The Patient Has Regained Conciousness

[ITguy2016]

You may recall that I recently purchased a dual 2.0GHz G5 system (this is the mid-2004 M9455LL/A - PowerMac7,3 - A1047 - 1969C model). It worked for the first day I had it then, after powering it off and coming back to use it the next day, it failed to boot.

Problem symptoms were a power light which illuminated, then flashed bright three times, then the sequence repeated. No chime, no video, nothing but the flashing power light. Attempts to revive the system by removing unnecessary parts (really just memory as there was nothing else installed), an SMC reset, PRAM reset was out (requires the system to at least start), and replacing the battery (with a brand new one purchased months earlier) all failed.

Searching online I came across the "About the Power Mac G5 (Late 2005) diagnostic LEDs" document which stated the power light flashing three times meant there was incompatible memory installed. I thought this odd because the system worked the day before with the exact memory it was now indicating was incompatible. I tried swapping the memory with another, functioning, G5 and the result was the same...power light flashing three times. The memory from the failed system worked fine in the functioning G5. Obviously memory was not the issue.

Then I stumbled across an explanation which ended up being the problem: A failed motherboard. Apparently there is a problem with the G5 motherboard whereas some PGA chip, on the topside of the motherboard (hidden as it faces down and you cannot see it without removing the motherboard), for which the solder joints fail. This chip is right next to the processors and between the memory slots (again you cannot see the chip itself, just the square area where the chip is mounted from the other side). I was able to confirm this was the problem as I applied pressure to the affected area and the Mac began to boot until I released pressure at which time the failure symptoms returned.

Unfortunately a motherboard replacement is beyond what this system is worth. So I decided to try reflowing the solder connections. Some people have pulled the motherboard and put it in an oven. Being the lazy guy I am I felt this was too much effort for the system. Instead I opted to try using a heat gun to selectively apply heat to the affected area. My first attempt at doing this failed...I think I was a little too conservative in heat application. I decided to give it a second go and apply more heat and for a longer time. I figured at this point what did I have to lose? I set the heat gun on the lowest setting (around 500 degrees) and held it about an inch from the affected area. I used a circular motion and applied the heat for about 15 minutes. The board was nice and toasty to the touch and I figured I would either not get it hot enough (i.e. the reflow would fail) or I would get it too hot and possibly damage something. The end result being the system would still be non-functional.

Well, I'm happy to report that it worked and, after a cooling down period, the system powered on and I heard the familiar POST chime. I assembled the remaining parts, hooked it up to keyboard, mouse, and monitor and I am happily typing this response on the system. The question now becomes: Will it continue to function? I'm going to play with it over the next week or so to see but it's starting off on a positive note. The only issue is the cooling fans speeding up and slowing down (same behavior as I reported with my dual 2.7GHz system so I'll try the recommendations provided for that on this system).

I'd post a photo of the affected area but it doesn't look like I can upload them to the site...has to be a URL.

 

[jbarley]

Glad to hear that you got it working, and without having to resort to mouth-to-mouth resuscitation.

 

[CooperBox]

Glad to hear that you got it working, and without having to resort to mouth-to-mouth resuscitation.

I enjoy tales like this too, with a positive result. I hope it turns out to be a long-term fix.

 

[ITguy2016]

It is with sadness that I must report the patient has slipped back into a coma once again. This afternoon I pressed the power button and she responded as expected. Unfortunately minutes later she flat lined with a frozen image on the screen. After powering off the system and attempting to restart the flashing power button has returned.

At this point I'm going to honor the DNR symptoms and write it off. Anyone in the Denver area want a dead 2.0GHz Power Mac G5?
[doublepost=1471894964][/doublepost]Update: While transplanting some of the memory from the 2.0GHz system to the 2.7GHz system I discovered I had forgotten to install the cooling fans which have to be removed to access the memory. I didn't realize this until I went to add the memory to the 2.7GHz system. With this oversight I've decided I'm going to give it another go with the heat gun. While I don't expect these fans to have any material impact on this problem I can't give up on it knowing there's the possibility it might have been the cause for the recurrence (I now know why the fans continued to speed up and slow down).

I'll keep everyone updated once I've had a chance to apply another heat treatment.

 

[ITguy2016]

UPDATE: Unfortunately the second heat treatment didn't revive the patient for any length of time. After subjecting the system to another reflow effort the system powered on, started to boot, and then froze. Power cycling the system resulted in the same failure symptoms as before. Perhaps leaving the fans out helped the situation by keeping heat on the affected area.

Given this failure I'm going to end efforts to revive it and write it off as broken as cease any further efforts to fix it (barring anyone handing me a system board).

 

[Orizence]

Thats because the issue is usually a dead chip, not the solder. Sucks but the oven trick doesn't fix it, only stops the bleeding of a slow and unfortunate death

 

[ITguy2016]

Thats because the issue is usually a dead chip, not the solder. Sucks but the oven trick doesn't fix it, only stops the bleeding of a slow and unfortunate death

That's interesting as everything I've learned through this indicates there's a physical connection issue. I say that because I am able to get the system to boot if I apply pressure to the affected area of the system board. This suggest the pressure connects a broken pathway(s). I've never heard of pressure "fixing" a failing chip. In addition the heat causes the traces to expand thus further leading to the conclusion there's a physical connection failure. The problem seems to manifest itself after the system has cooled off for a period of time (such as overnight).

Regardless I've reached the conclusion it's not worth fixing...shame as I would have liked to continue using it (even if on occasion). Looks like it's going to the recycler at some point.

 

[hellothere231]

That's interesting as everything I've learned through this indicates there's a physical connection issue. I say that because I am able to get the system to boot if I apply pressure to the affected area of the system board. This suggest the pressure connects a broken pathway(s). I've never heard of pressure "fixing" a failing chip. In addition the heat causes the traces to expand thus further leading to the conclusion there's a physical connection failure. The problem seems to manifest itself after the system has cooled off for a period of time (such as overnight).

Regardless I've reached the conclusion it's not worth fixing...shame as I would have liked to continue using it (even if on occasion). Looks like it's going to the recycler at some point.

You should think about keeping the case, I've read online that you can use it for a PC build. Maybe even Hackintosh it...

 

[ITguy2016]

You should think about keeping the case, I've read online that you can use it for a PC build. Maybe even Hackintosh it...

Sadly the case is in rough shape and I wouldn't be interested in doing that kind of work anyway. I'm trying to give it away as a complete system (save for hard drive as I moved it to my 2.7GHz system) to anyone who may want a parts system. If I can't find any takers I'll probably pull the more useful components (memory, graphics card, optical drive, possibly the power supply) and recycle it.

 

[Orizence]

That's interesting as everything I've learned through this indicates there's a physical connection issue. I say that because I am able to get the system to boot if I apply pressure to the affected area of the system board. This suggest the pressure connects a broken pathway(s). I've never heard of pressure "fixing" a failing chip. In addition the heat causes the traces to expand thus further leading to the conclusion there's a physical connection failure. The problem seems to manifest itself after the system has cooled off for a period of time (such as overnight).

Regardless I've reached the conclusion it's not worth fixing...shame as I would have liked to continue using it (even if on occasion). Looks like it's going to the recycler at some point.

It may be a solder issue, but in a lot of cases I've seen its an issue with the underfill of the chip. It depends on the case, the reason why I thought it was a dead chip because I don't recall you stating it works with pressure put onto the chips.

 

[ITguy2016]

It may be a solder issue, but in a lot of cases I've seen its an issue with the underfill of the chip. It depends on the case, the reason why I thought it was a dead chip because I don't recall you stating it works with pressure put onto the chips.

No worries. It just seemed unusual to me as in my research I did not come across anything mentioning the chip itself was the issue and it seemed to be at odds with everything I did find and my diagnostics of the system.

 

[Orizence]

No worries. It just seemed unusual to me as in my research I did not come across anything mentioning the chip itself was the issue and it seemed to be at odds with everything I did find and my diagnostics of the system.

The reason some chips die like I though yours could've is that there is an underfill in a lot of bga chips that starts to crack and bubble (Nvidia chips and some ATI chips were super bad for this around 2007/2008), this is because even with a heatsink/and or heat spreader the heat produced in the chip itself isn't entirely even and causes the bubbles, most of the time when you bake or heat gun a chip you are not actually melting the solder in anyway as the temperatures most people use isn't simply hot enough to get the solder at any type of melting point, when baking or heat gunning a chip you are just melting and smoothing out the underfill. But because the problem is the chip it self, it will simply die again... BUT in some cases a proper reflow (Note the word proper... not the oven or a heat gun) that actually melts the solder can fix some issues, Ive heard of some intel chips solder cracking or lifting from the board and needing a reflow but thats rare.

But as I said most of the time DIY fixes aren't actually fixing the solder, and your only putting a bandaid on a gushing wound.

There may be an off chance of solder being the actual issue but I don't think that is very common.

 

[ITguy2016]

The reason some chips die like I though yours could've is that there is an underfill in a lot of bga chips that starts to crack and bubble (Nvidia chips and some ATI chips were super bad for this around 2007/2008), this is because even with a heatsink/and or heat spreader the heat produced in the chip itself isn't entirely even and causes the bubbles, most of the time when you bake or heat gun a chip you are not actually melting the solder in anyway as the temperatures most people use isn't simply hot enough to get the solder at any type of melting point, when baking or heat gunning a chip you are just melting and smoothing out the underfill. But because the problem is the chip it self, it will simply die again... BUT in some cases a proper reflow (Note the word proper... not the oven or a heat gun) that actually melts the solder can fix some issues, Ive heard of some intel chips solder cracking or lifting from the board and needing a reflow but thats rare.

But as I said most of the time DIY fixes aren't actually fixing the solder, and your only putting a bandaid on a gushing wound.

There may be an off chance of solder being the actual issue but I don't think that is very common.

The heat gun has two settings: One is 500 degrees, the other 1,000 degrees. Both well above the melting point of solder (though not as concentrated as a soldering iron. I've only used the 500 degree setting. Maybe I will give the 1,000 degree setting a try. Can't do any harm.

 

[Orizence]

The heat gun has two settings: One is 500 degrees, the other 1,000 degrees. Both well above the melting point of solder (though not as concentrated as a soldering iron. I've only used the 500 degree setting. Maybe I will give the 1,000 degree setting a try. Can't do any harm.

You have to be careful with that though because when you try to heat up a chip like that you have a higher chance of hurting the chip then you do actually melting the solder, also you would have a risk of hurting another component around the chip. But I say go for it, you have nothing to loose.

 

[ITguy2016]

Having some free time I decided to try reflowing the solder again with the heat gun. This time, figuring I had nothing to lose, I cranked the heat gun up to the 1,000 degree setting. I applied the heat for approximately five minutes at a distance of approximately half an inch. I was certain I was frying all of the components but hey...it didn't work anyway so...what did I have to lose?

Reassembled the system, plugged it in, and heard the self test passed chime. Currently it's transcoding a video which takes approximately one hour to complete. At this time the results are nothing new compared to previous attempts. Where failures appear is the next day after the system has had an opportunity to cool down. I'm going to run the transcode and then shut it down to let it cool. I'll try to fire it up tomorrow morning and see if it works. If not I've found someone to take it off my hands "as is".

 

[PowerMac G5]

Having some free time I decided to try reflowing the solder again with the heat gun. This time, figuring I had nothing to lose, I cranked the heat gun up to the 1,000 degree setting. I applied the heat for approximately five minutes at a distance of approximately half an inch. I was certain I was frying all of the components but hey...it didn't work anyway so...what did I have to lose?

Reassembled the system, plugged it in, and heard the self test passed chime. Currently it's transcoding a video which takes approximately one hour to complete. At this time the results are nothing new compared to previous attempts. Where failures appear is the next day after the system has had an opportunity to cool down. I'm going to run the transcode and then shut it down to let it cool. I'll try to fire it up tomorrow morning and see if it works. If not I've found someone to take it off my hands "as is".

Is it still working?

 

[ITguy2016]

Is it still working?

Sadly it is not. I was never able to get it to work once the system had a chance to cool down to room temperature. It would work immediately after the initial treatment. It continued to do so for the duration of time the system remained on. However once I shut it off and it had an opportunity to cool down it would fail in the same manner. I ended up cannibalizing some parts from it and then gave the chassis and remaining parts to someone off of Craigs List.

 

[0248294]

From what I know, in these cases the failure comes from the fact that the motherboard slightly flexes and bends. It's not much, but enough to eventually cause the chip's connections to come loose. The heat methods is only a temporary fix because the solder connection made with the heatgun isn't very sturdy. The only fix is to straight up remove the chip, clean the entire BGA pad, reball the original chip, add solder paste, and very carefully solder on the chip with heat. And, somehow you'd need to find a way to lessen the flexing to prevent it from happening again. I agree with Orizence that in such cases with GPUs, it is often failed chips instead of bad physical contacts. However, that's mostly only true for GPUs, not so much for low-heat components such as the memory controller in the PMG5.

 

[havokalien]

I second Daniels statement and add using the Apple Hardware Test to calibrate then using a fan control app to turn on fans sooner can keep the heat down.

 

[ITguy2016]

I second Daniels statement and add using the Apple Hardware Test to calibrate then using a fan control app to turn on fans sooner can keep the heat down.

My experience was just the opposite wrt cooling. The system would work fine as long as it remained warm. Once it cooled down to room temperature it would fail to start. I suspect if I left it running 24 / 7 it might have continued to operate the entire time.

That's not to say heating was not the initial cause of the problem. However, like Daniels mentioned, that area looks as if it flexes when RAM is added / removed which could also lead to the failure.

 

[mzs.112000]

You will likely need to remove the chip(without breaking it), clean the board(and the chip), re-ball the chip and use a heatgun and/or reflow tool to solder it back in. This will be expensive(reflow tool), so your best bet if you want to save money is to just buy a new motherboard.

 

[ITguy2016]

You will likely need to remove the chip(without breaking it), clean the board(and the chip), re-ball the chip and use a heatgun and/or reflow tool to solder it back in. This will be expensive(reflow tool), so your best bet if you want to save money is to just buy a new motherboard.

I concluded the system wasn't worth the cost of repair. I could purchase a fully functional one for less than the cost to repair this one. Therefore I cannibalized certain parts and then gave the remained to someone from Craigs List.