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 |
#31
|
|||
|
|||
As noted previously, motherboards would be manufactured for
many other configurations. But Intel 29835002.pdf recommends only making a motherboard logic ground connection at far end, and adjacent to other devices that also make a chassis ground connection - IO peripheral cards. Intel design guidelines recommend 'single point ground' mounting via two mounting screws adjacent to IO card slots. Intel guidelines also demonstrate that other mounting holes not make an electrical chassis connection. Furthermore, note extra grounding layers that Intel recommends in the single point region for reducing ESD (chassis) noise into logic ground. Motherboard is best mounted using a single point ground connection. It was always good pc board practice to line every mounting hole with copper feedthroughs. This does not mean all mounting holes need or should make a connection to other grounds. Those plated feedthroughs mechanically strengthen each mounting hole. But Intel only recommends that two holes make an electrically conductive connection to chassis. Reasons why have been posted previously. JAD wrote: um if this is the case, what are MB manufactures trying to do? By putting ground rings around each mount hole(on both sides in some cases), they are in fact saying this point needs to be grounded. If they did not want a ground point there, the ring should be non conductive. Makes little sense and To point out 1 single scenario on one particular board, and if it was DM stating it, knowing him he probably stayed up all night finding the one 1 single situation that supports whatever his point is. |
#32
|
|||
|
|||
w_tom wrote:
David Maynard provides an Intel example of the single point ground technique. Intel recommends two screws at the far end It is not at the 'far end'. They are as near the potential ESD entry poin= t=20 as possible but HALF way across the motherboard, over 6 inches, from each= =20 other. of a motherboard so that, again, electrical currents in the chassis do not pass through motherboard logic ground. The single point ground, IO board connections, and power input connect to chassis at the same single point - at the far end of a motherboard. Where I recommended only one screw connection, Intel recommends two in the same location to create the single point ground at the farthest edge of motherboard. "Two" is not a "single point." Demonstrated by page 30 of that Intel 815E Chipset Platform document, no currents across electronic sections Correct, because they are shunted to chassis *there* instead of seeking=20 your 'single point ground'. The motherboard ground also transverses the PSU power cable to the PSU's = chassis ground; which makes for THREE chassis ground points in just this = limited example. And then, each AGP/PCI card with I/O will have it's chassis end plate=20 connected to ground. And then there's the drives which have logic to=20 chassis ground connections. There's logic/chassis ground connections all = OVER the place. of motherboard means no computer crashes due to static electric discharges to chassis. Best not to mount motherboard with conductive connections everywhere. Intel recommends only two connection, both at the far edge to make same single point ground - adjacent to IO slots and power connector. Intel document does not recommend conductive standoffs everywhere on motherboard - as I had posted. The issue was not about a connection 'everywhere'. The issue was your=20 patently false claim that only ONE is the only suitable means. =20 David Maynard wrote: =20 ... Look up Intel 29835002.pdf, Intel=AE 815E Chipset Platform Design Guide= , section 2.2 Electrostatic Discharge Platform Recommendations where it specifies connecting *two* motherboard mounting holes to ground, one on each side of the rear I/O area, so that any ESD there is shunted to case in the shortest possible path. ... |
#33
|
|||
|
|||
Actually one screw to chassis is a single point just as two
screws were a single point. The single point was the entire region that included IO slots, single screw to chassis, and power supply connection. This single point (or region) is located at the far end of PC board to that no transient current passes through sensitive electronics regions. Again I emphasis an Intel recommendation that the single point area have additional grounding layers. Just more reason for any ESD currents to remain in the single point region and not taken scenic tours through the sensitive digital electronic areas. I noted an Intel suggestion of more grounding at the chassis connection area. I did not bother to explain why while wondering if you understood the significance of that additional motherboard grounding. I guess not. Intel demonstrates single point grounding. This so that transient currents in chassis do not pass across logic ground plane where transients could cause computer crashes. Computer assemblers are advised to heed Intel recommendations. Do not connect motherboard to chassis at every mounting point - to increase computer reliability. A technique long practiced by experienced and learned engineers. A practice that many computer assemblers still deny - even after an Intel *pdf document demonstrates the concept. Again it is a so simple concept. Connect logic ground to chassis ground so that transient current on chassis will not pass through motherboard. It is called a single point ground. It is even demonstrated by the cited Intel document that YOU provided. I must assume you did not comprehend what Intel wrote. For some, the complexities of grounding were previously unknown. Grounding is but another reason why ESD solutions can appear complex. BTW I did not say "that only ONE is the only suitable means". Read back. I posted other means of obtaining the single point ground. Another was to connect every mounting hole the a chassis plate, and then have only one electrical connection from chassis plate to chassis. Why did you forget this and other single point grounding solutions? Is being wrong (again) so hard to admit? We keep doing this. I provide facts and electrical concepts. You keep denying it; all without demonstrating knowledge of engineering principles. This time it is more interesting. You provided the Intel document that demonstrates what I had posted. Interesting new way to deny basic electrical principles. Deny them and then cite an Intel document that disagrees with your denial. Yes, I am amused. David Maynard wrote: It is not at the 'far end'. They are as near the potential ESD entry point as possible but HALF way across the motherboard, over 6 inches, from each other. of a motherboard so that, again, electrical currents in the chassis do not pass through motherboard logic ground. The single point ground, IO board connections, and power input connect to chassis at the same single point - at the far end of a motherboard. Where I recommended only one screw connection, Intel recommends two in the same location to create the single point ground at the farthest edge of motherboard. "Two" is not a "single point." Demonstrated by page 30 of that Intel 815E Chipset Platform document, no currents across electronic sections Correct, because they are shunted to chassis *there* instead of seeking your 'single point ground'. The motherboard ground also transverses the PSU power cable to the PSU's chassis ground; which makes for THREE chassis ground points in just this limited example. And then, each AGP/PCI card with I/O will have it's chassis end plate connected to ground. And then there's the drives which have logic to chassis ground connections. There's logic/chassis ground connections all OVER the place. of motherboard means no computer crashes due to static electric discharges to chassis. Best not to mount motherboard with conductive connections everywhere. Intel recommends only two connection, both at the far edge to make same single point ground - adjacent to IO slots and power connector. Intel document does not recommend conductive standoffs everywhere on motherboard - as I had posted. The issue was not about a connection 'everywhere'. The issue was your patently false claim that only ONE is the only suitable means. David Maynard wrote: ... Look up Intel 29835002.pdf, Intel® 815E Chipset Platform Design Guide, section 2.2 Electrostatic Discharge Platform Recommendations where it specifies connecting *two* motherboard mounting holes to ground, one on each side of the rear I/O area, so that any ESD there is shunted to case in the shortest possible path. ... |
#34
|
|||
|
|||
w_tom wrote:
He cannot challenge the science, so his reasoning goes something like this. The manufacturer does this and says nothing. Therefore the manufacturer expected you to do that. Scientific proof? Of course not. That leaves out the circumstantial evidence (see my prior post) which IMO is very convincing. snipped nonsense Path: newssvr11.news.prodigy.com!newsswing.news.prodigy. com!prodigy.net!newsmst01a.news.prodigy.com!prodig y.com!border1.nntp.dca.giganews.com!nntp.giganews. com!local1.nntp.dca.giganews.com!nntp.nni.com!news .nni.com.POSTED!not-for-mail NNTP-Posting-Date: Fri, 26 Nov 2004 13:17:48 -0600 Message-ID: Date: Fri, 26 Nov 2004 14:17:45 -0500 From: w_tom X-Mailer: Mozilla 4.75 [en] (WinNT; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: alt.comp.hardware.pc-homebuilt Subject: Installing MoBo References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 54 NNTP-Posting-Host: 216.107.36.162 X-Trace: sv3-pBj2fsu/HSA7x2OeTviT5G0xkTB8G5Zle8VY1mre2vTl21Q4LzzvnlRNmq oWAhUMJazZNlYVg0G99uf!Wgqy8ptBNjcDiCOSeDK5rnMY25NY RP/Kjsal3gAYk7XcJloB9OR65HtDgv3g9y6j X-Complaints-To: X-DMCA-Complaints-To: X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly X-Postfilter: 1.3.20 Xref: newsmst01a.news.prodigy.com alt.comp.hardware.pc-homebuilt:423066 |
#35
|
|||
|
|||
On Thu, 25 Nov 2004 20:01:08 -0500, w_tom wrote:
Motherboards are designed for numerous configurations. Some configurations may want single point ground located elsewhere. Some configurations may attach the entire motherboard to chassis plate, and insulate chassis plate so that single point ground is between chassis plate and chassis. Furthermore, all mounting holes must be plated through. Computer assembler is expected to bring basic electrical knowledge to the assembly. He - not the motherboard manufacturer - decides where the single point ground will be made. Motherboard manufacture makes products with numerous options. Computer assembler decides how to implement those options. Cross talk and other problems are why computer boards are multi-layer assemblies using a copper plane for ground and other planes for Vcc. Same problems also why Intel (not the PC board designer) defines exact copper layout for PC traces to CPU. The layout is that critical. Electronic nature even of copper is but one reason why a CPU power supply is adjacent to the CPU and why newer ATX power supplies have a second power cable. Measure the conductivity between too ends of that motherboard ground plane. To meter, DC voltage is a perfect short. In reality, electric currents are RF electricity. No copper - even the ground plane - is a perfect conductor. All copper has significant impedance. Impedance (not resistance) is why Intel provides a layout for connections to their CPUs. Impedance is why bypass capacitors are scattered all over the motherboard. Conductors are electronic components when discussing RF currents to computers and in external generated transients. RF is why copper is not truly conductive. If copper were truly conductive, then antennas would not transmit and receive. Best antennas have near zero resistance and a specific impedance. The impedance (not resistance) makes multiple grounding of a motherboard problematic. Just another reason why grounds are at a single point. All conductors are electronic components. As Gary notes: ... at the frequencies modern computer systems operate at they have to use multiple grounds otherwise the systems become inoperable/unreliable because of cross talk. Problems are not just limited cross talk. Cross talk alone is not the reason for all those bypass capacitors. All PC traces and even the PC board ground plane are electronic components. Some currents (ie static electricity) through a ground plane can cause catastrophic consequences especially in hospital equipment. All grounds must be interconnected. So again, the interconnections between various grounds use the single point concept. Hospitals take the concept even one step farther. Hospitals do single point grounding on a room and on a building wide level. Bottom line for computer assemblers who learn new tricks - a single point connection between motherboard and chassis plate is best adjacent to IO slots and power connection. This post demonstrates another technical reason why: all conductors are electronic components; especially when dealing with RF frequencies in computers and in static electric discharges. Just another reason why more reliable computers uses the single point grounding technique. "We never did that before. Therefore we don't have to do that." People with this mentality need not reply. Unlike other posts that disagree, these posts technically explain why, provide examples, and provide experiments to demonstrate the problem. IOW these posts are based upon the science and not upon the myths so common among computer assemblers. These posts are for those who keep learning new techniques. Single point grounding of motherboard to chassis is the preferred mounting method. Provided here is but another reason why: RF currents make copper an electronic component. The relevant parameter is impedance. Here's a URL from *NASA* at the Marshall space flight center regarding the use of single point VS multiple point grounding and they show the frequency cutoff regions between single point and multiple point grounding. Scroll down to figure 4. http://klabs.org/DEI/References/desi...s/1214msfc.pdf With a 1 foot ground plane at frequencies up to 3 MHz they recommend single point grounding. With the same size ground plane up to 10 MHz they recommend hybrid grounds, At frequencies at and above 30 MHz they recommend multiple point grounds. I've got an amateur radio transceiver about the size of a bible that on transmit puts out 55 watts of power on our 2 meter VHF band ( 144 - 148 megahertz ) It'll also receive from 136 to 174 megahertz. The receive section is a double conversion super heterodyne with a first intermediate frequency of 15 megahertz and a 2nd intermediate frequency of 450 Kilohertz. It's synthesized with most likely a phase locked loop. With that amount of power and the frequency conversion sections It's no doubt got all kinds of RF potentially floating around inside. Recently it developed a popping sound through the speaker on transmit and some of the radios also had the popping sound on their transmitted audio. I did a search on one of the Yahoo groups concerning this particular radio and some of the them had left the factory without the grounding screws being sufficiently tightened. The cure was easy. According to the instructions you merely had to loosen the eight *grounding* screws on the circuit board and retighten them nice and snug so that they made a good ground contact and maybe even use some star washers for extra good grounding. I just loosened the eight screws, then tightened them nice and snug ( eight grounding points ) and the popping noise went away. Now this is a commercially constructed radio from one of ham radio's premier brands ( Icom ) and they had *eight* grounding points from the circuit board to chassis ground. I'm not going to get into a real lengthy debate as if you want to use just one ground point on your mother board feel free to do so, But when I see and use commercially built equipment that has to deal with all kinds of radio frequencies potentially floating around inside using multiple ground points and see soldered through holes on computer motherboards that have ground traces connected to them and brass stand offs that match the location of the grounded circles it seems painfully obvious that the motherboard manufacturer intended for those soldered circle to be connected to ground. Someone *has* to make a *conscious* effort to run a ground trace to those soldered circles. If they were just there for support of the motherboard they could easily just leave the ground trace off of it. You can have the last word as you seem steeped in the concept that single point grounding is the cure all concept covering every device that operates from direct current to the frequency of light. Single point grounding does have its place but when it comes to radio frequencies in the hundreds of megahertz + , you're dealing with an entirely different animal than direct current or low frequency RF. Gary Gary wrote: I read the article and it mainly deals with analog devices and/or A/D converters and low frequency PIC chips. While it is true that multiple grounds can create ground loops creating hum in audio or telephone circuits we're dealing with radio frequencies in the hundreds of megahertz for the system bus / memory and in the gigahertz region in the case of the processor. What the article didn't address and is more of a concern to motherboard manufacturers is cross talk. One example of cross talk is noise from one set of data / memory lines at 100/133/400 megahertz bleeding over onto another set of data / memory lines. In computer circuitry this is disastrous. What may be a ground at DC or in low level audio or radio circuits can be a great antenna for radiating noise at 100+ megahertz. I'm an amateur radio licensee and I've got an antenna that's at DC ground potential. If you measured the resistance between the two leads to it, its a dead short and would be a short even up to the range of several megahertz. But if is a great antenna at 144 thru 148 Megahertz. ( Our 2 meter VHF band ) I've gone over this years ago with another person and I did a search and found a web site of a computerized medical equipment manufacturer ( not too much room for error there ) and they stated just what I have above and that is at the frequencies modern computer systems operate at they have to use multiple grounds otherwise the systems become inoperable/unreliable because of cross talk. The late 70's era Radio Shack Model 1's expansion interface was a classic example of cross talk. It was housed in a plastic case and radiated almost as much cross talk inducing noise as a radio station and was notoriously unstable. It only ran at around 1 megahertz if memory serves me correctly. You'd also have to ask yourself if motherboard manufacturers didn't want the motherboards to be grounded to the brass standoffs why do they conveniently put a couple of soldered circles right in the exact spot where those standoffs are ? As an experiment you can take and attach the power leads part way onto the motherboard ( so the tips of your ohmmeter can connect to the metal surfaces ), then take an ohmmeter and measure the resistance between those soldered circles and the ground wires and you'll find that it's a dead short, at least it was on any motherboard I've ever checked. HTH Gary |
#36
|
|||
|
|||
w_tom wrote:
Actually one screw to chassis is a single point just as two screws were a single point. The single point was the entire region that included IO slots, single screw to chassis, and power supply connection. This single point (or region) is located at the far end of PC board to that no transient current passes through sensitive electronics regions.=20 =20 Again I emphasis an Intel recommendation that the single point area have additional grounding layers. Just more reason for any ESD currents to remain in the single point region and not taken scenic tours through the sensitive digital electronic areas. =20 I noted an Intel suggestion of more grounding at the chassis connection area. I did not bother to explain why while wondering if you understood the significance of that additional motherboard grounding. I guess not. =20 Intel demonstrates single point grounding. This so that transient currents in chassis do not pass across logic ground plane where transients could cause computer crashes. Computer assemblers are advised to heed Intel recommendations. Do not connect motherboard to chassis at every mounting point - to increase computer reliability. A technique long practiced by experienced and learned engineers. A practice that many computer assemblers still deny - even after an Intel *pdf document demonstrates the concept. =20 Again it is a so simple concept. Connect logic ground to chassis ground so that transient current on chassis will not pass through motherboard. It is called a single point ground. It is even demonstrated by the cited Intel document that YOU provided. I must assume you did not comprehend what Intel wrote. For some, the complexities of grounding were previously unknown. Grounding is but another reason why ESD solutions can appear complex. =20 BTW I did not say "that only ONE is the only suitable means". Read back. I posted other means of obtaining the single point ground. Another was to connect every mounting hole the a chassis plate, and then have only one electrical connection from chassis plate to chassis. Why did you forget this and other single point grounding solutions? Is being wrong (again) so hard to admit? =20 We keep doing this. I provide facts and electrical concepts. You keep denying it; all without demonstrating knowledge of engineering principles. This time it is more interesting. You provided the Intel document that demonstrates what I had posted. Interesting new way to deny basic electrical principles. Deny them and then cite an Intel document that disagrees with your denial. Yes, I am amused. LOL. Well, with your song and dance about 'two' being 'single point' and = using 'point', 'area', and 'region' as if they are 'equivalent', that mak= es=20 two of us, then. As for what you said, and I quote, "Motherboard logic ground best joins=20 chassis ground at a single point using a single conductive standoff." Now= ,=20 for the rational people of the world, 'single' means 'one' and a 'single = conductive standoff' is one screw electrically contacting the motherboard= =20 mounting hole ground and the one grounded standoff; not two, or more. (Yo= ur=20 'alternative' that one can somehow modify their ATX cases so that there i= s=20 "one electrical connection from chassis plate to chassis" is so comical a= s=20 to not need further discourse). Your miss characterization of the Intel PDF is equally scrambled. The=20 section is titled "Electrostatic Discharge Platform Recommendations" and = that is what it deals with: ESD and ESD only. The recommendation for two = connections in the motherboard rear I/O area to the chassis plate is for = that purpose and says nothing about 'single point' (because it isn't) nor= =20 'no more anywhere else', as you erroneously claim. You've gone from "single conductive standoff" to declaring 'two' conducti= ve=20 standoffs, separated by over 6 inches, (which in and of itself is enough = to=20 disprove the 'single conductive standoff' argument) are really a 'single = point' to now claiming the entire side of the board, including I/O slots,= =20 is a 'single point REGION'. (For those rightly baffled by w-tom's jumblin= g=20 of the terms, creating a single (constant) *potential* 'region' is the=20 purpose of a ground plane [or any power plane] and is not a 'single point= '=20 of anything. If one then connects something [such as another 'region', e.= g.=20 analog ground] to that 'region' through a single conductive path then you= =20 have made a 'single point' connection because current can only flow betwe= en=20 the two regions at the 'single point' of contact.) At this rate you'll be claiming the whole bloody system is a 'single poin= t=20 region' in two or three more go rounds. I only mentioned that particular PDF because you appear to have little=20 awareness of anything but ESD and that's what it dealt with. And, to a=20 rational person, two (not to mention an entire row of PCI end plates) not= =20 being equal to one, or a 'single point', would have been obvious enough a= nd=20 sufficient. There are, of course, other considerations and, if we look to Intel=20 document 24333402.pdf, Design For EMI, we find in section 3.0 Board EMI=20 Design Recommendations, 3.1 Grounding Considerations: "Provide multiple=20 direct metal-to-metal contacts for circuit board grounds to the chassis connections, unless the circuit ground must float. Unintended insulation = formed by paint overspray, washers, or non-conductive coatings degrade th= e=20 ground connection and increase radiation levels." As John Doe intuitively, and correctly, surmised, if the motherboard=20 manufacturer went to the trouble of putting ground pads around the mounti= ng=20 hole then they intend for it to be grounded there. Now he knows why. David Maynard wrote: =20 It is not at the 'far end'. They are as near the potential ESD entry point as possible but HALF way across the motherboard, over 6 inches, from each other. of a motherboard so that, again, electrical currents in the chassis do not pass through motherboard logic ground. The single point ground, IO board connections, and power input connect to chassis at the same single point - at the far end of a motherboard. Where I recommended only one screw connection, Intel recommends two in the same location to create the single point ground at the farthest edge of motherboard. "Two" is not a "single point." Demonstrated by page 30 of that Intel 815E Chipset Platform document, no currents across electronic sections Correct, because they are shunted to chassis *there* instead of seeking your 'single point ground'. The motherboard ground also transverses the PSU power cable to the PSU's chassis ground; which makes for THREE chassis ground points in just this limited example. And then, each AGP/PCI card with I/O will have it's chassis end plate connected to ground. And then there's the drives which have logic to chassis ground connections. There's logic/chassis ground connections all OVER the place. of motherboard means no computer crashes due to static electric discharges to chassis. Best not to mount motherboard with conductive connections everywhere. Intel recommends only two connection, both at the far edge to make same single point ground - adjacent to IO slots and power connector. Intel document does not recommend conductive standoffs everywhere on motherboard - as I had posted. The issue was not about a connection 'everywhere'. The issue was your patently false claim that only ONE is the only suitable means. David Maynard wrote: ... Look up Intel 29835002.pdf, Intel=AE 815E Chipset Platform Design Gui= de, section 2.2 Electrostatic Discharge Platform Recommendations where it= specifies connecting *two* motherboard mounting holes to ground, one on each side of the rear I/O area, so that any ESD there is shunted t= o case in the shortest possible path. ... |
#37
|
|||
|
|||
w_tom wrote:
snip of already disproved mumbo jumbo It was always good pc board practice to line every mounting hole with copper feedthroughs. This does not mean all mounting holes need or should make a connection to other grounds. Those plated feedthroughs mechanically strengthen each mounting hole. But Intel only recommends that two holes make an electrically conductive connection to chassis. Reasons why have been posted previously. This is just more nonsensical babble. In the first place, most, if not all (I can't swear to having seen every motherboard), motherboard mounting holes are not plated through. And, even if they were, there is nothing whatsoever that 'requires' the hole plating, nor the surrounding annulus (if there is one), to be electrically connected to anything in the motherboard circuitry. So, if the motherboard manufacturer wanted to 'strengthen' the mounting holes they can do so without needing any 'insulating' washers when you mount the thing but, when you see a mounting hole with a surrounding annulus and 6 to 9 'bumps' in it, those are via connections to the ground plane showing you that, by golly, they WANT that sucker electrically connected. JAD wrote: um if this is the case, what are MB manufactures trying to do? By putting ground rings around each mount hole(on both sides in some cases), they are in fact saying this point needs to be grounded. If they did not want a ground point there, the ring should be non conductive. Makes little sense and To point out 1 single scenario on one particular board, and if it was DM stating it, knowing him he probably stayed up all night finding the one 1 single situation that supports whatever his point is. |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Need Help Installing HSP56 PRO PCI Modem on ECS L7VMM3 MOBO | [email protected] | General | 8 | June 23rd 04 03:50 AM |
Help with overclocking Duron on gigabyte mobo! | matiii | Overclocking AMD Processors | 9 | June 8th 04 11:06 AM |
Need Help Installing ECS L7VMM3 MOBO | [email protected] | General | 3 | June 2nd 04 11:19 PM |
Desperately need help installing OS with RAID on an Intel mobo | Nate | General | 10 | January 1st 04 08:17 PM |
Desperately need help installing OS with RAID on an Intel mobo | Nate | Homebuilt PC's | 15 | January 1st 04 08:17 PM |