EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Solares on June 12, 2021, 02:38:53 pm
-
I work in IT - we upgrade Windows and Mac computers.
I am not an electronics expert.
However, I have noticed very interesting trend.
We get in old systems to upgrade to SSD. Previously, they would have been using spinning mechanical hard disks. So, it might be an Windows desktop PC or an iMac for example. The systems would be performing terribly slow. We put in a new SSD and the system is given a new lease of life. Where once the operating system and applications would run painfully slowly - after the SSD upgrade, they would run really fast.
However, anywhere from 3 to 6 months later, the motherboard would just fail on these systems. Not all of them, but about 30 per cent of them.
My hypothesis: It's as if the system was used to operating at a slow speed all it's life. With an SSD the electronic components get a shock, can no longer handle the speed and the motherboard fails because of overstress.
Sorry for being over simplistic here. But does anybody have any theories on why this happens?
(But please don't state the bleeding obvious by saying "they're old systems anyway")
-
How the motherboards fail? Have you tried to pinpoint what exactly fails? Because just to say that it "fails" is very abstract.
-
(But please don't state the bleeding obvious by saying "they're old systems anyway")
I'm not sorry, I'm going to.
How old are these machines? Of what sort of quality are they? (and don't say 'oh, well some are Macs' - they're not special.)
-
I've converted enough (mostly laptops) systems to know that I don't have a 30% failure rate with quite a high confidence interval. Actually, I've never had one over maybe a dozen such conversions as well as custom builds. What OS ends up on these systems and how does the motherboard 'fail'? Other than the capacitor plague and an occasional physical mishap, I haven't seen many failed motherboards. Of course, I don't work on very many Apple products.
-
I do the same job, but mainly on macs.I installed ssd in macs computers for 10 years or so. 1 to 10 per week.
my 2 co-workers do almost the same amount of installations.
only a few ssd failed ( max 5 I can remember) and no motherboard died (except 2011 imac with known graphic card problems)
I used mainly samsung ssd may be it's the reason ?
-
This particularly affects iMacs which are between 2010-2012.
We will not take on any systems below 2010.
The Windows systems would be typically 7-8 years. These are branded systems like HP.
SSD brands we would use would be WD Blue and SanDisk. BTW, just to re-emphasize the SSDs would be fine.
We often do not have time to pinpoint exact failure of motherboard.
And thanks kripton for bringing up the issue of iMac 2011 - yes, this graphics chips on these seem to be a big problem.
-
This particularly affects iMacs which are between 2010-2012.
Thermally abused fashion statements with unreliable MXM cards, in other words.
There's also the distinct possibility you're inducing the failures with poor handling. Do you take ESD precautions at all?
-
Get a DC current clamp, splice into the laptop power supply cable......note dc current in when with HDD and then with SSD.
Compare.
Its likely its using too much power with ssd
Laptops are crammed like mad inside, and cooling is at a critical point....a slight increase in power, and bang goes the motherboard.
Also, you can do the same test with a thermal cam
Use a thermal cam with max temp hold...otherwise your shaking hand will invalidate the reading...obviosuly you need to take it apart for thermal cam, and point at the motherboard or its heatsink.
-
Get a DC current clamp, splice into the power suppply cable......note dc current in when with HDD and then with SSD.
Compare.
Its likely its using too much power with ssd
Rubbish, treez. The average power consumption of an SSD is massively lower than an HDD.
-
yes but the motherboard running it with the increased speed, may overheat.
Good point though...so if the hdd is pulling more current, then you wont tell by scoping the dc current in....so you need to use thermal cam on the motherboard.
I seriously believe that laptops were never meant to be...theres not enough room inside for cooling......but their convenience means that they get sold anyway
-
Spurious correlation:
(https://storage.ning.com/topology/rest/1.0/file/get/2808309983?profile=original)
-
yes but the motherboard running it with the increased speed, may overheat.
Good point though...so if the hdd is pulling more current, then you wont tell by scoping the dc current in....so you need to use thermal cam on the motherboard.
It's spending far more time waiting on the user than anything else. You're not going to overheat a ten year old computer by putting a slow (yes, slow) SSD in it.
-
Are you using ESD precautions when you open up the computer.?
-
Yes, ESD wrist strap and ESD mat always used.
But, these failures rarely happen immediately.
It's usually anywhere between 1 month and 3 months after SSD installation.
Maybe it is a spurious correlation, but it just seems something happens to old motherboards after an SSD has been installed.
-
ESD failures are not always immediate. Often far from it.
How many machines of this age do you not open and upgrade, and then monitor for the next year?
-
This particularly affects iMacs which are between 2010-2012.
There might be a specific issue with the upgrade, but I doubt there's a general issue of overworking the MB due to the increased performance. All of the applications that I can think of that use a lot of CPU or GPU--HEVC encoding, web browsers with 100 tabs open--would not be significantly affected (powerwise) by drive performance.
We often do not have time to pinpoint exact failure of motherboard.
That would be the place to start if nobody else has observed a similar correlation. Even if your failure rates are abnormally high compared to a baseline of unimproved systems--not a foregone conclusion b/t/w--the cause could be any number of issues.
Maybe it is a spurious correlation, but it just seems something happens to old motherboards after an SSD has been installed.
Here's a thought. Were these systems in use right up to the time of upgrade or have they been taken out of service some time back, left in a closet and now you are resurrecting them?
-
Unless you're using defective SSDs then absoutely no way possible.
When you say "terribly slow" it could be the systems were already in a pre-failure state.
-
Get a DC current clamp, splice into the power suppply cable......note dc current in when with HDD and then with SSD.
Compare.
Its likely its using too much power with ssd
Rubbish, treez. The average power consumption of an SSD is massively lower than an HDD.
treez? :o What gives? treez = ocset?
-
Get a DC current clamp, splice into the power suppply cable......note dc current in when with HDD and then with SSD.
Compare.
Its likely its using too much power with ssd
Rubbish, treez. The average power consumption of an SSD is massively lower than an HDD.
treez? :o What gives? treez = ocset?
It appears he's just changed his name. Because his posting habits are really difficult to identify. ::)
Maybe he works for Tesco now.
-
Sigh ::)
-
Maybe it is a spurious correlation, but it just seems something happens to old motherboards after an SSD has been installed.
Perhaps it does, but William of Occam would probably side with an explanation of "machines are more likely to fail after someone has partially disassembled and reassembled them". Or as BD139 (who is on a break so isn't here to comment) would have put it "If you monkey around with systems they break".
The only way to be confident that your hypothesis is true would be to do a controlled experiment. Upgrade to SSDs in some machines (group "monkey involved type S"), remove and reinstall the HDDs in some of them (group "monkey involved type H"), and leave some alone (group "no monkey involved in process"). My hypothesis is that you would see higher failure rates in both the groups with "monkey involved" and normal failure rates in the "no monkey involved in process" group. Of course this all falls apart if your sample size is too small compared to the normal failure rates of machines anyway - as is probably already be the case for your anecdotal analysis. You'd need thousands of machines to get a reliable result in a reasonable period of time.
-
>>Here's a thought. Were these systems in use right up to the time of upgrade or have they >>been taken out of service some time back, left in a closet and now you are resurrecting them?
They would be used right up to time of upgrade.
However, as anybody who deals with end-users knows, they can lie. And some are very good at it!
So, what if they had been in storage?
And thanks Cerebus for your comment the hard disk upgrade process is actually a very unobtrusive procedure. For example, with old iMacs, it just a case of removing glass screen, disconnecting the screen and power connector to the LCD and then have direct access to the disk bay. No need to remove or alter any other components.
-
Hmmm, I really doubt it. There are good motherboards, and crummy ones. I have been using used Dell Optiplex systems from eBay for well over a decade. I have one on my web server, one on my main desktop, and a few others here and there. Most have SSDs in them, and have been running for anywhere from15 years down to just a few years. Other that one where I blew up the parallel port with a 12V wire touching, none of them have died. I have seen a few power supplies go bad.
The Optiplex is Dell's commercial line of computers. If you look inside one, you can't SEE anything that looks different from their consumer line, but there IS something different about them. This machine I'm typing on was bought used and has been running for 9 years straight. I just replaced the SSD as I was afraid it was going to suffer the "wear out" problem sometime, and also I needed a bigger one.
Dell did get hit with the capacitor plague about 15 years ago, and that did cause a lot of motherboards to crap out.
Jon
-
Yes, my second computer is also an Optiplex - great workhorse computer. Always starts - never any surprises with it.
If I had to visit the international space station for one year and could only take one computer, it would probably be an Optiplex.
I would certainly not take an iMac. @Monkeh knows what I'm talking about!
-
Aren't Macs of that era known for having BGA soldering issues? I suspect if there is any cause-effect relation going on here, it's that simply disturbing an old system internally and putting some mechanical stress on the board (plugging/unplugging cables) is causing cracks in already weak solder joints that results in a failure after some number of thermal cycles. There is no reason that a faster disk would make the motherboard fail.
-
Aren't Macs of that era known for having BGA soldering issues? I suspect if there is any cause-effect relation going on here, it's that simply disturbing an old system internally and putting some mechanical stress on the board (plugging/unplugging cables) is causing cracks in already weak solder joints that results in a failure after some number of thermal cycles. There is no reason that a faster disk would make the motherboard fail.
Yes, you're right. iMacs from the early 2010s are known for week soldering.
I guess I'm just trying to investigate a link with SSD installation and precipitated failure.
-
The only thing I have noticed retrofitting new SSDs to old systems is old Marvel SATA2 controllers generating link errors with new SSDs causing very puzzling system failures until the SSD is moved to a different controller.
-
The only thing I have noticed retrofitting new SSDs to old systems is old Marvel SATA2 controllers generating link errors with new SSDs causing very puzzling system failures until the SSD is moved to a different controller.
thanks David.
Can you give an example of a puzzling failure?
And when you say move to a different controller do you mean physically or virtually?
-
-I can tell my experience. I have put SSD (mostly fast Samsung 860 Evo, but there were a bit older 840-850 Pro models and 1-2 non-Samsung too) into up to 10 old PCs (and laptops), and only one MB died (it worked several years with several SSDs until death).
-
Here's a thought. Were these systems in use right up to the time of upgrade or have they been taken out of service some time back, left in a closet and now you are resurrecting them?
There was a study that tried to show the relationship between drinking alcohol and health. It concluded that a few alcoholic drinks per day promoted better health than none or too many. Later the study methodology was reviewed. Turns out that in the original study it was hard to find people who didn’t drink alcohol so all the people in that group were former alcoholics.
-
There was a study that tried to show the relationship between drinking alcohol and health. It concluded that a few alcoholic drinks per day promoted better health than none or too many. Later the study methodology was reviewed. Turns out that in the original study it was hard to find people who didn’t drink alcohol so all the people in that group were former alcoholics.
Seems like that wouldn't be all that hard. IIRC both Mormons and Muslims typically do not drink alcohol, neither is particularly hard to come by on a global scale.
-
ESD failures are not always immediate. Often far from it.
How many machines of this age do you not open and upgrade, and then monitor for the next year?
And a third group would also shed light: machines of the same age disassembled to the same degree needed for an SSD swap, except simply reassembled using the original components.
-
My hypothesis: It's as if the system was used to operating at a slow speed all it's life. With an SSD the electronic components get a shock, can no longer handle the speed and the motherboard fails because of overstress.
Sorry for being over simplistic here. But does anybody have any theories on why this happens?
(But please don't state the bleeding obvious by saying "they're old systems anyway")
Yeah, no.
Since this is not a widespread problem (I follow the Mac world closely), it’s most likely you are causing the damage somehow. ESD damage is the most likely, but could also be things like incorrect (dis)assembly, for example stressing boards incorrectly, causing damage to solder joints.
-
Get a DC current clamp, splice into the power suppply cable......note dc current in when with HDD and then with SSD.
Compare.
Its likely its using too much power with ssd
Rubbish, treez. The average power consumption of an SSD is massively lower than an HDD.
treez? :o What gives? treez = ocset?
Pop treez in a members search and hunt through a few of his posts looking for quoted ones. ;)
Good spotting by Monkeh. :-+
-
The only thing I have noticed retrofitting new SSDs to old systems is old Marvel SATA2 controllers generating link errors with new SSDs causing very puzzling system failures until the SSD is moved to a different controller.
thanks David.
Can you give an example of a puzzling failure?
And when you say move to a different controller do you mean physically or virtually?
I physically moved the SSD from the motherboard's SATA2 Marvel controller to the JMicron controller, but I think the Intel controller would have worked just as well. The same SSDs have had no issues on new SATA2 systems.
The problem took more than a year to resolve. The symptom was simply black screen computer crashes, which were eventually daily, with no diagnostic information coming from Windows, but it finally generated a blue screen which mentioned something about the page file. Further investigation showed that the SSD was accumulating UDMA errors which is an interface problem. What was going on is that page-in operations were failing not with corrupted data but unavailable data, which obviously causes the operating system to become extremely unhappy. Each crash took time to manifest because paging operations increase as the memory load increases over time.
The above was not the only thing I changed to attempt to fix the problem. Originally I thought the crash was caused by hyperthreading because the original blue screen error had to do with interprocessor communications and disabling HT stopped the crashes for months. Now however I have been able to enable HT and have not had a single problem since moving the SSD.
-
I agree that the ESD concern is possible.
Another concern is simply "schrodinger's pc" -- if you just leave it on (or maybe cycle it daily) but don't move it, don't really touch it, don't open it, don't pull cables, flip it over, etc. etc. as you'd do to upgrade a drive to SSD, it's mechanically and electrically in a "relaxed" stable state. Physically disturb that and then you're taking the risk that the vibration / shock / stress of just mechanically working with it has maybe done something like cracked solder joints / vias / capacitors, caused capacitors to move / start leaking, etc.
And SSDs absolutely can put a higher thermal stress on the system than mechanical hard drives. Mechanical drives may have a high transient and steady power consumption but a slow data transfer rate and the thermal heat is spread over a larger volume which usually has some heatsinking due to the mounting bay.
Good SSDs, though, like the little NVME ones run so hot under load they need heatpads / heatsinks right on them unless they overheat and thermally throttle etc.
Then the much higher PCIE transfer rate can cause the motherboard chipset or CPU to run quite a bit hotter due to the fast PCIE SSD traffic -- same for faster PCIE GPUs.
Newer / better PCIE4 motherboards have cooling fans right on the chipset chip largely due to the PCIE based heat from NVME drives, GPUs, etc.
Depends on your I/O profile but it could be significant in cases where good advantage is taken of the SSD's better bandwidth.
Some systems could fail simply by a ball of dust or a loose screw / whatever moving around from place to place when the system moves and unfortunately blocking something in a worse way than it was doing before.
Thank you for the brilliant and concise set of possibilities.
-
There was a study that tried to show the relationship between drinking alcohol and health. It concluded that a few alcoholic drinks per day promoted better health than none or too many. Later the study methodology was reviewed. Turns out that in the original study it was hard to find people who didn’t drink alcohol so all the people in that group were former alcoholics.
Seems like that wouldn't be all that hard. IIRC both Mormons and Muslims typically do not drink alcohol, neither is particularly hard to come by on a global scale.
I think the point is that some studies and experiments are set up more rigorously than others and the quality of the undertaking might be reflected in the conclusion.
Kinda like this:
https://www.eevblog.com/forum/beginners/ssd-disks-seem-to-accelerate-motherboard-failure-on-computers/msg3586921/#msg3586921 (https://www.eevblog.com/forum/beginners/ssd-disks-seem-to-accelerate-motherboard-failure-on-computers/msg3586921/#msg3586921)
Could be how the data was collected an/or how it was analyzed; sometimes the conclusion just isn't correct.
-
The problem with all academic research into alcohol, other recreational drugs, and all the other things that tend to attract a 'moral' response from a significant number of people, is that it tends to reflect the views, whether they be pro- or anti-, of the researchers or their funders. Treat any research in any of these areas with utmost scepticism. When people do good, unbiased research in these fields and produce valid results that other people don't like it can go badly for them - see what happened to Professor David Nutt, a former government drugs adviser who was sacked when is research findings didn't support the government's position.
-
There was a study that tried to show the relationship between drinking alcohol and health. It concluded that a few alcoholic drinks per day promoted better health than none or too many. Later the study methodology was reviewed. Turns out that in the original study it was hard to find people who didn’t drink alcohol so all the people in that group were former alcoholics.
This appears to be a garbled recollection of this study: https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1530-0277.2010.01286.x (https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1530-0277.2010.01286.x)
Background: Growing epidemiological evidence indicates that moderate alcohol consumption is associated with reduced total mortality among middle-aged and older adults. However, the salutary effect of moderate drinking may be overestimated owing to confounding factors. Abstainers may include former problem drinkers with existing health problems and may be atypical compared to drinkers in terms of sociodemographic and social-behavioral factors. The purpose of this study was to examine the association between alcohol consumption and all-cause mortality over 20 years among 1,824 older adults, controlling for a wide range of potential confounding factors associated with abstention.
Methods: The sample at baseline included 1,824 individuals between the ages of 55 and 65. The database at baseline included information on daily alcohol consumption, sociodemographic factors, former problem drinking status, health factors, and social-behavioral factors. Abstention was defined as abstaining from alcohol at baseline. Death across a 20-year follow-up period was confirmed primarily by death certificate.
Results: Controlling only for age and gender, compared to moderate drinkers, abstainers had a more than 2 times increased mortality risk, heavy drinkers had 70% increased risk, and light drinkers had 23% increased risk. A model controlling for former problem drinking status, existing health problems, and key sociodemographic and social-behavioral factors, as well as for age and gender, substantially reduced the mortality effect for abstainers compared to moderate drinkers. However, even after adjusting for all covariates, abstainers and heavy drinkers continued to show increased mortality risks of 51 and 45%, respectively, compared to moderate drinkers.
Conclusions: Findings are consistent with an interpretation that the survival effect for moderate drinking compared to abstention among older adults reflects 2 processes. First, the effect of confounding factors associated with alcohol abstention is considerable. However, even after taking account of traditional and nontraditional covariates, moderate alcohol consumption continued to show a beneficial effect in predicting mortality risk.
Of course there have been many other studies which support or refute the conclusions :-//
This one in particular got a lot of press: https://www.medrxiv.org/content/10.1101/2021.05.10.21256931v1 (https://www.medrxiv.org/content/10.1101/2021.05.10.21256931v1)
Conclusions No safe dose of alcohol for the brain was found. Moderate consumption is associated with more widespread adverse effects on the brain than previously recognised. Individuals who binge drink or with high blood pressure and BMI may be more susceptible. Detrimental effects of drinking appear to be greater than other modifiable factors. Current ‘low risk’ drinking guidelines should be revisited to take account of brain effects.
Cheers!
"abstainers had a more than 2 times increased mortality risk, heavy drinkers had 70% increased risk, and light drinkers had 23% increased risk."
"Abstainers may include former problem drinkers with existing health problems and may be atypical compared to drinkers in terms of sociodemographic and social-behavioral factors."
uh, which part was garbled?
-
uh, which part was garbled?
1. "it was hard to find people who didn’t drink alcohol"
2. "all the people in that group were former alcoholics"
3. The implication that the study results were invalidated because of a review
oh, yes, clearly garbled
-
I agree that the ESD concern is possible.
Another concern is simply "schrodinger's pc" -- if you just leave it on (or maybe cycle it daily) but don't move it, don't really touch it, don't open it, don't pull cables, flip it over, etc. etc. as you'd do to upgrade a drive to SSD, it's mechanically and electrically in a "relaxed" stable state. Physically disturb that and then you're taking the risk that the vibration / shock / stress of just mechanically working with it has maybe done something like cracked solder joints / vias / capacitors, caused capacitors to move / start leaking, etc.
And SSDs absolutely can put a higher thermal stress on the system than mechanical hard drives. Mechanical drives may have a high transient and steady power consumption but a slow data transfer rate and the thermal heat is spread over a larger volume which usually has some heatsinking due to the mounting bay.
Good SSDs, though, like the little NVME ones run so hot under load they need heatpads / heatsinks right on them unless they overheat and thermally throttle etc.
Then the much higher PCIE transfer rate can cause the motherboard chipset or CPU to run quite a bit hotter due to the fast PCIE SSD traffic -- same for faster PCIE GPUs.
Newer / better PCIE4 motherboards have cooling fans right on the chipset chip largely due to the PCIE based heat from NVME drives, GPUs, etc.
Depends on your I/O profile but it could be significant in cases where good advantage is taken of the SSD's better bandwidth.
Some systems could fail simply by a ball of dust or a loose screw / whatever moving around from place to place when the system moves and unfortunately blocking something in a worse way than it was doing before.
The load a NVME (or any SSD) is going to put on the PCIE bus is nothing compared to a 16x Graphics Card. The graphics card is constantly transfeting it's max data rate when you're gaming. The SSD works in short bursts. PCIE v1x16 graphics cards transfer 8GB/s. No SSD is there yet. They need 1/4 the bandwidth (4 lanes)
-
Motherboards don't fail from "hard use". In general motherboards rarely fail, a 30% failure rate on motherboards that are 5-10 years old is astonishing, and pretty much an indication that you are damaging them.
There are a few options but my money is on overheating. This could happen a couple ways. You could unplug a case fan either accidentally or deliberately and forget to plug it back in. If you try to clean the years of accumulated dust out of the case you could be actually damaging a fan causing it to fail within a few weeks and overheat. Blowing around dust with compressed air can drive more dust into a bearing. Or you could rearrange things in a way that that obstructs airflow. When you get a failure look to see if any fans are stopped, too slow, or show abnormal friction when off.
It could also easily be mechanical damage or ESD. Or it could be something else entirely. Since you haven't given details of the symptoms it could even be that the motherboard is fine, something other hardware or software is the problem.
-
The graphics card is constantly transfeting it's max data rate when you're gaming.
Utter rubbish. What do you think it has RAM on it for?
(https://cdn.mos.cms.futurecdn.net/v2aJ5uoqM8tZ95MCY5q4fM-970-80.png)
E: Oh, and SSDs are there. They've been there for several years. You've just got no way to use them with toy desktops.
-
I had a recent crypto mining adventure that used of SSDs and ran them flat out for days moving about 30 TB of data in a day.
SSDs can consume a considerable amount of power indeed, but its still less than a HDD. An actively working SSD does consume more than a idle HDD (laptop drives typically reduce RPM when under low utilization), but when a HDD is actively working it consumes more power than a actively working SSD. Still these powers are in the order of a few watts, so its not going to really break anything.
What i have noticed however is that hammering a bunch of SATA SSDs full on did make the southbridge run noticeably more toasty (This is where the SATA comes from) so i added a fan to help cool it.
Modern SSDs can indeed saturate a PCIe link because they typically only use up to a PCIe 4x link. Yes GPUs can transfer even more data since they have a PCIe x16 link while they indeed don't usually make use of all of it, the trick to making the GPU hog all of it is to let it run out of VRAM, this will make it start swapping into main RAM and clog that x16 link. But this is unlikely to make something overheat, if anything a big GPU itself would cook things to death before the PCIe power use would, the PCIe lines go straight to the CPU anyway. However... there are also chipsets out there that provide extra PCIe lanes from the southbridge chip, these are typically used for any extra onboard peripherals such as perhaps a fancy high performance network interface, WiFi, extra SATA controllers, and M.2 slots. So in this way a M.2 NVME SSD could put extra thermal load into the southbridge chip.
An SSD will also rise the peak CPU utilization since during boot the OS is spending much less time waiting for data to arrive from disk and much more time actually processing the data it just read. But a laptop should not die just because the CPU utilization hits 100% every so often. That's like a cars engine breaking down if you floor the gas pedal for a little bit too many times. But yeah holding 100% CPU utilization on a laptop tends to not be good for them since they typically don't have enough cooling to handle it unless you live in Antarctica. But this is not something a SSD would do.
I don't really think this is the cause of the machines giving up the ghost. Its likely just that they die from old age. All the machines you happened to upgrade ware likely machines that got used a lot (hence why they ended up upgraded) and they just slowly died on there own. A lot of apple laptop models are famous to just die after a certain number of years due to the soldering giving up.
-
I've done a couple hundred upgrades from mechanical to SSD in the PC environment (including a bunch of them the time I worked in Limerick :-+), both in desktops and laptops, and I can confirm that hasn't happened to me either.
Motherboards and any series of computer components, on the other hand, can develop a typical problem affecting models or batch of units as a whole over time, where a pattern of failures starts to show. As a wise guy mentions before there could be a correlation and not causality.
-
Motherboards don't fail from "hard use". In general motherboards rarely fail, a 30% failure rate on motherboards that are 5-10 years old is astonishing, and pretty much an indication that you are damaging them.
Over the past 30+ years I have had extended personal experience with ... at least 20 motherboards, and the only ones that failed without reason had nVidia nForce2 chipsets; those three, all being different makes and models purchased at about the same time, were 100% at about 5 years. The only other failures were motherboards with embedded lithium backup batteries which ran dead.
-
So the problem is only with a limited model types of MAC.
Could it be the material SSDs chassis are made with cause more stress from thermal expansion or the change of weight changes how the motherboard vibrates and it's mechanical harmonics. Could be just a particular connector getting lose from vibrations and shorting the whole thing to death.
It is well known fact that Macbook solder is known to crack. But 3-4 months with 30% failure rate is surprising. I've never seen such high failure rates even with shittiest of any computer shop.
What failure mode BTW.
-
It is well known fact that Macbook solder is known to crack.
What utter tosh! If that is true, prey tell us exactly what is special about "Macbook solder" that makes it so fragile and how a quality obsessed company like Apple would continue to use it after it became a "well known fact"?
-
It is well known fact that Macbook solder is known to crack.
What utter tosh! If that is true, prey tell us exactly what is special about "Macbook solder" that makes it so fragile and how a quality obsessed company like Apple would continue to use it after it became a "well known fact"?
well, work hardening?
Also, there's a long list of defects in their products that they know, exists but don't bother fixing it. Like butterfly keyboard and touch bars that die easily
-
What utter tosh! If that is true, prey tell us exactly what is special about "Macbook solder" that makes it so fragile and how a quality obsessed company like Apple would continue to use it after it became a "well known fact"?
I can't answer either of the technical questions, but there was a class-action lawsuit and a recall of Macbooks for this very issue.
-
There was a very particular range of AMD GPUs in Macbooks which suffered from cracked solder balls. Nobody seems much fussed by the fact that AMD are the ones who put the solder balls on the packages, of course, as Apple are the easy target. It is not a widespread concern of Macbooks.
Christ, I just defended Apple, I must be coming down with something.
-
Such things can easily happen even to the big guys.
For example Microsoft had one hell of a situation with the early generations of Xbox 360 consoles. They made the console considerably more compact compared to the previous generation, while packing in lots of processing power. As a result it ran pretty hot and the thermal cycling eventually cracked solder balls on a lot of them causing the famous "Red ring of death" error after a year or two of regular use(sometimes even weeks or months). This was not a small percentage of units either, it happened to a lot of them.
It cost them huge amounts of money to replace units under warranty. Microsoft never officially released what caused the issue but sources say that part of the problem was using the wrong lead free solder alloy that was particularly prone to cracking under these hot conditions. Some point at Microsoft sticking there fingers into the graphics ASIC design too much and causing it to run hotter than it should.
Then again there are also plenty of laptops that run very hot but don't have reliability issues. Just that with MacBooks there is a lot of them out there, so its easier to draw some statistics on there reliability.
-
Such things can easily happen even to the big guys.
For example Microsoft had one hell of a situation with the early generations of Xbox 360 consoles. They made the console considerably more compact compared to the previous generation, while packing in lots of processing power. As a result it ran pretty hot and the thermal cycling eventually cracked solder balls on a lot of them causing the famous "Red ring of death" error after a year or two of regular use(sometimes even weeks or months). This was not a small percentage of units either, it happened to a lot of them.
It cost them huge amounts of money to replace units under warranty. Microsoft never officially released what caused the issue but sources say that part of the problem was using the wrong lead free solder alloy that was particularly prone to cracking under these hot conditions. Some point at Microsoft sticking there fingers into the graphics ASIC design too much and causing it to run hotter than it should.
Then again there are also plenty of laptops that run very hot but don't have reliability issues. Just that with MacBooks there is a lot of them out there, so its easier to draw some statistics on there reliability.
Yeah, bad solder was quite common when RoHs was first implemented. Asus is also known to be using the bad alloy, but only Apple refuse to service and fix arising from it.
-
Such things can easily happen even to the big guys.
For example Microsoft had one hell of a situation with the early generations of Xbox 360 consoles. They made the console considerably more compact compared to the previous generation, while packing in lots of processing power. As a result it ran pretty hot and the thermal cycling eventually cracked solder balls on a lot of them causing the famous "Red ring of death" error after a year or two of regular use(sometimes even weeks or months). This was not a small percentage of units either, it happened to a lot of them.
It cost them huge amounts of money to replace units under warranty. Microsoft never officially released what caused the issue but sources say that part of the problem was using the wrong lead free solder alloy that was particularly prone to cracking under these hot conditions. Some point at Microsoft sticking there fingers into the graphics ASIC design too much and causing it to run hotter than it should.
Then again there are also plenty of laptops that run very hot but don't have reliability issues. Just that with MacBooks there is a lot of them out there, so its easier to draw some statistics on there reliability.
Yeah, bad solder was quite common when RoHs was first implemented. Asus is also known to be using the bad alloy, but only Apple refuse to service and fix arising from it.
Except it's a myth and had nothing to do with solder balls. Repair by reballing the chips was a scam. What failed were bumps between silicon die and substrate, heating restored the contact for a while, so many people stupidly assumed it was solder. Even though temperature needed to restore the operation was much lower than solder melting point.
-
Except it's a myth and had nothing to do with solder balls. Repair by reballing the chips was a scam. What failed were bumps between silicon die and substrate, heating restored the contact for a while, so many people stupidly assumed it was solder. Even though temperature needed to restore the operation was much lower than solder melting point.
Yeah this is what gave rise to the Xbox 360 "towel fix". People would wrap the console in a towel to make it get as hot as possible before it shuts itself off. This would actually help, but only for some time before it would die again.
I don't think Microsoft actually attempted to fix the failed boards, they likely just ripped out the old boards, threw a few truckloads of them into a landfill and put in the brand new revision boards that didn't have this problem.
Cracks in the BGA balls do still happen sometimes (Even tho most cases are actually the microballs on the die, as those see more stress). One of the changes on the new revision 360 motherboards was added glue. You often also see under fill glue on BGAs in iPhones and similar products, its not strictly necessary but it seams they found it does provide a meaningful reliability improvement (mechanical and keeping dirt out) to be worth it.
-
Maybe it's just users doing actual work instead of waiting for HDD cause higher CPU/thermal load ?
Or maybe users are angry you didn't give them new machine and are mistreating them :-DD
-
My guess is that, given that those are exactly the models affected by some GPU solder (or bumps) failures, that the forces involved during SSD installation (such as when removing and inserting connectors) flex the board just enough for further weaken the already iffy solder joints/bumps, such that subsequent thermal cycling causes them to fail.
Again, this is easily tested: do the same process as for installing the SSD, but reinstall the original parts. That would conclusively reveal whether it’s the part or the installation process that’s causing the subsequent failures.
-
I guess, if this thing was really true, there's no way that Louis Rossman wouldn't have encountered it.
-
I guess, if this thing was really true, there's no way that Louis Rossman wouldn't have encountered it.
Louis Rossman clearly said:
https://www.youtube.com/watch?v=1AcEt073Uds (https://www.youtube.com/watch?v=1AcEt073Uds)
-
Yeah, bad solder was quite common when RoHs was first implemented.
People say that, but I don’t think it actually happened. Heck, in some aspects, lead-free solder has superior characteristics, so it’s possible that those have actually negated the negative aspects in real-world products.
Look at https://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol38-1/paper14.pdf (https://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol38-1/paper14.pdf)
-
This particularly affects iMacs which are between 2010-2012.
We will not take on any systems below 2010.
The Windows systems would be typically 7-8 years. These are branded systems like HP.
SSD brands we would use would be WD Blue and SanDisk. BTW, just to re-emphasize the SSDs would be fine.
We often do not have time to pinpoint exact failure of motherboard.
And thanks kripton for bringing up the issue of iMac 2011 - yes, this graphics chips on these seem to be a big problem.
2010 imac?
I though that original post was about laptops. Now that I know it's desktop, I'm even more sure that the problem is mechanical.
Those things had huge 3.5 inch drives.
Replacing them with tiny SSD would need a mount. Maybe the Mount the you're using is bigger than what it's supposed to be and putting mechanical stresses all over the place
-
That doesn’t make any sense, since SSDs are smaller than HDDs, and the adapters necessarily have to fit inside the same dimensions as an HDD.
-
usually adapters are made to fit inside a metal drive cage in the front portion of chassis that can flex so the adapter being a bit bigger might not be a problem.
But whatever...We all can go on forever, guessing what the cause could be... nothing can be said without looking at the damaged computer itself
-
a quality obsessed company like Apple
A perceived quality obsessed company. Their products have to look high quality on the outside, the actual engineering on the inside has been shown to be pretty crap in a number of cases.
-
a quality obsessed company like Apple
A perceived quality obsessed company. Their products have to look high quality on the outside, the actual engineering on the inside has been shown to be pretty crap in a number of cases.
Unless you see them as engineered to fall. Same goes for Asus, which is a trap company for enthusiasts.
Asus puts their fans in laptop in such a way, that you'd have to remove cpu-gpu combo heatsink to get them out. Big problem when you're the service tech cleaning them
HP is even worse with their fans sucking air from the gaps between keyboard keys along with user's sweat.
-
a quality obsessed company like Apple
A perceived quality obsessed company. Their products have to look high quality on the outside, the actual engineering on the inside has been shown to be pretty crap in a number of cases.
That’s really quite an exaggeration. There are a few exceptions, but they’re a tiny, tiny minority.
-
Unless you see them as engineered to fall.
Except that reliability stats show them to be exceptionally good, above average for the industry.
-
Speaking of HDD replacement, Mac and thermal issues, it reminds me of something: some iMac (2011) use proprietary HDDs with customized firmware that repurposed SATA pin #11 (activity indicator signal) as a temperature sensing PWM signal, this is used by the logic board embedded controller for thermal management. Replacing the HDD with another HDD or SSD will cause erratic thermal behavior: the logic board finds the invalid temp. and enters fallback - maximal fan speed :bullshit: :bullshit: But this can't be the cause of OP's issue here as it increases cooling, not decreases it, but unusual enough (perhaps not for Apple) to have a mention.
The workaround was simply enough... pin #11 is open-drain, so bridge it to pin #12, ground, so the controller reads a 0 but not invalid, and you're done, but at your own risk: although an SSD is unlikely to overheat (and even then, has its own throttling), it's unknown whether the controller uses the HDD temperature together with other temperature to make a decision and whether bypassing the HDD temp. sensing makes the fan management less responsive. For those particularly care about the problem, iFixit sells an upgrade kit that recreates this signal by a microcontroller.
-
Speaking of HDD replacement, Mac and thermal issues, it reminds me of something: some iMac (2011) use proprietary HDDs with customized firmware that repurposed SATA pin #11 (activity indicator signal) as a temperature sensing PWM signal, this is used by the logic board embedded controller for thermal management. Replacing the HDD with another HDD or SSD will cause erratic thermal behavior: the logic board finds the invalid temp. and enters fallback - maximal fan speed :bullshit: :bullshit: But this can't be the cause of OP's issue here as it increases cooling, not decreases it, but unusual enough (perhaps not for Apple) to have a mention.
The workaround was simply enough... pin #11 is open-drain, so bridge it to pin #12, ground, so the controller reads a 0 but not invalid, and you're done, but at your own risk: although an SSD is unlikely to overheat (and even then, has its own throttling), it's unknown whether the controller uses the HDD temperature together with other temperature to make a decision and whether bypassing the HDD temp. sensing makes the fan management less responsive. For those particularly care about the problem, iFixit sells an upgrade kit that recreates this signal by a microcontroller.
Can turn the thing into a vacuum cleaner, blocking all the heatsinks with dusts. thus counterintuitively, leading to overheat failure.
As far as I know, that thing is impossible to clean for a normie.
-
To share some own experience: I also upgraded lots of PCs (mainly at work) from HDD to SSD to get some more life out of them, and this usually went well.
In my experience (from about 25 years of building and repairing PCs) PCs can fail after some period of no or minimum use due to several reasons.
Those include dust and dirt buildup that causes heat issues and/or attracts humidity with leakage currents, failing powersupplies that run unstable, and especially in systems that ran 24/7 for some time, minor cracks in the PCB itself or dry electrolytic caps. After getting cold, on power-on there would be some issues...
But this should not amount to about 30% failure rate.
Some good ideas when upgrading those old systems:
- Clean for dust and dirt with compressed air
- put new thermal paste between CPU and heatsink
- check all fans for smooth operation
- check all internal cables for tight fit- sometimes it is worth unplugging and replugging them a few times to get rid of some corrosion/oxidation
- same goes sometimes for RAM modules- in earlier times where not all contacts were gold plated, the SIMM or PS/2 modules were cleaned with a eraser/rubber- blue side, if needed...
Other story: A year ago, I built myself a new PC from scratch and left my old box in the corner (minus the HDD/SSDs). Two weeks ago I wanted to repurpose that machine as a Unix workstation for some fun stuff related to photography. So took out that system, cleaned it a bit, hooked the original SSD (120 GB Samsung before they were called EVO...) and tried to power on. Nothing. Some beeps pointed to memory issues.
Did the whole stuff of swapping out memory, removing everything that would not be needed for initial powerup- nada.
Put in a CPU from another system with the same socket 1155, and on it went like a charm.
Conclusion: In about the roughly 10 months the system stood aside something happened with the CPU- a genuine Xean 1230 v3 that worked about 7 years or so without a problem...
Mainboard was/is Gigabyte, one of the first models of their "Ultra Durable Series" with a H77 chipset- solid stuff, this combo worked long hours without any hiccups...
-
Put in a CPU from another system with the same socket 1155, and on it went like a charm.
Conclusion: In about the roughly 10 months the system stood aside something happened with the CPU- a genuine Xean 1230 v3 that worked about 7 years or so without a problem...
Did you try putting the original CPU back in? Even LGA suffers from bad contacts. ISA, PCI, PCIe, SIMM slots, DIMM slots.. every single low pressure gold-plated contact in a PC I've had problems with.
-
Yes. Old CPU now does not even gets warm, System Fan is also not running- exchange CPU in, system happily gets going, CPU heats up, fan is spinning, happy single beep after a few seconds.
-
Yes. Old CPU now does not even gets warm, System Fan is also not running- exchange CPU in, system happily gets going, CPU heats up, fan is spinning, happy single beep after a few seconds.
Your 'Other story' exactly describes a build I did a few years back and also parked for a few months only to also be greeted with with a system that wouldn't fire up ! :rant:
Then not being able to get the same mainboard but a lesser spec and now older CPU socket version the build was recovered using the older but still new CPU and its run fine for some years now. :phew:
The dud board was a ASUS B85M-G but was already out of warranty when we had these issues. >:(
-
Yeah, bad solder was quite common when RoHs was first implemented.
People say that, but I don’t think it actually happened. Heck, in some aspects, lead-free solder has superior characteristics, so it’s possible that those have actually negated the negative aspects in real-world products.
Look at https://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol38-1/paper14.pdf (https://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol38-1/paper14.pdf)
Tin wiskers was a real problem when lead free first started. Does not take much to short out a chip in many cases, reflowing most likely melted a wisker long before the rest of the solder. Some people recomended uping the voltage to blow the wiskers away.
-
Yeah, bad solder was quite common when RoHs was first implemented.
People say that, but I don’t think it actually happened. Heck, in some aspects, lead-free solder has superior characteristics, so it’s possible that those have actually negated the negative aspects in real-world products.
Look at https://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol38-1/paper14.pdf (https://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol38-1/paper14.pdf)
Tin wiskers was a real problem when lead free first started. Does not take much to short out a chip in many cases, reflowing most likely melted a wisker long before the rest of the solder. Some people recomended uping the voltage to blow the wiskers away.
won't reflowing the wiskers, introduce oxides into the joint?
-
Tin wiskers was a real problem when lead free first started. Does not take much to short out a chip in many cases, reflowing most likely melted a wisker long before the rest of the solder. Some people recomended uping the voltage to blow the wiskers away.
won't reflowing the wiskers, introduce oxides into the joint?
Still way better to have a slightly crusty looking joint compared to two pins being shorted together.
I think the modern alloys are better in terms of tin wiskers so it is pretty rare. Tho it doesn't necessarily need to come from the joint itself. I heard of a server room that suddenly started having the servers die at a abnormally high rate and they had no idea why, but after a while they found out that the tin plated air ducting in the server room has started growing a tin wisker fuzz on it. The airflow would eventually pick up some of these tiny tin needles and carry it into a server where it would kill it.
-
Tin wiskers was a real problem when lead free first started. Does not take much to short out a chip in many cases, reflowing most likely melted a wisker long before the rest of the solder. Some people recomended uping the voltage to blow the wiskers away.
won't reflowing the wiskers, introduce oxides into the joint?
Still way better to have a slightly crusty looking joint compared to two pins being shorted together.
I think the modern alloys are better in terms of tin wiskers so it is pretty rare. Tho it doesn't necessarily need to come from the joint itself. I heard of a server room that suddenly started having the servers die at a abnormally high rate and they had no idea why, but after a while they found out that the tin plated air ducting in the server room has started growing a tin wisker fuzz on it. The airflow would eventually pick up some of these tiny tin needles and carry it into a server where it would kill it.
Tin plated air duct seems rather unlikely. However, zinc whiskers from raised flooring is a well known problem.
-
Tin plated air duct seems rather unlikely. However, zinc whiskers from raised flooring is a well known problem.
Sheet metal steel air ducting will be galvanized, it'd only last 10 minutes if it wasn't, but tin galvanization would be pretty rare nowadays instead of zinc galvanization.
Who knows, perhaps the HVAC boys who installed it were tinkers rather then the usual bunch of cowboys? :)
-
The explanation could be as simple as poor repair technique. For example, over tightening screws and causing stress damage to the PCB. If you do this consistently, then you will see a pattern that others do not see because they don't introduce that particular variable.