Print

[dokworm]

After reading some of the excellent detective work here re the red lights of death and going over the couple of dead consoles here with a microscope, I've come to the conclusion that in most cases the red lights of death is due to a simple design flaw, not to overheating or poor soldering.

We seem to be seeing consoles where the motherboard has flexed and a few of the solder balls or pads have given way at the corner of the CPU or GPU as pointed out by SMTRework and others.

Now why would a motherboard flex?

The answer is the heatsink clamp design.

If you look at the clamp design,  *all* of the pressure to hold the heatsink down is in one tiny point. The centre of the X clamp has a plastic 'spacer button' that sits hard against the bottom of the motherboard and that causes all of the stress to push upwards  (towards the bottom centre of both the GPU and the CPU.)

What is the reaction to that upward force? the corners of the motherboard flexing in the opposite direction.
Because the motherrboard is not screwed down to the chassis in the areas around the CPU and GPU the natural reaction is for the board to flex downwards slightly because of the pressure in the middle pushing upwards.

When the console gets hot, the solder balls may become slightly softer allowing the board to flex further and the break occurs.

Even if the soldering was perfect the poor design of the heatsink clamps means the corners of the CPU and GPU are going to be under tension.

So perhaps the fix would be to remove the clamps altogether, fit 'screw in' pegs to the chassis and screw down the motherboard and the heatsink so rather than tension being on the board to flex, it is actually encouraged to stay flat. The existing holes in the board where the clamp attached would be ideal.

So extra cooling and other measures may not do squat unless you remove the tension of the giant pimple trying to push up from the bottom.

[oZKa]

nice piece of info m8 all of this is so true, i've read about this many times though

[dokworm]

Still going strong since chucking the crappy x clamps in the bin where they belong...

[RBJTech]

ah sorry mate - just posted a thread with an almost identical conclusion (mod section - I guess should have been in chat ... )

....

So what are you using to apply pressure onto the die if you've got rid of the x clamps .. ?

This post has been edited by RBJTech: Feb 26 2007, 01:25 PM

[dogdirt2000]

QUOTE(dokworm @ Feb 23 2007, 04:02 AM)

After reading some of the excellent detective work here re the red lights of death and going over the couple of dead consoles here with a microscope, I've come to the conclusion that in most cases the red lights of death is due to a simple design flaw, not to overheating or poor soldering.

We seem to be seeing consoles where the motherboard has flexed and a few of the solder balls or pads have given way at the corner of the CPU or GPU as pointed out by SMTRework and others.

Now why would a motherboard flex?

The answer is the heatsink clamp design.

If you look at the clamp design,  *all* of the pressure to hold the heatsink down is in one tiny point. The centre of the X clamp has a plastic 'spacer button' that sits hard against the bottom of the motherboard and that causes all of the stress to push upwards  (towards the bottom centre of both the GPU and the CPU.)

What is the reaction to that upward force? the corners of the motherboard flexing in the opposite direction.
Because the motherrboard is not screwed down to the chassis in the areas around the CPU and GPU the natural reaction is for the board to flex downwards slightly because of the pressure in the middle pushing upwards.

When the console gets hot, the solder balls may become slightly softer allowing the board to flex further and the break occurs.

Even if the soldering was perfect the poor design of the heatsink clamps means the corners of the CPU and GPU are going to be under tension.

So perhaps the fix would be to remove the clamps altogether, fit 'screw in' pegs to the chassis and screw down the motherboard and the heatsink so rather than tension being on the board to flex, it is actually encouraged to stay flat. The existing holes in the board where the clamp attached would be ideal.

So extra cooling and other measures may not do squat unless you remove the tension of the giant pimple trying to push up from the bottom.

i agree with you for the most-part as i have had this problem with my 360 before. However, i do not think the main cause of the motherboard flexing is from the Heatsink clamps unless there are abnormalities in the shaping and design of the heatsink and/or x-clamp, which may shift the pressure point "off-center".

Like you said, "The centre of the X clamp has a plastic 'spacer button' that sits hard against the bottom of the motherboard and that causes all of the stress to push upwards". But because the clamp is only physically attached the heatsink on the otherside off the mobo and nothing else, the same force is exerted back through to the pressure point caused by the X clamp as stated by Newton's third law of reciprocal actions (For every action force there is an equal, but opposite reaction).

So to go over that again, the center point of the x-clamp exerts an upwards force on the motherboard, which in turn causes the attached heatsink to be pulled downwards with exactly the same force as the upwards force (assuming the heatsink and x-clamp are perfectly 'shaped' and manufactured). Seeing that the heatsink applies pressure on top of the cpu/gpu, and the x-clamp applies pressure underneath the cpu/gpu, i don't believe the x-clamp and heatsink design are to be the blame........

.......unless (like what mentioned a second ago) there are even minor abnormalities in the shaping of the heatsink and/or x-clamp, which may shift the pressure point "off-center" (eg, if one of the four heatsink clamping pins is ever so slightly different in height to the other three, then the pressure point applied by the heatsink and the x-clamp will not match.....and that's where you may get problems), which means a manufacturing flaw, etc other than a design flaw.

This post has been edited by dogdirt2000: Feb 26 2007, 06:47 PM

[dogdirt2000]

as true as my post above is, i've just remembered/realised that the heatsink actually screws onto the case and is not left to act like a normal x-clamp would in a pc. The reasoning i made remains the same though - the designs are probably 100% correct, but very small abnormalities in the shaping of the heatsink could potentially cause issue whether or not it is screwed onto the 360 metal case anyway.

(apologies for my stupidness)    

[dokworm]

The problem is that it is way beyond the tolerance these things are manufactured to to expect that you get a perfect alignment of force, the slightest bit out and you have an untenable situation.

It is a design flaw as the motherboard area around the CPU and GPU absolutely should be screwed through to the case (via pegs/standoffs) to keep that area of board flat as a tack.
With the heat and airflow on that thing, a single pressure point clamp, and a motherboard only secured via the corners is a recipe for disaster.

A simple design with the mobo secured to the case at the four points around the perimeter of both the CPU and GPU solves the problem. It is just poor design.

Mine is still glitch free so far.


To be clearer it is a combination effect of the x clamp system *and* the motherboard effectively floating that combines to cause the problem.
I can't believe they didn't really lock the mobo down to the case similar to a PC, especially on something designed to run vertically.

[fahrenheit]

With the weight of the CPU heatsink, I think it needs to have a standoff beneath it which will replace the center of the X-clamp. Although the standoff shouldn't apply upward pressure, it needs to be there inorder to prevent the center of the CPU part of the board from drouping.

I wouldn't be at all surprised if the CPU heatsink's weight+heat+horizontal orientation is what finished my unit off. With the front right of my GPU slightly higher than all other corners, a downward movement of the board under the CPU is highly possible.

I've done away with the X-clamps, but I'm far from done with redesigning the mounting. I have a 0020 error to overcome also, so it might take a while, but I've been taking pictures every step of the way and will post them up when I'm happy I've made progress.

[Spawman1926]

Please do post some pics cause I dont even know what the GPU is. I really wanna try what you guys are suggesting if it did indeed work.

[Helliano]

QUOTE(Spawman1926 @ Feb 28 2007, 02:57 PM)

Please do post some pics cause I dont even know what the GPU is. I really wanna try what you guys are suggesting if it did indeed work.

GPU means Graphic Processing Unit, which explains what it stand for, its basicly the brain of the graphics.

[dokworm]

QUOTE(fahrenheit @ Feb 28 2007, 11:08 AM)

With the weight of the CPU heatsink, I think it needs to have a standoff beneath it which will replace the center of the X-clamp. Although the standoff shouldn't apply upward pressure, it needs to be there inorder to prevent the center of the CPU part of the board from drouping.

I wouldn't be at all surprised if the CPU heatsink's weight+heat+horizontal orientation is what finished my unit off. With the front right of my GPU slightly higher than all other corners, a downward movement of the board under the CPU is highly possible.

I've done away with the X-clamps, but I'm far from done with redesigning the mounting. I have a 0020 error to overcome also, so it might take a while, but I've been taking pictures every step of the way and will post them up when I'm happy I've made progress.

I agree, I put a plastic 'peg' under the middle of the CPU as a support to keep that whole section of the board nice and flat.

Does anyone make aftermarket heatsinks for the 360? It would be nice to get some real copper designs with proper clamps and board support from one of the 3rd party vendors and then we could kiss this problem goodbye for everybody.

[Elemino]

I'm not discounting your theory or anything, just trying to help come up with the right answer. So please don't take this the wrong way.

If this issue was caused by the board flexing, how come it does not affect PCs? These days, heat sinks are getting heavier and chunkier all the time. Some of the large ones made by Cooler Master and such companies are huge. They hold on to a processor and board and dangle that weight horizontally. They've gotten so large that some need bracing that goes on the under side of the board to keep the heatsink from ripping the socket off the board. To add to that, in my PC I'm using Cooler Master's Aqua Gate Mini (Seen below) and the pump/heatsink that sits on the processor has a bracing that goes underneither to CPU to litteraly sandwhich, hold and apply force to the CPU to make sure it has good contact as well as a secure mount.

(IMG:http://www.coolermaster-usa.com/ProductImages/1177_1.jpg)

This post has been edited by Elemino: Mar 1 2007, 11:50 PM

[bojngles]

this theory does sound interesting...

[dokworm]

QUOTE(Elemino @ Mar 1 2007, 11:56 PM)

I'm not discounting your theory or anything, just trying to help come up with the right answer. So please don't take this the wrong way.

If this issue was caused by the board flexing, how come it does not affect PCs? These days, heat sinks are getting heavier and chunkier all the time. Some of the large ones made by Cooler Master and such companies are huge. They hold on to a processor and board and dangle that weight horizontally. They've gotten so large that some need bracing that goes on the under side of the board to keep the heatsink from ripping the socket off the board. To add to that, in my PC I'm using Cooler Master's Aqua Gate Mini (Seen below) and the pump/heatsink that sits on the processor has a bracing that goes underneither to CPU to litteraly sandwhich, hold and apply force to the CPU to make sure it has good contact as well as a secure mount.

(IMG:http://www.coolermaster-usa.com/ProductImages/1177_1.jpg)

I'm all for discussing the whys and why nots to try and get the proper answer.

There are multiple reasons you don't get the problem on PCs.

1) The PC mainboard is properly secured to the baseplate via pegs/standoffs at multiple points across the board. This keeps the board nice and flat and prevents warping. On the 360 the board is only secured around the edges leaving it lots of room to warp and flex.

2) Most of the PC heatsink clamping methods I have seen put the clamping pressure *outside* the perimeter of the CPU, i.e. it attaches to the board or baseplate at 4 anchor points which surround the chip(which is what should be done on the 360) i.e. the clamp pressure is not focussed on the bottom side of the board under the centre of the CPU.

If the CPU/GPU was just a flat plate that sat directly on the board perhaps you wouldn't get as much of a problem, but the CPU/GPU sit above the board on top of the solder balls. So there is an air gap and therefore no equivalent downwards pressure from the top.
So the upwards pressure point under the centre of the CPU/GPU does not have the CPU/GPU sitting hard against the top of the board so the board wants to flex upwards at the centre point (towards the chips) causing the rest of the board to want to flex downwards (away from the chips).

You can simulate the effect quite easily and see it happen.

Clamping the chip down properly (like on a PC) and securing the boards properly (like on a PC) makes the problem go away.

Last night I put the original clamps back on and had the ring of death within 20 minutes, put my PC style clamps back on and played for the rest of the night without incident. I fired it up again today and it is still running just fine.

So it is looking like the culprit, but really I will only know if my 360 is still running in 6 months time I guess.

[fahrenheit]

@Elemino -

You have to factor in the quality of the PCB that is used to produce the 360 motherboards. Its poor. Some PC motherboards cost more than the 360 itself and are of far higher quality componentry. The 360 is a high volume, mass produced product that is made as cheaply as possible because it is made at a loss to the supplier (MS).

You aren't comparing apples with apples.