logic speeds up when it gets hot.

[snip]

Beta certainly increases, but speed changes with temperature will
depend on logic type... properly designed PECL is pretty flat.

I'll disagree. IBM's ECL was pretty well designed[*] and was
significantly faster at its design temperature (85C) than cooler.
Enough so that thermal insulators were put on the cooler chips to
get the temperature *up*.

[*] unless you're begging the question and "well designed" == flat.

 

[[email protected]]

We routinely blast our prototype circuits with freeze spray,
especially FPGAs and other fast logic that may have timing margin
issues. That gets them down to about -55C, and we've never seen a part
quit working down there. We heat them, too, to see what margins are...
all logic families I know of slow down when it gets hot. We usually go
to 120C or timing failure, whichever comes first.

I guess some linear parts may crash cold... bias networks or whatever.
Jim may know.

I'm eager to get a real, big, LN2-boosted enviro chamber for serious
system testing, but currently we don't have space for it.

The last time I had free access to an enviromental testing chamber with
a decent sized port for running cables in and out, was at George Kent,
from 1973-76.

EMI Central Research must have had one somewhere, but we got by with
freezing spray and hot-air guns. Cambridge Instruments never bothered
to get one - and on one occasion I found myself relying on the way the
building heated up during the day to get a handle on the temperature
sensitivity of a crucial circuit.

Haffmans BV had two big enviromental test chambers, but they were used
for everything - including roasting silica gel to get it dry - so they
weren't all that accessible. We mostly used temperature controlled
baths when we wanted to look at the temperature behaviour of our
circuits. Wouldn't have worked for big boards.

 

[John Larkin]

Bipolar logic speeds up when it gets hot.

10K ecl and EclipsLite, on real boards, slow down a bit with
increasing temp, usually a ps or two per K per package. Some of that
may be pcb prop delay, I guess, but I always wind up compensating for
drift in that direction.

John

 

[Tim Wescott]

We routinely blast our prototype circuits with freeze spray,
especially FPGAs and other fast logic that may have timing margin
issues. That gets them down to about -55C, and we've never seen a part
quit working down there. We heat them, too, to see what margins are...
all logic families I know of slow down when it gets hot. We usually go
to 120C or timing failure, whichever comes first.

I guess some linear parts may crash cold... bias networks or whatever.
Jim may know.

I'm eager to get a real, big, LN2-boosted enviro chamber for serious
system testing, but currently we don't have space for it.

John

The place where I got my experience with cryo-cooled circuits has a
couple of such chambers. One of their suppliers is small and got some
covert snickers when they admitted to testing parts with dry ice dumped
in a styrofoam cooler.

Now they've moved and Engineering no longer has access to the chambers,
so they're thinking of going down to the local Baskin & Robbins for some
dry ice...

My point being that you can do some amazing things with a cheap cooler,
some nichrome wire and various frozen substances.

 

[Jim Thompson]

[snip]

Now they've moved and Engineering no longer has access to the chambers,
so they're thinking of going down to the local Baskin & Robbins for some
dry ice...

[snip]

Some Safeway outlets around here have dry ice.

...Jim Thompson

 

[John Larkin]

The place where I got my experience with cryo-cooled circuits has a
couple of such chambers. One of their suppliers is small and got some
covert snickers when they admitted to testing parts with dry ice dumped
in a styrofoam cooler.

Now they've moved and Engineering no longer has access to the chambers,
so they're thinking of going down to the local Baskin & Robbins for some
dry ice...

My point being that you can do some amazing things with a cheap cooler,
some nichrome wire and various frozen substances.

The real pain with going cold on the cheap is condensation. What I'd
really like is a LN2-cooled chamber, full of nice dry nitrogen. NOT
walk-in size, of course.

John

 

[martin griffith]

On Wed, 12 Oct 2005 11:46:25 -0700, in sci.electronics.design Tim

snip

The place where I got my experience with cryo-cooled circuits has a
couple of such chambers. One of their suppliers is small and got some
covert snickers when they admitted to testing parts with dry ice dumped
in a styrofoam cooler.

Now they've moved and Engineering no longer has access to the chambers,
so they're thinking of going down to the local Baskin & Robbins for some
dry ice...

My point being that you can do some amazing things with a cheap cooler,
some nichrome wire and various frozen substances.

I seem to remember a trick with dry ice and alcohol?
What temp can you get down to with it?



martin