Yet another bulging-capacitors replacement
whit3rd
GHz).
This has already had the logic board replaced, as these machines
had some bad-filter-capacitor issues... but this time it was the
capacitors in the power supply, not on the logic board, that
were bulging and leaking electrolyte.
It took an hour or two of catalog work to find low-ESR replacements
for the
nine low-V high-I filter capacitors in the power supply, in form
factors
that would fit the cramped footprint of the originals. So, I thought
I'd relate the parts list here, in case anyone else has need of such
info.
C40 and C52 10V 1000 uF
EKY-100ELL102MH20D
C45, C55 and C56 2200 uF 10V
UHM1A222MPD
C47 16V 1200 uF
UHE1C122MPD
C49 10V 3300 uF
UHN1A332MHD ** this is slightly larger diameter than the
original, but it fits **
UHZ0J332MPM **right size, but less voltage margin**
C59 35V 330 uF
ELXV350ELL331MJ20S
C64 15V 1000 uF
EEU-FC1E102B
These were all in stock at Mouser Electronics, if that matters.
Meat Plow
Hell yes it matters. I'm going to fix a year old Coolmax 650 watt PC PSU
and will be looking for some replacement caps. I like to keep a spare and
I need 650 with this new AMD 120 watt quad core PhenomII 3.2 ghz CPU and
Asus M4A78E-T mobo. With Asus overclocking friendly special settings I'm
able to run it at 4.0 ghz for each core. Makes video encoding on an
application supporting multicore encoding really fly. Not unusual to get
over 350 frames/sec out of NTSC 740x480 avi's. I can make a high quality
20 chapter DVD with all the bells and whistles in about an hour. Used to
take 24 hours on a 2ghz single core AMD :)
--
Live Fast, Die Young and Leave a Pretty Corpse
Arfa Daily
"Meat Plow" <mhy...@yahoo.com> wrote in message
news:pan.2010.08...@hahahahahahahah.nutz.I.am...
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've
recently been working with some games machines that have two very powerful
processors on the board, and have been having some thermal issues when using
'standard' white silicon grease on them, which appears to be what the
manufacturer used originally. I have today reassembled one using some Arctic
Silver compound instead, and it seems to be doing a fine job. I have always
resisted using this stuff, because it's so messy, and so hard to remove
unless you use the complementary cleaner, but if it really is that much more
effective, then I might be prepared to live with these shortcomings. Anyone
else got any constructive comments on the subject of thermal interfacing of
coolers to high power chips ?
Arfa
Jeff Liebermann
<arfa....@ntlworld.com> wrote:
>Anyone
>else got any constructive comments on the subject of thermal interfacing of
>coolers to high power chips ?
In a past life, I used to design HF SSB marine radios. The typical
transmitter was Class AB 150 watts with about 30% efficiency. That's
two devices, dissipating about 125 watts each, over an area of about
70 sq-cm. Oh yes, no fan allowed.
This is quite a bit more dissipation than the average desktop, causing
some things to be more critical. In the process of getting it to
work, I learned a few things.
1. The less silicon grease used, the better. The idea behind silicon
grease is to fill in the gaps, scratches, and gouges in the power
transistor base and aluminum heat sink. Cross sectional
microphotographs show metal to metal contact on the peaks, but huge
gaps, filled with silicon grease, in between. Under ideal
circumstances, maximum metal to metal contact, with minimum gaps is
the target practice.
2. All heat sinks and transistor bases are NOT flat. I made a
dramatic improvement to the measured thermal resistance by polishing
flat the base of the xsistor and the face of the heat sink. That
meant removing the gold from the copper base, but that's what was
necessary. I used a Moire pattern to measure flatness. A mirror
finish was best, but difficult to achieve. To prevent corrosion, I
plated the exposed copper with electroless tin or silver. For the
aluminum heat sink, I just used abrasive polish and a glass polishing
plate to obtain a mirror finish and flat surface.
3. Compression pressure is important. None of the standard spring
clip CPU heat sink holders come even close to optimum. Compression
adjusts for the bends, and also provides some level of galling to
provide metal to metal contact. If done correctly, adding silicon
grease actually increases the thermal resistance. However, this is
difficult to do with a CPU that has components on the bottom side,
thus preventing compression. Applying pressure only on the top center
of the CPU, will cause the substrate to bend, and eventually break. I
have some ideas, but nothing that can be retrofitted to an existing
motherboard and CPU socket. This is close, but not optimum:
<http://www.frostytech.com/articleview.cfm?articleID=2273>
Note the comments on base finish and flatness.
So, if you want the best head sinking, polish flat the CPU top
(removing all the laser scribbled markings, polish the heat sink face,
use very very very very little silicon grease, and compress the
sandwich until it nearly breaks the CPU.
--
Jeff Liebermann je...@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
N_Cook
news:jh3e761hp6mg3b0sb...@4ax.com...
Any opinions on silipads? From my limited trials mica sheet and absolute
minimal white grease has better thermal transfer. Emphasis again on minimal.
Having to decide this week whether to spend out on 6 MJ series TO3 devices
in an old Carver amp failed due to one of the driver TO3 being pushed
through a heap of white grease , so grease on pins getting inside the TO3
socket housing , so insulating partially and eventually burning up the pin .
Archon
the fishtank type noise!), CPU and GPU a few degrees above room temp
even at full load. Never go back to jet engine graphics card fans.
JC
Meat Plow
You answered your own question. The AMD heatsink / quad core PhenomII 955
Black Edition package comes with Artic Silver already applied. I'm using
an Antec server case that has a hole and tube in the side cover. The end
of the tube fits directly over the CPU heat sink so it draws air directly
from the outside. In back is a pair of 120mm fans controlled by the
mainboard. If the CPU temp goes up all three fans increase according to
the temp. Or you can set them to run at full speed all the time. The 650
watt PSU also has a temp sensing 120mm fan. So the box is really quiet
most of the time. But when rendering video and the CPU usage hovers
around 50% fan speed increases slightly. Video rendering with an
application that takes advantage of multi-core processors seem to use the
most CPU percentage. I've never seen it go over 50%. Most of the time it
doesn't go over 10%.
Meat Plow
> Any opinions on silipads? From my limited trials mica sheet and absolute
> minimal white grease has better thermal transfer. Emphasis again on
> minimal. Having to decide this week whether to spend out on 6 MJ series
> TO3 devices in an old Carver amp failed due to one of the driver TO3
> being pushed through a heap of white grease , so grease on pins getting
> inside the TO3 socket housing , so insulating partially and eventually
> burning up the pin .
I've seen pads used in a lot of high power amps. Soundcraftsmen, BGW,
Carver, Peavey, Crown all used them at some time or another. If the
devices are torqued down properly I don't have a problem with them. You
are talking about the rubberized pads with embedded compound right?
Jeff Liebermann
wrote:
>Any opinions on silipads? From my limited trials mica sheet and absolute
>minimal white grease has better thermal transfer. Emphasis again on minimal.
Do the math. Silicon impregnated rubber TO-3 pads have a thermal
resistance of about 0.4K/watt. 0.0002" Mica, with silicon grease
smeared on both sides is about 0.6K/watt. Depending on your total
power dissipation, that's hardly any difference. However, if your
heat sink is undersized, buried inside a cabinet, or located in a
place where there's no air flow, it might make a difference.
<http://www.wakefield.com/LinkClick.aspx?fileticket=1ULQQwz8xmU%3d&tabid=64>
See Page 6-8. The example shown is for a TO-3 packaged device.
>Having to decide this week whether to spend out on 6 MJ series TO3 devices
>in an old Carver amp failed due to one of the driver TO3 being pushed
>through a heap of white grease , so grease on pins getting inside the TO3
>socket housing , so insulating partially and eventually burning up the pin .
Keep it stock. However, if this is a push-pull type of amp, where the
thermal balance of the xsistors has an effect on the bias point,
crossover distortion, and possibly linearity, I would make sure that
whatever you do to one side, the same gets done to the other.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:jh3e761hp6mg3b0sb...@4ax.com...
Thanks for the insights Jeff. All interesting stuff. These are dedicated
games machines, not based on a PC in any way. The power supply is specced to
deliver 12v at 23 amps, yes, that's twenty three amps ...
Almost all of this is potentially going into these two processors, so not
far off 300 watts between them. No mean task shifting the heat off them !
Arfa
Arfa Daily
Yes, seems to be 'busy' video rendering that causes all the problems on the
machines I am working on. Q & D calcs show that at max chat, the two
processors are potentially using close to 300 watts of input power between
them, and the heat that this generates in them, takes some shifting ...
Arfa
Jim Yanik
news:dgZdo.79946$Pa3.38201@hurricane:
what processors(microprocessors?) run at 12V?
ISTR that today's uPs run mostly on 3.3V
Most other digital logic runs at 5V,I believe.
I think you wil find that most of your power is going into the video drive
(or LCD backlight) circuitry.
--
Jim Yanik
jyanik
at
localnet
dot com
Jeff Liebermann
<arfa....@ntlworld.com> wrote:
>Thanks for the insights Jeff.
There was quite a bit of "Learn By Destroying(tm)" involved. Measuring
flatness and thermal resistance were a major exercise, but settled all
kinds of lab arguments.
Incidentally, you might be interested in how Arctic Silver works.
<http://www.arcticsilver.com/products.htm>
<http://www.arcticsilver.com/msds.htm>
Silver has a much higher thermal conductivity (410 W/m*K) as compared
to zinc oxide (21 W/m*K) and aluminum oxide (30 W/m*K) which is what's
in common thermal compound.
<http://en.wikipedia.org/wiki/List_of_thermal_conductivities>
However, if your shove an ohms-guesser into a puddle of Arctic Silver,
it's not conductive. That's because the particles of silver are so
far and few, that the bulk of the solution is polyolester or mineral
oil, which insulates the particles from each other, preventing mutual
contact. However, if you tear apart a CPU/heatsink that's been used
for a while, you'll notice that the Arctic Silver is a thick and dense
paste which is conductive. What has happened is that the polyol ester
mixture has evaporated sufficiently to provide contact between
particles. Since thermal conductivity is best through the silver
particles, the result is a superior thermal connection, with the
sliver particles filling the voids. You could do the same thing with
silver dust, but it would difficult to handle and apply. Meanwhile,
ordinary silicon grease does the same thing, but there's a difference.
The oil does not evaporate as easily, and the ceramic particles are
much larger and less compressible than the silver particles. Fewer
points of contact and lower thermal conductivity of ceramic, means a
worse thermal connection.
>All interesting stuff. These are dedicated
>games machines, not based on a PC in any way. The power supply is specced to
>deliver 12v at 23 amps, yes, that's twenty three amps ...
>
>Almost all of this is potentially going into these two processors, so not
>far off 300 watts between them. No mean task shifting the heat off them !
I don't believe it. The winner of the power hogging consumer CPU
contest was the DEC/Intel Alpha 21364 (EV79):
<http://en.wikipedia.org/wiki/Alpha_21364>
which burned 155 watts. Itanium II came close with 130 watts (per
core). I had an Alpha CPU machine to play with for a while, which
would burn my hand from the hot air coming out the back.
If you have a power line wattmeter or a Kill-A-Watt meter, I think a
measurement would be helpful.
Arfa Daily
>> dedicated games machines, not based on a PC in any way. The power
>> supply is specced to deliver 12v at 23 amps, yes, that's twenty three
>> amps ...
>>
>> Almost all of this is potentially going into these two processors, so
>> not far off 300 watts between them. No mean task shifting the heat off
>> them !
>>
>> Arfa
>>
>>
>
> what processors(microprocessors?) run at 12V?
> ISTR that today's uPs run mostly on 3.3V
>
> Most other digital logic runs at 5V,I believe.
>
> I think you wil find that most of your power is going into the video drive
> (or LCD backlight) circuitry.
>
> --
> Jim Yanik
Er no. There are no backlights. Or display processor. These are X-Box /
Playstation type boxes. Apart from some support circuitry in IC form - which
admittedly does gobble enough power to make it run hot enough that a degree
of heatsinking to the pcb shielding via thermal pads is required -
everything goes on in a pair of very large BGA processors, one of which is
the data processing engine, and the other of which is the graphics
processing engine. It is they which make use of the 12v, and they which
gobble the amps from it ...
The power supply does have other outputs, but these are all at very low
current availabilities, so will be for support logic and maybe some core
supplies for the two processors. Trust me when I say that the two big chips
is where all the power is going, and generating heat that needs shifting
:-)
Arfa
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:ebug76lk13plippid...@4ax.com...
Again, more interesting stuff Jeff. As to the power consumption of these
chips, see my reply to Jim above. Also, it is split between two chips, it's
actually not quite as much as 300 watts, and will of course be an 'on
demand' thing, depending on how hard the chips are being asked to work by
the processing task that's happening at the time, so the 23 amps is only a
worst case potential input current. However, that said, these two chips do
produce *very* considerable heat even when idling to produce nothing more
than the splash screen.
Arfa
Arfa Daily
"Arfa Daily" <arfa....@ntlworld.com> wrote in message news:...
>>> Thanks for the insights Jeff. All interesting stuff. These are
>>> dedicated games machines, not based on a PC in any way. The power
>>> supply is specced to deliver 12v at 23 amps, yes, that's twenty three
>>> amps ...
>>>
>>> Almost all of this is potentially going into these two processors, so
>>> not far off 300 watts between them. No mean task shifting the heat off
>>> them !
>>>
Jim Yanik
Odd that BGA processors are using 12V instead of logic level voltages.
I'm surprised they don't use some sort of liquid or heat-pipe plumbing to
remove all that heat. Wasn't it the CRAYs that used liquid Freon to flood
the processor cabinet to dissipate al the heat built up?
Michael A. Terrell
Jim Yanik wrote:
>
> Odd that BGA processors are using 12V instead of logic level voltages.
Then they would need around 100A at 3.3 volts. The voltage drop
would be a big problem. I'm sure there is a DC to DC converter near the
chip, like used on computer motherboards.
> I'm surprised they don't use some sort of liquid or heat-pipe plumbing to
> remove all that heat. Wasn't it the CRAYs that used liquid Freon to flood
> the processor cabinet to dissipate al the heat built up?
--
Politicians should only get paid if the budget is balanced, and there is
enough left over to pay them.
Meat Plow
Most run at 12. The core at 1.6. Both AMD and Intel boards have a 4 pin
Molex plug near the CPU for direct 12v from the PSU.
Arfa Daily
"Jim Yanik" <jya...@abuse.gov> wrote in message
news:Xns9DE25B60366AA...@216.168.3.44...
Yes, I was amazed when I read the PSU specs. One of the versions is actually
specced 12v at 32 amps !! The PSU plugs directly onto the board via a pair
of brass pins as thick as those on a power cord for a kettle. The heat is
removed via a pair of flat plates that are connected to a network of sealed
copper pipes, a bit like you see on some Technics amps. I've no idea what is
inside those pipes. This whole assembly is cooled by a centrifugal fan that
idles at a very low speed. The heatsinks have to get up to blisteringly hot
before the processor thinks that it might be a good idea to ramp up the
speed of the fan a bit. I guess they have done this to try and keep the
thing quiet, but personally, I think it is a really poor bit of design. I am
looking at ways to make the fan idle faster, without compromising the auto
ramp up beyond that, when the processor deems it necessary, but
unfortunately, it's not quite as easy as a simple 'analogue OR' function,
because the fan is controlled digitally. It is supplied with a constant 12v,
but a third wire has a PWM signal placed on it by the CPU, and the fan's
internal electronics respond to that to control the speed.
Yes, it was the Cray. It had a central octagonal bus backplane as I recall.
The met office here in the UK used to use one for weather data number
crunching, but I think it has been replaced now.
Arfa
Arfa Daily
"Michael A. Terrell" <mike.t...@earthlink.net> wrote in message
news:ipednVJmQcLfkOTR...@earthlink.com...
>
> Jim Yanik wrote:
>>
>> Odd that BGA processors are using 12V instead of logic level voltages.
>
>
> Then they would need around 100A at 3.3 volts. The voltage drop
> would be a big problem. I'm sure there is a DC to DC converter near the
> chip, like used on computer motherboards.
Yes indeed Michael. There are in fact 6 of them. Three on each side of the
board ...
Arfa
JW
wrote in Message id: <ebug76lk13plippid...@4ax.com>:
>I don't believe it. The winner of the power hogging consumer CPU
>contest was the DEC/Intel Alpha 21364 (EV79):
><http://en.wikipedia.org/wiki/Alpha_21364>
>which burned 155 watts. Itanium II came close with 130 watts (per
>core).
Check again.
http://ark.intel.com/Product.aspx?id=43410&processor=9350&spec-codes=SLBMX
185W! Gotta love that price as well.
Jeff Liebermann
I stand corrected.
<http://techreport.com/discussions.x/18445>
Some of the reader comments are rather interesting. Still, with any
of these "powerful" processors, a conventional air cooled machine is
going to have a very hot breath and a rather large power supply. I
just don't see this kind of power dissipation in a "dedicated game
machine". Measuring the AC mains power consumption should settle the
matter.
stra...@yahoo.com
<snip>
> You answered your own question. The AMD heatsink / quad core Phenom
II 955
> Black Edition package comes with Artic Silver already applied. I'm
using
> an Antec server case that has a hole and tube in the side cover.
The end
> of the tube fits directly over the CPU heat sink so it draws air
directly
> from the outside. In back is a pair of 120mm fans controlled by the
> mainboard. If the CPU temp goes up all three fans increase
according to
> the temp. Or you can set them to run at full speed all the time.
The 650
> watt PSU also has a temp sensing 120mm fan. So the box is really
quiet
> most of the time. But when rendering video and the CPU usage hovers
> around 50% fan speed increases slightly. Video rendering with an
> application that takes advantage of multi-core processors seem to
use the
> most CPU percentage. I've never seen it go over 50%. Most of the
time it
> doesn't go over 10%.
>
> --
> Live Fast, Die Young and Leave a Pretty Corpse
Sony Vegas software will push your processor usage up to nearly 100%
and stay there for minutes. My Phenom II 955 machine normally idles
115-120 Watts but will peak about 100 more while Vegas is rendering a
file. I'm almost tempted to try a 6 core processor to see what
happens.
G²
Meat Plow
I use Kino in linux to render raw DV capture. Can't find a suitable linux
app to create dvd containers/structure etc.. in linux so I use a fairly
inexpensive app called DVDtoX from VSO. Fast 10x frame rates on VBR video
encoding usually 2000KB/s. Easy to use nice output. tried Sony DVD
Architect trial, more than I needed. Linux is much more efficient at
processor usage but applications aren't up to par.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:phnn76d5fjpcls7lf...@4ax.com...
> On Mon, 30 Aug 2010 05:35:00 -0400, JW <no...@dev.null> wrote:
>
>>On Fri, 27 Aug 2010 20:19:44 -0700 Jeff Liebermann <je...@cruzio.com>
>>wrote in Message id: <ebug76lk13plippid...@4ax.com>:
>>
>>>I don't believe it. The winner of the power hogging consumer CPU
>>>contest was the DEC/Intel Alpha 21364 (EV79):
>>><http://en.wikipedia.org/wiki/Alpha_21364>
>>>which burned 155 watts. Itanium II came close with 130 watts (per
>>>core).
>
>>Check again.
>>http://ark.intel.com/Product.aspx?id=43410&processor=9350&spec-codes=SLBMX
>>185W! Gotta love that price as well.
>
> I stand corrected.
> <http://techreport.com/discussions.x/18445>
> Some of the reader comments are rather interesting. Still, with any
> of these "powerful" processors, a conventional air cooled machine is
> going to have a very hot breath and a rather large power supply. I
> just don't see this kind of power dissipation in a "dedicated game
> machine". Measuring the AC mains power consumption should settle the
> matter.
>
> --
> Jeff Liebermann je...@cruzio.com
Have you any idea just how much processing power it takes to run a
user-interactive story in real time, and then to 3D render the graphics in
real time ? Do you think that they rate the 12v PSU for 23.5 amps in one
version, and 32 amps in the other, for fun ? Those are not real questions,
because I know full well when you stop and think about it, you know the
answers, Jeff.
I've just looked at the rating plate on the bottom of one of the cases, and
it is 240v (nominal UK line voltage) at 1.8 amps. I make that a maximum
input power of around 430 watts. It's a switching PSU, so I reckon that we
can rate that as being at the very worst 80% efficient, so that's still 345
watts potentially going somewhere. I'm prepared to go with 45 watts into
ancillary circuitry on the board, which still leaves around 300 watts going
somewhere. Perhaps I'm being naive, but my best guess is that it's
disappearing into the two bloody great BGAs which the manufacturers are
trying their utmost to heatsink. If you try to run one of these machines
with the heatsinking not in place, it goes into thermal protect in about 5
seconds - and all it's doing then is booting. The heatplates on the BGAs are
at this point hot enough to take your fingerprints off ...
Nope, I'm pretty sure that these two puppies are good for 150 watts apiece,
when the machine is doing some real work.
Arfa
Jim Yanik
>
>
> "Jeff Liebermann" <je...@cruzio.com> wrote in message
> news:phnn76d5fjpcls7lf...@4ax.com...
>> On Mon, 30 Aug 2010 05:35:00 -0400, JW <no...@dev.null> wrote:
>>
>>>On Fri, 27 Aug 2010 20:19:44 -0700 Jeff Liebermann <je...@cruzio.com>
>>>wrote in Message id: <ebug76lk13plippid...@4ax.com>:
>>>
>>>>I don't believe it. The winner of the power hogging consumer CPU
>>>>contest was the DEC/Intel Alpha 21364 (EV79):
>>>><http://en.wikipedia.org/wiki/Alpha_21364>
>>>>which burned 155 watts. Itanium II came close with 130 watts (per
>>>>core).
>>
>>>Check again.
>>>http://ark.intel.com/Product.aspx?id=43410&processor=9350&spec-codes=S
>>>LBMX 185W! Gotta love that price as well.
>>
>> I stand corrected.
>> <http://techreport.com/discussions.x/18445>
>> Some of the reader comments are rather interesting. Still, with any
>> of these "powerful" processors, a conventional air cooled machine is
>> going to have a very hot breath and a rather large power supply. I
>> just don't see this kind of power dissipation in a "dedicated game
>> machine". Measuring the AC mains power consumption should settle the
>> matter.
>>
>> --
>> Jeff Liebermann je...@cruzio.com
>
>
> Have you any idea just how much processing power it takes to run a
> user-interactive story in real time, and then to 3D render the
> graphics in real time ? Do you think that they rate the 12v PSU for
> 23.5 amps in one version, and 32 amps in the other, for fun ? Those
> are not real questions, because I know full well when you stop and
> think about it, you know the answers, Jeff.
>
> I've just looked at the rating plate on the bottom of one of the
> cases, and it is 240v (nominal UK line voltage) at 1.8 amps. I make
> that a maximum input power of around 430 watts. It's a switching PSU,
> so I reckon that we can rate that as being at the very worst 80%
> efficient, so that's still 345 watts potentially going somewhere. I'm
> prepared to go with 45 watts into ancillary circuitry on the board,
> which still leaves around 300 watts going somewhere.
that assumes that all the power of the supply is actually used.
I'm sure there is some reserve capacity there.
"max input power" is not "actual used power".
> Perhaps I'm being
> naive, but my best guess is that it's disappearing into the two bloody
> great BGAs which the manufacturers are trying their utmost to
> heatsink. If you try to run one of these machines with the heatsinking
> not in place, it goes into thermal protect in about 5 seconds - and
> all it's doing then is booting. The heatplates on the BGAs are at this
> point hot enough to take your fingerprints off ...
>
> Nope, I'm pretty sure that these two puppies are good for 150 watts
> apiece, when the machine is doing some real work.
>
> Arfa
>
>
--
Jeff Liebermann
<arfa....@ntlworld.com> wrote:
>Have you any idea just how much processing power it takes to run a
>user-interactive story in real time, and then to 3D render the graphics in
>real time ?
Well, no. I'm not a power user. What little rendering I do is with
bacon fat.
>Do you think that they rate the 12v PSU for 23.5 amps in one
>version, and 32 amps in the other, for fun ? Those are not real questions,
>because I know full well when you stop and think about it, you know the
>answers, Jeff.
Actually, I don't know. I don't have any customers with such machines
and have had zero experience with high power graphic workstations
(other than early 1980's Applicon CAD stations) or game machines. I
have worked on various network servers, which do burn such power
levels. I have looked at a 3D MRI image processor, which had some
manner of dedicated processor inside, but it certainly wasn't belching
400 watts of heat (my estimate by the amount of fan noise).
>I've just looked at the rating plate on the bottom of one of the cases, and
>it is 240v (nominal UK line voltage) at 1.8 amps. I make that a maximum
>input power of around 430 watts. It's a switching PSU, so I reckon that we
>can rate that as being at the very worst 80% efficient, so that's still 345
>watts potentially going somewhere. I'm prepared to go with 45 watts into
>ancillary circuitry on the board, which still leaves around 300 watts going
>somewhere. Perhaps I'm being naive, but my best guess is that it's
>disappearing into the two bloody great BGAs which the manufacturers are
>trying their utmost to heatsink. If you try to run one of these machines
>with the heatsinking not in place, it goes into thermal protect in about 5
>seconds - and all it's doing then is booting. The heatplates on the BGAs are
>at this point hot enough to take your fingerprints off ...
>
>Nope, I'm pretty sure that these two puppies are good for 150 watts apiece,
>when the machine is doing some real work.
Ok, I stand corrected. I've been assuming that the CPU's are doing
most of the power dissipation. I didn't think of a dedicated graphics
processor or whatever the BGA chips are doing. Do you have a gun
style IR thermometer? I use that to determine if anything is running
hot. I use a black (non-reflective) cardboard tube attached to the
lens to prevent it from picking up adjacent components. Incidentally,
I have yet to find one where the laser dot actually points to where
the device is measuring when in close proximity. You can also get a
rough idea of how much effort is going into cooling. If the BGA's
burn more power than the CPU's, then they're going to need more
massive heat sinks and better air cooling. At 400 watts, I would
think they would have gone to heat pipes and external radiators or
maybe liquid cooling.
Incidentally, I repaired a P4 motherboard yesterday which used Artic
Silver. My guess is that there was about 5 times as much Artic Silver
smeared over the CPU (and down the sides where it does nothing) as
necessary. The stuff down the sides was still fluid, so at $10 for
3.5 grams, I saved the excess.
Meat Plow
> ncidentally, I repaired a P4 motherboard yesterday which used Artic
> Silver. My guess is that there was about 5 times as much Artic Silver
> smeared over the CPU (and down the sides where it does nothing) as
> necessary. The stuff down the sides was still fluid, so at $10 for 3.5
> grams, I saved the excess.
Heh...reminds me of a previously repaired (not by me) QSC PLX series amp
I opened up and scraped about a pound of white paste out of it.
William R. Walsh
> This is quite a bit more dissipation than the average desktop, causing
> some things to be more critical.
Well, it used to be. ;-)
The Pentium 4 "Pres-hot" didn't earn that derogatory nickname for nothin'. I
cringe to think of multi-processor systems and how much heat they were
dumping into the air. I had a 3.4GHz Prescott P4 in a Dell Dimension 8300.
On hot days, it had no problem equaling the sound volume of a small canister
vacuum cleaner.
> 1. The less silicon grease used, the better.
I remember reading that somewhere. I'm not sure that everyone--including
some major manufacturers--got the memo. After removing the STK-2038 II
module from my Techics SA-310 receiver, I found a massive amount of heatsink
compound behind it. Wow.
> 2. All heat sinks and transistor bases are NOT flat.
Somtimes not by a *long* shot!
> 3. Compression pressure is important. None of the standard spring
> clip CPU heat sink holders come even close to optimum.
Really? I find that extremely surprising, especially as firmly as some of
them hold on. They really do *seem* to be doing a good job.
Sometimes the heatsink compound has established a tight enough bond that the
processor comes out firmly glued to the bottom of the heatsink, without so
much as releasing the ZIF socket lever. I've seen that on Socket 478 and
AM2+ boxen before. It's kind of scary to look down and realize the processor
isn't where it should be!
William
Arfa Daily
"Jim Yanik" <jya...@abuse.gov> wrote in message
news:Xns9DE5620083A92...@216.168.3.44...
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.
Anyway, just think about that premise for a moment. When have you ever known
a manufacturer of a piece of domestic grade electronics, to over-rate any
aspect of it, let alone the power supply, by more than the few percent
required to just about let it scrape by? It's all about cost, and they are
not going to rate the rectifiers and magnetics and filter caps and output
connectors and so on, for any more than they have to, to keep the
manufacturing costs as low as possible ...
Arfa
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:60aq76t79k66vhdrm...@4ax.com...
> Skype: JeffLiebermann AE6KS 831-336-2558
All of the processing power is in those two BGAs Jeff. They *are* the
processors. One is a dedicated engine that runs the game (or plays a BluRay
disc), as well as handling all the disc I/O - optical and hard - and
internet / network access. On top of this, it manages all of the
housekeeping tasks, so it's doing a lot of work, especially when it's
actually running a game. Modern games have come a long way since the days of
Doom. Most maintain a highly complex 3D 'reality' in which the game is set,
and the gameplay takes place. Just consider for a moment, the highly complex
calculations that have to go on, to work out how potentially many actions
all at once, interact with the 3D model, and the knock-on effects that these
might have on both the gameplay and the graphical environment. And remember
that this is taking place in real time. The second BGA is a dedicated
graphics engine. Again, consider how these games now look. Most are quite
close to reality, and some scenes would have you hard pressed to tell if you
were looking at a photo, or a piece of virtual reality. Given all that, just
imagine the billions of calculations that are going on, again in real time,
to work out the texturisation and surface rendering of all the visible
objects, and how the light and shadows interact with those objects as they
move within the scene. It really is mind-boggling just how sophisticated all
of this is now. I can recall 25 years ago when I worked on high-end graphics
systems, rendering the famous 3D conch shell image took a dedicated graphics
terminal, hosted by a VAX mainframe, around 20 minutes. That's one image,
not moving. Now think about a moving HD image in an HD background in real
time. That's a LOT of processing power, needing a lot of amps to perform ...
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.
Arfa
JW
You sure that someone hadn't left their marshmallows in there?
Meat Plow
That's a possibility. The 3402's did get hot enough under full load
to roast them.
Michael A. Terrell
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. :)
Jeff Liebermann
<arfa....@ntlworld.com> 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.
Arfa Daily
"Michael A. Terrell" <mike.t...@earthlink.net> wrote in message
news:m5ednfqzQtE_8ePR...@earthlink.com...
>
> 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.
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 ?
Arfa
Michael A. Terrell
Arfa Daily wrote:
>
> "Michael A. Terrell" ?mike.t...@earthlink.net? wrote in message
> news:m5ednfqzQtE_8ePR...@earthlink.com...
> ?
> ? 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?
Meat Plow
Half what a US kettle would draw?
Archon
> On Wed, 01 Sep 2010 16:55:34 -0400, Michael A. Terrell wrote:
>
>> Arfa Daily wrote:
>>>
>>> "Michael A. Terrell" ?mike.t...@earthlink.net? wrote in message
>>> news:m5ednfqzQtE_8ePR...@earthlink.com... ?
>>> ? 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 ?
>>
>>
>> much current do those kettles draw?
>
> Half what a US kettle would draw?
>
>
>
minutes for the lousy thing to boil.
JC
Arfa Daily
"Michael A. Terrell" <mike.t...@earthlink.net> wrote in message
news:Jo2dnfqSRvF9IOPR...@earthlink.com...
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 ? 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.
Arfa
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:o4vs76df7mcfo4o21...@4ax.com...
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
Mike Tomlinson
<je...@cruzio.com> 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.
--
(\__/)
(='.'=)
(")_(")
Mike Tomlinson
<arfa....@ntlworld.com> 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.
Mike Tomlinson
<arfa....@ntlworld.com> 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.
Mike Tomlinson
<arfa....@ntlworld.com> 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.
Mike Tomlinson
Terrell <mike.t...@earthlink.net> 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.
Michael A. Terrell
> ?
> ? Arfa Daily wrote:
> ??
> ?? news:m5ednfqzQtE_8ePR...@earthlink.com...
> ?? ?
> ?? ? Arfa Daily wrote:
> ?? ??
> ?? ?? Well Jim, that was why I used the word "potentially", but judging by
> ?? ?? 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. :)
Arfa Daily
"Mike Tomlinson" <mi...@jasper.org.uk> wrote in message
news:6GldFyBD...@jasper.org.uk...
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
Meat Plow
> Arfa Daily wrote:
>>
>> "Michael A. Terrell" ?mike.t...@earthlink.net? wrote in message
>> news:Jo2dnfqSRvF9IOPR...@earthlink.com... ?
>> ? Arfa Daily wrote:
>> ??
>> ?? "Michael A. Terrell" ?mike.t...@earthlink.net? wrote in message
>> ?? news:m5ednfqzQtE_8ePR...@earthlink.com... ?? ?
>> ?? ? 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
Jeffrey Angus
> 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
Meat Plow
Some parts of the country that's a real possibility. And you wouldn't
even have to flash your lights.
Michael A. Terrell
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.
Jeff Liebermann
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
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.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:avlv76lbbv730nj64...@4ax.com...
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
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 ...
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>
Mike Tomlinson
<arfa....@ntlworld.com> 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.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:11n086hmsc5jvt998...@4ax.com...
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
Arfa Daily
"Mike Tomlinson" <mi...@jasper.org.uk> wrote in message
news:R+UXbOA7...@jasper.org.uk...
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
Meat Plow
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.
Jeff Liebermann
<arfa....@ntlworld.com> 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
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...
Mike Tomlinson
<arfa....@ntlworld.com> 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.
Jeff Liebermann
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.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:gbg286hlovdiatsoh...@4ax.com...
All agreed
Arfa
Jeff Liebermann
<arfa....@ntlworld.com> 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 je...@cruzio.com
# http://www.LearnByDestroying.com AE6KS
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:bp7386l3k3gbikbmb...@4ax.com...
> On Sat, 4 Sep 2010 02:07:59 +0100, "Arfa Daily"
> <arfa....@ntlworld.com> 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
Jeff Liebermann
<arfa....@ntlworld.com> 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.
whit3rd
> On Sat, 4 Sep 2010 09:44:11 +0100, "Arfa Daily"
>
> >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.
Jeff Liebermann
wrote:
<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.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:e09586l6jf1dgfbkg...@4ax.com...
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
Jeff Liebermann
<arfa....@ntlworld.com> 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.
Arfa Daily
"Jeff Liebermann" <je...@cruzio.com> wrote in message
news:uis586hjidp9kgih2...@4ax.com...
Blimey, and I thought 200 quid was expensive for a new tooth crown ... !!
Arfa
Daniel Prince
>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.
jurb...@gmail.com
But now that you mentioned cats :
http://www.craigslist.org/about/best/col/3440629699.html