IBM Intellistation power problem

[Ginchy]

I have an old IBM Intelliestation E Pro desktop machine. In fact I have 2.
Model 6893

When you first attatch the kettle lead to the back the machine turns on
immediately without touching the power button at the front.

This happens like this on both machines and the machine wont boot or show
any pic on the monitor.

In the past I have put the power off using the front button and when you
switch it back on everything works fine.

At least "I think" thats the way I done it. What I do know is there was a
specific thing I had to do to get it working and now I have forgot.

Today I switched on one machine after connecting the psu kb and m. As usual
it switches on and the cd rom spins etc but no boot. The problem this time
is that I cannot switch it off using the power button at the front of the
machine. If i press it in and let go it just stays on with the green light
showing. If I hold it in for 6 secs it powers down but as soon as let the
button go it powers up again. I tried letting go really quick as soon as I
hear the machine powering down but no joy it still starts up.

I tired switching the mains on and of at the wall socket but no use.

Anyone got any ideas? Like I say I KNOW I have done this before and had the
exact same problem but I always got pout of it I just cannot remember how.
Either that or I am remembering correct but something else is wrong. All I
know is that if I can switch it off with the power button then it WILL start
up properly.

 

[Paul]

Hi,

Just wondering if you ever managed to solve your problem as my IBM
Intellistation Z Pro 6221 has exactly the same problem, only mine
started after I connected an ipad 2 to one of the front USB ports, which
I've done before with no problems, but this time for whatever reason a
message came up regarding some error or issue with the usb device or
port and then the computer shut down, so I didn't get a chance to read
exactly what the message was. I also smelt a burning smell just when
this occurred. Now it won't boot up and there is no display on the
monitors. The power however still comes on and the fans run along with
the lights on the front but thats it. Display is blank and the power
button doesn't work as I have to unplug the power cord it to turn off or
it turns itself on when I plug it in again.

At first I thought the power supply was faulty so I got another one and
it didn't change anything. I also inspected the top of the motherboard
for any burn marks or burnt components but it all looked ok to me. If
you did fix or if anyone else knows how to fix this problem please let
me know. I'm hoping it's not the motherboard or a processor that's
fried.

It probably uses an ICH4. The ICH5 Southbridge, is the one that was
notorious for latchup failures. ICH4 was supposed to be affected,
but I never ran into any reports by users of one failing. I'm assuming
the chipset is E7505 and I have the right diagram.

http://img.tomshardware.com/us/2004..._than_one_intel_p4_extreme_platform/e7505.gif

Check your Southbridge, for a burn mark like this. This is an ICH5
after a latchup failure (internal). The chip gets so hot, the plastic
melts over top of the fault. Maybe if you look closely, you can see
a blemish like this.

http://onfinite.com/libraries/179057/2ea.jpg

Paul

 

[Paul]

Thanks for the reply. I checked the chip I think you mean and took a
photo of it. It appears to be ok. There is a tiny blemish on it but it
looks more like a scratch than a burn mark. If it is this chip, can it
be replaced?:

14672


+-------------------------------------------------------------------+
|Filename: IMG_0753[1].jpg |
|Download: http://phorums.com.au/attachment.php?attachmentid=14672 |
+-------------------------------------------------------------------+

Take this picture:

http://onfinite.com/libraries/179057/2ea.jpg

and compare it to your chip, when your chip is rotated.

http://img33.imageshack.us/img33/9661/img0753aburnedich4maybe.jpg

It looks like the two tiny "marks" are in the same relative position.
Yours just didn't burn as bad. Note that, in the majority of
ICH5 latchup failure cases, there is no burn mark to be
seen, and in those cases, all the USB ports are dead (because
the bond wire powering the USB pads, is burned out). Only when
the bond wires or lead frame continue to have integrity, does
the plastic really get a good roasting - and then the chip
ends up completely dead, as it can't take that kind of heat.
So in a minor failure case, you end up with no USB ports (and
the motherboard still POSTs). In a major failure case, the
motherboard will no longer POST. No BIOS screen, no nothing.
Not possible for it to even "beep". No instructions can be executed,
if the firmware hub the BIOS flash chip is connected to, is actually
connected to the burned chip. That's why the processor can't start.
No code to run.

I can't be sure about this. I think I'm reading something like
82801DB as the chip number (all I can really see is "DB"), and the
letters on the end identify which Southbridge model it is. As
far as I'm concerned, it's unusual for an ICH4 to fail like that.
Even though it was listed in the one and only advisory (on the Gigabyte
web site), there are way more ICH5 failures, than failures on ICH4.
In fact, if your chip says "DB" on the end of the part number, this
might be the only time I've run into a poster with a dead "DB".

http://en.wikipedia.org/wiki/I/O_Controller_Hub

When I look at the ICH5 datasheet and the ICH4 datasheet, I can't really
see the correspondence. At one time, I'd read an explanation of why
the burn mark was exactly there, but I can't see by looking at the
ballout diagram right now, why the mark is at that point.

You can try taking a sniff, right next to that IC, and see if
it has developed an unusually strong odor, for something which should
have outgassed long ago. That might be confirmation of the type of fault.
That chip is normally low power, and shouldn't smell.

The top of the IC, shouldn't really have any marks on it, because
that would be the seating plane for a heatsink. The top should be
flat and blemish free, as much as possible.

The IC can be replaced. One poster reported, he sent his broken
motherboard to Asus, and they sent back the same motherboard, but
with a new ICH5 soldered to it. So they can do that kind of repair,
but I was really surprised to read of such a thing. I have a company
in my town, which does those repairs, but they charge $1000 to
replace a chip. They include an XRay inspection, after the new chip
is in place, to prove their soldering job is good. In China, they can
do that for $25 or less, worth of materials. I wouldn't expect
we could match them in price. (My local guys are a rip-off, but because
they're not working every day, they still have to pay the rent. They've
probably left town by now anyway, as what they did was only viable
during the industry peak. Considering the economy, fringe activities
like that, just can't survive.)

To do the repair, you need a hot air repair station. With a vacuum wand
to pick up the IC, when the balls are molten underneath. The hot air
repair machine, has to have a metal fitting that fits over top exactly,
the shape of the IC. That's how you get precision heating of just
that chip. The machine also provides a bit of heat from the bottom.
When all the solder is a liquid, then the chip can be picked up from the
top. The only problem with this procedure, is the quality of the motherboard
makes a big difference. Computer motherboards are "crap" in terms of
physical durability. Which is why I wouldn't want to bet on the odds
of the motherboard, surviving such a repair. On the PCBs my company used
to make, we could change a chip like that up to three times, before the
PCB would be discarded. But our PCBs were like "rock", in terms of
hardness. I've seen many a ham-handed engineer attack one of our
PCB boards with a soldering iron, and the board survived. The
Chinese motherboards are made for a tenth of the price, and the
physical qualities of the materials, just aren't the same. When I
repaired a Chinese motherboard here, with a soldering iron, I managed
to lift a pad - and I was being careful, too. The copper foil on these
boards, comes off too easy.

For more info on latchup failures, you can start with an article
like this. Since Intel never sent out their own admission of a problem
like this, we can't really be sure of the diagnosis of why it burns.
Only Gigabyte (a motherboard maker), sent out a warning in the form
of a web page. That's the only confirmation we've got.

http://en.wikipedia.org/wiki/Latchup

Paul

 

[Paul]

Thanks for the very informative reply. Looks like it will be more
practical and economical to just replace the whole motherboard. I've
tracked down a couple on ebay ranging from $150-$220 so I might go with
that. The model is a IBM 59P2607.

You can try pulling the RAM (with the power off), and testing without
any RAM present. If the computer beeped, that would tell you the processor
was still running, and it wasn't a total loss. If the board is still silent,
then chances are the motherboard is cooked.

In the case of a minor failure (all USB ports dead, computer still boots),
you can fix that with a $10 USB2 card. The motherboard might continue to
function, once all the USB ports are dead (because just the I/O pins die, and
not the logic blocks - the USB ports are "blind" rather than "dead").
But if it's a major failure, there was the burnt smell, and now it won't
POST at all (no BIOS screen), then generally, it's "replace the motherboard time".
Only if the motherboard was under an extended warranty, might you get it fixed
for free.

It was unclear to me, whether Intel gave any money to the motherboard
makers, to cover the need for repairs. As far as I know, this is an
Intel issue, and if it meant more expense for the motherboard makers,
Intel should have compensated them. But Intel themselves, never publicly
admitted there was a problem. Leaving the customers stuck with the problem.
I also, never saw a new revision of ICH5, which might have been considered
to be an admission of guilt. The 865G chipset was shipped for some time
after those failures started showing up, so you'd think they'd attempt
to ship good ICH5 chips to go with those. And if the chip was changed,
the letters on the top should have been altered as well (for inventory
tracking purposes).

I have a motherboard with ICH5, but mine never died One of the
things I don't do with that computer, is use the USB ports Note that,
even avoiding the usage of the motherboard USB ports isn't enough. Of
the twenty or so reports of this problem I've reviewed and discussed,
at least one happened, with no USB involvement at all. So if you were
that person, you could avoid using the motherboard USB ports, use
a PCI USB2 card instead, and it might still fail on you. That's how
sensitive this problem is for some people.

The record holder, was somebody who had five motherboards fail in the
same way. You can't get more unlucky than that. All with burnt Southbridge.
That person didn't give any details of the conditions around each
failure, but he wasn't very happy. I guess it all depends, on whether
there's any warranty left, as to the financial impact.

I see no evidence of a repeat of that failure, on any other chipsets.
So if you were worried about buying a new motherboard, it might be
less of a concern. The last big screwup at Intel, was some SATA ports
that were predicted to die, because some transistor in the I/O pad
wasn't connected right or something. Intel did a big, showy, recall
for those defective products. And showered the motherboard makers
with money to fix it. An entirely different response than the ICH5
problem.

Paul

 

[Flasherly]

Well I received a replacement motherboard yesterday that had been
refurbished and tested. Same model and eveything. The computer powers up
now with monitor output but I get a series of beeps. 2-2-3 & 3-1-4 and
it starts to boot up but shuts down after a list of error messages with
the bios comes up. I checked all the connections and the ram (which did
show full memory up in post) and all looks ok.

I looked it up and it seems to be the battery on the board. Going to get
another one today and give it a try. Since it's a replacement board,
I've also read that the bios should be reset and cmos cleared by moving
two jumpers. Is there a need to do this as well as change the battery?

I recently swapped a battery (with a spare, although turned out the
pre-existing was OK) and flashed with a newer BIOS. Although the BIOS
update routine may stipulate a hard boot, power-off condition, as well
as the routine resetting the new BIOS to defaults, the subsequent
power-up may not indicate such was the case when/if reading the new
BIOS. Whereas taking the power entirely away through the shorted
pins, say for 5 minutes, should ensure it is. Sort of fixed a problem
I was having when getting a different CPU up, better than before,
anyway, or dealing with HD related transfer problems. An improvement
for what I can't remember exactly the problem(s). I'm possibly
disappointed for its age, which is perhaps beginning to show for a 5-
year-oldish ASUS 478 setup. I also used to know people running IBM
desktops in a small business. May God have mercy on those things.
I'd take a generic Pacific-Rim MB any day -- Gigabyte, another Asus,
something, and cross my fingers. (My longest build/sale to date I can
still verify is someone at 10 years + and still working -- AMD,
probably an early Athlon 800Mhz-ish and doing "ok" for XP. Want to
say a MSI MB. Cost him $1000 I do recall when I took him with me to
pick up and pay for delivery of the parts from a wholesale supply
house. The one he just wanted, I built nice, everything shipped in,
up to and included in a case, for $200. An Intel/MSI S775 @ 3.4Ghz.)

 

[Paul]

Well I received a replacement motherboard yesterday that had been
refurbished and tested. Same model and eveything. The computer powers up
now with monitor output but I get a series of beeps. 2-2-3 & 3-1-4 and
it starts to boot up but shuts down after a list of error messages with
the bios comes up. I checked all the connections and the ram (which did
show full memory up in post) and all looks ok.

I looked it up and it seems to be the battery on the board. Going to get
another one today and give it a try. Since it's a replacement board,
I've also read that the bios should be reset and cmos cleared by moving
two jumpers. Is there a need to do this as well as change the battery?

If the BIOS knows the battery is flat, it should restore
defaults when loading up CMOS RAM. If the battery is
good, then it does the opposite, and just reads the
CMOS RAM (and the BIOS settings are in there).

Motherboards keep inventory information in the DMI and
ESCD sections of BIOS flash. If the hardware changes,
the motherboard should be able to compare DMI and ESCD
against current motherboard hardware. So it can tell if
things are different or not.

You clear the CMOS if:

1) The motherboard is behaving strangely, and you suspect
bogus data is stored in CMOS RAM. Perhaps after a sudden
power failure or the like.

2) You've just upgraded the BIOS flash with new code.
While some BIOS flashing tools, have code to clear CMOS
of their own, there are still cases where clearing CMOS
with the jumper (with all power removed), is needed before
the new BIOS code works properly. This might happen if the
new map for the CMOS RAM usage, differs from the old map.

3) You overclock a board, and it's so poorly designed,
you need to clear CMOS to get "nominal" settings back
in the BIOS again. In fact, some motherboards were so
sensitive, certain RAM settings would give you a black
screen and no future joy (whether you clear CMOS or not).
Fortunately, there aren't many chipsets like that.

Just change the battery, and see what happens

Could be more "beeps" waiting for you...

HTH,
Paul

 

[Paul]

Ok, I swapped the battery for a new one and the error list became
smaller. The 2-2-3 beeping has also now gone and the 3-1-4 one was at
first still there, but after trying it again the next day it seems to
have gone. Now no beeping and the PC boots up but only one error
messages still appears:

01298001 - No update for processor one.

Apart from the message, the main problem now is that it the PC keeps
automatically switching off. Sometimes I can get as far as the bios
screen (F1) and sometimes not, and then it just switches itself off. The
startup screen did show that there was 2 processors and I even reseated
it but no change. It just shuts down. Sometimes I get as far as updating
the time and date in the bios and then bang!>>> power off it goes..

I downloaded the most recent bios update and put it on a floppy disc.
The bios on the motherboard I received has a very old version and a new
version I read may resolve the error message. I removed the cmos jumper
and it boots up to the disc but the dreadded power problem strikes again
and the PC just shuts down so I can't even properly update the bios.

As a last resort, I borrowed a spare working power supply and swapped it
over, but the problem was still the same so it's not the power supply.
Could it be the power button itself or just something very simple. I
checked and everything is plugged correctly. I even reseated the ram to
no avail. Very frustrating....

One last interesting thing I noticed when I compared the original board
with the replacement is that the bios chip is different on each board.
Everything else is exactly the same on the board including the printed
text, etc, but both have different bios/cmos chips. The chip on the
original board is soldered directly onto the board and just has a
sticker with a code/model/version number printed. The new board however
has a plug in type socket with a Phoenix bios chip plugged in.

I interpret "No update for processor one" to mean there is no microcode
patch information for the processor. As if the processor is of a different
vintage than the generation of motherboard. Updating the BIOS might fix that.

Intel issues microcode patches, after the processor is manufactured. Installing
a patch in runtime, isn't absolutely essential in all cases, but it helps.
A typical BIOS has support for about eight different families of processor.
It's possible to extract the file inside the BIOS, and actually list
the processors that are supported from a microcode perspective.

You should *not* attempt to flash update the motherboard, if the power won't
stay on. You could corrupt the flash, half way through the update, and
be left with no POST afterwards. While the BIOS design has a so-called "boot block"
for recovery from a bad flash, it seldom works due to stupid shortcuts
done with the flashing tools.

If the new board has a socketed BIOS, you can get a new chip from
http://www.badflash.com/ . You tell them the part number of the motherboard,
part number of the flash chip, they obtain a blank chip of the same type.
You tell them where to find the BIOS flash file, and they copy the
BIOS file into the chip for you, then send the new chip to you. The
BIOS file should be in a simple format - some web sites have the
file in some kind of download .exe that is an all-in-one installer,
and that makes it harder for them to process (they may complain).
A proper BIOS file is typically power_of_two bytes in size.

If you had a *good* computer shop in town, they may be able to carry
out the same kind of process, using your existing chip. We had an expensive
programmer at work, that would program just about anything, but most
computer shops cannot afford a thing like that. Ours may have been
in the $6000 region or so. There are also $100 programming devices
available on the web, but they're typically for some specific flash
part, and not general purpose.

That is one way to rescue a motherboard, that has a bad BIOS flash.
Or, provide BIOS flash upgrade support, when the computer is unstable.

On modern motherboards, the flash can be an eight pin, serial interface
chip, and some of those are soldered down. Similarly, some older motherboards
have DIP or PLCC flash chips, but they're soldered down as well. That's to
save on the cost of a socket, which is poor economics from the customer's
perspective. Removing a chip like that, from the motherboard, isn't
exactly easy, because of the danger you'd damage the copper foil.
By comparison, pulling a chip from a socket, is much easier. I've
pulled around a hundred PLCC chips from PLCC sockets, with nothing
more than a pointed "pick" tool. You can also get the extractor
from Radio Shack, if you want a fancy PLCC remover. The DIP packages,
I used to slide a staple remover with pointed tip under those, to
pry them up. That was my favorite tool for that job.

http://www.radioshack.com/product/index.jsp?productId=2062619

In terms of the power button:

1) The power button is momentary contact.
2) The motherboard converts the momentary contact info, into a solid
level called PS_ON#. That signal is what controls the power supply.
3) The motherboard chip that listens to the front power button,
has the ability to "condition" the signal. For example, when the OS
is running, a four second filter is used, such that you have to press
and hold the power button for four seconds, to turn off the motherboard.
At the BIOS level, the filter time is much shorter.
4) If the front power button is "jammed on", that may cause things
to switch off. If you suspected that, you could unplug the two wires
from the front switch, and use a multimeter to check the switch itself.
Also, instead of using the switch, you can touch a screwdriver tip to the
two pins that normally host the POWER cable pair. I've turned motherboards
on and off that way, without a switch. Be sure you're only touching the
two pins in question. You need steady hands, to do it that way.
5) If your motherboard has both POWER and RESET header pairs, you can unplug
the POWER cable, then move the RESET cable to the POWER pair of pins.
Both switches are momentary contact, normally open type.
That makes the RESET switch, your new power switch. You'd do that, if
you suspected the POWER switch was defective, and lacked a multimeter
for test.

The processors have been protected for some time, by THERMTRIP. If the
processor gets too hot, it can assert THERMTRIP, and that is a gating
logic term for PS_ON#. The power supply will immediately shut off,
with no help at all from the processor or processors. If you have a
dual socket board, both THERMTRIPs would be tied into the logic.
You would then have to figure out, why it was overheating, such as
check that the CPU heatsink is still in place, that there is thermal
paste between CPU heatsink and CPU top surface, as that aids in heat
transfer. A tube of Arctic Silver or the like (thermal paste),
can be obtained to solve that kind of problem. One tube lasts
a long time, as only a rice grain sized portion is needed per
installation.

"Arctic Silver 5" $10
http://www.newegg.com/Product/Product.aspx?Item=N82E16835100007

A motherboard can also choose to switch off, for BIOS program reasons.
If the BIOS has code added, to switch off when something unexpected
happens, then that could be a reason for the power to go off as well.
And then, you're left guessing as to why.

That's a few ideas, but I don't know what's the most cost effective
next step in your journey.

While it could be the power supply that is switching off on overload,
I somehow doubt it. The new motherboard could have a short in it, for
example. I have a clamp-on DC ammeter I can use, to check for excessive
current. (You clamp that around wires on the main wire harness.)
Or even if you had a Kill-O-Watt meter, and checked total consumption
going into the PC, that might also hint if something was totally out
of whack.

Paul

 

[Paul]

When I swapped over the motherboard, I didn't replace the thermal paste
on both processors and heatsinks (I read that I should have). I used the
same paste that was on before as it looked ok (big mistake I guess).
Maybe this is the problem as I noticed when I could bring up the bios
screen before the PC switched itself off that the CPU temperature seemed
very hot on processor 1. The bios showed processor 1 was running at
around 250 degrees F whereas processor 2 hovered around 80-90 degrees F.
Going to get some paste with 25% silver which says is recommended for PC
processors. I'll completely clean the old paste off and apply the new
one as per the maintenance manual I found online for the Intellistation
Z Pro 6221. Hopefully this could be the problem of the PC shutting off
all the time where this morning I fired it up again (after being off all
night) and it stayed on for about 3-4 minutes, then after restarting
again a couple of times, the time switched on got shorter.

OK. Sounds like we have a match on symptoms

Once it's stable, you can look at BIOS flashing again.

As for installation technique, arctic silver has installation
instructions on their site. Probably a few PDF files there
for your usage.

I can give you what little I've learned by experimentation.

1) For your first attempt, install a half grain of paste, then
squash the processor into it and do up the clamp. This is a
test install, to "calibrate" the amount to use. Now take the
heatsink off. You should see a circle of paste as a result.
If you use too much, it will ooze out. On the calibration
run, we want to err on the "too little" side.

2) When you compute the rough amount needed (say, a full rice grain
sized helping), add a tiny bit more. Not only should the paste reach
the edge of the surface. You want enough excess that the paste
"wets" the joint. Then, the white color of the paste should be
visible from the side. You want enough that you can see it from
the side, but not so much it runs down onto sensitive components.
Depending on the tools you've got (I have a dental mirror on the end of
an extendible wand), you can inspect along the sides of the heatsink,
to look for the distinctive white color of a properly filled joint.

3) In terms of cleaning, isopropyl isn't the right solvent, but will
provide a liquid you can dampen a rag with while cleaning. The paste
is an organic, so a non-polar solvent is probably better if you
actually needed to solvate the paste. You can look up the MSDS for
this cleaning kit, to see the solvents they use in it. The solvent
used, should not attack plastic (gasoline is out of the question).
Maybe orange oil would be close.

http://www.newegg.com/Product/Product.aspx?Item=N82E16835100010

Some thermal interface materials are solids. I have one heatsink here,
where a phase change material was used, and it's relatively hard at
room temperature. Solvent was getting me no where fast, so I actually
had to scrape it off to get the majority of it. Not all TIMs are
easy to clean.

If dealing with LGA sockets, or ZIF for that matter, you
definitely don't want to get the paste into the socket. The
pins on an LGA are impossible to clean (you'd need a good matching
solvent and plenty of it - like the board cleaning machine they use
at the factory). While the top can be torn off a ZIF (like S478 or
S939 perhaps), I don't know if you can fit it back on later. That's
why there's all this care in calibrating the amount of material to use.
To prevent gross over-application.

At one time, I used to use the "credit card technique" and spread the
material out. That seemed to work well for the AthlonXP-M I had (no
lid on that one). Most modern desktop processors have a metal top now,
so the "squash" technique might work better to avoid air bubbles.
The squash also lends itself to calibration, whereas the credit
card technique can ooze if you use too much (you don't know how
thick to put it on). The purpose of paste is to displace the air
layer, not to make an Oreo cookie The paste takes the space
between two less than perfectly flat surfaces. Once the air
is displaced, no further paste is really helping. You want a
thin layer, for least thermal resistance.

Inside the processor, they use a TIM as well. Between the lid and the
silicon die, there is a material. Up until recently, Intel used a low
temperature solder, between the top of the silicon die and the metal top.
But I was just reading the other day, the latest generation has returned
to using a "gravel" material. Which I don't like, because the consistency
isn't necessarily that uniform. There was at least one processor product,
where the lack of consistency actually caused a problem (hot spots), but
that was a number of years ago. In cases where a processor mysteriously
"cannot be cooled down", one possibility is a TIM failure underneath the
lid. The Intel solder technique was a good solution, but I suppose
somebody at the factory has to be a hero by "cost reducing" the
manufacturing process. Low temperature solders are expensive.

With my AthlonXP-M, there's no lid, and it's all up to
the user to do a good job of cooling. You can't blame the
manufacturer if one of those burns up.

Both Intel and AMD use THERMTRIP now, so we no longer
have to worry about that movie on the Tomshardware site
any more. The power supply on the computer will be switched
off, if a processor gets too hot. The support is inside
the processor for it, so it doesn't cost anything to give
customers that protection.

HTH,
Paul

 

[Paul]

Here is the link to the Z Pro 6221 maintenance manual. The thermal
grease instructions are on page 78.

ftp.software.ibm.com/systems/support/system_x_pdf/71p7639.pdf

I'm guessing that's their way of spreading it. By laying
it out in a matrix like that.

All I can say is, just make sure you don't get the paste
all over the place. The paste is there, as a preferential
material to having an "air gap". Some of the particles
in the paste are boron nitride, a ceramic with good
thermal properties. So the whole thing isn't silver.

Apparently, even industrial diamond can be used in thermal
paste, but it isn't too common. It's surprising some
of the materials with good thermal properties. There
is one other material, but I've sworn not to promote it,
because I don't think it's good to have "floating
around the house" (unknown toxicity). The AS5 probably
wouldn't be good for you if you ate it, but there are
probably still a few materials that are worse for you.

http://www.amazon.com/Antec-Diamond...8290689&sr=8-3&keywords=diamond+thermal+paste

Paul