If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#21
|
|||
|
|||
Help! Major Memory Installation Issue
On 21 Apr, 18:55, Paul wrote:
wrote: The video card is an NVIDIA GEFORCE 9500GT according to the receipts. It's interesting that you mention the video card. There is one small thing I noticed when I was doing my tests. I was surprised to see a rather large fan mounted to the bottom of the card. I may have seen that before at some point but I'd completely forgotten that it had a fan of its own, over and above the CPU fan and the case fans. Anyway, just out of curiousity, I tried spinning the fan with my finger and found that it didn't turn. I had assumed it would freewheel when not being driven by electricity but it didn't budge. I didn't apply any real pressure for fear of breaking it but I was surprised that it didn't turn. Now that I think of it, I don't think that fan EVER turned during the couple of hours I spent doing these tests. It didn't dawn on me until now that it probably should have been turning all or most of the time. (I told you I'm not a hardware guy! More proof of why I'm not a hardware guy and shouldn't ever pretend to be ;-) There is no wire running to the video card from elsewhere in the computer. There is one pair of wires running from underneath the fan to a spot a short distance from the fan on the card itself and that's the only wire connected to the card. I tried powering up again (with all four memory slots in use) and that video card fan is definitely not turning. Before powering it up, I tried turning that fan with my finger and pushed somewhat harder than I did yesterday; it's definitely not turning. I wonder if it has a brake of some kind that only releases when it has power? Does the *CPU fan turning off have significance in diagnosing the problem? Is there any more I can do for myself before bringing it in for that diagnostic at the computer store? Are there any components we can be sure have not failed, like the motherboard or CPU, or is the diagnostic the best way of determining all that? I'm trying to formulate alternatives. If I'm looking at building a whole new computer (or pretty close), I wonder if I might be better to remove the drives, mount them in an enclosure of some kind and access them from my laptop and do without a desktop. That's got a major problem associated with it - the desktop is XP SP3 and the laptop is Windows 8 - but maybe I'd be able to see and access most of the data anyway. But I'm probably getting ahead of myself. Let's figure out how bad this is first before declaring the desktop dead and moving on.... I gave you my diagnosis, and I'll repeat it again. You did. I suppose I'm saying I didn't really understand it. ;-) 1) B channel on CPU memory controller has failed. To regain * * usage of all four slots for memory DIMMs, you'll need to test * * with a replacement processor. If a test processor determines a working * * A1/A2 channel and B1/B2 channel, then we know the original CPU * * had a dead B channel. If for some reason the replacement processor * * did not repair the problem, then it's motherboard. But the motherboard * * contributes next to nothing, to memory operation on the M3A. Only * * if a pin was bent, or the VDIMM regulator blew up, would the * * replacement processor fail to fix it. You've already told me * * a visual check of the DIMM slots reveals nothing. I don't remember doing any formal inspection. I certainly didn't look at them carefully. I _did_ find that the memory cards went into some slots more easily than others. Some were smooth as silk but others needed some extra firmness and felt like there might be a bit of grit in the track somewhere. I just did a thorough inspection of all four DIMM slots and the memory cards in them and can't see anything amiss in any of them. There was one speck of dust in the B2 slot but a little puff of air blew that away. The A2 slot was the one that took a little extra work to get the card settled in. And your A1/A2 * * work, which proves VDIMM is working. Ipsofacto - replace CPU ! * * If you want to run with only two DIMM slots, use only A1/A2, then * * hold off on buying the replacement CPU for now. We have to work * * on our video. CPU is not a priority at the moment. But it has * * to be fixed eventually, some day. We want working video, to prove * * we're making progress. So you think we might have a working computer with only the original memory cards if we can get some video working to prove it. That sounds encouraging! 2) The BIOS beep pattern is beeping about a video problem. You * * have discovered a video card fan which no longer rotates. This * * is a bad sign. Uh oh! In some cases, on fan failure, the heat from * * the GPU, melts the plastic body of the fan (which is confirmation of * * evil spirits :-) ) And you think this fan failed because of my mishap with changing the memory with the power still on? Why would that affect the video fan? We term this a "double fault". Does that mean we had two unrelated failures at the same time? Coincidences do happen - I've seen quite a few - although I'm inclined to think that two simultaneous failures with different causes seems unlikely. But my knowledge of electronics is negligible so maybe this is less implausible than it sounds to my ignorant ear. In any case, I'm not all that concerned about why it happened. Maybe the fan was already dead and has worked for weeks; I only noticed that it wasn't turning as a result of trying to fix the other issue. And only testing, either temporary testing (using computer shop test CPU and test video card), or you purchase "worthy" replacement parts and test with them. If you do it that way, you risk spending more for the overall repair, than going through the computer shop. I repair my own systems... because I'm an idiot, and I wouldn't do it any other way. My worst case experience, I bought one new item, of everything. That's an expensive way to do repair, but it's part of the gamble of repairing your own stuff by yourself. Most of the time, I narrow it down to just the defective item, but I don't get lucky every time. Even a computer shop will guess incorrectly, and sometimes it leads to the computer shop suffering burnt out test components. That's the risk they take, and why the repair fee is higher than it should be - it helps pay for the stuff they wreck. Agreed. They're not superhuman. They're bound to make mistakes too. I was pleasantly surprised that they didn't laugh at me when I told them what I'd done. Perhaps it was just them not wanting to **** off a customer but I'd like to think it was because they've made a mistake too on occasion. ******* The bearing on the video fan has probably failed. The fan draws power from the two pin header on the video card PCB. The video fan does not need to be plugged into the motherboard, because there is a perfectly good fan header right on the video card. It will be a two wire fan, with no RPM measurement capabilities. Too bad the video card company didn't spend a few cents on a fan failure indicator! It would only do you some good if the indicator appeared where you'd see it, like an idiot light on your car's dash ;-) There is no point "just replacing the fan", because the card is giving us evidence it is dead. We won't know it's really dead, until a test card has been substituted in its place. Now you need to borrow a video card for testing. With same slot type as 9500 GT. I've got previous computers but I don't imagine any of them would be PCI Express. I've certainly never seen anything that looked like that clamp arrangement on any of my previous PCs. The guy at the computer store said he probably had a spare CPU that would fit that mobo. I don't think he said anything about video cards but there's at least a chance. Maybe if I go there tomorrow and just see if he can swap out the video card for me, he could do that while I wait - and free - rather than putting it in his service queue.... The idea would be, you test to see if you can enter the BIOS, or get the BIOS screen up and running. If you boot into Windows with the new card in place, you'll end up running at a low resolution (like 800x600), depending on the driver situation. Some driver work may be needed, to get full resolution back. Uninstall original driver from Add/Remove, install new driver from video card boxed CD. We don't know the video card is dead. If the card is naturally a low power card, perhaps a fan failure won't kill it. On high power cards, there have been cases where a fan failure, leads to the fan being melted, as the chip underneath overheats :-) If we look at your motherboard chipset info... http://en.wikipedia.org/wiki/AMD_700_chipset_series#770 * * "770 * * *One physical PCIe 2.0 x16 slot" So you should be able to use any PCI Express video card you can find in that slot. Subject to your power supply being able to provide enough power. If you buy a $500 card, the supply may not be adequate. If you buy (or borrow) a cheap $50 card, the power is likely to be less than 50W. For example, a Geforce 210 card is available for $30. Not a gamer card by any stretch of the imagination. And your local shop will likely charge a bit more than $30 for their Geforce 210. This is good enough to be a "frame buffer", but may not be strong enough to run "The SiMMs" latest version. http://www.newegg.com/Product/Produc...82E16814134157 I'm not much of a gamer. My main game is Hoyle Puzzle Games 2003, which has Mahjong and other similar board and card games. ;-) I _do_ have Flight Simulator 2003 but haven't even tried it in years; I can never get past a point in the third lesson before washing out. But I do think I might upgrade to Windows 7 or 8 on the desktop so I'd like a video card that is compatible with that. For now, I'll see if I can get the guy at ComputersCanada to put in another video card and see if it fires up with memory in the A1/A2 slots. By the way, I said I assumed I was out of luck if I'd fried the new memory and he said not necessarily since I'd only just purchased it. He said if I went back to the manufacturer and said the right words, they'd probably give me my money back. So maybe I'll give that a whirl. If the B1/B2 slots aren't getting recognized anyway, I don't need the extra memory unless I go for a new CPU. Thanks very much for your help so far, Paul!! You've been great with answering my dumbass questions and helping me go in a sensible direction. I really appreciate that. |
#23
|
|||
|
|||
Help! Major Memory Installation Issue
On 21 Apr, 22:26, Paul wrote:
wrote: On 21 Apr, 18:55, Paul wrote: We term this a "double fault". Does that mean we had two unrelated failures at the same time? Yes. Remember, we're relying on a sick computer, to tell us what's wrong. Only you can listen to the beep pattern right now, and decide whether that's a video beep error or not. Maybe the beep pattern is just a fake, and the computer is crazy... :-) That fan did not stop on its own. On my NVidia 7900GT, the fan runs at 100% speed, until a Windows driver loads and slows it down. (On Linux, it just wails at 100% the whole time. Only the Nvidia tainted Linux driver will slow it down, under Linux. As a hardware guy, I "fixed" this (evil grin). ) You don't have an OS running right now, which means the fan should be at full power (such as it is). I'm not aware of any NVidia video card implementations, with fully automated fan control - such that a cool GPU would not require cooling and the fan could actually stop rotating. Some motherboard/CPU combinations work that way. There have been a few AMD processor setups, where at startup, the CPU fan won't rotate. It has to heat up a bit. Then it starts to spin. Coincidences do happen - I've seen quite a few - although I'm inclined to think that two simultaneous failures with different causes seems unlikely. But my knowledge of electronics is negligible so maybe this is less implausible than it sounds to my ignorant ear. At the moment, I have no theory, to correlate your RAM installation, finishing off the video card. I'm just dealing with the observations, as thin as they are. I think you have a video beep error, you're telling me the fan stopped spinning. That's not encouraging as a set of symptoms. I'm trying to get us past the point, of being stopped by a video beep error. If a substitute video card doesn't do that, then we're back to changing out the CPU, and working on it from that angle. (The theory being in that case, that the memory operation is flaky enough, it's causing the video test to fail.) Maybe you could blame a power supply failure, for causing the symptoms, but then, we are able to get beep codes, and if severe damage was done, we'd be getting no beeps at all. The presence of any beeps, is a very encouraging sign, as a lot of hardware on the core of the system has to work, to get beeps out of it. It proves a lot of onboard regulators have to be working. snip Now you need to borrow a video card for testing. With same slot type as 9500 GT. I've got previous computers but I don't imagine any of them would be PCI Express. I've certainly never seen anything that looked like that clamp arrangement on any of my previous PCs. The guy at the computer store said he probably had a spare CPU that would fit that mobo. I don't think he said anything about video cards but there's at least a chance. Maybe if I go there tomorrow and just see if he can swap out the video card for me, he could do that while I wait - and free - rather than putting it in his service queue.... Remember to install one of your working A1/A2 DIMM configs, before the trip. If you want to confirm the video card error, pull the video card (with the power off!) and switch on the system, with a working A1/A2 DIMM configuration. The beep pattern should match the beep pattern we think indicates a 9500GT problem. This is a test, to (partially) confirm the beep pattern is what the bioscentral table says it is. Now I'm really confused. I made sure the green mobo LED was off and removed the video card. I also removed the new memory cards from the B1 and B2 DIMM slots. Then I tried powering on the computer. Nothing happened. The green LED won't come on any more! I checked to make sure the AC cord was secure and tried it in a different outlet. I even tried another AC cord - I have a few kicking around - but still nothing. I also put the video card back in (securing the screw on the offchance it was grounding the card and the PC wouldn't power on otherwise). The computer won't power up now, with or without a video card, with the original or a different AC cord, in the same or different outlet. Why would my problem get worse? Is this now a triple fault?? I have a strong suspicion that the next step is the fully $40 diagnostic and that we may now have THREE active problems: a) the dead fan on the video card b) the dead B channel for the memory, possibly attributable to the CPU c) a dead power supply This is starting to sound very expensive..... The idea would be, you test to see if you can enter the BIOS, or get the BIOS screen up and running. If you boot into Windows with the new card in place, you'll end up running at a low resolution (like 800x600), depending on the driver situation. Some driver work may be needed, to get full resolution back. Uninstall original driver from Add/Remove, install new driver from video card boxed CD. I'm not much of a gamer. My main game is Hoyle Puzzle Games 2003, which has Mahjong and other similar board and card games. ;-) I _do_ have Flight Simulator 2003 but haven't even tried it in years; I can never get past a point in the third lesson before washing out. But I do think I might upgrade to Windows 7 or 8 on the desktop so I'd like a video card that is compatible with that. For now, I'll see if I can get the guy at ComputersCanada to put in another video card and see if it fires up with memory in the A1/A2 slots. By the way, I said I assumed I was out of luck if I'd fried the new memory and he said not necessarily since I'd only just purchased it. He said if I went back to the manufacturer and said the right words, they'd probably give me my money back. So maybe I'll give that a whirl. If the B1/B2 slots aren't getting recognized anyway, I don't need the extra memory unless I go for a new CPU. Thanks very much for your help so far, Paul!! You've been great with answering my dumbass questions and helping me go in a sensible direction. I really appreciate that. In terms of video card test options: 1) Borrow a PCI Express x16 card. This test case, would be a check * * that the 9500 GT is at fault. Now, say the replacement card doesn't * * work either. Then, we do (2). 2) Borrow a *PCI* video card. I have an FX5200 that fits in a PCI slot. * * I use it, when flashing the BIOS on a "better" video card. The * * card is a stinker in terms of performance, and so it does not * * stay inside any computer. But it does function as a "screen" to use, * * while debugging. If the PCI video card works, but the PCI Express * * one doesn't, then the PCI Express x16 slot is blown, and that * * means the AMD 770 chipset is damaged (new motherboard). This is * * highly unlikely to be the case. However, if you want a morale * * booster, there's nothing like trying the PCI card and getting * * to see a BIOS screen. So if you had a PCI card in your stock, * * you could try that. In a few years, they won't be putting PCI slots * * in systems any more. PCI is removed from a few new systems, already. This is the cheapest PCI test card. A 6200. $40. Similar to my FX 5200. http://www.newegg.com/Product/Produc...82E16814130466 This is the "most useful" PCI test card, in that driver support might last for a while. GeForce 210 PCI. $53 http://www.newegg.com/Product/Produc...82E16814187206 A GT 630 for $60. This would allow some gaming. http://www.newegg.com/Product/Produc...82E16814130792 A GT 640 for $85. This would allow some gaming. http://www.newegg.com/Product/Produc...82E16814162113 Video cards can come with GDDR3 or GDDR5 memory, and generally the ones with GDDR5 are better (give better memory bandwidth). For similar video cards (say, cards from $80 to $130) with the same model number, there could be up to a factor of four difference, in the performance of the memory subsystem. So while you might think a model number is a "constant", you actually have to pick through them, to find a "good value" card. The above examples were picked solely on price, and don't represent a "full shopping analysis". However, if you're motivated by price, then just using price as a selector is fine. This list, gives simple-minded single point analysis of cards. We don't know how a Passmark, correlates with game play. No idea. http://www.videocardbenchmark.net/gpu_list.php GeForce FX 5200 * * * * 7 *--- My old test card. Wow. The test is * * * * * * * * * * * * * * * * *obviously using a feature the 5200 does * * * * * * * * * * * * * * * * *not have. Software emulation is used. GeForce 6200 * * * * * 46 *--- A screamer... Not. GeForce 210 * * * * * 184 *--- Note: A gaming stinker, OK as a tester only. GeForce 7900 GT/GTO * 240 *--- My best video card. Yeah! GeForce 9500 GT * * * 291 *--- Your current card. Rest in peace. GeForce GT 630 * * * *718 * * * $60 GeForce GT 640 * * * 1306 * * * $85 None of the example cards above, have a PCI Express 2x3 power connector. Similar I think, to your 9500 GT. I'm trying to select cards, which have no extra requirements. The GeForce 210 is OK for testing, but unimpressive otherwise. Your existing card, beats mine by a bit. And the other two, offer a slight improvement. The 630/640 might even be cooler at idle. HTH, * * *Paul |
#24
|
|||
|
|||
Help! Major Memory Installation Issue
On Mon, 22 Apr 2013 14:03:59 -0700 (PDT), wrote:
This is starting to sound very expensive..... Might be cheaper to cut your losses and buy a new computer. I'm not sure about your area, but where I live (Australia) you can buy a new AMD powered PC, box only, ie, no monitor. keyboard or mouse, for under $300. |
#25
|
|||
|
|||
Help! Major Memory Installation Issue
On 22 Apr, 18:35, Gettamulla Tupya wrote:
On Mon, 22 Apr 2013 14:03:59 -0700 (PDT), wrote: This is starting to sound very expensive..... Might be cheaper to cut your losses and buy a new computer. *I'm not sure about your area, but where I live (Australia) you can buy a new AMD powered PC, box only, ie, no monitor. keyboard or mouse, for under $300. I'm definitely not going to rule that out, although money is really tight for me right now and I'd still prefer to spend a lot less than that to get back up and running. |
#26
|
|||
|
|||
Help! Major Memory Installation Issue
|
#27
|
|||
|
|||
Help! Major Memory Installation Issue
|
#28
|
|||
|
|||
Help! Major Memory Installation Issue
On 22 Apr, 21:22, Paul wrote:
wrote: Why would my problem get worse? Is this now a triple fault?? I have a strong suspicion that the next step is the fully $40 diagnostic and that we may now have THREE active problems: a) the dead fan on the video card b) the dead B channel for the memory, possibly attributable to the CPU c) a dead power supply This is starting to sound very expensive..... The $40 diagnosis, is beginning to sound pretty good about now. Head off to the store, and see what they say. If the repair list is too expensive, then bail. ******* I'm just back from taking the tower to the computer store. I went over the situation with the same guy as before, whose name is Richard. Unfortunately, he, had no recollection of our previous conversation but gradually started to recall it as we went along. Richard was really quite helpful and immediately started to try some things to isolate the problem. I think he really wanted to figure out what had failed so that he could sell me whatever I needed and get back up and running immediately. All the fun, is in the guessing right now. Your diagnostics guy, will be able to figure this out for sure. I'm guessing +5VSB is being shorted by the motherboard, for some reason. One of the first things he did was disconnect a large connector from the mobo and attach it to a digital tester. (Don't ask me what the tester is called; it looked like an oversized pager and gave digital readouts on the display but that's all I can tell you.) He tested 12 V first and said that part was fine, although the reading may have been just a wee bit on the low side. I mentioned that you had concerns about the 5 V circuit and he tested that next. He found the numbers unreasonably low on the 5 V rail and seemed to be homing in on the power supply as at least one of our culprits. I would pull the supply, put it on the bench, turn it on, then connect PS_ON# to COM with a jumper wire. (I have a home-made load tester, with a switch to do that.) I thought about doing this but: a) I've never tried removing a power supply and with all the wires coming out of it, found the prospect a bit daunting for fear that I would not be able to put it back together again b) the only test equipment I have is one very cheap voltmeter c) I'm not sure where to find PS_ON# or COM d) I'm not sure what a jumper wire is or if I could construct one out of materials at hand. I'm guessing it's just a normal length of wire with insulation on it and bare ends, which I could find but if it's something else, I might not have it My guess is, with no load, the +5VSB rail (the thing that runs the green LED), will come back. But, as soon as the power supply is connected to the motherboard, and the motherboard draws more than two or three amps from +5VSB, both the LED will be off and the supply (main section) will be off. I have a clamp-on DC ammeter, that I can test for that here. (Check the current draw on each rail.) I think that's essentially what Richard did with his tester, as reported above. The next test case, is placing a load on the supply that at least manages to spin its cooling fan. I have a power supply here, that passes the simple switch test (join PS_ON# to COM). But, as soon as you draw even a tiny bit of current from +12V (more than 0.1 amps), the +12V rail drops to nothing. If I were to connect my load box (maybe 100 watts of loading), that supply is useless. So there will be cases, where the problem shows ("weak supply"), only after you connect a representative load. The voltages from that supply are normal, when nothing is connected to it as a load. I mentioned that the fan on the video card wasn't moving and couldn't be made to turn by gentle finger pressure. He agreed that it was seized. In answer to my question, he said that if I wasn't working the card very hard, that fan could have seized months ago and not shown any symptoms. He tried another video card and we were able to get into the BIOS. He said the 5 V and 12 V numbers looked good in there. However, we had a message about the CPU fan. He changed a bunch of settings with regards to the CPU fan and then we got a lot of beeps on the next bootup. He seemed confident that the CPU fan should be replaced and showed me one that he said would be perfectly good for $15. It was starting to look like a new video card and a new CPU fan might be the extent of the problems. However, there were still doubts about the power supply. I gave him a printout of my diagnostic tests with the memory. We discussed this and he was concerned about the B1/B2 DIMM slots. He said there might be an issue with the motherboard that tied the various problems together but that was not trivial to diagnose. When he saw that the CPU fan WOULD turn just fine in cases where we got no beep codes, I think he realized that things were more complicated than he first hoped. I was certainly confused since he'd indicated possible problems with the power supply, CPU fan, motherboard, as well as the video card. So I suggested maybe the full diagnostic for $40 was the way to confirm exactly which components had failed and he agreed. So I left him the tower and memory and paid for the diagnostic. He's going to try to have the results for me for Thursday but he's got quite a lot in his queue so it might take longer. He's even going to take a whirl at fixing the video card. I think he wants to try cleaning and lubricating the fan to see if that gets it behaving again. Your diagnostics guy will know how to do these tests, and come to some conclusion. Considering the number of things wrong with the system now, just the runtime alone for all the tests, is going to soak up the entire $40. I suspect it's going to feel like money well-spent once we have a relatively definitive set of results. Maybe, just maybe, I'll still get out of this without having to spend a fortune.... I'll let you know what he reports. I'm sure you're curious to know what the problem ultimately was ;-) -- Rhino |
#29
|
|||
|
|||
Help! Major Memory Installation Issue
|
#30
|
|||
|
|||
Help! Major Memory Installation Issue
On 23 Apr, 16:05, Paul wrote:
wrote: I'm just back from taking the tower to the computer store. I went over the situation with the same guy as before, whose name is Richard. Unfortunately, he, had no recollection of our previous conversation but gradually started to recall it as we went along. Richard was really quite helpful and immediately started to try some things to isolate the problem. I think he really wanted to figure out what had failed so that he could sell me whatever I needed and get back up and running immediately. All the fun, is in the guessing right now. Your diagnostics guy, will be able to figure this out for sure. I'm guessing +5VSB is being shorted by the motherboard, for some reason. One of the first things he did was disconnect a large connector from the mobo and attach it to a digital tester. (Don't ask me what the tester is called; it looked like an oversized pager and gave digital readouts on the display but that's all I can tell you.) He tested 12 V first and said that part was fine, although the reading may have been just a wee bit on the low side. I mentioned that you had concerns about the 5 V circuit and he tested that next. He found the numbers unreasonably low on the 5 V rail and seemed to be homing in on the power supply as at least one of our culprits. I would pull the supply, put it on the bench, turn it on, then connect PS_ON# to COM with a jumper wire. (I have a home-made load tester, with a switch to do that.) I thought about doing this but: a) I've never tried removing a power supply and with all the wires coming out of it, found the prospect a bit daunting for fear that I would not be able to put it back together again b) the only test equipment I have is one very cheap voltmeter c) I'm not sure where to find PS_ON# or COM d) I'm not sure what a jumper wire is or if I could construct one out of materials at hand. I'm guessing it's just a normal length of wire with insulation on it and bare ends, which I could find but if it's something else, I might not have it My guess is, with no load, the +5VSB rail (the thing that runs the green LED), will come back. But, as soon as the power supply is connected to the motherboard, and the motherboard draws more than two or three amps from +5VSB, both the LED will be off and the supply (main section) will be off. I have a clamp-on DC ammeter, that I can test for that here. (Check the current draw on each rail.) I think that's essentially what Richard did with his tester, as reported above. The next test case, is placing a load on the supply that at least manages to spin its cooling fan. I have a power supply here, that passes the simple switch test (join PS_ON# to COM). But, as soon as you draw even a tiny bit of current from +12V (more than 0.1 amps), the +12V rail drops to nothing. If I were to connect my load box (maybe 100 watts of loading), that supply is useless. So there will be cases, where the problem shows ("weak supply"), only after you connect a representative load. The voltages from that supply are normal, when nothing is connected to it as a load. I mentioned that the fan on the video card wasn't moving and couldn't be made to turn by gentle finger pressure. He agreed that it was seized. In answer to my question, he said that if I wasn't working the card very hard, that fan could have seized months ago and not shown any symptoms. He tried another video card and we were able to get into the BIOS. He said the 5 V and 12 V numbers looked good in there. However, we had a message about the CPU fan. He changed a bunch of settings with regards to the CPU fan and then we got a lot of beeps on the next bootup. He seemed confident that the CPU fan should be replaced and showed me one that he said would be perfectly good for $15. It was starting to look like a new video card and a new CPU fan might be the extent of the problems. However, there were still doubts about the power supply. I gave him a printout of my diagnostic tests with the memory. We discussed this and he was concerned about the B1/B2 DIMM slots. He said there might be an issue with the motherboard that tied the various problems together but that was not trivial to diagnose. When he saw that the CPU fan WOULD turn just fine in cases where we got no beep codes, I think he realized that things were more complicated than he first hoped. I was certainly confused since he'd indicated possible problems with the power supply, CPU fan, motherboard, as well as the video card. So I suggested maybe the full diagnostic for $40 was the way to confirm exactly which components had failed and he agreed. So I left him the tower and memory and paid for the diagnostic. He's going to try to have the results for me for Thursday but he's got quite a lot in his queue so it might take longer. He's even going to take a whirl at fixing the video card. I think he wants to try cleaning and lubricating the fan to see if that gets it behaving again. Your diagnostics guy will know how to do these tests, and come to some conclusion. Considering the number of things wrong with the system now, just the runtime alone for all the tests, is going to soak up the entire $40. I suspect it's going to feel like money well-spent once we have a relatively definitive set of results. Maybe, just maybe, I'll still get out of this without having to spend a fortune.... I'll let you know what he reports. I'm sure you're curious to know what the problem ultimately was ;-) -- Rhino I think he's doing well so far. About the only area he might be going off the rails, is with regard to the B1/B2 symptoms. An AMD motherboard, where the memory controller is inside the processor, the motherboard contributes precious little to the test results. The motherboard is just a bunch of copper wires in that path. The motherboard copper wires, join the memory controller inside the CPU, to the DIMM slots. The fact that you got into the BIOS, means A1/A2 are working, means A1/A2 had working VDIMM and Vtt supplies. In other words, the motherboard has proved its merits. His next step in resolving the B1/B2 problem, is to install his test AM2 processor, and note that B1/B2 start working again. Then, his conclusion will be, new CPU. That's my best guess at this point. ******* A computer motherboard is tremendously complex. Many kinds of failures are theoretically possible (and I've seen a bunch under lab conditions). The thing about production components, motherboards made in the millions, a number of fault types just don't happen. This is a function of the degree of control the factory has, over the quality. That leaves only a certain set of classes of failure, to be seen in the field. 99% of the faults will fall into those classes. For example, things exposed to the user, where the user can "zap" things with static electricity, that opens some fault possibilities. But those faults are "outward facing", things like blown USB ports on a USB2 NEC chip. Internal signals, like signals from Southbridge to Ethernet controller, you never find those blown out. Similarly, you never find cracked PCB copper traces. It just doesn't happen. 99.5% of the time, it's going to be something else which is damaged, not the traces. I don't think he considers damaged traces very likely, more of an outside possibility. He said he wasn't going to go over the whole mobo with a voltmeter testing each circuit. For one thing, he said he doesn't have a voltmeter ;-) But mostly the issue is time; checking every circuit would take a LONG time. Basically, he said it just isn't done; instead, the mobo is replaced if it turns out to be the problem. I had a poor quality PCB shop in town, which used to make board blanks for us. We ordered small quantities from them. A board blank came in, we built it up, soldered components to it, tested it, stored it in a shelf PCB holder for later. That board tested as good. Three months later, I tested it again and it failed. My boss sent it out for bed-of-nails testing, and the tester showed that over 200 copper traces had gone open circuit. That means the blank PCB was defective, (open circuits happened while the PCB was cold and stored unpowered in the lab) and was fabricated improperly. Now, our production boards, *never* do that. Neither do the three million boards a month that Asus might make. The shop in town was disqualified as a supplier, and we shopped elsewhere. PCB is the material the circuit board is made of, right? The basic green plastic? Poly-something-something? My knowledge of electronics and the techniques used to make chips and boards is negligible. I projected a film that talked about the process once when I had a part- time gig showing films on campus and that's been pretty much my whole exposure to electronics, I'm afraid ;-) And that was 30+ years ago so even that isn't very up-to-date ;-) So when I listen to a set of symptoms here, I tend to dismiss some fault types as very unlikely. In the case of your DIMM problem, I see evidence the DIMMs are being powered OK. Since the A1/A2 are working, that leaves connectivity on B1/B2 (bent pins, broken trace, or bad CPU). You've verified no bent pins, the symptoms affect both B1/B2, and that says bad CPU. As to when it blew, we'll never know. If could be, that the B-bus was bad on the CPU from day one. But more likely, something happened when you were plugging in DIMMs. So while lots of things could happen (as in my example, broken traces), the vast amount of evidence out there, is broken traces just don't happen on production boards. Production systems use closed loop feedback, to tune production and watch for problems. For example, in our production, we got a bad batch of boards, and it was detected by impedance coupon testing, and the boards didn't even get loaded into the pick and place machine. The boards were put aside, to be cross-sectioned and checked. they were out of spec, and the supplier ate the loss. They're actually supposed to check the impedance coupon themselves, as part of their checks. No, that makes perfect sense. Look for things that actually happen in the real world; don't worry about things that are only theoretically possible but never seen. ******* On the video card, I'd inspect the plastic fan body for signs of melting. If it's melted, that means the chip temperature went too hot. (The fact the bearing is seized, means something, but I don't know what.) I don't see any evidence right now, that the 9500 GT can be recovered. The fact his shop video card worked, rules out a bad PCI Express x16 slot. (Note that, it's possible to blow lanes on an x16 slot, and the slot will still work. The negotiation just narrows down the lane set, and you get reduced slot bandwidth. What the slot can't handle well, is if lane 0 fails, then that would be deadly. Lane 15 isn't nearly as important, and if lane 15 fails, it can run in x8 mode.) I expect Richard will do something along those lines. I was a little suprised that he even considered fixing the card. He'd earlier said you can't really fix them and pointed out that a decent replacement card (I don't recall which one) was only about $45. Then suddenly he talked about repairs. I suspect he was really hoping to diagnose the whole problem with what he did today and is offering to try fixing the video card as sort of a consolation for him having to ding me $40 for the diagnostic.... In any case, I have reasonable comfort that he is competent which is a big hurdle. It wouldn't be hard for him to tell me a bunch of BS but I think he was actually being straight with me. Here's hoping he is! -- Rhino |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Major Issue With P5K-E Not Detecting SATA [Non-RAID] | Martin Hughes | Asus Motherboards | 1 | November 11th 07 04:44 PM |
PowerPath SLES9 SP1 installation issue | Seb | Storage & Hardrives | 2 | July 29th 05 10:31 AM |
Major Issue with 8400 | llamaPie | Dell Computers | 4 | January 29th 05 04:07 AM |
Decoder installation issue | snowball | Ati Videocards | 1 | May 13th 04 08:36 AM |
Remote Wonder Installation Issue | Andrew | Ati Videocards | 1 | April 3rd 04 06:42 PM |