Jump to content

Kirby Smith

  • Content Count

  • Joined

  • Last visited

Everything posted by Kirby Smith

  1. I suggest, ExRoadie, that you not embrace such a narrow logic, nor guess about mine in particular. By my logic, there are several alternatives, four of which follow. a) DFI repair and return the board to its original owner, taking proper ESD precautions. This is the ESD-fair approach, but would likely delay board returns. But the customer should be given the chance to decide. B) DFI establish a testing regime for refurbished boards that can determine ESD damage, assuming that the types of ESD damage that could occur in their specific design would result in predictable and testable behavior. This may not be possible, which was my point. c) DFI guarentee refurbished boards for at least as many years as they would claim for their Japanese capacitors' MTBF. Else why bother with the caps, just do like Abit and buy caps that slowly degrade. You do end up with bigger capacitor cases than you bought, afterall. d) DFI reset the standard warrenty to time = 0 when a refurbished board is provided to a customer. Sometime in the future there might be a cost to the customer of labor trying to figure out weird behavior before he gives up and RMAs his board again, but hey, he's only the customer. My further point is that if DFI are following an RMA process that has at least the appearance of a deficiency, they should be able to say why there is no deficiency, or that such failures are as statistically rare as early new board failures, or otherwise defend their practice. Would you be satisfied with this method if it were applied to your car you brought in for service? Please note that I am not rantng against DFI here. I think that they have a superior product. They support this forum and appear to have much better than average service. However, the defense posed earlier that "everyone does it" is a very weak defense. I would prefer a technical one. kirby
  2. There is a risk with refurb'd boards due to not knowing where its been, so to speak. But in this case, the issue is not infection. The issue is electrostatic discharge (ESD). The board is shipped in ESD packaging, so one must assume that with all components soldered in place, there is still some hazard. And the hazard is that ESD can not only damage ICs in ways that are obvious in testing, but in ways that are subtle and only manifest themselves in particular circumstances. ESD can also initiate degradation that gets worse over time. That is why my employer, among many others, demands maximum proper anti-ESD practice in all laboratories and manufacturing areas. So, never-takes-direction junior wippersnapper buys a board and igores the ESD warnings because he doesn't understand why he should follow them, and zaps his board while assembling his computer. He sends the board in for repair and gets a refubished one. Meanwhile, his damaged board is repaired until it checks out and is shipped to the next person needing replacement. What is unclear here is the risk that the refurbished board is less than fully functional, or will become so in the future. I would like to hear from an engineer at Diamond Flower Electric Instruments, Inc. why this is not a problem. kirby
  3. I see other options, which I apologize for being vague about. a) Buy a new BIOS chip with a BIOS loaded. There is someone on these fora who provides them inexpensively. B) Buy (not inexpensive, but worth it if you value your time) grc.com's SpinRite6 and use it to check out the hardrive that Windows boots from unless you are certain that the drive is functioning perfectly. c) Be sure no other hard drives are connected. I've noticed that if the boot drive is absent for a single bootup, the BIOS picks the next drive as the boot drive and then complains that it is not a boot drive. However, this should be obvious if it is the problem. d) Perhaps I missed it in your initial question, but did you force the BIOS to re-acquire the properties of the hard drive and did it do so correctly? Or is it that you can't access the BIOS because it is "broken"? kirby
  4. There is a long thread at the top of the forum that has pictures of where the PSU hooks to the board. There is a MEMTEST86+ built into the BIOS. I think it is in the Genie section at the bottom. Or, you can download one from the web onto a floppy and boot with that. kirby
  5. You may want to search for check out ProGold (now called DeoxIT Gold, I notice). http://store.caig.com/s.nl It is a contact resistance reducer. In the same area of whatever electronics site (Amazon, Newark, etc,) you find it, there will be contact cleaners also. I find ProGold to be a checkbook saver on contacts on the older Audis I drive and have driven. This stuff is not inexpensive, but a tiny bottle will last your lifetime. kirby
  6. I think a better answer is that keeping the impedance low enough for good stability across the entire circuit board without these auxilliary connectors would have required another layer to the circuit board, adding significant expense in a price competitive market, or forcing the removal of components (features) so that larger traces could be run. Imagine what could be added with twice the ATX form factor for space. The point made before about the 24 pin connector also has merit; even if massive wire were used between the connector and the PSU, and even if the pins were silver, there is a limit to the size of the traces that can be run to the connector pins, and hence a limit to how low the circuit impedance can be. Ac impedance is also determined by the size of the capacitors used near each IC on the board. That too is an economic decision, for which I suggest more PSU connections is the lower cost, and probably more reliable approach. I myself appreciate the capabilities of my motherboard for its cost. I appreciate that the choice by DFI to stay away from defective capacitor manufacturers added cost that I was very much willing to bear. (In fact, this is a primary reason for my choosing a DFI board.) No matter what one's time is worth, it doesn't take long trying to troubleshoot a computer with a defective motherboard to eat the cost differential. And I have an Abit KT-7 handy if anyone wants to see the multiple examples of defective capacitor manufacture that has cost me a lot of time and money. kirby
  7. Thanks, but no thanks. I'm old school. I still believe that if the method is obvious to someone skilled in the art it shouldn't be awarded a patent. Not that this stops the modern Patent Office. The new theory is that anything can be patented, no matter how trivial. Let any conflict be fought out in court, and may the company with the biggest wallet win. In the meantime, both parties have to keep shoveling money at the Patent Office to maintain their patents. Maybe someone will copy it and OC'ers will gain the advantage of mass production. Those skilled in this art could probably reproduce it sooner than I could write the disclosure. kirby
  8. DFI Fans: There have been postings recommending that fans not be connected to the motherboard headers. There have also been postings about the CPU fan not starting, as I have observed myself. The forum recommendation is to connect the fans to an external fan controller or directly to the PSU. Yet, one might want to utilize the motherboard's ability to measure the temperature of the relevant parts, and demand more fan output when needed. External fan controllers tend to have a single speed setting. The desired properties are" - Fan should have a minimum voltage to ensure starting - Fan rpm should still be sensible by the motherboard - Fan voltage should not be less than the motherboard fan controller requests based on component temperature To accommodate these requirements, I assembled the circuit shown in the figure below. It is merely one of myriad ways of accomplishing the goals. The circuit passes the rpm signal to the motherboard, provides my desired minimum fan voltage of 7.4 volts (the zener diode voltage + 0.6V), and it tracks the motherboard commanded voltage + 0.6V. Maximum output voltage is 10.5 volts with 12 Vdc input. I don't know what the maximum is that the motherboard commands. Unless it uses an elegant circuit, such as used in rail-to-rail op amps, it too will not reach the 12 Vdc rail, and hence would never power the fan to its maximum voltage, so I doubt I have lost much there. I suspect all the parts in the diagram are available in some form from Radio Shack; I used Digi-Key for whatever I couldn't scrounge from my ancient stockpile of parts. Two of the circuits were built on Vectorboard, one for the CPU fan and one for a fan I have in the case aimed at the motherboard voltage converter fins. The circuit has run for a week now, uneventfully, but is not yet tested for long-term full power loads. It does, however, have plenty of margin for any likely fans. If you have one that requires an ampere of current at 7.4 volts, the heat sink and/or air flow will have to be appropriate for 4.6W. kirby
  9. Try this. Disconnect all harddrives except the boot drive whereever it is. Put the windows installation cd in a cd drive. Assuming a boot order of cd before harddrive, let the bios report booting from cd but let it time out. Then it should find the harddrive with windows. Boot up windows and restart it once after it settles down. Now remove the installation cd and boot again. If all goes well, you should be able to add drives during each shutdown and still have windows boot from the desired drive. kirby, who went thru all this very recently.
  10. I discovered today that there is a tertiary menu in the BIOS to determine the boot order of like-type drives. Actually, there are several such menus, one for harddrives, one for which CD player is first, and one for things like IOmega drives. These are in addition to the secondary menu setting boot order by type of device, i.e., floppy, cd, harddrive. I can't see the BIOS menus while in windows to explain exactly where it is, but it's there. kirby
  11. Not at all. I learned something new. Once I removed the data harddrive, I was not only able to boot with the boot harddrive on SATA1, but now I could use it back on IDE1 where it was originally embedded in the Arco EzRAID. I didn't get it to work there by itself originally because of the data drive and my not discovering the CD timeout trick. Now to get the data harddrive and IOmega drive back into play and all I'll have to do is RMA the EzRAID. Thanks again kirby
  12. Two points... a) The shutdown temperature is to shut down the computer if the CPU is too hot. The 'shut off fan below X degrees C' is fan specific. B) I have observed that the Panaflo fan I use can be off even when the CPU is above 25 degrees. This is because the voltage the MB controller feeds it when the chip is not too hot is below the voltage needed to start the fan. I just measured 3.5 something volts a while ago, for example. You can see that there is some voltage there by hand starting the fan blades. If the fan runs, the voltage is below the starting voltage, but above zero. Maybe the MB designers assumed fans are modeled as brushless dc torque motors. I believe somewhere on Panasonic's web site, or maybe was NH Ball Bearings web site, the voltage minimum for that fan is specified. And I think it was 7 volts. Anyway, I decided to overcome the low voltage problem with a circuit that leaves the MB in control at the hot end, limits the low end voltage to 7.4 volts, and pulls most of the power from the PSU. I'll explain further when I can formalize the schematic and have more time on the circuit in situ. Two such circuits are working at the moment for both CPU fan and fan no. 2, which I have blowing on the PWM power circuit fins. kirby
  13. oundx98: I'm not sure what you mean by IDE drives and power. Presently, there are only CD/DVD drives on IDE2. I think I may have fixed it as a result of your first response. I did not find any mention of RAID in the BIOS. However, I did find a tertiary menu specifying the boot order of each type of drive (not the order of types). I think the BIOS was finding the data harddrive (then on SATA1) and not finding a boot partition, and would stop before finding the harddrive on SATA3. I moved the boot harddrive to SATA1, removed the other hard drive and IOmega drive temporarily, and got the BIOS to boot without an install CD present. Strangely, the boot harddrive is listed in the harddrive boot order menu as channel 4. I am unclear how SATA1 becomes channel 4, but wadda I know. Thanks again, and please clarify your last remark. I am depowering the computer when moving drives on the SATA ports. kirby [drive config in signature not presently correct]
  14. The quick answer is no, because the EZraid was on IDE1 and I wasn't running any RAID on the SATA. I don't think I installed the nVidia raid drivers. The BIOS shows the four SATA drives as individuals, but maybe there is some place in the BIOS I haven't looked to check. I'll confirm next reboot, which should be very soon. I think I'll try making the boot drive SATA1 so it is the first drive seen. The BIOS might be confused by the order of drives in this temporary setup. Thanks kirby
  15. I need a suggestion here, please. I had a hard drive failure or ezRaid failure or both today, but found one good drive that I can get to boot up via a less than normal process. The BIOS sees it correctly. I have run chkdsk on it successfully. The drive is presently on SATA 3. If I attempt a normal boot, the BIOS reports that there is a boot drive failure. If I attempt a normal boot except that the Windows 2000 installation CD is in the CD player, and I let the "hit any key to boot from the CD" time run out, Windows will boot from the hard drive. Note that relative to my sig below, there is nothing presently hooked to IDE1. What's going on here? Does the drive have to be on SATA1 to be seen as the boot drive. Have I forgotten something in the BIOS that needs adjustment? Thanks for any suggestions kirby
  16. I had that problem after turning on hibernate mode at the urging of my APC UPS software. And it went away when I turned off hibernate mode. I can't claim stability yet, even though the board is not oc'd and I have no problems. I wont run serious stability tests (other than MEMSAV, which is ok) until I finish a fan controller I'm building to offload the fans from the MB and ensure that they start positively, but Azureus runs for days without a problem with a large list of files and allocated 200 kBps up speed (on the nVidia port). And I can watch DVDs while it is doing this without any burps. Board/CPU/Memory/GPU etc have made a great improvement in performance over my previous setup. kirby
  17. Are you sure that these are not simple right (red) left (white) audio cables passing analogue signals? If so, then you will need a combiner to convert RCA plugs to a stereo mini-phone plug (e.g., from Radio Shack) and plug it into the Karajan 'line in' jack or its counterpart on your sound card, if used. kirby
  18. While you should check any particular ESD foam with a megohm meter to be sure, most of it is not sufficiently conductive to short out anything on the back of a board. However, the foam should be used for test purposes only in the open as it is not very resistant to heat. Don't put ESD sensitive electronics on non-ESD foam. There can be significant static on the surface. You can't see it or feel it. Do use an ESD mat and wrist strap. I wonder how much of the RMA activity described in this forum is due to lack of ESD procedures. Most materials (and humans) have capacitance and can store charge dangerous to modern ICs. kirby
  19. Thanks for the help, but I'm running Win2k SP4 and not XP. As it happens, I tried letting Windows find a driver for the extra USB with the question mark, which it did. Sometime after that (and probably after one or more reboots) I noticed that I now have only one USB controller line, with four sub-items. These include USB 2.0 drivers. So with a bit of help, it fixed itself. I am only using USB to communicate with my UPS, so testing its bandwidth is not feasible. kirby
  20. With respect to the original question, my Win2k SP4 build has the same apparent triplication. However, if one dives into the details of each one, they seem to be diferent. I have no idea what is supposed to be normal. kirby
  21. And in my limited, but recent experience, OCZ have excellent support. kirby
  22. I recall installing the USB drivers from the DFI CD. However, I see in Device Manager the following disconnect. Am I missing some driver? If so, where do I get it? Thanks kirby
  23. Put together the components in my sig today after the PSU arrived. I followed the nice photo steps that started this thread, as well as Ed Jakobson's instructions. Initial testing went well. I switched to the optimized defaults as directed. Tested the first RAM bank in the outer orange slot with both the BIOS version of MEMTEST and ver. 1.4 that I had on a floppy. (Without a much better grounding in Linux, I have no way to make a ver. 3.2 copy at home, until this setup is completed.) Tests were uneventful. Obviously, in the above process, I determined that the Floppy drive worked. I then tested that WD's HDD formatting floppy would work with a single drive in IDE1. With these step accomplished, I tested the other bank of RAM in the outer orange slot. No problems after 3 passes. Heh heh, I said to myself, so far so good. Then I put in the first tested bank into the second orange slot and ran MEMTEST86+ again. Opps. With two 1 GB banks running, I got many errors in one bit (00040000) at location 1268.0 MB. These started after pass zero. Temps were all good at this point. (36, 40, 45) So I swapped the two RAM banks. Now the same error occurred at location 769.8, suggesting to me that the two banks are read in some mirror image mode. I then pulled one bank out (call it the good one) and ran the other in the outer orange slot. Now the same error occurred at location 384.4 MB. I think these numbers are consistent, but without knowing how the board and/or MEMTEST define memory locations, I can't be sure. The final test was to pull this RAM bank and put in the "good" one. At this writing only two passes have passed (so to speak), but there are no errors. I conclude from all this that one RAM bank failed one bit during testing and that the problem does not lie with the Athlon CPU. (One bit out of a billion bytes isn't so bad, ppm wise, but without software that finds these bits and relocates code, it is one bit too many.) Unless one of you sees another interpretation, or has another test in mind that should be tried, I intend to RMA the bad DIMM to OCZ. kirby
  24. The goal here is to filter short transients. It is unlikely that the value 0.22 uF was determined by simulation and analysis, and it is even more unlikely that the value must be precise.. If you can't find a 0.22 uF, buy two 0.1 uF caps and put them in parallel. Use them temporarily if desired until the more elegant solution of a single 0.22 uF cap is available. And no, don't use electrolytics unless you want to delve into the circuit characteristics a lot deeper. They might have to be bypassed anyway with a smaller cap due to high ESR at higher frequencies. From what I've read here, this problem appears to be a PSU design oversight -- not anticipating noise on the control line. kirby
  • Create New...