MP 1,1-5,1 Mac Pro A1289 Broken Chip Advice

[mgaudio]

Mac Pro A1289 Broken Chip Advice:

Hi all . . .
I've already used many of the various member's tips/protocols in setting up my macs, and was just in the process of troubleshooting frequent crashes on a recently-acquired Mac Pro - when I think I made a collossal screw-up.

Quick Summary:
I knocked this tiny chip off the processor tray. Is the entire tray now completely hosed ? or is it fixable ?
The mac does not boot after pressing the power button anymore.
The broken-off chip number says "K72WO". It looks like this:

K72WO,marking code K72WO,smd marking K72WO,information K72WO,branding K72WO,-szgolon pcb assembly

www.szgolon.com

I cannot tell where it came from on the CPU tray. (see attached jpg for actual photo)

Extended History:
I recently got a fantastic deal on a couple of Mac Pros.
1. 4,1 Dual 4-Core 2.26 GHz A1289 (2009) - to be used as a backup rig for my identical (already upgraded) main Pro Tools computer.
2. 5,1 Dual 6-Core 2.66 GHz A1289 (2010) - to be used as a Dolby Atmos Renderer.
They both checked out fine initially.
I updated them both with SSD drives.
I bought a 12-core 3.46GHz CPU Upgrade kit for the 4,1 and successfully updated the machine firmware to a 5,1 before install.
I then updated to the newer Boot ROM via a clean install of macOS High Sierra 10.13.6.
I then upgraded the ATI Radeon HD 5770 to a mac-flashed GeForce GTX680 video card, and on another partition installed macOS Mojave 10.14.6 - flashing again to the latest 144.0.0.0.0 ROM during install.

I was experimenting with OpenCore combined with dosdude's patcher for Catalina, and had finally managed to get High Sierra, Mojave, and Catalina all booting from separate partitions, and all seeing my flashed Titan Ridge TB3 card at a speed of 40Gbps. (using dosdude's Catalina install version brought my built-in Airport card back to life in that OS)

The 2.66 is working flawlessly, but the 3.46 kept seizing up the mouse and screen after 30-40 mins of use.
I thought that perhaps I had not done a proper job of installing the new CPUs, so I had been in the process of cleaning them up again, and thought I should also do the Northbridge heatsink, and re-apply Arctic Silver. Seemed to go smoothly. Machine booted.
But still no luck with the freezes. Carried out deep RAM tests, swapped RAM, zapped PRAM, no luck. So I disassembled again, and then suddenly noticed this tiny little item on my worksurface. Under microscopic examination I could see the letters "K72WO" on the top, and six tiny pins. Crap ! I've broken off a microscopic chip from somewhere unknown on the CPU tray.

I would appreciate any advice that anyone could share on how best to deal with this.

Thanks in advance . . .

 

[tsialex]

Mac Pro A1289 Broken Chip Advice:

Hi all . . .
I've already used many of the various member's tips/protocols in setting up my macs, and was just in the process of troubleshooting frequent crashes on a recently-acquired Mac Pro - when I think I made a collossal screw-up.

Quick Summary:
I knocked this tiny chip off the processor tray. Is the entire tray now completely hosed ? or is it fixable ?
The mac does not boot after pressing the power button anymore.
The broken-off chip number says "K72WO". It looks like this:

K72WO,marking code K72WO,smd marking K72WO,information K72WO,branding K72WO,-szgolon pcb assembly

www.szgolon.com

I cannot tell where it came from on the CPU tray. (see attached jpg for actual photo)

Extended History:
I recently got a fantastic deal on a couple of Mac Pros.
1. 4,1 Dual 4-Core 2.26 GHz A1289 (2009) - to be used as a backup rig for my identical (already upgraded) main Pro Tools computer.
2. 5,1 Dual 6-Core 2.66 GHz A1289 (2010) - to be used as a Dolby Atmos Renderer.
They both checked out fine initially.
I updated them both with SSD drives.
I bought a 12-core 3.46GHz CPU Upgrade kit for the 4,1 and successfully updated the machine firmware to a 5,1 before install.
I then updated to the newer Boot ROM via a clean install of macOS High Sierra 10.13.6.
I then upgraded the ATI Radeon HD 5770 to a mac-flashed GeForce GTX680 video card, and on another partition installed macOS Mojave 10.14.6 - flashing again to the latest 144.0.0.0.0 ROM during install.

I was experimenting with OpenCore combined with dosdude's patcher for Catalina, and had finally managed to get High Sierra, Mojave, and Catalina all booting from separate partitions, and all seeing my flashed Titan Ridge TB3 card at a speed of 40Gbps. (using dosdude's Catalina install version brought my built-in Airport card back to life in that OS)

The 2.66 is working flawlessly, but the 3.46 kept seizing up the mouse and screen after 30-40 mins of use.
I thought that perhaps I had not done a proper job of installing the new CPUs, so I had been in the process of cleaning them up again, and thought I should also do the Northbridge heatsink, and re-apply Arctic Silver. Seemed to go smoothly. Machine booted.
But still no luck with the freezes. Carried out deep RAM tests, swapped RAM, zapped PRAM, no luck. So I disassembled again, and then suddenly noticed this tiny little item on my worksurface. Under microscopic examination I could see the letters "K72WO" on the top, and six tiny pins. Crap ! I've broken off a microscopic chip from somewhere unknown on the CPU tray.

I would appreciate any advice that anyone could share on how best to deal with this.

Thanks in advance . . .

View attachment 1823864

You damaged the early-2009 or the mid-2010? You don't clearly tell and the dual CPU trays are different between early-2009 and mid-2010/mid-2012.

Btw, if you ripped PCB pads or traces, it's dead - if the pads/traces are intact, maybe you can solder it back.

 

[flashy-cat]

The fact you're even asking this question clearly demonstrates you're out of your depth.

 

[mgaudio]

obviously, that's why people ask questions - in the hope of getting some advice from an expert. the fact that you felt compelled to respond in such a demeaning manner clearly demonstrates that you are out of your depth in answering it.

 

[mgaudio]

You damaged the early-2009 or the mid-2010? You don't clearly tell and the dual CPU trays are different between early-2009 and mid-2010/mid-2012.

Btw, if you ripped PCB pads or traces, it's dead - if the pads/traces are intact, maybe you can solder it back.

Thanks for the response. As mentioned the 2.66 (2010 machine) is working flawlessly with the various upgrades. It's the 2009 machine that I upgraded the CPU's to 3.46, which is now unbootable. The entire CPU board on this now-unbootable machine looks in fairly good shape cosmetically. I did not slip with any tools, so I'm not sure how the transistor came off. I cannot even see a possible location, even under magnification.

I suspect that this component may have been unstable before (hence the seizes after 30-40 mins), possibly through overheating - and maybe this tiny transistor was already on the edge of disconnecting. It might have been a dry joint. The machine was full of dust when I bought it, but I cleaned it out thoroughly - all fans, heatsinks, CD drive area etc. I was simply hoping that someone here had some experience with such a situation and could offer advice as to whether I should give up in the hope of such a microscopic soldering repair, or if there's a professional that they've dealt with in the past who has successfully carried it out such a delicate repair. It's beyond my soldering skills.

I did also notice some discoloration on the CPU tray surface in certain areas, which appeared to me as if some component had leaked or a spill had occurred in the past. Here's a screenshot of the area:

 

[Macschrauber]

From the picture you see traces at the chip. So something is ripped off.

check further, even on the backside of the boards.

 

[KeesMacPro]

I did also notice some discoloration on the CPU tray surface in certain areas, which appeared to me as if some component had leaked or a spill had occurred in the past

IMO this is normal; on the heatsink in this area (voltage regulators) there are heatpads and they "sweat" a bit (and dust sticks to it).

 

[mgaudio]

From the picture you see traces at the chip. So something is ripped off.

check further, even on the backside of the boards.

ok, I'll see if I can see any possible location on the rear of the board.
I'll try to take some better magnified photos of the chip too, so that I can show them to any potential technician for evaluation.

 

[mgaudio]

IMO this is normal; on the heatsink in this area (voltage regulators) there are heatpads and they "sweat" a bit.

Oh OK, many thanks for that info-tip. I had no idea that kind of "sweating" happened.
It gives me some hope that this board is not a complete turkey !

 

[mgaudio]

for anyone interested, here's some slightly closer shots.

 

[Analog Kid]

I wouldn't apply power again until you figure out where this came from and replace it. It was there for a reason, now it's not there. It's entirely possible that powering the system without it will damage something else.

 

[tsialex]

At least one pad/trace was ripped, maybe more:

This need to be evaluated by a professional, maybe this damage can still be repaired by someone very competent with micro soldering and some kynar wire jumpers. To be repaired correctly, this won't be cheap, so I'd look at a replacement CPU tray before. Being an early-2009, it's a lot cheaper than a mid-2010 dual CPU tray.

 

[mgaudio]

At least one pad/trace was ripped, maybe more:

View attachment 1824173

This need to be evaluated by a professional, maybe this damage can still be repaired by someone very competent with micro soldering and some kynar wire jumpers. To be repaired correctly, this won't be cheap, so I'd look at a replacement CPU tray before. Being an early-2009, it's a lot cheaper than a mid-2010 dual CPU tray.

wow. you've got sharp eyes tsialex. many thanks.
ok, I'll reach out to a few people for a quote, but I think you're right about the cost.
looks like I'll need a CPU tray - hopefully I could just get the raw board without CPUs, heatsinks.
anything to watch out for eg. slight changes/variations in models for the 2009 ? or are they all essentially the same ?

 

[mgaudio]

I wouldn't apply power again until you figure out where this came from and replace it. It was there for a reason, now it's not there. It's entirely possible that powering the system without it will damage something else.

thanks.
yeah, I only tried twice.
now the machine is sitting sadly in the corner, relegated to the substitute bench of Leeds United. ?

 

[tsialex]

wow. you've got sharp eyes tsialex. many thanks.
ok, I'll reach out to a few people for a quote, but I think you're right about the cost.
looks like I'll need a CPU tray - hopefully I could just get the raw board without CPUs, heatsinks.
anything to watch out for eg. slight changes/variations in models for the 2009 ? or are they all essentially the same ?

It's just one part number for the early 2009 dual CPU tray, 661-4998.

The issue is that the initial run of early-2009 dual CPU trays frequently don't work with X5690 (X5690 was released much later on, Q1-2011) and have some trouble with some X5680, there are lot's of reports of this issue here over the years, and you can't decipher the build date from the CPU tray serial number and no dates are marked on the PCB. So, you have to test.

The early made CPU trays that have trouble with X5690 usually work perfectly fine with X5675.

 

[mgaudio]

holy ****. how do you know all this stuff ?!
LOL.
Many thanks dude. respect.

 

[Macschrauber]

The most people scratch off the chips north of the GPU power connectors. Take a look there if one is missing.

if you cant see with the bare eyes take a good camera.

 

[Macschrauber]

this is the area, when pulling the gpu cables those chips sometimes get off. I have no board here to look atm.
If not someone else jumps in I can inspect it Monday eve.

 

[tsialex]

The most people scratch off the chips north of the GPU power connectors. Take a look there if one is missing.

if you cant see with the bare eyes take a good camera.

Nope, the two voltage/amperage sensors for the PCIe AUX connectors are 8-pin chips with a even smaller encapsulament/pitch than the ripped one.

There is another chip with a very similar size, but it's a 5-pin one. I don't think that is something on that region of the backplane.

 

[Macschrauber]

Just an idea, have no 4.1/5.1 here during the weekend.

 

[tsialex]

@mgaudio, since the damage is done and you right now still don't know which is the defective part, install the mid-2010 CPU tray on the early-2009 Mac Pro, the fans will run at full RPM/full time without any way of software control and the noise will be insupportable with the SMC firmware version mismatch between the backplane/CPU tray, but will work fine for a test.

Check if you can use the early-2009 Mac Pro "normally", forget the jet engine noise - if you can, the problem should be on the CPU tray. If you can't, then the problem can be elsewhere. You can also do the reverse, install the early-2009 dual CPU tray in the mid-2010 Mac Pro and check if you can boot.

 

[Syncretic]

Could be Q8900, about 5cm behind slot 4. Alternatively, maybe Q2610, near Little Frank. Lots of places to hide a tiny 6-pin SOIC.

 

[mgaudio]

@mgaudio, since the damage is done and you right now still don't know which is the defective part, install the mid-2010 CPU tray on the early-2009 Mac Pro, the fans will run at full RPM/full time without any way of software control and the noise will be insupportable with the SMC firmware version mismatch between the backplane/CPU tray, but will work fine for a test.

Check if you can use the early-2009 Mac Pro "normally", forget the jet engine noise - if you can, the problem should be on the CPU tray. If you can't, then the problem can be elsewhere. You can also do the reverse, install the early-2009 dual CPU tray in the mid-2010 Mac Pro and check if you can boot.

oh ok. I actually had tried putting in the 2010 tray a day ago but quickly shut down once I heard the "jet noise" in case I was going to cause some damage, so I didn't even get far enough in that test. I might try that again. At least I'd then truly know where the problem lies. And I'd never even thought of vice-versa.

I do actually have yet another machine which is a fully-operational 2009 4,1>5,1/2.29>3.46 machine, but it's my main working Pro Tools computer. This one has been a rockstar for several years. I was scared to troubleshoot by interchanging any parts in case I totally screwed the pooch on my working rig. Do you think it's too dangerous ?

I'm also going to inspect that area which Macschrauber suggested, since I had actually been in that vicinity while unplugging the power cables to my GTX680/swapping out the card - during my troubleshooting whether the video card was part of the problem.

Many thanks to everyone for all these constructive ideas and help ! It's truly appreciated.

 

[tsialex]

I do actually have yet another machine which is a fully-operational 2009 4,1>5,1/2.29>3.46 machine, but it's my main working Pro Tools computer. This one has been a rockstar for several years. I was scared to troubleshoot by interchanging any parts in case I totally screwed the pooch on my working rig. Do you think it's too dangerous ?

Use the mid-2010 components, no need to risk your main Mac Pro, the SMC mismatch won't interfere with anything else other than the fan management - it will run at full RPM.

I'm also going to inspect that area which Macschrauber suggested, since I had actually been in that vicinity while unplugging the power cables to my GTX680/swapping out the card - during my troubleshooting whether the video card was part of the problem.

There are no CIs there that are 6-pin SOIC/SOJ/SOP, just 5 or 8-pin. I looked at my own mid-2012 Mac Pro to confirm it before writing.

There are 6-pin SOIC/SOJ/SOP on other places of the backplane, like @Syncretic wrote, but none that you will rip removing the mini-PCIe power cables.

 

[mgaudio]

Use the mid-2010 components, no need to risk your main Mac Pro, the SMC mismatch won't interfere with anything else other than the fan management - it will run at full RPM.

There are no CIs there that are 6-pin SOIC/SOJ/SOP, just 5 or 8-pin. I looked at my own mid-2012 Mac Pro to confirm it before writing.

There are 6-pin SOIC/SOJ/SOP on other places of the backplane, like @Syncretic wrote, but none that you will rip removing the mini-PCIe power cables.

ok, I just put the 2010 CPU tray into the 2009 machine, and went for a quick flight to Europe and back to LA in Concorde . . . ?‍✈️
It booted just fine and brought me to the desktop. Everything works, but I didn't do any extended tests beyond this.
I also tried vice-versa, and the 2010 machine with the 2009 tray inserted wouldn't boot. (nice jet-thrust too though)
After putting my 2010 back together, it's perfectly quiet once again and boots fast. At least I didn't cause any damage by installing the failed tray.
This seems to indicate definitively that the 2009 tray is the problem.
Now to track down the original location of my orphaned microchip on the 2009 CPU tray.