Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.

tsialex

Contributor
Original poster
Jun 13, 2016
13,454
13,601
This will be a wiki about the early-2009 dual CPU tray issues, from the Rev. A not supporting 130W Westmere Xeons to the MOSFETs issues with cooling, damaged MOSFETs, replacement for the degraded thermal pads and etc.

A thing to note, early-2009 factory refurbished trays, with that long 17-digit serial number are not Rev. A.
 
Last edited:

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
A mistake, you should always de-lid when installing to a dual CPU early-2009 tray. Besides the risk of damaging the CPU socket pins, the power plane is not cooled correctly because of the height difference.

No, exactly the opposite, BH9 is a mid-2010/mid-2012 backplane, not compatible with a early-2009 CPU tray.

Useless at this moment.

Yes, all mid-2010/mid-2012 backplanes work with Westmere Xeons independent of the firmware installed, but not with your early-2009 CPU tray.

You should repair your current BootROM image or buy a new CPU tray.

You can't mix a mid-2010/mid-2012 backplane with an early-2009 CPU tray.
whats the best way to delid the CPUs?
 

BrianRecchia

macrumors member
Jan 17, 2008
35
3
For me, pre-heating the IHS with hot air/hot plate to around 180ºC then removing it with a vice is the fastest and less aggressive way to do it.

See the various ways to do it and try with a cheap Xeon first instead of a still somewhat expensive X5690.
thanks for the hot air tip! I’ll try that with the 5675 I’ve got coming in the mail!

I’ve just been chucking them in the vise (one vise jaw is on the lip of the IHS, the other is on the PCB edge) and cranking down extremely slowly until I feel it start to give a little, and then continuing even slower with steady pressure and my hand around the CPU.

Really, the hardest part for me is removing the solder thermal interface material from the top of the CPU die. I used a razor blade before and did a reasonably good job, but the die looked a little scratched up afterwards, definitely less than ideal. I also only have double-edge razor blades handy, so I kind of cut my hand up and don’t want to do that again 😅

I got a couple 5675’s in the mail to delid and install, but in lieu of a double-edged razor blade I tried scraping with a folding knife and ended up knocking loose one of the SMD components. It might have been possible to solder it back, but for like $11 or less a piece on eBay I decided to just order another one (and look into a better way to remove the STIM before risking the other chips…)

I had an email chat with one of the people from rockitcool.com about their Quicksilver STIM removal product, and the person I spoke with said that they’ve used that product themselves dozens of times to delid Xeons for Mac Pros, so I’m comfortable that it’s compatible. Now I just have to order it and impatiently wait!

I’m definitely a little worried about it, because I can only assume that it dissolves solder similarly to the way gallium eats aluminum… I definitely will make sure to clean the crap out of the die before putting the chip anywhere near the aluminum Mac Pro!
 
  • Like
Reactions: JedNZ and tsialex

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
For me, pre-heating the IHS with hot air/hot plate to around 180ºC then removing it with a vice is the fastest and less aggressive way to do it.

See the various ways to do it and try with a cheap Xeon first instead of a still somewhat expensive X5690.

I've been using a lid 5690 for about 2 years now, and its never given me an issue, is that because I was on Mojave? I just updated to Monterey 12.6 with OC 0.8.0 using TB4 Titan Ridge card. You think I'll have some issues? What's the best way to diagnose? Is there an app that can evaluate the lifespan of the CPU?
 

tsialex

Contributor
Original poster
Jun 13, 2016
13,454
13,601
I've been using a lid 5690 for about 2 years now, and its never given me an issue, is that because I was on Mojave? I just updated to Monterey 12.6 with OC 0.8.0 using TB4 Titan Ridge card. You think I'll have some issues? What's the best way to diagnose? Is there an app that can evaluate the lifespan of the CPU?

Like I wrote before, the damage happens to the CPU tray, not to the Xeon. The MOSFETS of the power plane are not cooled when you install lidded Xeons to an early-2009 dual CPU tray, there is no contact between the heatsink and the transistors because of the added height of the IHS.

What's the best way to diagnose? Is there an app that can evaluate the lifespan of the CPU?

There is no diagnostic for this, or you compensate the height difference with additional thermal pads for the power plane, or you will get a damaged CPU tray in the long run.

Besides that, there is always the risk of damaging the pins of the CPU socket.
 

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
Like I wrote before, the damage happens to the CPU tray, not to the Xeon. The MOSFETS of the power plane are not cooled when you install lidded Xeons to an early-2009 dual CPU tray, there is no contact between the heatsink and the transistors because of the added height of the IHS.



There is no diagnostic for this, or you compensate the height difference with additional thermal pads for the power plane, or you will get a damaged CPU tray in the long run.

Besides that, there is always the risk of damaging the pins of the CPU socket.
I just looked it up, I've been using lids for about 3 years. I started with High Sierra and followed this vid below:

This step by step process suggested I needed extra thermal pads and I did add a thick layer to compensate for height. Maybe that's the way its been helping?

Either way you have me curious, If I open up the tray, do you know what signs I can look for to see if my tray is being effected? Any pics so I can compare if I have damage?

BTW, I've talked with you here many years ago, so again, after many years of your insight, THANK YOU!
 

BrianRecchia

macrumors member
Jan 17, 2008
35
3
I just looked it up, I've been using lids for about 3 years. I started with High Sierra and followed this vid below:

This step by step process suggested I needed extra thermal pads and I did add a thick layer to compensate for height. Maybe that's the way its been helping?

Either way you have me curious, If I open up the tray, do you know what signs I can look for to see if my tray is being effected? Any pics so I can compare if I have damage?

BTW, I've talked with you here many years ago, so again, after many years of your insight, THANK YOU!
It’s better than nothing for thermals, but really not nearly as safe as delidding the chips, even without talking about the sockets. i’m not sure guy in the video you linked also is putting thermal pads on everything that needs them, and stacking thermal pads is a VERY BAD IDEA (you’ll end up with tiny air gaps between them that destroy their effectiveness). It’s also not a great idea to combine paste and pads, but not AS bad…

If you really want to stick with your stock CPUs, I’d order thicker thermal pads (3mm) and some 2.5mm washers to put around the screws so you can’t over-tighten them. Otherwise you really risk damaging the CPUs and sockets.

Or you can just delid them. Or just try your luck, which might be alright! But I really can’t recommend it at all with everything I know about thermal interfaces and CPU sockets.

There’s no way to check the components for thermal damage. You could monitor the VRM voltages using software like iStat Menus (for macOS) or HWiNFO64 (for Windows) while putting the CPUs under a heavy load, but that’ll miss most of the picture. Unless things look incredibly out of whack using software to monitor it, assume that the VRMs and MOSFETs are fine for now and plan to redo the thermal interface components some time soon…
 
Last edited:

Macschrauber

macrumors 68030
Dec 27, 2015
2,980
1,487
Germany
We getting way off topic here.

But I would nail it down to this: the board design is for lidless CPUs. Getting off the specs is not good for a long lasting working cpu board.

Adding thermal stripes for the regulators will add thermal resistance. Adding none and Leaving a gap between is just nasty.
 
  • Like
Reactions: tsialex

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
We getting way off topic here.

But I would nail it down to this: the board design is for lidless CPUs. Getting off the specs is not good for a long lasting working cpu board.

Adding thermal stripes for the regulators will add thermal resistance. Adding none and Leaving a gap between is just nasty.
Thanks for the reminder on the off-topic. Maybe we need an influencer to do a video? on this topic.

Also, I bought/upgraded a different 2009 tray (off ebay) and am seeing things on MOSFETs after I updated to Monterey. My first workhorse, I have not opened to check but on this second one, I'm able to see some things that might look like moisture. Not sure, but I wana share some pics of what I found. What's the best thread to start this convo? Any suggestions?
 

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
It’s better than nothing for thermals, but really not nearly as safe as delidding the chips, even without talking about the sockets. i’m not sure guy in the video you linked also is putting thermal pads on everything that needs them, and stacking thermal pads is a VERY BAD IDEA (you’ll end up with tiny air gaps between them that destroy their effectiveness). It’s also not a great idea to combine paste and pads, but not AS bad…

If you really want to stick with your stock CPUs, I’d order thicker thermal pads (3mm) and some 2.5mm washers to put around the screws so you can’t over-tighten them. Otherwise you really risk damaging the CPUs and sockets.

Or you can just delid them. Or just try your luck, which might be alright! But I really can’t recommend it at all with everything I know about thermal interfaces and CPU sockets.

There’s no way to check the components for thermal damage. You could monitor the VRM voltages using software like iStat Menus (for macOS) or HWiNFO64 (for Windows) while putting the CPUs under a heavy load, but that’ll miss most of the picture. Unless things look incredibly out of whack using software to monitor it, assume that the VRMs and MOSFETs are fine for now and plan to redo the thermal interface components some time soon…
Mac 2009 Tray1 copy.png

So, this is the 2nd 2009 cMP tray I've worked on in the last week.

The MOSEFTS seem to be showing moisture? Do you think its from the thermal pads I'm using? I'm stacking those thin pads so its matches about 3mm.

Also, what do you think is showing on one of the CPUs? Do you see a bit of discolor, I've circled on the image.
 

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
It’s better than nothing for thermals, but really not nearly as safe as delidding the chips, even without talking about the sockets. i’m not sure guy in the video you linked also is putting thermal pads on everything that needs them, and stacking thermal pads is a VERY BAD IDEA (you’ll end up with tiny air gaps between them that destroy their effectiveness). It’s also not a great idea to combine paste and pads, but not AS bad…

If you really want to stick with your stock CPUs, I’d order thicker thermal pads (3mm) and some 2.5mm washers to put around the screws so you can’t over-tighten them. Otherwise you really risk damaging the CPUs and sockets.

Or you can just delid them. Or just try your luck, which might be alright! But I really can’t recommend it at all with everything I know about thermal interfaces and CPU sockets.

There’s no way to check the components for thermal damage. You could monitor the VRM voltages using software like iStat Menus (for macOS) or HWiNFO64 (for Windows) while putting the CPUs under a heavy load, but that’ll miss most of the picture. Unless things look incredibly out of whack using software to monitor it, assume that the VRMs and MOSFETs are fine for now and plan to redo the thermal interface components some time soon…
I see a different reply in my email, I think it was lost during the transfer of the thread into a new wikipost, but can you share the info on vise again? Not just for me but maybe someone else who might want to research later cuz I have info in my email but its missing from this thread.

thanks
 
  • Like
Reactions: BrianRecchia

BrianRecchia

macrumors member
Jan 17, 2008
35
3
I see a different reply in my email, I think it was lost during the transfer of the thread into a new wikipost, but can you share the info on vise again? Not just for me but maybe someone else who might want to research later cuz I have info in my email but its missing from this thread.

thanks
I edited that post a few times and then ended up just rewriting it, so I apologize! I’ll rewrite how to do it below. I watched a video on YouTube to learn how (can’t recall which one), but it’s simple enough and I’ll outline the steps again. I apologize in advance for repeating myself a lot through this, but I really want to stress certain points to minimize any risk.

I use a heavy cast iron vise, but it’s just what I have around. (Got it for free at a garage sale years ago, and it’s come in handy for so many things… a used heavy duty vise is an amazing tool to have, but I digress!)

First, open the vise jaws so that it’s about as wide as you’d need to fit the CPU in flat. You actually want to put it in at a bit of an angle, though: one jaw goes on the edge of the CPU’s PCB, the other jaw goes on the lip of the IHS. You probably can go “flatter” by putting the vise on the lower edge of the IHS closer to the PCB, but I worry about damaging it that way and haven’t had a problem.

Look at the engraving on the IHS. When the text is right-side up, the right-hand edge of the IHS is where you want to push from. That is, you want to slide the lip of the right side of the IHS towards the left edge of the CPU module’s PCB. Forcibly sliding the IHS like this will break the solder and epoxy bonds holding them together, so this is the goal, and there are tiny components under the IHS that you want to avoid because there’s a chance things can dismount sort of violently risking damage to those small components. the right side of the CPU die is clear of them, so this orientation will clear that.

[ Intel Xeon ] ← push the IHS this way.

Anyway, mount the CPU in the vise UPSIDE DOWN so that the LGA pads are pointing up, the lip of the right edge of the IHS is against one vise jaw, and the CPU’s PCB is against the other jaw, and tighten it just enough to keep it in place. (Please look up a photo or YouTube tutorial because I worry I’m not being clear enough!)

I’m left-handed, so for me the vise crank is on the left side at this point with that jaw against the PCB, and I’m going to be using my right hand to steady it and prepare to catch the CPU and IHS. If you’re right handed you’ll probably want to turn the vise around, but you still need to really make sure that the CPU is oriented correctly!

At this point, optionally heat the CPU with a hot air gun or hair dryer or something to soften things. Do not use anything like a lighter or a flame, and you definitely don’t want to get it too hot to touch it. I didn’t use heat at all for all four I’ve done and it was fine, but it’ll be more controlled and less violent (and probably require less force!) if you do heat it. I’m going to heat up the X5675 that I delid tomorrow.

Once the cpu is mounted in the vise and is optionally heated up, place your hand around the chip, preparing to catch it, because it will probably fall out when you’re done and you want it to not go flying! Don’t grab it, definitely don’t squeeze it, just prepare to catch it if it falls or violently flies.

I hold my palm down against the vise, with my fingers curled gently around the entire CPU module. Don’t squeeze, and don’t get your fingers too close to the vise jaws. You don’t want to hurt yourself. You just want to catch it when it falls, with your fingers around the edges so that nothing can fly away…

Oh, and if you have an ESD wrist strap you should probably be wearing it, because I don’t see this being possible to do without touching the LGA pads.

Begin to tighten the vise jaws slowly and steadily, feeling how much force it takes to tighten. I need to use both hands plus the table I’m working on to keep it steady, so I have to be mindful to keep my hand in the right spot so that I can catch it.

After you’ve tightened it a good bit more (you might notice the PCB flexing very slightly) you’ll feel the amount of force you need to tighten the vise will drop considerably! Stop tightening! This is when you have to really be careful.

Take a breath, and then turn the crank SLOWLY and with a very constant pressure, with your hand in the right spot. If you’re lucky it will gently come apart into your hand, possibly with a satisfying sound. If you’re less lucky it’ll violently fly, possibly sooner than you expect. If you’re REALLY unlucky or did something wrong you’ll be going on eBay to get another CPU :)

At this point the IHS has been removed but you are NOT done! In between the IHS and the silicon CPU die is a special thermal compound called solder thermal interface material, or STIM. It’s soft metal that is adhered to the silicon, and you need to remove it in order for the heat sink to make direct contact with the CPU die. It has to be as flat and smooth as possible.

There are a few ways to remove the STIM, and the way I did for the first couple of chips was by scraping it off with a razor blade. This is kind of dangerous because you can slip and hurt yourself or damage the CPU.

A better way is to use a liquid metal that will dissolve the STIM, and then you can clean them both off. I’ve heard really good things about Rockit Cool Quicksilver, and I have a large syringe full on order, but I haven’t personally tried it yet. Its main ingredient is gallium, and you might be able to do this with pure elemental gallium which is relatively inexpensive and not too hard to find online. (You can buy a test tube of gallium on Amazon.com for like $10.)

If you use a liquid metal to “eat” the STIM, you need to be very aware that it will also eat aluminum and is extremely potent. Do not get it anywhere near your Mac Pro tower, or you’ll be really sorry. Follow directions included, or look up a tutorial specific to it, because I definitely don’t feel comfortable trying to come up with that yet!

Once you have it all cleaned up, you’re set! I take a metallic Sharpie permanent marker and write the model of the chip on the PCB after, because all of the identifying markings are on the IHS you’ve removed.

Apply a small amount of thermal compound to the CPU die only when reassembling.
 

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
I edited that post a few times and then ended up just rewriting it, so I apologize! I’ll rewrite how to do it below. I watched a video on YouTube to learn how (can’t recall which one), but it’s simple enough and I’ll outline the steps again. I apologize in advance for repeating myself a lot through this, but I really want to stress certain points to minimize any risk.

I use a heavy cast iron vise, but it’s just what I have around. (Got it for free at a garage sale years ago, and it’s come in handy for so many things… a used heavy duty vise is an amazing tool to have, but I digress!)

First, open the vise jaws so that it’s about as wide as you’d need to fit the CPU in flat. You actually want to put it in at a bit of an angle, though: one jaw goes on the edge of the CPU’s PCB, the other jaw goes on the lip of the IHS. You probably can go “flatter” by putting the vise on the lower edge of the IHS closer to the PCB, but I worry about damaging it that way and haven’t had a problem.

Look at the engraving on the IHS. When the text is right-side up, the right-hand edge of the IHS is where you want to push from. That is, you want to slide the lip of the right side of the IHS towards the left edge of the CPU module’s PCB. Forcibly sliding the IHS like this will break the solder and epoxy bonds holding them together, so this is the goal, and there are tiny components under the IHS that you want to avoid because there’s a chance things can dismount sort of violently risking damage to those small components. the right side of the CPU die is clear of them, so this orientation will clear that.

[ Intel Xeon ] ← push the IHS this way.

Anyway, mount the CPU in the vise UPSIDE DOWN so that the LGA pads are pointing up, the lip of the right edge of the IHS is against one vise jaw, and the CPU’s PCB is against the other jaw, and tighten it just enough to keep it in place. (Please look up a photo or YouTube tutorial because I worry I’m not being clear enough!)

I’m left-handed, so for me the vise crank is on the left side at this point with that jaw against the PCB, and I’m going to be using my right hand to steady it and prepare to catch the CPU and IHS. If you’re right handed you’ll probably want to turn the vise around, but you still need to really make sure that the CPU is oriented correctly!

At this point, optionally heat the CPU with a hot air gun or hair dryer or something to soften things. Do not use anything like a lighter or a flame, and you definitely don’t want to get it too hot to touch it. I didn’t use heat at all for all four I’ve done and it was fine, but it’ll be more controlled and less violent (and probably require less force!) if you do heat it. I’m going to heat up the X5675 that I delid tomorrow.

Once the cpu is mounted in the vise and is optionally heated up, place your hand around the chip, preparing to catch it, because it will probably fall out when you’re done and you want it to not go flying! Don’t grab it, definitely don’t squeeze it, just prepare to catch it if it falls or violently flies.

I hold my palm down against the vise, with my fingers curled gently around the entire CPU module. Don’t squeeze, and don’t get your fingers too close to the vise jaws. You don’t want to hurt yourself. You just want to catch it when it falls, with your fingers around the edges so that nothing can fly away…

Oh, and if you have an ESD wrist strap you should probably be wearing it, because I don’t see this being possible to do without touching the LGA pads.

Begin to tighten the vise jaws slowly and steadily, feeling how much force it takes to tighten. I need to use both hands plus the table I’m working on to keep it steady, so I have to be mindful to keep my hand in the right spot so that I can catch it.

After you’ve tightened it a good bit more (you might notice the PCB flexing very slightly) you’ll feel the amount of force you need to tighten the vise will drop considerably! Stop tightening! This is when you have to really be careful.

Take a breath, and then turn the crank SLOWLY and with a very constant pressure, with your hand in the right spot. If you’re lucky it will gently come apart into your hand, possibly with a satisfying sound. If you’re less lucky it’ll violently fly, possibly sooner than you expect. If you’re REALLY unlucky or did something wrong you’ll be going on eBay to get another CPU :)

At this point the IHS has been removed but you are NOT done! In between the IHS and the silicon CPU die is a special thermal compound called solder thermal interface material, or STIM. It’s soft metal that is adhered to the silicon, and you need to remove it in order for the heat sink to make direct contact with the CPU die. It has to be as flat and smooth as possible.

There are a few ways to remove the STIM, and the way I did for the first couple of chips was by scraping it off with a razor blade. This is kind of dangerous because you can slip and hurt yourself or damage the CPU.

A better way is to use a liquid metal that will dissolve the STIM, and then you can clean them both off. I’ve heard really good things about Rockit Cool Quicksilver, and I have a large syringe full on order, but I haven’t personally tried it yet. Its main ingredient is gallium, and you might be able to do this with pure elemental gallium which is relatively inexpensive and not too hard to find online. (You can buy a test tube of gallium on Amazon.com for like $10.)

If you use a liquid metal to “eat” the STIM, you need to be very aware that it will also eat aluminum and is extremely potent. Do not get it anywhere near your Mac Pro tower, or you’ll be really sorry. Follow directions included, or look up a tutorial specific to it, because I definitely don’t feel comfortable trying to come up with that yet!

Once you have it all cleaned up, you’re set! I take a metallic Sharpie permanent marker and write the model of the chip on the PCB after, because all of the identifying markings are on the IHS you’ve removed.

Apply a small amount of thermal compound to the CPU die only when reassembling.
Great breakdown! BUT, if you will be performing this task soon, I bet many would love to see the video.
 

aclarubicin

macrumors member
May 27, 2018
76
20
Any tips on what to do with trays that have CPU B slot not working, but CPU A does?

Got two such boards, both look pristine and no bent/damaged CPU pins ...
 

tsialex

Contributor
Original poster
Jun 13, 2016
13,454
13,601
Any tips on what to do with trays that have CPU B slot not working, but CPU A does?

Still boots, but SMC will run fans at max RPM, no one in the right mind will tolerate the noise.

Got two such boards, both look pristine and no bent/damaged CPU pins ...

Inspect the MOSFETs, probably you will find one damaged on the CPU B power plane.
 
Last edited:

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
Still boots, but SMC will run fans at max RPM, no one in the right mind will tolerate the noise.



Inspect the MOSFETs, probably you will find one damaged on the CPU B power plane.
This happened on my 2nd cMP I updated, but am using MAC Fans Control to control the fan speed so it runs with ambient temps. Do you suggest against using this type of app for that issue "fans running at full speed"

 

tsialex

Contributor
Original poster
Jun 13, 2016
13,454
13,601
When the SMC detects a failure with the CPU socket, enters fail-safe mode and accelerate the fans to the maximum rotation, no app can control it.
 

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
When the SMC detects a failure with the CPU socket, enters fail-safe mode and accelerate the fans to the maximum rotation, no app can control it.

I'm able to control my cMP fans with the app above. Is it because its using OC 0.8.0 and Monterey 12.6? I can post a video to show you when the app is off and you can hear the fans running full speed but then with the app I can switch the fan speed by taking it out of AUTO and into sensor-based, ambient.
 

tsialex

Contributor
Original poster
Jun 13, 2016
13,454
13,601
I'm able to control my cMP fans with the app above. Is it because its using OC 0.8.0 and Monterey 12.6? I can post a video to show you when the app is off and you can hear the fans running full speed but then with the app I can switch the fan speed by taking it out of AUTO and into sensor-based, ambient.
If you can control via an app, the failure is not the one I'm referring and it's not fail-safe mode. You are writing about something else.
 

chuco915

macrumors member
May 12, 2020
61
3
El Chuco, Tejas
If you can control via an app, the failure is not the one I'm refering and it's not fail-safe mode.

That makes sense. Any idea why the fans are running at full speed? After I updated the CPUs to 5690, both sockets, is when this happened. My first cMP I did the upgrade to does not do that, but for this one not sure what I did wrong. I did exactly same thing from the video and on my first cMP.

Any ideas what it could be?
 

tsialex

Contributor
Original poster
Jun 13, 2016
13,454
13,601
That makes sense. Any idea why the fans are running at full speed? After I updated the CPUs to 5690, both sockets, is when this happened. My first cMP I did the upgrade to does not do that, but for this one not sure what I did wrong. I did exactly same thing from the video and on my first cMP.

Any ideas what it could be?
Nope, too much variables. Start running ASD and see what it detects.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.