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
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
Norm Why wrote:
The safety ground on the three prong plug, should be joined to the chassis ground. You are allowed to test the system with no SATA cables in place. You can bring up a machine and enter the BIOS, while no hard drives or optical drives are connected. This would allow you to observe system operation as SATA cables are added to a successfully running (so far) system. Paul Thanks Paul, I've even removed the beep speaker to find the short. Ohm meter says short is low resistance. Step one in two-page diagnostic troubleshooting flow chart must be wrong , or it is a bad, bad short. Assuming Step One was wrong. I carried on down to Plug in Power supply. Turn power on. No Beep, momentary power, system drops dead. Short must be real and getting worse, or system is learning how to be more uncooperative. The short to ground goes through the 4pin 12V connector. Did you reverse it ? The 2x2 has shaped enclosures for the pins, and they are only intended to fit one way. The two pieces that form a latch, should be on the same side. If a power supply has become weak with age, it can shut down when less than the rated load is present. You'd need a clamp-on ammeter, to easily be able to make measurements of current flows on the various rails. For example, if the main PSU connector has four +5V wires, you put all four within the jaws of a DC clamp-on ammeter, and the meter gives the total current flow in the four wires. For the first 35 milliseconds, the supply ignores overloads. This gives time for the main outputs to charge any capacitors present. It's after that point, that it starts looking for overloads. I do my initial builds, on the kitchen table, using a phone book as an "insulating base". Once the components are proved working, then I start installing the stuff in a computer case. and this gives an opportunity to compare symptoms between "insulated" and "grounded". Paul |
#22
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
You are allowed to test the system with no SATA cables in place.
You can bring up a machine and enter the BIOS, while no hard drives or optical drives are connected. This would allow you to observe system operation as SATA cables are added to a successfully running (so far) system. Paul Thanks Paul, I've even removed the beep speaker to find the short. Ohm meter says short is low resistance. Step one in two-page diagnostic troubleshooting flow chart must be wrong , or it is a bad, bad short. Assuming Step One was wrong. I carried on down to Plug in Power supply. Turn power on. No Beep, momentary power, system drops dead. Short must be real and getting worse, or system is learning how to be more uncooperative. The short to ground goes through the 4pin 12V connector. Did you reverse it ? No. Even removed I still have short to ground. The 2x2 has shaped enclosures for the pins, and they are only intended to fit one way. The two pieces that form a latch, should be on the same side. If a power supply has become weak with age, it can shut down when less than the rated load is present. You'd need a clamp-on ammeter, to easily be able to make measurements of current flows on the various rails. For example, if the main PSU connector has four +5V wires, you put all four within the jaws of a DC clamp-on ammeter, and the meter gives the total current flow in the four wires. For the first 35 milliseconds, the supply ignores overloads. This gives time for the main outputs to charge any capacitors present. It's after that point, that it starts looking for overloads. I do my initial builds, on the kitchen table, using a phone book as an "insulating base". Once the components are proved working, then I start installing the stuff in a computer case. and this gives an opportunity to compare symptoms between "insulated" and "grounded". Paul I am now pondering my silly collection of MOBO mounting screws. https://www.aliexpress.com/item/32957035974.html Shows nylon screws that can be used for mounting or standoffs. Standoffs don't work for me because I need PCI VGA card. Size: M2 (Thread 6mm) M2.5 (Thread 6mm) M3 (Thread 6mm) M4 (Thread 6mm) I guess 'size' is pitch, because all are 6mm thread. What size would work? Length: 5mm (50pcs) 6mm (50pcs) 7mm (50pcs) 8mm (50pcs) 9mm (50pcs) 10mm (50pcs) 11mm (50pcs) 12mm (50pcs) 13mm (50pcs) 14mm (50pcs) 15mm (50pcs) 16mm (50pcs) 17mm (50pcs) 18mm (50pcs) 19mm (50pcs) 20mm (50pcs) 21mm (50pcs) 22mm (50pcs) 23mm (50pcs) 24mm (50pcs) 25mm (50pcs) 26mm (50pcs) 28mm (50pcs) 30mm (50pcs) 32mm (50pcs) 35mm (50pcs) 40mm (50pcs) 45mm (50pcs) 50mm (50pcs) What Length would work? |
#23
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini OnboardCase Buzzer Board Beep Alarm NEW."
Norm Why wrote:
You are allowed to test the system with no SATA cables in place. You can bring up a machine and enter the BIOS, while no hard drives or optical drives are connected. This would allow you to observe system operation as SATA cables are added to a successfully running (so far) system. Paul Thanks Paul, I've even removed the beep speaker to find the short. Ohm meter says short is low resistance. Step one in two-page diagnostic troubleshooting flow chart must be wrong , or it is a bad, bad short. Assuming Step One was wrong. I carried on down to Plug in Power supply. Turn power on. No Beep, momentary power, system drops dead. Short must be real and getting worse, or system is learning how to be more uncooperative. The short to ground goes through the 4pin 12V connector. Did you reverse it ? No. Even removed I still have short to ground. The 2x2 has shaped enclosures for the pins, and they are only intended to fit one way. The two pieces that form a latch, should be on the same side. If a power supply has become weak with age, it can shut down when less than the rated load is present. You'd need a clamp-on ammeter, to easily be able to make measurements of current flows on the various rails. For example, if the main PSU connector has four +5V wires, you put all four within the jaws of a DC clamp-on ammeter, and the meter gives the total current flow in the four wires. For the first 35 milliseconds, the supply ignores overloads. This gives time for the main outputs to charge any capacitors present. It's after that point, that it starts looking for overloads. I do my initial builds, on the kitchen table, using a phone book as an "insulating base". Once the components are proved working, then I start installing the stuff in a computer case. and this gives an opportunity to compare symptoms between "insulated" and "grounded". Paul I am now pondering my silly collection of MOBO mounting screws. https://www.aliexpress.com/item/32957035974.html Shows nylon screws that can be used for mounting or standoffs. Standoffs don't work for me because I need PCI VGA card. Size: M2 (Thread 6mm) M2.5 (Thread 6mm) M3 (Thread 6mm) M4 (Thread 6mm) I guess 'size' is pitch, because all are 6mm thread. What size would work? Length: 5mm (50pcs) 6mm (50pcs) 7mm (50pcs) 8mm (50pcs) 9mm (50pcs) 10mm (50pcs) 11mm (50pcs) 12mm (50pcs) 13mm (50pcs) 14mm (50pcs) 15mm (50pcs) 16mm (50pcs) 17mm (50pcs) 18mm (50pcs) 19mm (50pcs) 20mm (50pcs) 21mm (50pcs) 22mm (50pcs) 23mm (50pcs) 24mm (50pcs) 25mm (50pcs) 26mm (50pcs) 28mm (50pcs) 30mm (50pcs) 32mm (50pcs) 35mm (50pcs) 40mm (50pcs) 45mm (50pcs) 50mm (50pcs) What Length would work? The trick is not to lose the hardware that came with the computer case. The standoffs are custom length, intended to elevate the PCB to be flush with the slot holes for the I/O cards, in the back. There's no way for me to guess any such measurement from here. There are as many styles of solutions, as there are days of the week. If you artificially elevate the motherboard, with too-tall standoffs, then your video card won't fit. It's one of the items on the buyers list, to check the fit and dress of standoffs before buying a computer case. And that's pretty hard to do. You can't always inspect this issue in advance. I used to look at computer cases at a mom&pop (single store), while they were in business. Basically, any store in town that put computer cases on display in a useful way, went out of business. And I did buy cases from them, so it's not like the effort on their part wasn't rewarded. For example, I don't buy "bump" style trays. That's where the metal of the tray is formed into a bump of the appropriate height, then a hole drilled, then the hole is threaded. While it saves the manufacturer $0.0001 per hole, it doesn't do me any good. Instead, I like the brass standoffs that have a male thread on one end and a female on the other, then the motherboard screw fits to fasten the board down. These have a better controlled contact area where it touches the ground ring on the motherboard. In terms of "places that short", there was one Asus motherboard where the plastic socket frame around the CPU socket, was fitted *before* the soldering step. The plastic went through a wave solder perhaps, and it swept up some solder underneath the plastic (hard to inspect) and it shorted out VCore. But that would only cause VCore with a current limiter to trip off. And a little repair work, you could find and remove the short. But for direct rail shorts, sometimes there can be a short inside the motherboard, but it's less likely there is a short, say, right at the 24 pin connector. Some Biostar motherboards, used to join the yellow 12V wires on the 2x2 ATX12V, to the 12V wires on the main 24 pin connector. This is not a good idea, and I have no idea why Biostar was doing that. The 12VA and 12VB should be separate from one another, in case some old PSU with actual separate transformers for those two 12V was in usage. The 12V on the ATX12V 2x2, could be a different 12V than the 12V on the main connector (fraction of a volt difference). Since you say you are observing a short with the ATX12V removed, then this suggests something on the rails on the main 24 pin is shorted. But I don't see a cosmetic root cause, because the transfer of some of those rails around the motherboard, is on a buried layer. The motherboards have 4,6, or 8 layers, with 6 layers used for RAMBUS boards, 4 layers for a lot of regular DDR motherboards, and 8 layers reserved for sockets with huge pincounts, where it's just too hard to route all the signals using a limited number of layers. Somewhere in that stack, can be some 2oz layers for routing power. At work, we would use a bed-of-nails tester for evaluating "ohms". After a board came off the line, sometimes tests were done that way too (JTAG), which isn't all that common in the computer industry. Those are known as structural tests. A company like Asus, would use a camera and automation to check for visible flaws (stuff in places it doesn't belong), can use an Xray machine for BGA verification, but for electrical test, there is only a 2 minute hand functional test (lady plugs boards into mobo and runs some test code for two minutes total time, then product goes in box). They rely on statistical sampling for quality, rather than every board being inspected with a magnifying glass. Every board receives the 2 minute functional test, but other steps are likely automated. To visually inspect a square foot of PCB, takes about 2 hours with a 10x magnifier, because I've done a few boards visually, and you do find stuff. (That's where I found a certain brand of resistors, with all the bodies cracked in half.) At one time, Asus made 5 million motherboards per month, and there's no time for that sort of inspection, and they can only sample motherboards and inspect them for manufacturing defects, then go back to the line and correct them. Not every motherboard can be treated like a Rolls Royce. Paul |
#24
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
[snippage]
I am now pondering my silly collection of MOBO mounting screws. https://www.aliexpress.com/item/32957035974.html Shows nylon screws that can be used for mounting or standoffs. Standoffs don't work for me because I need PCI VGA card. Size: M2 (Thread 6mm) M2.5 (Thread 6mm) M3 (Thread 6mm) M4 (Thread 6mm) I guess 'size' is pitch, because all are 6mm thread. What size would work? Length: 5mm (50pcs) 6mm (50pcs) 7mm (50pcs) 8mm (50pcs) 9mm (50pcs) 10mm (50pcs) 11mm (50pcs) 12mm (50pcs) 13mm (50pcs) 14mm (50pcs) 15mm (50pcs) 16mm (50pcs) 17mm (50pcs) 18mm (50pcs) 19mm (50pcs) 20mm (50pcs) 21mm (50pcs) 22mm (50pcs) 23mm (50pcs) 24mm (50pcs) 25mm (50pcs) 26mm (50pcs) 28mm (50pcs) 30mm (50pcs) 32mm (50pcs) 35mm (50pcs) 40mm (50pcs) 45mm (50pcs) 50mm (50pcs) What Length would work? The trick is not to lose the hardware that came with the computer case. The standoffs are custom length, intended to elevate the PCB to be flush with the slot holes for the I/O cards, in the back. There's no way for me to guess any such measurement from here. There are as many styles of solutions, as there are days of the week. If you artificially elevate the motherboard, with too-tall standoffs, then your video card won't fit. It's one of the items on the buyers list, to check the fit and dress of standoffs before buying a computer case. And that's pretty hard to do. You can't always inspect this issue in advance. I used to look at computer cases at a mom&pop (single store), while they were in business. Basically, any store in town that put computer cases on display in a useful way, went out of business. And I did buy cases from them, so it's not like the effort on their part wasn't rewarded. For example, I don't buy "bump" style trays. That's where the metal of the tray is formed into a bump of the appropriate height, then a hole drilled, then the hole is threaded. While it saves the manufacturer $0.0001 per hole, it doesn't do me any good. Instead, I like the brass standoffs that have a male thread on one end and a female on the other, then the motherboard screw fits to fasten the board down. These have a better controlled contact area where it touches the ground ring on the motherboard. In terms of "places that short", there was one Asus motherboard where the plastic socket frame around the CPU socket, was fitted *before* the soldering step. The plastic went through a wave solder perhaps, and it swept up some solder underneath the plastic (hard to inspect) and it shorted out VCore. But that would only cause VCore with a current limiter to trip off. And a little repair work, you could find and remove the short. But for direct rail shorts, sometimes there can be a short inside the motherboard, but it's less likely there is a short, say, right at the 24 pin connector. Some Biostar motherboards, used to join the yellow 12V wires on the 2x2 ATX12V, to the 12V wires on the main 24 pin connector. This is not a good idea, and I have no idea why Biostar was doing that. The 12VA and 12VB should be separate from one another, in case some old PSU with actual separate transformers for those two 12V was in usage. The 12V on the ATX12V 2x2, could be a different 12V than the 12V on the main connector (fraction of a volt difference). Since you say you are observing a short with the ATX12V removed, then this suggests something on the rails on the main 24 pin is shorted. But I don't see a cosmetic root cause, because the transfer of some of those rails around the motherboard, is on a buried layer. The motherboards have 4,6, or 8 layers, with 6 layers used for RAMBUS boards, 4 layers for a lot of regular DDR motherboards, and 8 layers reserved for sockets with huge pincounts, where it's just too hard to route all the signals using a limited number of layers. Somewhere in that stack, can be some 2oz layers for routing power. At work, we would use a bed-of-nails tester for evaluating "ohms". After a board came off the line, sometimes tests were done that way too (JTAG), which isn't all that common in the computer industry. Those are known as structural tests. A company like Asus, would use a camera and automation to check for visible flaws (stuff in places it doesn't belong), can use an Xray machine for BGA verification, but for electrical test, there is only a 2 minute hand functional test (lady plugs boards into mobo and runs some test code for two minutes total time, then product goes in box). They rely on statistical sampling for quality, rather than every board being inspected with a magnifying glass. Every board receives the 2 minute functional test, but other steps are likely automated. To visually inspect a square foot of PCB, takes about 2 hours with a 10x magnifier, because I've done a few boards visually, and you do find stuff. (That's where I found a certain brand of resistors, with all the bodies cracked in half.) At one time, Asus made 5 million motherboards per month, and there's no time for that sort of inspection, and they can only sample motherboards and inspect them for manufacturing defects, then go back to the line and correct them. Not every motherboard can be treated like a Rolls Royce. Thanks Paul From Google: Thread: What size/kind of screws do I use for mounting motherboard? Screw size depends on the size of the threads in the mobo standoffs, usually M3x0.5 (metric) or #6-32 (US standard). I personally would not go over 6mm or .25in in length. Head configuration is entirely up to you. Most well-stocked hardware stores should have just about any size and type you want. |
#25
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini OnboardCase Buzzer Board Beep Alarm NEW."
Norm Why wrote:
[snippage] I am now pondering my silly collection of MOBO mounting screws. https://www.aliexpress.com/item/32957035974.html Shows nylon screws that can be used for mounting or standoffs. Standoffs don't work for me because I need PCI VGA card. Size: M2 (Thread 6mm) M2.5 (Thread 6mm) M3 (Thread 6mm) M4 (Thread 6mm) I guess 'size' is pitch, because all are 6mm thread. What size would work? Length: 5mm (50pcs) 6mm (50pcs) 7mm (50pcs) 8mm (50pcs) 9mm (50pcs) 10mm (50pcs) 11mm (50pcs) 12mm (50pcs) 13mm (50pcs) 14mm (50pcs) 15mm (50pcs) 16mm (50pcs) 17mm (50pcs) 18mm (50pcs) 19mm (50pcs) 20mm (50pcs) 21mm (50pcs) 22mm (50pcs) 23mm (50pcs) 24mm (50pcs) 25mm (50pcs) 26mm (50pcs) 28mm (50pcs) 30mm (50pcs) 32mm (50pcs) 35mm (50pcs) 40mm (50pcs) 45mm (50pcs) 50mm (50pcs) What Length would work? The trick is not to lose the hardware that came with the computer case. The standoffs are custom length, intended to elevate the PCB to be flush with the slot holes for the I/O cards, in the back. There's no way for me to guess any such measurement from here. There are as many styles of solutions, as there are days of the week. If you artificially elevate the motherboard, with too-tall standoffs, then your video card won't fit. It's one of the items on the buyers list, to check the fit and dress of standoffs before buying a computer case. And that's pretty hard to do. You can't always inspect this issue in advance. I used to look at computer cases at a mom&pop (single store), while they were in business. Basically, any store in town that put computer cases on display in a useful way, went out of business. And I did buy cases from them, so it's not like the effort on their part wasn't rewarded. For example, I don't buy "bump" style trays. That's where the metal of the tray is formed into a bump of the appropriate height, then a hole drilled, then the hole is threaded. While it saves the manufacturer $0.0001 per hole, it doesn't do me any good. Instead, I like the brass standoffs that have a male thread on one end and a female on the other, then the motherboard screw fits to fasten the board down. These have a better controlled contact area where it touches the ground ring on the motherboard. In terms of "places that short", there was one Asus motherboard where the plastic socket frame around the CPU socket, was fitted *before* the soldering step. The plastic went through a wave solder perhaps, and it swept up some solder underneath the plastic (hard to inspect) and it shorted out VCore. But that would only cause VCore with a current limiter to trip off. And a little repair work, you could find and remove the short. But for direct rail shorts, sometimes there can be a short inside the motherboard, but it's less likely there is a short, say, right at the 24 pin connector. Some Biostar motherboards, used to join the yellow 12V wires on the 2x2 ATX12V, to the 12V wires on the main 24 pin connector. This is not a good idea, and I have no idea why Biostar was doing that. The 12VA and 12VB should be separate from one another, in case some old PSU with actual separate transformers for those two 12V was in usage. The 12V on the ATX12V 2x2, could be a different 12V than the 12V on the main connector (fraction of a volt difference). Since you say you are observing a short with the ATX12V removed, then this suggests something on the rails on the main 24 pin is shorted. But I don't see a cosmetic root cause, because the transfer of some of those rails around the motherboard, is on a buried layer. The motherboards have 4,6, or 8 layers, with 6 layers used for RAMBUS boards, 4 layers for a lot of regular DDR motherboards, and 8 layers reserved for sockets with huge pincounts, where it's just too hard to route all the signals using a limited number of layers. Somewhere in that stack, can be some 2oz layers for routing power. At work, we would use a bed-of-nails tester for evaluating "ohms". After a board came off the line, sometimes tests were done that way too (JTAG), which isn't all that common in the computer industry. Those are known as structural tests. A company like Asus, would use a camera and automation to check for visible flaws (stuff in places it doesn't belong), can use an Xray machine for BGA verification, but for electrical test, there is only a 2 minute hand functional test (lady plugs boards into mobo and runs some test code for two minutes total time, then product goes in box). They rely on statistical sampling for quality, rather than every board being inspected with a magnifying glass. Every board receives the 2 minute functional test, but other steps are likely automated. To visually inspect a square foot of PCB, takes about 2 hours with a 10x magnifier, because I've done a few boards visually, and you do find stuff. (That's where I found a certain brand of resistors, with all the bodies cracked in half.) At one time, Asus made 5 million motherboards per month, and there's no time for that sort of inspection, and they can only sample motherboards and inspect them for manufacturing defects, then go back to the line and correct them. Not every motherboard can be treated like a Rolls Royce. Thanks Paul From Google: Thread: What size/kind of screws do I use for mounting motherboard? Screw size depends on the size of the threads in the mobo standoffs, usually M3x0.5 (metric) or #6-32 (US standard). I personally would not go over 6mm or .25in in length. Head configuration is entirely up to you. Most well-stocked hardware stores should have just about any size and type you want. Yes, that's the screw. But the length of the standoff and what joins it to the motherboard, is up to the individual computer case. The standoffs can be discrete brass items, male threaded on one and and female threaded on top. And they go into a flat tray metalwork which has threaded holes in at least 9 standard places. Usually the tray has holes drilled and threaded for several PCB standards. But some trays dispense with the brass standoffs. Instead of the metal tray being perfectly flat, it's formed so that there are nine "bumps" in the metal, raised a certain height above the plane of the rest of the metal. The bump has a threaded hole in the top. This means they don't have to place nine finely crafted brass standoffs in the box-of-hardware. And I find the bump concept a bit worse, as the bump might be less precise where it touches the motherboard (larger footprint, in danger of exceeding the keep-out area around the PTH plated through hole). The height of the motherboard above the tray is "custom" to a computer case design. This is why the brass standoffs cannot be swapped from an Antec to a Compucase. You can see that in the collection this guy has on hand, to take this picture. "Various types of motherboard standoffs" https://en.wikipedia.org/wiki/Comput...MGP5029_wp.jpg ( https://en.wikipedia.org/wiki/Computer_case_screws ) To me, in that picture, I think I'm seeing some different diameter holes on the female ends of the depicted standoffs. Compare the center three brass items, to see how much difference there can be in the screw choices. The head of the motherboard fastener screw has to be small enough, to stay within the keepout zone maintained by the motherboard designer (the PCB has a multitude of keepouts, like a square-ish pattern around the CPU socket for the items placed on top of the CPU). That's why my advice is: "Don't lose the hardware that comes with your case" If you still have samples of some of the items, you can measure them and find substitute materials. But a brass standoff of the correct dimensions, would be like "hens teeth", not easy to find. Paul |
#26
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
[snippage]
What Length would work? The trick is not to lose the hardware that came with the computer case. The standoffs are custom length, intended to elevate the PCB to be flush with the slot holes for the I/O cards, in the back. There's no way for me to guess any such measurement from here. There are as many styles of solutions, as there are days of the week. If you artificially elevate the motherboard, with too-tall standoffs, then your video card won't fit. It's one of the items on the buyers list, to check the fit and dress of standoffs before buying a computer case. And that's pretty hard to do. You can't always inspect this issue in advance. I used to look at computer cases at a mom&pop (single store), while they were in business. Basically, any store in town that put computer cases on display in a useful way, went out of business. And I did buy cases from them, so it's not like the effort on their part wasn't rewarded. For example, I don't buy "bump" style trays. That's where the metal of the tray is formed into a bump of the appropriate height, then a hole drilled, then the hole is threaded. While it saves the manufacturer $0.0001 per hole, it doesn't do me any good. Instead, I like the brass standoffs that have a male thread on one end and a female on the other, then the motherboard screw fits to fasten the board down. These have a better controlled contact area where it touches the ground ring on the motherboard. In terms of "places that short", there was one Asus motherboard where the plastic socket frame around the CPU socket, was fitted *before* the soldering step. The plastic went through a wave solder perhaps, and it swept up some solder underneath the plastic (hard to inspect) and it shorted out VCore. But that would only cause VCore with a current limiter to trip off. And a little repair work, you could find and remove the short. But for direct rail shorts, sometimes there can be a short inside the motherboard, but it's less likely there is a short, say, right at the 24 pin connector. Some Biostar motherboards, used to join the yellow 12V wires on the 2x2 ATX12V, to the 12V wires on the main 24 pin connector. This is not a good idea, and I have no idea why Biostar was doing that. The 12VA and 12VB should be separate from one another, in case some old PSU with actual separate transformers for those two 12V was in usage. The 12V on the ATX12V 2x2, could be a different 12V than the 12V on the main connector (fraction of a volt difference). Since you say you are observing a short with the ATX12V removed, then this suggests something on the rails on the main 24 pin is shorted. But I don't see a cosmetic root cause, because the transfer of some of those rails around the motherboard, is on a buried layer. The motherboards have 4,6, or 8 layers, with 6 layers used for RAMBUS boards, 4 layers for a lot of regular DDR motherboards, and 8 layers reserved for sockets with huge pincounts, where it's just too hard to route all the signals using a limited number of layers. Somewhere in that stack, can be some 2oz layers for routing power. At work, we would use a bed-of-nails tester for evaluating "ohms". After a board came off the line, sometimes tests were done that way too (JTAG), which isn't all that common in the computer industry. Those are known as structural tests. A company like Asus, would use a camera and automation to check for visible flaws (stuff in places it doesn't belong), can use an Xray machine for BGA verification, but for electrical test, there is only a 2 minute hand functional test (lady plugs boards into mobo and runs some test code for two minutes total time, then product goes in box). They rely on statistical sampling for quality, rather than every board being inspected with a magnifying glass. Every board receives the 2 minute functional test, but other steps are likely automated. To visually inspect a square foot of PCB, takes about 2 hours with a 10x magnifier, because I've done a few boards visually, and you do find stuff. (That's where I found a certain brand of resistors, with all the bodies cracked in half.) At one time, Asus made 5 million motherboards per month, and there's no time for that sort of inspection, and they can only sample motherboards and inspect them for manufacturing defects, then go back to the line and correct them. Not every motherboard can be treated like a Rolls Royce. Thanks Paul From Google: Thread: What size/kind of screws do I use for mounting motherboard? Screw size depends on the size of the threads in the mobo standoffs, usually M3x0.5 (metric) or #6-32 (US standard). I personally would not go over 6mm or .25in in length. Head configuration is entirely up to you. Most well-stocked hardware stores should have just about any size and type you want. Yes, that's the screw. But the length of the standoff and what joins it to the motherboard, is up to the individual computer case. The standoffs can be discrete brass items, male threaded on one and and female threaded on top. And they go into a flat tray metalwork which has threaded holes in at least 9 standard places. Usually the tray has holes drilled and threaded for several PCB standards. But some trays dispense with the brass standoffs. Instead of the metal tray being perfectly flat, it's formed so that there are nine "bumps" in the metal, raised a certain height above the plane of the rest of the metal. The bump has a threaded hole in the top. This means they don't have to place nine finely crafted brass standoffs in the box-of-hardware. And I find the bump concept a bit worse, as the bump might be less precise where it touches the motherboard (larger footprint, in danger of exceeding the keep-out area around the PTH plated through hole). The height of the motherboard above the tray is "custom" to a computer case design. This is why the brass standoffs cannot be swapped from an Antec to a Compucase. You can see that in the collection this guy has on hand, to take this picture. "Various types of motherboard standoffs" https://en.wikipedia.org/wiki/Comput...MGP5029_wp.jpg ( https://en.wikipedia.org/wiki/Computer_case_screws ) To me, in that picture, I think I'm seeing some different diameter holes on the female ends of the depicted standoffs. Compare the center three brass items, to see how much difference there can be in the screw choices. The head of the motherboard fastener screw has to be small enough, to stay within the keepout zone maintained by the motherboard designer (the PCB has a multitude of keepouts, like a square-ish pattern around the CPU socket for the items placed on top of the CPU). That's why my advice is: "Don't lose the hardware that comes with your case" If you still have samples of some of the items, you can measure them and find substitute materials. But a brass standoff of the correct dimensions, would be like "hens teeth", not easy to find. Paul Thanks Paul, As an aside. nylon MOBO screws might not be a bad idea. But for it to work, I would be necessary to replace metal standoffs beneath the main board. One the PCI VGA card is in place, there is direct connection to the frame. |
#27
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
"Don't lose the hardware that comes with your case"
If you still have samples of some of the items, you can measure them and find substitute materials. But a brass standoff of the correct dimensions, would be like "hens teeth", not easy to find. Paul Thanks Paul, As an aside. nylon MOBO screws might not be a bad idea. But for it to work, I would be necessary to replace metal standoffs beneath the main board. One the PCI VGA card is in place, there is direct connection to the frame. At this point, I have removed al but one MOBO screw (yikes), the 4pin 12V connector and all PCI cards. I still get a 'short to ground' buzz on my multimter. I shorted the CLEAR_CMOS. This is the worse it's ever been. I think I am being informed of a ground loop that might generate hum in audio. The ATX power cable pinout can be found he https://en.wikipedia.org/wiki/ATX The lines marked in blue are sense lines from the MOBO back to the PSU. Imagine we had a male and female socket pair. Then we would be able to meter the sense lines. We would be able to cut some sense lines that were causing trouble. How dangerous and absurd is this idea? If it is not absurd then someone may have thought of it first and there might be a product on the market? |
#28
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini OnboardCase Buzzer Board Beep Alarm NEW."
Norm Why wrote:
"Don't lose the hardware that comes with your case" If you still have samples of some of the items, you can measure them and find substitute materials. But a brass standoff of the correct dimensions, would be like "hens teeth", not easy to find. Paul Thanks Paul, As an aside. nylon MOBO screws might not be a bad idea. But for it to work, I would be necessary to replace metal standoffs beneath the main board. One the PCI VGA card is in place, there is direct connection to the frame. At this point, I have removed al but one MOBO screw (yikes), the 4pin 12V connector and all PCI cards. I still get a 'short to ground' buzz on my multimter. I shorted the CLEAR_CMOS. This is the worse it's ever been. I think I am being informed of a ground loop that might generate hum in audio. The ATX power cable pinout can be found he https://en.wikipedia.org/wiki/ATX The lines marked in blue are sense lines from the MOBO back to the PSU. Imagine we had a male and female socket pair. Then we would be able to meter the sense lines. We would be able to cut some sense lines that were causing trouble. How dangerous and absurd is this idea? If it is not absurd then someone may have thought of it first and there might be a product on the market? There is one pin with "remote sense". It is the 3.3V pin. The remote sense is present, because of the relatively low voltage of that pin. To do a good job of regulating really low voltages, you use remote sense wires for the job. It's better than sampling the voltage using a circuit inside the PSU, as remote sensing measures the voltage as seen at the load. A remote sense is typically a higher impedance feedback line. It's not a power conductor and may use a slightly thinner gauge of wire. This isn't an ATX supply, and this just demonstrates the impedances involved. Maybe some bench supplies for the lab are built this way. FB+ ------- 1K ohms ------+ | VCC -----------------------+-X Load GND -----------------------+-X Load | FB- ------- 1K ohms ------+ The FB (Feedback) wires can't short anything, because they have a series resistance. The series resistance is small, compared to the input impedance of the amp that picks up the diff between FB+ and FB-. You need to take the difference, in case the supply is "elevated" or "floated" with respect to the load. The message here, is the FB isn't hurting anything, can't hurt anything, and *does not* need to be cut. Also, if you break a feedback loop, by cutting a feedback wire, the circuit will go "open loop" and potentially rail, blowing up the load with an excessive output voltage. (Um, don't to that, OK?) When you're making these prognostications of yours, remember that "capacitors", especially the large caps used to provide bulk decoupling, they appear as a dead short to the beeper on your meter. This is why ohming out an item like that (a mobo rail) can give a deceiving result. This is why I mentioned my clamp-on ammeter, which makes current measurements without disturbing any conductors. It works a treat on the older systems where the wires have all the right colors, and it's visually easier to grab all the wires-of-same-color and put them in the jaws of the clamp-on DC ammeter. Then, you get a total current reading, and you can use your knowledge of motherboard design era, to figure out whether the current value measured is reasonable or not. My clamp-on ammeter also has a "peak detect", and can capture the top current flow value over a period of time. When my previous car had a starter problem, I wanted to measure the current, and by setting the clamp-on meter to "peak detect", I was able to determine it drew as much as 150 amps from the 12V battery. Using a second regular-multimeter with trough detection, I got a snapshot of 9V as the lowest value while the battery tries to turn the car over and start it. Using these values, you get some idea of the impedances involved (3V drop/150 amps, in ohms, or about 20 milliohms). My clamp-on ammeter can measure up to 400 amps, and for motherboard work, I set it to 40 amps full scale. An Athlon motherboard can draw up to 25 amps from +5V, as an example of a pretty high current reading from that era. Modern motherboards aren't even close to that. Since the power supplies now are sleeved and the wires are all black in color, this has put a big damper on my measurement fun. I must use my 24 pin to 24 pin extension cable (which has loose wires), to gain access to the wires to make a current flow measurement. Then I don't have to cut the sleeving to get at individual wires. By measuring current flows, i can avoid being deceived by the multimeter "buzzer" trying to charge up capacitors. This could inadvertently show something as shorted, which is not shorted. Instead, I measure current flows to get some idea whether a load is abnormally high. Because I don't cut any conductors to insert a measurement shunt, and the meter uses magnetic fields for the measurement, it's possible to quickly review the loading (one rail at a time) as a diagnostic procedure. The clamp-on DC ammeter cost me around $300, years ago, and I don't know how much those have dropped in price since. My regular-multimeter was $100, and it has an RS232 output for recording readings on a PC. The RS232 output is optoisolated, so voltages you're working on, cannot jump the RS232 interface and blow up your computer :-) I think that's pretty neat, and is one of the reasons I spent $100 on a multimeter, rather than about $20 for a HarborFreight one. I don't have a lot of test equipment, and that's the best of what I've got. I don't own a scope. Paul |
#29
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
As an aside. nylon MOBO screws might not be a bad idea. But for it to work, I would be necessary to replace metal standoffs beneath the main board. One the PCI VGA card is in place, there is direct connection to the frame. At this point, I have removed al but one MOBO screw (yikes), the 4pin 12V connector and all PCI cards. I still get a 'short to ground' buzz on my multimter. I shorted the CLEAR_CMOS. This is the worse it's ever been. I think I am being informed of a ground loop that might generate hum in audio. The ATX power cable pinout can be found he https://en.wikipedia.org/wiki/ATX The lines marked in blue are sense lines from the MOBO back to the PSU. Imagine we had a male and female socket pair. Then we would be able to meter the sense lines. We would be able to cut some sense lines that were causing trouble. How dangerous and absurd is this idea? If it is not absurd then someone may have thought of it first and there might be a product on the market? There is one pin with "remote sense". It is the 3.3V pin. The remote sense is present, because of the relatively low voltage of that pin. To do a good job of regulating really low voltages, you use remote sense wires for the job. It's better than sampling the voltage using a circuit inside the PSU, as remote sensing measures the voltage as seen at the load. A remote sense is typically a higher impedance feedback line. It's not a power conductor and may use a slightly thinner gauge of wire. This isn't an ATX supply, and this just demonstrates the impedances involved. Maybe some bench supplies for the lab are built this way. FB+ ------- 1K ohms ------+ | VCC -----------------------+-X Load GND -----------------------+-X Load | FB- ------- 1K ohms ------+ The FB (Feedback) wires can't short anything, because they have a series resistance. The series resistance is small, compared to the input impedance of the amp that picks up the diff between FB+ and FB-. You need to take the difference, in case the supply is "elevated" or "floated" with respect to the load. The message here, is the FB isn't hurting anything, can't hurt anything, and *does not* need to be cut. Also, if you break a feedback loop, by cutting a feedback wire, the circuit will go "open loop" and potentially rail, blowing up the load with an excessive output voltage. (Um, don't to that, OK?) When you're making these prognostications of yours, remember that "capacitors", especially the large caps used to provide bulk decoupling, they appear as a dead short to the beeper on your meter. This is why ohming out an item like that (a mobo rail) can give a deceiving result. This is why I mentioned my clamp-on ammeter, which makes current measurements without disturbing any conductors. It works a treat on the older systems where the wires have all the right colors, and it's visually easier to grab all the wires-of-same-color and put them in the jaws of the clamp-on DC ammeter. Then, you get a total current reading, and you can use your knowledge of motherboard design era, to figure out whether the current value measured is reasonable or not. My clamp-on ammeter also has a "peak detect", and can capture the top current flow value over a period of time. When my previous car had a starter problem, I wanted to measure the current, and by setting the clamp-on meter to "peak detect", I was able to determine it drew as much as 150 amps from the 12V battery. Using a second regular-multimeter with trough detection, I got a snapshot of 9V as the lowest value while the battery tries to turn the car over and start it. Using these values, you get some idea of the impedances involved (3V drop/150 amps, in ohms, or about 20 milliohms). My clamp-on ammeter can measure up to 400 amps, and for motherboard work, I set it to 40 amps full scale. An Athlon motherboard can draw up to 25 amps from +5V, as an example of a pretty high current reading from that era. Modern motherboards aren't even close to that. Since the power supplies now are sleeved and the wires are all black in color, this has put a big damper on my measurement fun. I must use my 24 pin to 24 pin extension cable (which has loose wires), to gain access to the wires to make a current flow measurement. Then I don't have to cut the sleeving to get at individual wires. By measuring current flows, i can avoid being deceived by the multimeter "buzzer" trying to charge up capacitors. This could inadvertently show something as shorted, which is not shorted. Instead, I measure current flows to get some idea whether a load is abnormally high. Because I don't cut any conductors to insert a measurement shunt, and the meter uses magnetic fields for the measurement, it's possible to quickly review the loading (one rail at a time) as a diagnostic procedure. The clamp-on DC ammeter cost me around $300, years ago, and I don't know how much those have dropped in price since. My regular-multimeter was $100, and it has an RS232 output for recording readings on a PC. The RS232 output is optoisolated, so voltages you're working on, cannot jump the RS232 interface and blow up your computer :-) I think that's pretty neat, and is one of the reasons I spent $100 on a multimeter, rather than about $20 for a HarborFreight one. I don't have a lot of test equipment, and that's the best of what I've got. I don't own a scope. Paul Thanks for the advice. I have re-installed all metal MOBO screws with labor. All my work with a Multimeter buzz for short, was useless. I shook assembly and one screw came out. It is amazing what's to be found when one's nose is face down in rug. I removed, tested and re-installed the 3.3V CMOS battery. I have re-installed all useful connectors, even the speaker beep connector. Computer stills shuts down immediately - no beep codes, now. My problem is worse than when I began. If I am not to blame, maybe there is blown capacitor? I thought Gigabyte used plastic film caps. Maybe that came later? I have looked at the caps and none are exploded. However when I search the board, I see a cap with + nearby suggesting it is electrolytic. I have intermittence problems with xBox 360 and Samsung wide screen with my be due to keep caps. I am unable to find gibabyte circuit diagram. Because they are numerous, maybe I should replace all with ceramic, plastic film or tantalum? WWW says shelf life of electrolytic capacitor is two years. After 11 years storage they are discharged. Maybe if I could use external 5V supply, the caps could be repolarized. Where do test resistance so I don't start a fire? |
#30
|
|||
|
|||
"Speaker For PC Interanal BIOS Computer Motherboard Mini Onboard Case Buzzer Board Beep Alarm NEW."
Norm Why wrote:
Thanks for the advice. I have re-installed all metal MOBO screws with labor. All my work with a Multimeter buzz for short, was useless. I shook assembly and one screw came out. It is amazing what's to be found when one's nose is face down in rug. I removed, tested and re-installed the 3.3V CMOS battery. I have re-installed all useful connectors, even the speaker beep connector. Computer stills shuts down immediately - no beep codes, now. My problem is worse than when I began. If I am not to blame, maybe there is blown capacitor? I thought Gigabyte used plastic film caps. Maybe that came later? I have looked at the caps and none are exploded. However when I search the board, I see a cap with + nearby suggesting it is electrolytic. I have intermittence problems with xBox 360 and Samsung wide screen with my be due to keep caps. I am unable to find gibabyte circuit diagram. Because they are numerous, maybe I should replace all with ceramic, plastic film or tantalum? WWW says shelf life of electrolytic capacitor is two years. After 11 years storage they are discharged. Maybe if I could use external 5V supply, the caps could be repolarized. Where do test resistance so I don't start a fire? The service life of electrolytic capacitors is 17 years (from the webpage of Illinois Capacitor). At the 17 year point, the rubber seal in the bottom can dry out. However, I have a motherboard which is close to 20 years old, and it still works. Electrolytic capacitors with an internal pH problem, the liquid inside will eat through the casing of the capacitor in as little as two years. This is a defective product, plain and simple. The electrolytic formula was a "stolen" chemical formula, where the buffering agent was missed. Some materials were supposed to be added to stabilize the liquid and make it compatible with the materials around it. Electrolytic reliability is proportional to temperature. At 105C or greater temperatures, the life of some caps is listed as short as 2000 hours. But modern designs, and the recommended computer air temp of no higher than 35C to 45C or so, and the exponential in the reliability math equation, ensures the life is many times 2000 hours. The reliability formula is a curve-fitted Arrhenius equation. Normal liquid chemistry reactions, the Arrhenius rate doubles for every 7C. On capacitors, the curve fit to the reliability data, has a doubling of rates every 15C or so. So while in chemistry class, we're taught there is just one coefficient for that exponential, electrical engineering shows that many items follow the general form of the relationship, except the coefficient needs to be curve fitted to the data. Between 45C and 105C would be 60C or four doublings or a factor of 16. So instead of 2000 hours, we'd get 32000 hours. And the only time a PC is 45C, is in summer, with the AC turned off. The time is even larger than that, because of the chilled temperatures in winter months. When working with plane-to-plane shorts inside PCBs, those can be hard to find. You would, at the very least, remove all components and carry out a visual inspection, as a starting point. And a bench magnifier, to make things a bit bigger, helps at a time like that. In two places I worked, we had a binocular microscope (maybe $2000 to $3000 or so), which was a great aid for stuff like that. You don't need too much magnification. It's not a job for "oil immersion" or the like. (I think the strongest microscope I've used, was 1200X or so, and used oil immersion to some cover slips for biology work. The objective was spring loaded because... students. You can easily crush the glass while adjusting one of those.) PCB inspection is fine at around 10X. If the flaw is inside the PCB, you won't spot that visually unless the board heats up and a burn mark appears on the surface. The worst computer motherboard failure I've ever seen, a ring was "charred" all the way around the CPU socket. That's a VCore-related failure, where the current limiter didn't work properly. Paul |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
I need your advice - new build: fans and lights, "no signal", nopost beep | ToolPackinMama | Homebuilt PC's | 44 | November 28th 08 11:04 AM |
Searching Desktop motherboard for "Core Duo" CPU ? Only for Notebooks ? Diff to "Core 2 Duo" ? | Paolo Severin | Asus Motherboards | 0 | December 27th 06 07:47 AM |
Alarm sounds from the computer..." | ZR | Abit Motherboards | 1 | August 28th 06 08:59 PM |
Packard Bell PB420 motherboard - BIOS error "cannot read/write to CMOS" | [email protected] | Packard Bell Computers | 5 | June 28th 06 08:16 PM |
"Alarm Siren" sound when plugging a Hard disk (!!!?) | [email protected] | Storage (alternative) | 2 | January 12th 06 09:32 AM |