Motherboard, CPU, RAM Diagnostics Flowchart
Warning! You must unplug your ATX power supply from the wall before working inside the case.
Copyright 2008 by Morris Rosenthal
All Rights Reserved
The 2nd edition of the eBook can downloaded anywhere in the world for $9.95 or the new 3rd edition of the book can ordered from Amazon for $17.00, or from Amazon UK for £13.95 or through any retail store by the title "Computer Repair with Diagnostic Flowcharts Third Edition" or ISBN which is: 0-9723801-8-3.
CPU, RAM and Motherboard Troubleshooting
Note that these steps correspond with decision points on the flowchart and are reached directly by clicking on the diamond symbols. The text below cannot be read sequentially.
Do you get a live screen? A message saying "No Video Signal" or anything similar doesn't count as a live screen in this case. You need to get at least as far as a BIOS screen, either the system BIOS or an adapter BIOS loading.
Does the system power up? Do you hear any beeps, drives spinning up, fans, etc. If the power isn't coming on, proceed to Power Supply Failure. If the power supply diagnostics sent you back here, follow through these diagnostics as a double-check before giving up on the motherboard.
If you haven't performed the Video Failure diagnostics for a dead screen yet, do so now. Don't ignore the obvious steps, like checking the power cord and the outlet. If you skip the video diagnostics and continue with the motherboard flowchart, you could easily end up buying replacement parts for hardware that's not bad.
One of the most common failures following motherboard or RAM upgrades is improper insertion of memory modules. The levers should be lowered before inserting the memory module, and should raise themselves up and lock in place when the module is correctly seated. If you're using obsolete RIMM (Rambus Inline Memory Module) memory, the modules in a bank must be matched, and you must install CRIMMs (Continuity RIMMs) in the empty sockets. If you're using older SIMM (Single Inline Memory Modules), each bank needs a matched pair. In both cases, matched doesn't just mean capacity and speed, it also means manufacturer. I have a new page up for upgrading laptop memory with SODIMMs. For replacing regular DIMM memory, see the illustrated replacing RAM.
There are a number of reasons for a system with a good power supply to refuse to power up which were covered in the power supply diagnostics. Another reason is a failed CPU insertion, whether it's a slot or socket CPU. With good lighting, using a flashlight if necessary, make sure that any socket CPU is sitting dead flat in the socket, which means that the heatsink should be perfectly parallel to the motherboard surface; the CPU may be so totally hidden beneath some heatsinks that you can't see the edges. This problem should really only be relevant if you just upgraded your CPU or installed a new motherboard, because the CPU socket locks the CPU in firmly and the heatsink adds another level of clamping. If a socket CPU is a new install, you have to remove the heatsink and CPU to visually inspect it for damage such as crushed or bent legs. A CPU will not seat correctly if the socket locking arm wasn't raised all the way up before the CPU was inserted, or wasn't lowered all the way down after. If your CPU won't sit down in the socket properly, either the socket is faulty or you have the wrong CPU for the motherboard! I haven't seen a CPU creep out of a socket due to thermal shock for over a decade.
It's pretty tough to tell if old slot type CPUs are seated by visual inspection, so when in doubt, I reseat them. On the plus side, you can remove and reseat a slot CPU without removing the heatsink, since they form an integral unit. Make sure you correctly identify release levers on a slot CPU package, which are normally located at the top of the CPU package, to the inside of the motherboard support structure.
A stone dead CPU is another reason for a system to fail. All modern CPUs require a heatsink, and most of these are an active heatsink, with a fan on top. You may encounter a heatsink without a fan in mass-manufactured brand-name systems where the manufacturer had the engineering talent in-house to do a thorough thermal analysis and determined that the airflow over a passive finned heatsink was enough to keep the CPU within the operating temperature range. When there is a fan on the heatsink, it must be hooked up to the correct power point on the motherboard for the BIOS to monitor its condition and turn it off and on. If you just installed a new CPU and powered the system up with no heatsink, it may have failed already. If the fan on your active heatsink isn't spinning up, replace it and hope for the best. Make sure you see the new heatsink fan operating since it could be the power point on the motherboard that's failed.
If you have a system that powers up, the next question is, do you hear any beeps coming from the motherboard speaker. If your motherboard doesn't have an integrated piezoelectric speaker but does have a speaker connection next to the power and reset connections (usually the front, left-hand corner of the motherboard) attach a case speaker. If you hear an unending string of beeps, it's often bad RAM, while a repeated sequence can be RAM or video. Other beep codes have been largely abandoned since they pertained to non-user replaceable surface mount components. Beeps or no beeps, I always reseat the video adapter and the RAM, paying special attention to the locking levers on the memory sockets
Are your motherboard settings on the defaults? Whether you just put in a new motherboard or have been fooling around with overclocking, restore the default settings. This is often accomplished with a single jumper or switch setting, but sometimes it involves moving several jumpers or switches. Get the default values from the motherboard documentation. If you can't find the original manual or locate the equivalent documentation on the Internet, you may have to skip this procedure. Sometimes, the silk screens on the motherboard are sufficiently detailed to work out the defaults, but you need really good eyes to figure it out.
Although we're repeating a little of the power supply diagnostics here, stripping down the system is the next step in a "no power-up" scenario. Unplug the power cord before each change in the case. Disconnect drives, one at a time, reconnecting power and trying power up after each. Next start removing adapters, saving the video adapter for last, reconnecting power and retrying after each change to ensure you discover which component is causing the failure.
Running the motherboard without a case is a common technique used by technicians to eliminate any weird grounding and shorting issues or mechanical stresses. It also makes it much easier to swap the CPU if that's required. I normally do my bench testing on top of a cardboard box, with a static free bag or foam between the bottom of the motherboard and the cardboard. You don't walk away from a test like this or you might come back to find the box on fire! If your motherboard powers up on the bench with the same power supply that you used in the case, you have a geometry problem. Ideally, you should have a spare power supply for bench testing if you're going to do regular repair and testing work.
Make sure some standoffs aren't higher than others, putting unacceptable stress on the motherboard. Check that every standoff appears under a screw hole. The easiest way to be sure is to count the standoffs, count the screws, and make sure there are no screws leftover after you install the motherboard. There could be a short caused by a misplaced standoff, a loose screw, metal chips from shoddy materials. I've encountered standoff shorts that produce an endless string of beeps like RAM failure, without damaging the motherboard. There's also the possibility that the case geometry is so messed up (out of square or level when the cover is forced on) that it's putting an unacceptable mechanical stress on the motherboard resulting in an open circuit. If you can't find the cause of the problem, don't hesitate to try another case and power supply.
If you still have a "no power" situation with the motherboard running out of the case, there's always the last refuge of a scoundrel. Swap in a known good CPU not forgetting to install a good heatsink and to connect the fan, even just for a quick test. I try to keep around some cheap old CPUs for this purpose, just in case the motherboard is a CPU eater. It's another good reason to leave all the motherboard settings on the default "Automatic" setting, so you don't have to fool around with them at this stage. If your old CPU is bad and the heatsink fan is dead, it's a pretty sure bet that the dead fan caused the CPU failure. If the heatsink fan is working, determining whether the CPU failure was due to poor heatsink contact, improper motherboard settings, or lousy power regulation from the motherboard is a guessing game. If the motherboard is an older make and you have a couple bucks to spare, replace the CPU and the motherboard together. Replacing just the CPU, even if the motherboard tests out OK, is kind of risky and usually tough to justify from a price/performance standpoint unless the system was practically new, say less than a half a year old.
If you still have a no power situation, not to mention no beeps and no video, you're probably looking at a bad motherboard. Again, this diagnosis assumes that you went through the Video Failure diagnostics, which would have forced you through the Power Supply Failure diagnostics as well. I still wouldn't be in a hurry to take a gun to the motherboard. Get your system operating with a replacement motherboard and all the identical parts that the old motherboard failed with before you make the trash can decision. I just added some illustrated instructions for replacing a motherboard to this site, including installing a new motherboard.
Does the system power right up, give a happy beep or two, then freeze on the BIOS screen? This can occur on an all text screen, during or after memory count, while checking for drives, or the feared "Verifying DMI Data Pool." The problem is very likely due to a conflict, most like between the adapters but also possibly between incompatible drives sharing a bus.
Strip the system down to bare-bones, just a power supply, motherboard, minimum RAM, CPU and heat sink, and video adapter. If the system no longer freezes when it's stripped down, but complains about the lack of a boot device, proceed to Conflict Resolution.
Try swapping the RAM around, reordering the banks if you have more than one bank of RAM installed, or moving the only module installed to a neighboring slot. If this doesn't cure the freeze-up, and you have some suitable known good RAM from another system, try it. If the RAM currently installed doesn't meet the motherboard manufacturer specs, you shouldn't be using it; even if it seemed to work until this point. Improperly selected RAM can be the cause of problems ranging from no-boot to intermittent lock-ups. Is the RAM seated correctly and in the proper quantities (i.e., number of modules, addition of continuity modules, or CRIMMs, if you are using RIMMs). Also make sure that the system didn't use tinned (silver color) contacts against gold contacts, or the dissimilar metals will cause corrosion over time due to a constant electrical current when the power is off. Replacing RAM at this point isn't a guaranteed proposition, but it's a good item to eliminate. Don't toss out the RAM you remove because you may find out later that it's actually good.
If you aren't using the default CMOS settings, try restoring them all at this point. You can usually restore these from a major CMOS menu item like "Restore Default Settings" or "BIOS Default Settings." The default settings usually put everything on autodetect and use the recommended timing for the RAM. This means if you're overclocking, stop it, at least until you get the system running again. It doesn't matter whether or not overclocking the exact same CPU or RAM in a friend's system worked without a hitch, you're exceeding the manufacturers recommendations so it's a gamble.
An overheating will cause the system to quickly lock up. Remove the existing heat sink and fan, make sure that the fan is working properly AND that the geometry of the bottom of the heat sink will bring it in full contact with the exposed CPU die or the top of the CPU package. See my illustrated guide for how to replace a CPU. Apply an approved thermal grease or thermal tape before reinstalling the heat sink. Don't put on too much thermal grease or you'll just make a mess. The thermal media is only there to fill the microscopic gaps between the die surface and the heat sink. Don't improvise your thermal material, go to a computer or electronics store and buy some. Installing heat sinks can be frustrating, but this isn't a "bash away at it" process. You can damage the CPU if you start cracking the heat sink against it in an attempt to get the heasink to sit right. Be patient, study the mechanical connections, make sure you aren't hitting some poorly placed component on the motherboard and check that your heat sink isn't so oversized it just won't fit on the particular motherboard.
Make sure the fan on that heat sink spins up the second that power comes on. If it doesn't, despite being connected to the correct power point (see the motherboard manual), replace it with a new active heat sink unit. Make sure the bottom surface of the new unit will make full contact with the exposed CPU die or the top of the CPU package. The only problem with replacing an active heat sink is it may be too late for your CPU. CPUs have an unfortunate tendency to damage themselves when they overheat. Some CPUs can go into thermal runaway and destroy themselves in a matter of seconds without proper cooling.
Computer Repair with Diagnostic Flowcharts | Power Supply Troubleshooting | Video Card Diagnostics | CPU, RAM Motherboard Troubleshooting | Hard Drive Failure | CD and DVD Troubleshooting | Modem Diagnostics | Sound Card Problem | Network Troubleshooting | Order Book