"Jeff Liebermann" wrote in message
news
On Wed, 1 Sep 2010 02:40:53 +0100, "Arfa Daily"
wrote:

All of the processing power is in those two BGAs Jeff.
(...)

Thanks for the details. I really don't know anything about such
dedicated game machines. I just assumed that all such machines used
common processors to make development easier.

That's a LOT of processing power, needing a lot of amps to perform ...

I found the Kill-o-watt meter and stuffed it in line with my Dell
Optiplex 960 (E8500 3.2Ghz). 43 watts at idle, 70 watts max when
playing a DVD (not including LCD monitor). Speedfan 4.40 says 31C for
both CPU cores after about an hour. The one large fan is barely
spinning and very quiet (which is why I bought this one). When I set
the fan to run full speed, it's quite loud.

The fan on these things *is* large, as is the heatsinking assembly, and
when
the processor finally decides to ramp the fan up, it sounds like a vacuum
cleaner. For this reason, at idle they tend to run it at below what I
would
consider a 'sensible' minimum, exacerbating the thermal stresses on the
chips, their (lead-free) soldering, and the board to which they are
attached.

Well, theory suggests that the life of a semiconductor device is
greatly affected by the number of thermal cycles it experiences
(thermal fatigue). I don't know if this also applies to CPU's or
whatever is in those BGA chips (FPGA/GPU?), but might be something
else to worry about. I would guess(tm) that the large aluminum heat
sink would moderate any abrupt changes in temperature, thus making it
less of a concern. However, that might not be the case for the solder
balls supporting the BGA.



--
Jeff Liebermann


The soldering under the BGAs letting go, is the commonest problem with both
the Sony PS3 ( "yellow light of death") and the X-Box 360 ("red ring of
death")

Those names for the conditions refer to the behaviour of the front panel
indicator LEDs when the faults that result, show themselves.

I am quite convinced that the slow running of the fan at idle, is a major
contributory factor in the failing of the BGA soldering.

Arfa

In article BDDdo.2319$3p1.124@hurricane, Arfa Daily
writes

Just as a matter of interest Meat, what is your preferred brand and type of
heatsink goop when working with these very high power processors?

I'd recommend Arctic Silver.

I have always
resisted using this stuff, because it's so messy, and so hard to remove
unless you use the complementary cleaner

nah, standard IPA works fine. The trick is to use clean tissue wetted
with IPA, wiping just once or twice then replacing with a fresh piece,
repeating until the CPU is clean. If you go back with used tissue, you
just spread the AS about more.

If it makes that much mess, you're using too much. See the application
tips on the AS website. You literally only need a half-a-grain-of-rice
sized blob in the centre of the CPU heat spreader, it'll spread out by
itself with pressure from the heatsink. Note AS say it can take a few
heat/cool cycles to become fully effective, and indeed this is what I
have found.

, but if it really is that much more
effective

It is. A colleague at work was struggling to cool a CCD (a big one!)
without having to resort to cryogenic cooling. He was experimenting
with a Peltier cooler and unable to transfer heat away from the CCD fast
enough. I suggested he try replacing the standard white goop with AS
and he was astonished at the massive improvement in heat transfer.

--
(\__/)
(='.'=)
(")_(")

In article , Jeff Liebermann
writes

I don't believe it. The winner of the power hogging consumer CPU
contest was the DEC/Intel Alpha 21364 (EV79):

I herded a fleet of 21264s (AlphaServer DS10) for a while. Impressive
heatsinks in those. One is still in use today.

--
(\__/)
(='.'=)
(")_(")

In article , Arfa Daily
writes

Why is that ? You have electric kettles in your kitchens - I've used them.

They do, they just take three times longer to boil. They're nowhere
near as ubiquitous in American kitchens as they are in the UK.

--
(\__/)
(='.'=)
(")_(")

In article , Michael A.
Terrell writes

3/16" is between AWG 5 & AWG 4. 4 mm is between AWG 7 & AWG 6. How
much current do those kettles draw?

2kW and 3kW kettles are common. 2kW is ~8.3A, 3kW is ~12.5A. UK plugs
and sockets (=receptacles over the pond) are rated for 13A.

--
(\__/)
(='.'=)
(")_(")

In article , Arfa Daily
writes

exacerbating the thermal stresses on the
chips, their (lead-free) soldering, and the board to which they are
attached.

Which is what caused the 'red ring of death' on Microsoft's Xbox.

--
(\__/)
(='.'=)
(")_(")

Arfa Daily wrote:

"Michael A. Terrell" ? wrote in message
m...
?
? Arfa Daily wrote:
??
?? "Michael A. Terrell" ? wrote in message
?? m...
?? ?
?? ? Arfa Daily wrote:
?? ??
?? ?? Well Jim, that was why I used the word "potentially", but judging by
?? the
?? ?? size of the pins used to couple the power supply's output into the
?? ?? board -
?? ?? if you've been following the thread, you will recall that I previously
?? ?? described them as being of the size you would find on the line cord
?? for a
?? ?? kettle - then I wouldn't say that there was too much in the way of
?? ?? reserve.
?? ?
?? ? "The size you would find on the line cord for a kettle" doesn't have
?? ? much meaning in the US.
??
?? Why is that ? You have electric kettles in your kitchens - I've used
?? them.
?
?
? I've never seen one. Even Coffee pots are rare these days.
?
?
?? Or don't you call them kettles ?. OK, anyway, if it's a better
?? description,
?? the size of the round ground pin on a line cord that has a three pin
?? plug.
?? Is that more meaningful ? 3/16" diameter maybe ? 4mm ?
?
?
? 3/16" is between AWG 5 ? AWG 4. 4 mm is between AWG 7 ? AWG 6. How
? much current do those kettles draw?
?
?
? --

Typical UK kettle is 2 - 3kW so 8 to 12 amps or thereabouts. Now, I'm really
confused that you say that you've never seen one. How do you boil water for
a cup of tea, or a cup of instant coffee?


On the gas or electric stove. A lot of people heat the water in a
microwave. Fast and efficient.


Whenever I come to Florida, I
stay in a private rental home, and although some have had a kettle that
heats from a ring on the cooker, I'm sure that I have also stayed in homes
that had an electric version. Or maybe I'm mistaken on this? Perhaps with
your line power at only 110v at a non 3 phase outlet, the current levels are
impractical with an element powerful enough to heat the water in short
order. Here, every home - and I really mean *every home* - has one. It is a
known problem for the electricity grid controllers, when TV ads come on in
the middle of the popular soaps. Short term demand goes through the roof, as
everyone rushes out to make a cup of tea or coffee, at the same time. The
controllers genuinely have to know the advert schedules in the TV
programmes, and factor this into their load shedding operations.


The same thing with water demand when people rush to the bathroom
during a commercial.


--
Politicians should only get paid if the budget is balanced, and there is
enough left over to pay them.

"Mike Tomlinson" wrote in message
...
In article BDDdo.2319$3p1.124@hurricane, Arfa Daily
writes

Just as a matter of interest Meat, what is your preferred brand and type
of
heatsink goop when working with these very high power processors?

I'd recommend Arctic Silver.

I have always
resisted using this stuff, because it's so messy, and so hard to remove
unless you use the complementary cleaner

nah, standard IPA works fine. The trick is to use clean tissue wetted
with IPA, wiping just once or twice then replacing with a fresh piece,
repeating until the CPU is clean. If you go back with used tissue, you
just spread the AS about more.

If it makes that much mess, you're using too much. See the application
tips on the AS website. You literally only need a half-a-grain-of-rice
sized blob in the centre of the CPU heat spreader, it'll spread out by
itself with pressure from the heatsink. Note AS say it can take a few
heat/cool cycles to become fully effective, and indeed this is what I
have found.

, but if it really is that much more
effective

It is. A colleague at work was struggling to cool a CCD (a big one!)
without having to resort to cryogenic cooling. He was experimenting
with a Peltier cooler and unable to transfer heat away from the CCD fast
enough. I suggested he try replacing the standard white goop with AS
and he was astonished at the massive improvement in heat transfer.

--
(\__/)
(='.'=)
(")_(")



Yes indeed. This is kind of what I'm finding. I in fact use a vanishingly
small amount of AS which as you say is easy to remove with IPA, but I come
across devices that have been 'excessed' on the AS by other people, and it
is very messy to remove compared to white compound. Until I really got into
using the stuff, I was of the same misconceived notion about the quantity to
use, as others seem to be. I have always been sparing with compound - and I
use a lot of it as I repair many big amps for a living - but it is a fact
that a very thin translucent layer of white, is not effective enough on a
standard non-flatted device face, and heatsink contact area, whereas with
AS, it would appear that it is. These BGAs are the size of a large graphics
chip, and I apply a very thin line of AS across the face, and then spread it
using an old credit card, rather than hoping that it will spread out across
the whole face on its own. This negates the tedious disassembly and
reassembly required to get at the heatsinking faces if the cooling turns out
to not be adequate. So far, this seems to be working well.

Arfa

On Thu, 02 Sep 2010 02:59:25 -0400, Michael A. Terrell wrote:


Arfa Daily wrote:

"Michael A. Terrell" ? wrote in message
m... ?
? Arfa Daily wrote:
??
?? "Michael A. Terrell" ? wrote in message
?? m... ?? ?
?? ? Arfa Daily wrote:
?? ??
?? ?? Well Jim, that was why I used the word "potentially", but judging
by ?? the
?? ?? size of the pins used to couple the power supply's output into
the ?? ?? board -
?? ?? if you've been following the thread, you will recall that I
previously ?? ?? described them as being of the size you would find on
the line cord ?? for a
?? ?? kettle - then I wouldn't say that there was too much in the way
of ?? ?? reserve.
?? ?
?? ? "The size you would find on the line cord for a kettle" doesn't
have ?? ? much meaning in the US.
??
?? Why is that ? You have electric kettles in your kitchens - I've used
?? them.
?
?
? I've never seen one. Even Coffee pots are rare these days. ?
?
?? Or don't you call them kettles ?. OK, anyway, if it's a better ??
description,
?? the size of the round ground pin on a line cord that has a three pin
?? plug.
?? Is that more meaningful ? 3/16" diameter maybe ? 4mm ? ?
?
? 3/16" is between AWG 5 ? AWG 4. 4 mm is between AWG 7 ? AWG 6.
How ? much current do those kettles draw? ?
?
? --

Typical UK kettle is 2 - 3kW so 8 to 12 amps or thereabouts. Now, I'm
really confused that you say that you've never seen one. How do you
boil water for a cup of tea, or a cup of instant coffee?


On the gas or electric stove. A lot of people heat the water in a
microwave. Fast and efficient.

I've evolved into the single cup K style or K-Cup Keurig machine made by
Cuisinart. Mine turns on at 5:am water is ready to brew in 2 minutes.
Place k-cup in head (could be coffee, chi-latte, hot choc, Earl Grey,)
close head and hit brew. In 45 seconds you have 12 oz of your favorite
brew. I shelved my Cuisinart Grind and Brew conventional 12 cup machine
several months ago. You can buy k-cups filled with your favorite or use
the k-cup adapter and spoon in your favorite grind.

Oh and about heating water in the microwave. There is a phenomena called
hyper-boil that I'm sure you know about. Got to be careful



--
Live Fast, Die Young and Leave a Pretty Corpse

On 9/2/2010 6:54 AM, Meat Plow wrote:
Oh and about heating water in the microwave. There is a phenomena called
hyper-boil that I'm sure you know about. Got to be careful

http://www.snopes.com/science/microwave.asp

Yeah, but your odds of having this happen are about the same
as flashing your high beams at oncoming traffic and getting
killed as a result of a gang initiation.

Jeff

On Thu, 02 Sep 2010 08:03:49 -0500, Jeffrey Angus wrote:

On 9/2/2010 6:54 AM, Meat Plow wrote:
Oh and about heating water in the microwave. There is a phenomena
called hyper-boil that I'm sure you know about. Got to be careful

http://www.snopes.com/science/microwave.asp

Yeah, but your odds of having this happen are about the same as flashing
your high beams at oncoming traffic and getting killed as a result of a
gang initiation.


Some parts of the country that's a real possibility. And you wouldn't
even have to flash your lights.





--
Live Fast, Die Young and Leave a Pretty Corpse

Meat Plow wrote:

On Thu, 02 Sep 2010 02:59:25 -0400, Michael A. Terrell wrote:

Arfa Daily wrote:

"Michael A. Terrell" ? wrote in message
m... ?
? Arfa Daily wrote:
??
?? "Michael A. Terrell" ? wrote in message
?? m... ?? ?
?? ? Arfa Daily wrote:
?? ??
?? ?? Well Jim, that was why I used the word "potentially", but judging
by ?? the
?? ?? size of the pins used to couple the power supply's output into
the ?? ?? board -
?? ?? if you've been following the thread, you will recall that I
previously ?? ?? described them as being of the size you would find on
the line cord ?? for a
?? ?? kettle - then I wouldn't say that there was too much in the way
of ?? ?? reserve.
?? ?
?? ? "The size you would find on the line cord for a kettle" doesn't
have ?? ? much meaning in the US.
??
?? Why is that ? You have electric kettles in your kitchens - I've used
?? them.
?
?
? I've never seen one. Even Coffee pots are rare these days. ?
?
?? Or don't you call them kettles ?. OK, anyway, if it's a better ??
description,
?? the size of the round ground pin on a line cord that has a three pin
?? plug.
?? Is that more meaningful ? 3/16" diameter maybe ? 4mm ? ?
?
? 3/16" is between AWG 5 ? AWG 4. 4 mm is between AWG 7 ? AWG 6.
How ? much current do those kettles draw? ?
?
? --

Typical UK kettle is 2 - 3kW so 8 to 12 amps or thereabouts. Now, I'm
really confused that you say that you've never seen one. How do you
boil water for a cup of tea, or a cup of instant coffee?


On the gas or electric stove. A lot of people heat the water in a
microwave. Fast and efficient.

I've evolved into the single cup K style or K-Cup Keurig machine made by
Cuisinart. Mine turns on at 5:am water is ready to brew in 2 minutes.
Place k-cup in head (could be coffee, chi-latte, hot choc, Earl Grey,)
close head and hit brew. In 45 seconds you have 12 oz of your favorite
brew. I shelved my Cuisinart Grind and Brew conventional 12 cup machine
several months ago. You can buy k-cups filled with your favorite or use
the k-cup adapter and spoon in your favorite grind.

Oh and about heating water in the microwave. There is a phenomena called
hyper-boil that I'm sure you know about. Got to be careful


I let things sit for 30 seconds or more before I remove them from a
microwave.

I don't drink coffee, and I can't find the tea I like, except as a
concentrate. The price has doubled in the last year, so when i run out
of what I have, I doubt that I'll buy more.


--
Politicians should only get paid if the budget is balanced, and there is
enough left over to pay them.

On Thu, 2 Sep 2010 02:41:15 +0100, "Arfa Daily"
wrote:

The soldering under the BGAs letting go, is the commonest problem with both
the Sony PS3 ( "yellow light of death") and the X-Box 360 ("red ring of
death")

Those names for the conditions refer to the behaviour of the front panel
indicator LEDs when the faults that result, show themselves.

I am quite convinced that the slow running of the fan at idle, is a major
contributory factor in the failing of the BGA soldering.

On the other foot, I suspect that a high air flow fan will make it
worse. The problem is NOT that the BGA is flexing with increasing
temperatures. It's that the PCB underneath the BGA is flexing.
Stabilizing the temperature of the BGA is probably useful, but unless
the PCB is also stabilized, it will bend, bulge, buckle, twist, or
otherwise go through various contortions trying deal with the
temperature difference between the BGA and the PCB. If the
differential temperature is large enough, the PCB may bulge enough to
tear way from the BGA. Again, the BGA does not move, the PCB does.

Now, add a high air flow fan into the picture and we have a larger
temperature differential. The air flow will probably do a fair job of
cooling the PCB because of the comparatively smaller mass of the PCB.
The thermal conductivity of G10/FR4 isn't all that wonderful,
resulting in a localized hot spot. With a larger difference between
the BGA area and the surrounding PCB, the result is a larger PCB bulge
with PCB air cooling. I've seen PCB's (usually motherboards) with
permanent bulges under BGA's from this effect.

For entertainment, take any PCB, heat it in the middle with a heat
gun, and watch the bulge form. It's that bulge that's ripping the
BGA's apart. Extra credit to laptop manufacturers, that add heat
sinks to the BGA, and then mechanically connects the heat sink to the
frame. When the board bends, it will literally tear the BGA off the
PCB, since the heat sink can't move with the board.

In the instructions for hot air reflowing of BGA's, there's usually a
section on pre-heating and slow cool down of the PCB. The idea is to
not tear the BGA ball apart from differential thermal expansion
between the large thermal mass of the BGA and the comparatively
smaller mass of the PCB. It's exactly like moving a solder connection
while it's cooling. You get a "cold" solder joint.

Incidentally, I once designed a 150 watt 2-30Mhz HF power amplifier.
After about a year of normal use, we started seeing failures caused by
the power transistor screws coming loose. Suspecting cold flow, I
worked on improving the grip with stainless inserts. This worked,
but now produced failures in the ceramic case power transistors. The
clue was when a PA module arrived with all the ceramic lids popped off
the transistors, but still working. Weird(tm).

After a dozen bad guesses, I determined that PCB expansion and
contraction was initially causing the loose screws. When the screws
were properly secured, the next weakest link was ripping the leads out
of the power transistor case, causing the glued lid to pop off. The
problem was solved by slightly pre-bending the power transistor leads
in a fixture so that PCB thermal expansion would be absorbed by the
bends. I still do this even on TO220 packages, which can have the
same problem. Too bad it can't be done with BGA packages.




--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Thu, 02 Sep 2010 09:54:44 -0700, Jeff Liebermann
wrote:

The thermal conductivity of G10/FR4 isn't all that wonderful,
(...)

Ok, let's do the numbers. The coefficient of thermal expansion for
G10/FR4 is:
1*10^-5 cm/cm/C
That means a 1 cm long piece of G10/FR4, will expand 1*10^-5 cm for
every degree C of temperature differential.

So, we have a big fat BGA chip, that's about 5cm across. It's running
hot with a bottom temperature of about 80C. Assuming the PCB is
running at room temp of 25C, that's a 55C differential temperature.
Over the diameter of the BGA, that's
125*10^-5 cm
movement of the PCB.

Solder balls come in all manner of sizes, but my guess(tm) that for a
1mm pitch BGA, a 0.4mm ball is appropriate. When soldered, the ball
will remain about the same diameter, but the height will be reduced to
about 0.1mm.

The angle that the ball moves over temperature is:
angle = arctan ( 125*10^-5 cm / 0.01cm ) = arctan 0.125
angle = 7 degrees
which is a fair amount of ball rotation. Do that often enough, and
the ball will "roll" itself off the pad. For a sanity check, solder a
rigid bar of something to a flat surface, and bend it back and forth
about 7 degrees. It will take a while, but it will eventually break.


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

"Jeff Liebermann" wrote in message
...
On Thu, 02 Sep 2010 09:54:44 -0700, Jeff Liebermann
wrote:

The thermal conductivity of G10/FR4 isn't all that wonderful,
(...)

Ok, let's do the numbers. The coefficient of thermal expansion for
G10/FR4 is:
1*10^-5 cm/cm/C
That means a 1 cm long piece of G10/FR4, will expand 1*10^-5 cm for
every degree C of temperature differential.

So, we have a big fat BGA chip, that's about 5cm across. It's running
hot with a bottom temperature of about 80C. Assuming the PCB is
running at room temp of 25C, that's a 55C differential temperature.
Over the diameter of the BGA, that's
125*10^-5 cm
movement of the PCB.

Solder balls come in all manner of sizes, but my guess(tm) that for a
1mm pitch BGA, a 0.4mm ball is appropriate. When soldered, the ball
will remain about the same diameter, but the height will be reduced to
about 0.1mm.

The angle that the ball moves over temperature is:
angle = arctan ( 125*10^-5 cm / 0.01cm ) = arctan 0.125
angle = 7 degrees
which is a fair amount of ball rotation. Do that often enough, and
the ball will "roll" itself off the pad. For a sanity check, solder a
rigid bar of something to a flat surface, and bend it back and forth
about 7 degrees. It will take a while, but it will eventually break.


--
Jeff Liebermann

Great info and insights in both posts as always Jeff. I will take them into
consideration. The temperature differential thing is something that I hadn't
considered, but following through your numbers, seems to be a very valid
point ...

Arfa

On Fri, 3 Sep 2010 02:04:18 +0100, "Arfa Daily"
wrote:



"Jeff Liebermann" wrote in message
.. .
On Thu, 02 Sep 2010 09:54:44 -0700, Jeff Liebermann
wrote:

The thermal conductivity of G10/FR4 isn't all that wonderful,
(...)

Ok, let's do the numbers. The coefficient of thermal expansion for
G10/FR4 is:
1*10^-5 cm/cm/C
That means a 1 cm long piece of G10/FR4, will expand 1*10^-5 cm for
every degree C of temperature differential.

So, we have a big fat BGA chip, that's about 5cm across. It's running
hot with a bottom temperature of about 80C. Assuming the PCB is
running at room temp of 25C, that's a 55C differential temperature.
Over the diameter of the BGA, that's
125*10^-5 cm
movement of the PCB.

Solder balls come in all manner of sizes, but my guess(tm) that for a
1mm pitch BGA, a 0.4mm ball is appropriate. When soldered, the ball
will remain about the same diameter, but the height will be reduced to
about 0.1mm.

The angle that the ball moves over temperature is:
angle = arctan ( 125*10^-5 cm / 0.01cm ) = arctan 0.125
angle = 7 degrees
which is a fair amount of ball rotation. Do that often enough, and
the ball will "roll" itself off the pad. For a sanity check, solder a
rigid bar of something to a flat surface, and bend it back and forth
about 7 degrees. It will take a while, but it will eventually break.

Great info and insights in both posts as always Jeff. I will take them into
consideration. The temperature differential thing is something that I hadn't
considered, but following through your numbers, seems to be a very valid
point ...

Well, I did manage to make one mistake. The 7 degrees is the worst
case bending angle assuming everything accumulates in one direction.
That's not the case as local heating of the PCB will be from the
center outward. Instead of 125*10^-5 cm of lengthening measured from
the edge, the PCB will elongate half that amount, measured from the
center of the BGA. Correcting accordingly:

The angle that the ball moves over temperature is:
angle = arctan ( 63*10^-5 cm / 0.01cm ) = arctan 0.063
angle = 3.5 degrees
That's still enough to tear apart the solder ball, but not as radical
as I previously suggested.

One solution is to use a BGA adapter socket. Obviously, this isn't
going to work inside a laptop, where vertical height is a major
limitation. Same with some desktops, where the CPU heatsink and fan
can only be so tall or air flow out the top of the heatsink and fan
will be constricted. I've never tried to retrofit one of these into
an existing motherboard, but it sure looks tempting.
http://www.advanced.com/bgastart.html
http://www.mill-max.com/products/newproducts_detail.cfm?pid=7
http://www.ironwoodelectronics.com/products/adapters/giga_snap.cfm


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

In article , Arfa Daily
writes

I have always been sparing with compound - and I
use a lot of it as I repair many big amps for a living - but it is a fact
that a very thin translucent layer of white, is not effective enough on a
standard non-flatted device face, and heatsink contact area, whereas with
AS, it would appear that it is.

Have you tried applying the white stuff to both surfaces, then scraping
it off with the edge of a card? That will fill in any valleys on both
surfaces, and you should get a good thermal bond with the minimum of
compound.

This is the method that AS suggest using, by the way.

--
(\__/)
(='.'=)
(")_(")

"Jeff Liebermann" wrote in message
...
On Fri, 3 Sep 2010 02:04:18 +0100, "Arfa Daily"
wrote:



"Jeff Liebermann" wrote in message
. ..
On Thu, 02 Sep 2010 09:54:44 -0700, Jeff Liebermann
wrote:

The thermal conductivity of G10/FR4 isn't all that wonderful,
(...)

Ok, let's do the numbers. The coefficient of thermal expansion for
G10/FR4 is:
1*10^-5 cm/cm/C
That means a 1 cm long piece of G10/FR4, will expand 1*10^-5 cm for
every degree C of temperature differential.

So, we have a big fat BGA chip, that's about 5cm across. It's running
hot with a bottom temperature of about 80C. Assuming the PCB is
running at room temp of 25C, that's a 55C differential temperature.
Over the diameter of the BGA, that's
125*10^-5 cm
movement of the PCB.

Solder balls come in all manner of sizes, but my guess(tm) that for a
1mm pitch BGA, a 0.4mm ball is appropriate. When soldered, the ball
will remain about the same diameter, but the height will be reduced to
about 0.1mm.

The angle that the ball moves over temperature is:
angle = arctan ( 125*10^-5 cm / 0.01cm ) = arctan 0.125
angle = 7 degrees
which is a fair amount of ball rotation. Do that often enough, and
the ball will "roll" itself off the pad. For a sanity check, solder a
rigid bar of something to a flat surface, and bend it back and forth
about 7 degrees. It will take a while, but it will eventually break.

Great info and insights in both posts as always Jeff. I will take them
into
consideration. The temperature differential thing is something that I
hadn't
considered, but following through your numbers, seems to be a very valid
point ...

Well, I did manage to make one mistake. The 7 degrees is the worst
case bending angle assuming everything accumulates in one direction.
That's not the case as local heating of the PCB will be from the
center outward. Instead of 125*10^-5 cm of lengthening measured from
the edge, the PCB will elongate half that amount, measured from the
center of the BGA. Correcting accordingly:

The angle that the ball moves over temperature is:
angle = arctan ( 63*10^-5 cm / 0.01cm ) = arctan 0.063
angle = 3.5 degrees
That's still enough to tear apart the solder ball, but not as radical
as I previously suggested.

One solution is to use a BGA adapter socket. Obviously, this isn't
going to work inside a laptop, where vertical height is a major
limitation. Same with some desktops, where the CPU heatsink and fan
can only be so tall or air flow out the top of the heatsink and fan
will be constricted. I've never tried to retrofit one of these into
an existing motherboard, but it sure looks tempting.
http://www.advanced.com/bgastart.html
http://www.mill-max.com/products/newproducts_detail.cfm?pid=7
http://www.ironwoodelectronics.com/products/adapters/giga_snap.cfm


--
Jeff Liebermann

This is something that I was talking about with a colleague just a few days
ago. I'll take a look at the links. Going back to the differential heating
issue, I've thought a bit more about it, and it seems that the greatest
source of heat is going to be the top surface of the BGA itself, which has
the bonded heat dissipation plate for interfacing with the heatsink
assembly. Heat getting into the PCB is going to be two ways i.e. by
conduction through the solder balls, and by direct radiation from the
underside of the chip. Neither of these are going to be particularly
efficient, and I would expect as much heat as possible to be directed
upwards into the plate, by design. So it seems to me that the board is going
to remain relatively cool, compared to the underside of the BGA, and more to
the point, the upper side. So the hotter that the BGA is allowed to run, the
greater will be the undesired thermal difference between board and chip.
Therefore, any help to the cooling of the upper surface of the chip, should
help to reduce the temperature differential rather than exacerbate it,
shouldn't it ? To take it to its logical conclusion, if you could remove all
heat that the chip was generating, then there would be none to heat the
board, so there would be no thermal differential, at all ??

Arfa

"Mike Tomlinson" wrote in message
...
In article , Arfa Daily
writes

I have always been sparing with compound - and I
use a lot of it as I repair many big amps for a living - but it is a fact
that a very thin translucent layer of white, is not effective enough on a
standard non-flatted device face, and heatsink contact area, whereas with
AS, it would appear that it is.

Have you tried applying the white stuff to both surfaces, then scraping
it off with the edge of a card? That will fill in any valleys on both
surfaces, and you should get a good thermal bond with the minimum of
compound.

This is the method that AS suggest using, by the way.

--
(\__/)
(='.'=)
(")_(")



Yes Mike. Prior to starting to use the AS, I have always treated both
surfaces when using white, contrary to much perceived wisdom where it is
insisted that only one surface should be coated. I believe in doing both
surfaces for the exact same reasons that you cite. I am also doing both
surfaces with AS, but very sparingly. There are always milling patterns on
the heatsink faces on these machines, which I think is a bit bad on the part
of the manufacturers anyway, given the huge thermal loads that are produced
by these chips ...

Arfa

On Fri, 03 Sep 2010 12:36:34 +0100, Arfa Daily wrote:

"Mike Tomlinson" wrote in message
...
In article , Arfa Daily
writes

I have always been sparing with compound - and I use a lot of it as I
repair many big amps for a living - but it is a fact that a very thin
translucent layer of white, is not effective enough on a standard
non-flatted device face, and heatsink contact area, whereas with AS, it
would appear that it is.

Have you tried applying the white stuff to both surfaces, then scraping
it off with the edge of a card? That will fill in any valleys on both
surfaces, and you should get a good thermal bond with the minimum of
compound.

This is the method that AS suggest using, by the way.

--
(\__/)
(='.'=)
(")_(")



Yes Mike. Prior to starting to use the AS, I have always treated both
surfaces when using white, contrary to much perceived wisdom where it is
insisted that only one surface should be coated. I believe in doing both
surfaces for the exact same reasons that you cite. I am also doing both
surfaces with AS, but very sparingly. There are always milling patterns
on the heatsink faces on these machines, which I think is a bit bad on
the part of the manufacturers anyway, given the huge thermal loads that
are produced by these chips ...

Arfa

I've never thought it necessary to coat both surfaces. If you use
sufficient paste on one it will suffice for both sides. Key word
sufficient but not overly so. I guess it's just something you develop
a knack for in knowing what is too much or not enough. This 120 watt
AMD 955 PhenomII chip in my PC runs in its normal temp range. Idles
around 43c. CPU fan runs at 2500rpm, half of 5000 at full speed automatic
control. What I'm getting at is the heatsink that comes with the chip
has a very thin coat of Arctic. And it seems to do very well being
applied to the heatsink side only.



--
Live Fast, Die Young and Leave a Pretty Corpse

On Fri, 3 Sep 2010 12:31:17 +0100, "Arfa Daily"
wrote:

This is something that I was talking about with a colleague just a few days
ago. I'll take a look at the links. Going back to the differential heating
issue, I've thought a bit more about it, and it seems that the greatest
source of heat is going to be the top surface of the BGA itself, which has
the bonded heat dissipation plate for interfacing with the heatsink
assembly.

Yes, but...
http://en.wikipedia.org/wiki/Ball_grid_array
Heat conduction
A further advantage of BGA packages over packages with discrete
leads (i.e. packages with legs) is the lower thermal resistance
between the package and the PCB. This allows heat generated by the
integrated circuit inside the package to flow more easily to the
PCB, preventing the chip from overheating.

Heat getting into the PCB is going to be two ways i.e. by
conduction through the solder balls, and by direct radiation from the
underside of the chip.

When the PCB is so close to the bottom of the BGA package, whatever
heat is produced is radiated directly to the PCB. Assuming a fairly
uniform case temperature (possibly a bad assumption) by conduction,
the radiated heat out the bottom of the BGA case has to go somewhere.
It can't accumulate or it would just continue to heat up until it
melts. So, it heats the PCB.

Neither of these are going to be particularly
efficient, and I would expect as much heat as possible to be directed
upwards into the plate, by design.

I think you'll find that unless there's a hidden insulator somewhere
in the package, the bottom case temperature will be fairly close to
the top case temperature. If it were otherwise, the case would
distort or in extreme cases, crack. I can work out the exact numbers,
using the thermal resistance, if you give me the exact case style and
dissipation in watts.

So it seems to me that the board is going
to remain relatively cool, compared to the underside of the BGA, and more to
the point, the upper side.

How much is "relatively"? Most (not all) BGA arrays have the chip
mounted on the base. For example, see Fig 2 the wire bonded example
at:
http://www.siliconfareast.com/bga.htm
The heat will be coming out of the base, which will be hotter than the
lid due to some thermal resistance in the case. Others have the chip
mounted on the top. These are easily identified by the epoxy blob or
metal cover on the bottom PCB side of the BGA. See:
http://www.intel.com/assets/pdf/pkginfo/Ch_14.pdf
http://www.intel.com/design/packtech/packbook.htm
for Intel's packaging handbook. Also see 14.10 section for a little
on thermal performance. There's a section on thermal package stress
at:
http://www.intel.com/Assets/PDF/pkginfo/ch_04.pdf
See section 4.2.1 under "Stresses generated during a thermal
excursion".

So the hotter that the BGA is allowed to run, the
greater will be the undesired thermal difference between board and chip.

True. Heat removal is not 100% efficient. Think of temperature as
the voltage across a string of resistors (thermal resistance). Crank
up the input power and each resistor has more voltage across it.
However, the ratio of the various voltages and temperatures remains
constant as long as the thermal resistances don't change. That means
that fairly small thermal resistances, such as between the heat sink
and the case, are not going to see much of a temperature change for
increase dissipation, while large thermal resistances, such as the
heat sink to the air, are going to see a large increase.

Therefore, any help to the cooling of the upper surface of the chip, should
help to reduce the temperature differential rather than exacerbate it,
shouldn't it ?

Sure. But the difference in temperature is still what's bending the
board and breaking the bonds. That's what my guess(tm) was causing
the Nvidia video chip failures in many laptops. The chip was
literally tearing itself away from the PCB because the board was
bending.

There's another problem with your analysis. If you assume that the
edges of the PCB are at room temperature, or at least at case
temperature, then the temperature gradient across the PCB will remain
fairly constant as you increase board heating. The result is just a
larger heat affected zone, and no real improvement in cooling. It
would be like putting a computah inside a plastic bag (for
waterproofing) and dumping it inside a bucket of cold water. The case
will be very cool, but the CPU will still burn up inside.

To take it to its logical conclusion, if you could remove all
heat that the chip was generating, then there would be none to heat the
board, so there would be no thermal differential, at all ??

True. If the thermal resistance between the chip and every component
of the thermal circuit path were zero, and the thermal mass of the air
were assumed to be infinite (a really bad assumption), then the chip,
heatsink, case, and air temperature would all be the same. However,
if any or all of these exhibit any thermal resistance, there will be a
temperature difference across it.



--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Fri, 03 Sep 2010 09:27:56 -0700, Jeff Liebermann
wrote:

http://www.intel.com/design/packtech/packbook.htm
for Intel's packaging handbook. Also see 14.10 section for a little
on thermal performance. There's a section on thermal package stress
at:
http://www.intel.com/Assets/PDF/pkginfo/ch_04.pdf
See section 4.2.1 under "Stresses generated during a thermal
excursion".

I just noticed table 4-14 on Page 4-24 of the above handbook. It's a
table of the number of power cycles a CPU is expected to endure before
failure.

4.2.2 Temperature Cycles in Operation
A microprocessor package is subjected to numerous heating and
cooling cycles in operation. When the device is powered up, its
temperature rises, and when it is shut down, its temperature drops.
The magnitude of the maximum temperature on the die surface depends
on the thermal solution employed, and is usually between 80 to
125°C. In addition to these power on and power off cycles
(maxi-cycles), the microprocessor is cycled between different
intermediate temperature values depending upon processor usage
(mini-cycles) in any application program. The Institute for
Interconnecting and Packaging Electronic Circuits [2] lists the
typical worst case usage conditions for personal computers and
consumer electronics as given below. This table is intended only as
a guideline, and individual companies use different field use
conditions based on their research.

Category Worst case use environment
Tmin °C Tmax °C DT °C Dwell (hrs) Cycle/yr Approx. Years in Service
Consumer 0 +60 35 12 365 1-3
Computers +15 +60 20 2 1460 5

As I read this, if you turn your computer on and off once a day for 5
years, the CPU could fail due to thermal fatigue. For consumer
electronics, it's 1-3 year. Lovely...




--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

In article , Arfa Daily
writes

There are always milling patterns on
the heatsink faces on these machines, which I think is a bit bad on the part
of the manufacturers anyway, given the huge thermal loads that are produced
by these chips ...

Indeed. I think they hope the thermal compound (or phase-changing pad)
will cover up a multitude of sins.

--
(\__/)
(='.'=)
(")_(")

On Fri, 3 Sep 2010 06:24:11 +0100, Mike Tomlinson
wrote:

Have you tried applying the white stuff to both surfaces, then scraping
it off with the edge of a card? That will fill in any valleys on both
surfaces, and you should get a good thermal bond with the minimum of
compound.

This is the method that AS suggest using, by the way.

I've always suspected that it's a conspiracy by the manufacturer to
consume more expensive Artic Silver. Kinda like washing your hair
twice with "pH balance" shampoo.

The best heat tranfer between heat sink and CPU is metal to metal
contact, with no grease. The problem is that neither the heat sink or
CPU lid are flat and have pits, holes, gouges, lumps, cavities, and
other problems that prevent good contact. Even without these problem,
and with a mirror finish base, the typical warped package and
non-stress relieved heat sink, will not produce proper metal to metal
contact (without extreme mechanical pressure). My guess(tm) is that a
typical "brushed" aluminum heat sink to a Pentium 4 package might have
30% or less metal to metal contact. This sucks.

The idea is to fill the pits, holes, gouges, lumps, cavities, etc with
something thermally conductive, thus eliminating the need for mirror
finished and flat CPU's and heat sinks. The trick is to only fill the
pits, holes, gouges, lumps, cavities, etc and still retain as much
metal to metal contact as possible. That's not going to happen if you
use too much. As a clue, see the thermal resistance spec for Artic
Silver at:
http://www.arcticsilver.com/as5.htm
Thermal Resistance:
0.0045°C-in^2/Watt (0.001 inch layer)
Notice the 0.001 inch (0.025mm) layer. That's really really really
thin. So thin, that you could probably not even see it on the surface
because most of the stuff is in the pits, holes, gouges, lumps,
cavities, etc. If it had been specified with a thicker layer, the
thermal resistance would have been much worse.

It's probably a good idea to smear on some Artic Silver on both sides
of the junction, but then wipe off everything except what's in the
pits, holes, gouges, lumps, cavities, etc leaving as much metal to
metal contact as possible. If you're dealing with a badly warped or
an unpolished casting, then a little more grease might justifiable.
However, packing it on in a thick layer, but doing both sides, is a
waste.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

"Jeff Liebermann" wrote in message
...
On Fri, 3 Sep 2010 06:24:11 +0100, Mike Tomlinson
wrote:

Have you tried applying the white stuff to both surfaces, then scraping
it off with the edge of a card? That will fill in any valleys on both
surfaces, and you should get a good thermal bond with the minimum of
compound.

This is the method that AS suggest using, by the way.

I've always suspected that it's a conspiracy by the manufacturer to
consume more expensive Artic Silver. Kinda like washing your hair
twice with "pH balance" shampoo.

The best heat tranfer between heat sink and CPU is metal to metal
contact, with no grease. The problem is that neither the heat sink or
CPU lid are flat and have pits, holes, gouges, lumps, cavities, and
other problems that prevent good contact. Even without these problem,
and with a mirror finish base, the typical warped package and
non-stress relieved heat sink, will not produce proper metal to metal
contact (without extreme mechanical pressure). My guess(tm) is that a
typical "brushed" aluminum heat sink to a Pentium 4 package might have
30% or less metal to metal contact. This sucks.

The idea is to fill the pits, holes, gouges, lumps, cavities, etc with
something thermally conductive, thus eliminating the need for mirror
finished and flat CPU's and heat sinks. The trick is to only fill the
pits, holes, gouges, lumps, cavities, etc and still retain as much
metal to metal contact as possible. That's not going to happen if you
use too much. As a clue, see the thermal resistance spec for Artic
Silver at:
http://www.arcticsilver.com/as5.htm
Thermal Resistance:
0.0045°C-in^2/Watt (0.001 inch layer)
Notice the 0.001 inch (0.025mm) layer. That's really really really
thin. So thin, that you could probably not even see it on the surface
because most of the stuff is in the pits, holes, gouges, lumps,
cavities, etc. If it had been specified with a thicker layer, the
thermal resistance would have been much worse.

It's probably a good idea to smear on some Artic Silver on both sides
of the junction, but then wipe off everything except what's in the
pits, holes, gouges, lumps, cavities, etc leaving as much metal to
metal contact as possible. If you're dealing with a badly warped or
an unpolished casting, then a little more grease might justifiable.
However, packing it on in a thick layer, but doing both sides, is a
waste.

--
Jeff Liebermann


All agreed

Arfa

On Sat, 4 Sep 2010 02:07:59 +0100, "Arfa Daily"
wrote:

All agreed
Arfa

Nobody ever agrees with me. I must have said something wrong.

See:
http://www.microsi.com/packaging/thermal_grease.htm
Notice what happens to the thermal resistance as the thickness of the
silicon grease layer increases. Also notice the comment about
"solvent evaporation" which is why Arctic Silver and other greases
takes a while to "break-in".


--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558
# http://802.11junk.com
# http://www.LearnByDestroying.com AE6KS

"Jeff Liebermann" wrote in message
...
On Sat, 4 Sep 2010 02:07:59 +0100, "Arfa Daily"
wrote:

All agreed
Arfa

Nobody ever agrees with me. I must have said something wrong.

See:
http://www.microsi.com/packaging/thermal_grease.htm
Notice what happens to the thermal resistance as the thickness of the
silicon grease layer increases. Also notice the comment about
"solvent evaporation" which is why Arctic Silver and other greases
takes a while to "break-in".


--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060


OK. So here's the thing. The articles that you linked to are very
interesting, and at least one says that "the pcb is the primary heatsink in
the case of BGAs". Given that is true, as it was Intel I think that said it,
is this true for all BGAs ? If it is, then what is the point of fixing an
elaborate heatsinking system to the *tops* of the BGAs, and force cooling
this with a blower of over 2 amps rating, capable of ramping up to vacuum
cleaner levels ? When it gets going a bit, it actually exhausts pretty hot
air from these things. I would say that the heatsink gets *much* hotter than
the pcb, and if you try to run the board even at idle without the heatsinks
being placed, the unit goes into thermal protect inside a few seconds. If
the pcb was really the "primary heatsink" in the case of these particular
BGAs, I would have thought that at least when just idling, they would have
run ok 'naked' ??

Arfa

On Sat, 4 Sep 2010 09:44:11 +0100, "Arfa Daily"
wrote:

OK. So here's the thing. The articles that you linked to are very
interesting, and at least one says that "the pcb is the primary heatsink in
the case of BGAs".

There are about 100 assorted BGA packages, most of which do not
require a heat sink. You see them on video cards, cell phones, glue
chips, game machines, and most commonly on memory cards. There is NO
WAY your large BGA package, which probably has a big FPGA burning 200
watts inside, is going to work with just heat sinking to the PCB. The
leads are the primary heat sink for the small packages, not for the
monsters.

Given that is true, as it was Intel I think that said it,
is this true for all BGAs ?

Absolutely not. Size matters.

If it is, then what is the point of fixing an
elaborate heatsinking system to the *tops* of the BGAs, and force cooling
this with a blower of over 2 amps rating, capable of ramping up to vacuum
cleaner levels ?

Desperation? If you can't get the heat out via the leads, you do
whatever else is necessary.

When it gets going a bit, it actually exhausts pretty hot
air from these things.

I think you'll be surprised at how close to meltdown your BGA's are
running. Even a small heat producer will accumulate heat if the box
isn't adequately vented. The problem is that air really sucks as a
thermal conductor. It takes an awful lot of air to do very little
cooling. Give me some numbers to work with. Incidentally, you might
try using an IR thermometer on the heat sink, BGA, and exhaust air for
a sanity check.

I would say that the heatsink gets *much* hotter than
the pcb, and if you try to run the board even at idle without the heatsinks
being placed, the unit goes into thermal protect inside a few seconds. If
the pcb was really the "primary heatsink" in the case of these particular
BGAs, I would have thought that at least when just idling, they would have
run ok 'naked' ??

Yep. For BGA's without heat sinks, the primary heat conduction path
is through the vias in the substrate, to the solder balls, and then to
the PCB. For larger BGA's, it's through the case to a heat sink.



--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Sep 4, 10:19*am, Jeff Liebermann wrote:
On Sat, 4 Sep 2010 09:44:11 +0100, "Arfa Daily"

wrote:
OK. So here's the thing. The articles that you linked to are very
interesting, and at least one says that "the pcb is the primary heatsink in
the case of BGAs".

... NO
WAY your large BGA package, which probably has a big FPGA burning 200
watts inside, is going to work with just heat sinking to the PCB. *The
leads are the primary heat sink for the small packages, not for the
monsters.

Like Intel says, it's primary.
'Primary' does not mean the heatsink with the largest heat flux. It
means the FIRST heatsink, the one that all designers start with.
BGA packages have quite a lot of thermal conductivity through those
soldered-down feet, it's not something to be ignored.
In related news, 'prime rib' is a rib roast with the rib #1 included.

On Sat, 4 Sep 2010 12:28:55 -0700 (PDT), whit3rd
wrote:

On Sep 4, 10:19*am, Jeff Liebermann wrote:
On Sat, 4 Sep 2010 09:44:11 +0100, "Arfa Daily"

wrote:
OK. So here's the thing. The articles that you linked to are very
interesting, and at least one says that "the pcb is the primary heatsink in
the case of BGAs".

... NO
WAY your large BGA package, which probably has a big FPGA burning 200
watts inside, is going to work with just heat sinking to the PCB. *The
leads are the primary heat sink for the small packages, not for the
monsters.

Like Intel says, it's primary.
'Primary' does not mean the heatsink with the largest heat flux. It
means the FIRST heatsink, the one that all designers start with.
BGA packages have quite a lot of thermal conductivity through those
soldered-down feet, it's not something to be ignored.
In related news, 'prime rib' is a rib roast with the rib #1 included.

http://www.intel.com/assets/pdf/pkginfo/Ch_14.pdf
The exact quote is:
A considerable increase in thermal effectiveness of a BGA
package can be obtained by using boards that are thermally
efficient, increasing the airflow, or providing thermal paths
from the board. Remember, with PBGAs, the board is your
primary heatsink.

PBGA is a plastic ball grid array. I guess "primary" does make sense,
since the vias going through the base are closer to the heat source
than the package lid. Therefore, heat will try to exit through the
leads before the lid.

Thermally conductive PCB material:
http://www.bergquistcompany.com/thermal_substrates/
http://www.bergquistcompany.com/thermal_substrates/t-clad-product-overview.htm

It's a wonder they don't unsolder themselves. Oh wait... Nvidia
laptop video chips do that.
http://www.tgdaily.com/hardware-features/39045-nvidia-gpu-failures-caused-by-material-problem-sources-claim
According to our sources, the failures are caused by a solder bump
that connects the I/O termination of the silicon chip to the pad
on the substrate. In Nvidia’s GPUs, this solder bump is created
using high-lead. A thermal mismatch between the chip and the
substrate has substantially grown in recent chip generations,
apparently leading to fatigue cracking. Add into the equation a
growing chip size (double the chip dimension, quadruple the stress
on the bump) as well as generally hotter chips and you may have the
perfect storm to take high lead beyond its limits. Apparently,
problems arise at what Nvidia claims to be "extreme temperatures"
and what we hear may be temperatures not too much above 70 degrees
Celsius.

Note the "thermal mismatch". I have a Dell XPS1210 laptop on the
bench with exactly this problem and am waiting to justify the expense
of a hot air SMT rework machine.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

"Jeff Liebermann" wrote in message
...
On Sat, 4 Sep 2010 12:28:55 -0700 (PDT), whit3rd
wrote:

On Sep 4, 10:19 am, Jeff Liebermann wrote:
On Sat, 4 Sep 2010 09:44:11 +0100, "Arfa Daily"

wrote:
OK. So here's the thing. The articles that you linked to are very
interesting, and at least one says that "the pcb is the primary
heatsink in
the case of BGAs".

... NO
WAY your large BGA package, which probably has a big FPGA burning 200
watts inside, is going to work with just heat sinking to the PCB. The
leads are the primary heat sink for the small packages, not for the
monsters.

Like Intel says, it's primary.
'Primary' does not mean the heatsink with the largest heat flux. It
means the FIRST heatsink, the one that all designers start with.
BGA packages have quite a lot of thermal conductivity through those
soldered-down feet, it's not something to be ignored.
In related news, 'prime rib' is a rib roast with the rib #1 included.

http://www.intel.com/assets/pdf/pkginfo/Ch_14.pdf
The exact quote is:
A considerable increase in thermal effectiveness of a BGA
package can be obtained by using boards that are thermally
efficient, increasing the airflow, or providing thermal paths
from the board. Remember, with PBGAs, the board is your
primary heatsink.

PBGA is a plastic ball grid array. I guess "primary" does make sense,
since the vias going through the base are closer to the heat source
than the package lid. Therefore, heat will try to exit through the
leads before the lid.

Thermally conductive PCB material:
http://www.bergquistcompany.com/thermal_substrates/
http://www.bergquistcompany.com/thermal_substrates/t-clad-product-overview.htm

It's a wonder they don't unsolder themselves. Oh wait... Nvidia
laptop video chips do that.
http://www.tgdaily.com/hardware-features/39045-nvidia-gpu-failures-caused-by-material-problem-sources-claim
According to our sources, the failures are caused by a solder bump
that connects the I/O termination of the silicon chip to the pad
on the substrate. In Nvidia's GPUs, this solder bump is created
using high-lead. A thermal mismatch between the chip and the
substrate has substantially grown in recent chip generations,
apparently leading to fatigue cracking. Add into the equation a
growing chip size (double the chip dimension, quadruple the stress
on the bump) as well as generally hotter chips and you may have the
perfect storm to take high lead beyond its limits. Apparently,
problems arise at what Nvidia claims to be "extreme temperatures"
and what we hear may be temperatures not too much above 70 degrees
Celsius.

Note the "thermal mismatch". I have a Dell XPS1210 laptop on the
bench with exactly this problem and am waiting to justify the expense
of a hot air SMT rework machine.

--
Jeff Liebermann


Assuming that you're talking a 'standard' SM rework station with hot air
pencil, and not a multi-thousand dollar fixed rework station, then the one I
recently purchased direct from China, was just 55 quid - about $85. Bit of
postage to add on of course, but at that sort of money, not too much
justification required, I would suggest ? Look on eBay for KADA 852D. Very
good value for money. I'm very pleased with mine. The eBay shop I bought
mine from (dragondirectmall I think it was), has a video on the site of them
building one, so you can get an idea of the quality.

Arfa

On Sun, 5 Sep 2010 01:17:10 +0100, "Arfa Daily"
wrote:

Assuming that you're talking a 'standard' SM rework station with hot air
pencil, and not a multi-thousand dollar fixed rework station, then the one I
recently purchased direct from China, was just 55 quid - about $85. Bit of
postage to add on of course, but at that sort of money, not too much
justification required, I would suggest ? Look on eBay for KADA 852D. Very
good value for money. I'm very pleased with mine. The eBay shop I bought
mine from (dragondirectmall I think it was), has a video on the site of them
building one, so you can get an idea of the quality.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=370427912032
$124. I saw your previous reply to someone asking about SMD rework
stations. The problem with the KADA 852D is that it only includes 5
generic circular nozzles (2-10mm). I need the big square BGA nozzle
assortment, which are about $100 extra from other vendors. I also
can't seem to find any listing for KADA parts. The eBay listings does
include one spare heater and soldering iron element. I've got two
off-brand soldering irons I bought at various hamfests for which I
can't find tips or repair parts. Kada looks good, but not good
enough.

What I'm looking at is, at $230.
http://www.circuitspecialists.com/prod.itml/icOid/9766
It's twice as expensive, but has all the features I want (or could
possibly want later). Also, lots of parts available. The tips are a
useful assortment, but I'll still need to buy some QFP nozzles at
about $18/ea. I borrowed this model for about 2 weeks and really
liked using it.

This is another possibility, as it includes 20 nozzles for $239:
http://www.circuitspecialists.com/prod.itml/icOid/8227
However, it leaves out the soldering iron and desoldering iron, so
it's not really a fair comparison.

What's stopping me is an impending $2,000 dental bill, which will
greatly reduce my ability to buy new toys and tools.



--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

"Jeff Liebermann" wrote in message
...
On Sun, 5 Sep 2010 01:17:10 +0100, "Arfa Daily"
wrote:

Assuming that you're talking a 'standard' SM rework station with hot air
pencil, and not a multi-thousand dollar fixed rework station, then the one
I
recently purchased direct from China, was just 55 quid - about $85. Bit of
postage to add on of course, but at that sort of money, not too much
justification required, I would suggest ? Look on eBay for KADA 852D. Very
good value for money. I'm very pleased with mine. The eBay shop I bought
mine from (dragondirectmall I think it was), has a video on the site of
them
building one, so you can get an idea of the quality.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=370427912032
$124. I saw your previous reply to someone asking about SMD rework
stations. The problem with the KADA 852D is that it only includes 5
generic circular nozzles (2-10mm). I need the big square BGA nozzle
assortment, which are about $100 extra from other vendors. I also
can't seem to find any listing for KADA parts. The eBay listings does
include one spare heater and soldering iron element. I've got two
off-brand soldering irons I bought at various hamfests for which I
can't find tips or repair parts. Kada looks good, but not good
enough.

What I'm looking at is, at $230.
http://www.circuitspecialists.com/prod.itml/icOid/9766
It's twice as expensive, but has all the features I want (or could
possibly want later). Also, lots of parts available. The tips are a
useful assortment, but I'll still need to buy some QFP nozzles at
about $18/ea. I borrowed this model for about 2 weeks and really
liked using it.

This is another possibility, as it includes 20 nozzles for $239:
http://www.circuitspecialists.com/prod.itml/icOid/8227
However, it leaves out the soldering iron and desoldering iron, so
it's not really a fair comparison.

What's stopping me is an impending $2,000 dental bill, which will
greatly reduce my ability to buy new toys and tools.



--
Jeff Liebermann


Blimey, and I thought 200 quid was expensive for a new tooth crown ... !!

Arfa

Mike Tomlinson wrote:

nah, standard IPA works fine. The trick is to use clean tissue wetted
with IPA, wiping just once or twice then replacing with a fresh piece,
repeating until the CPU is clean.

http://en.wikipedia.org/wiki/IPA_%28disambiguation%29

Says that IPA can mean Isopropyl alcohol. Is that what you meant?
If so, do you use 70 percent or 99 percent?
--
When a cat sits in a human's lap both the human and the cat are usually
happy. The human is happy because he thinks the cat is sitting on him/her
because it loves her/him. The cat is happy because it thinks that by sitting
on the human it is dominant over the human.

Talk about awakening the dead (thread) lol.

But now that you mentioned cats :

http://www.craigslist.org/about/best...440629699.html