Grounding

[nihili]

Prompted by some comments from Hoot in Cases and Power Supplies:

What are the relative merits of floating the DC ground about the safety ground? How does the grounding of the power supply relate to the choice of grounding the motherboard via metal standoffs?

nihili

[Tweaked!]

Actually. I believe that all the components within the pc are grounded by the psu's ground, thus grounding the psu completes the grounding. I've heard alot of debates about grounding psu's, hsf's, chassis's, etc... and I've personally experimented around with various different ways of grounding, and grounding either the chassis or the psu are basically the same thing. But by doing so, I've noticed a small but noticeable difference with the majority of my pcs. Especially older ones that have been running for quite a while. I can't give any true technical data to whether or not this is beneficial, but like I say, I've noticed a difference:)

[Dissolved]

I dunno if this matters or not.. But the Psu IS grounded To the Pc Case. ive tested myself via multimeter.

[nihili]

Yes, the question is whether the Ground rail (black wires) should be directly connected to the Safety Ground (third prong of the plug). THe safety ground is typically conected to case, and the Ground rail is typically connected to this. But in some PSU's the safety ground and Ground rail are isolated. I understand that there are arguments in favor of each way of doing it, but I don't know what those arguments are. Hence the question.

nihili

[Dissolved]

I see.. id probly think it should be connected to the Safey gorund then in that case.

[nihili]

Ok, remember, this is Technical Discussion. WHat I want is the electronic theory behind this. WHat are the theoretical areguments behind the two methods of doing this?

[UnLoadeD]

There's a big difference between AC ground and DC negative. In DC negative pole is needed to complete the circuit and allow it to work. In AC the ground is there to provide path of least resistance to earth, in case there is a short. If there is no ground on an AC system and there is a short, YOU might become the path of least resistance if you touch any part of it.

Personally I can't see a good reason to have ground and negative connected directly. If there is a short in the AC part, it could use this path to fry all your DC components.

peace.
unloaded

[nihili]

But as a matter of fact nearly all PSUs connect the DC and AC ground. I've used 4 different brands of PSUs, three of the four connected AC and DC ground. The only one that didn't was an NMB server supply. Antec and Sparkle both connect them. Hoot's remarks that I referred to earlier implied that it was his opinion that AC and DC grounds should always be connected.

Also, does anyone know if whether your PSU connects the two affects the advisability of gounding the motherboard to the case via metal standoffs. It seems to me that perhaps one would want to isolate the mobo if the two grounds were not connected. Instead, perhaps one should provide a DC ground to the standoff holes, but keep them isolated from the case.

But frankly I don't know a lot of electronic theory, which is why I'm posting. I'd like to know a lot more.

nihili

[Greedy Guido]

Ground is what is considered 0V wwhich is also the ground of AC outlets. You ground the DC to this in case of a transfromer fault to avoid making your system and case go live.

If this happens it should throw the mains power.

This is common in industry to ensure very low touch potential and avoid zapping people during fault conditions.

[Paxmax]

If you mount the motherboard in a case the MB will by default be grounded through standoffs to case to PS to ground. If the PS digital 0V is connected to ground and then the MB is grounded too(through a different path) you might in theory create a ground loop. Thats nothing good. But I think we can disregard that ground loop, it isn't very weak so there isn't much room for influences. If it was my choice I'd go for the PS with no connection between oV and Ground, but it wouldn't bother me if the PS 0V was grounded inside PS.
I can explain a bit further of switching PS if anyone's intrested, but I don't know if I have anything more to say about 0V and grounds. Any questions/suggestions?

Anyways, I seldom connect my PS to a grounded outlet. Cause usually there is just a socket at home with Neutral(N) and Live (L).
BTW I live in Sweden. It has come to my attention (many times) that when I skip the grounded outlet, my whole PC tends to float at approx half of input voltage 230V. So, the whole case(and electronics) is loaded with 115V !!!! The current supplying that 115V is however very low, it comes from a split between two small caps in the PS. I think that those caps are on all PS.
So, the connection for those caps are: Live wire--cap--ground--cap--neutral wire
I can certainly feel the 115V if I happen to brush the underside of the arms(thin skin) against a unpainted corner of the chassie. It's a stinging/burning sensation.

When I think about it, I'm putting my components in great peril when handling them, but still nothing has broken. All MB electronics are hovering at 115 volts, and I could well be grounded and stick my hand in there messing with fanplugs, mems and who-know-whats. Effectivly delivering 115V jolts here and there. Luckily, I'm not perfectly grounded, and the charge in the caps are low(though not so low as it would spare electronics)
I'd strongly recommend pulling the plug out before changing add-on cards and such. I tend to forget when I go at it... even though I work with/construct electronics all day long. Well it's just like that saying: Don't do as I do! Do as I say!

[Since87]

I've been avoiding wading into this one because I can say a fair amount about the subject, but for PC's it mostly adds up to it to, "It usually doesn't matter much."

I'm going to repeat a bunch of stuff others have already said largely without attribution cause I'm recovering from the flu and feeling lazy.

Connecting the earth ground of the power outlet to chassis serves two purposes:

Safety - except for some extremely bizarre circumstance the case would never be at a dangerous voltage regardless of component failures inside it.

EMI shielding - Tthe case acts as a very effective shield for the EMI being radiated by the circuitry inside, when it is connected to earth ground. If it was not connected to earth ground, the case would be acting as an antenna for some of the high frequency signals floating around inside.

Ok, that's the easy part which you probably already knew anyway.

So, why connect 0V to earth ground?

First, as Paxmax mentioned, if the 0V line of a power supply is not connected to earth ground somehow, that 0V line will actually be at an AC voltage with respect to earth ground. (The capacitance that conducts the AC current is generally the capacitance within a transformer - between the primary and secondary windings.) The capacitance is small enough that it's not dangerous to people, but some electronic components can easily be toasted by even the 60Vac signals (mains/2) we tend to see in North America. This floating AC voltage can also damage other equipment the PC may get connected to.

Second, suppose there is a major fault in the PSU, and 0V is suddenly shorted to the hot AC line. With the metal standoffs connecting MOBO ground to AC ground, a huge current is going to flow through the through the traces of the PCB if there isn't a lower resistance path to earth ground within the PSU. Are the MOBO traces going to blow up before the circuit breaker does? Maybe, maybe not. A low resistance connection between 0V and earth ground within the PSU avoids the question altogether.

So, why not to connect 0V to earth ground?

In many circumstances, such as radio equipment, high end audio equipment, or (in my particular experience) test instrumentation, extreme care needs to be taken with what is considered to be 0V, and what paths currents are allowed to flow through. Many non-PC supplies have output voltages which are not directly connected to earth ground, in order to let the engineer designing it in, have control over what paths currents can take. Paxmax mentioned ground loops. High frequency current traveling through loops such as below can radiate "a lot" of noise.

PSU output->MOBO component->MOBO ground->MOBO standoff->case->PSU chassis->PSU ground->PSU output filter capacitor->PSU output

Forcing the return current to follow virtually the same path as the outgoing current reduces radiation. (and susceptibility to radiation) That is why twisted pair and coax are used for high frequency (or highly sensitive) signals. It's generally considered bad engineering to design in ground loops.

Compared to analog signals, digital signals aren't very sensitive to the relatively weak signals that can come from this sort of ground loop, but as system speeds go up, even the digital circuitry is become more and more sensitive to EMI inside the case. Also, high frequency signals flowing in the case itself can radiate, particularly around slots in the case.

So, why are there conductive standoffs between the MOBO and the case? And ground connections on the MOBO for the standoffs to connect to?

My best guess, is that when the IBM PC was originally designed, connecting the MOBO to the case was considered to be the only hope of keeping Joe Sixpack from static zapping the S out of the IC's. Even if the supply was unplugged from the Mobo, the odds were good on the original PC that you would make contact with some portion of the case before making contact with the MOBO, thereby reducing the risk of static damage.

So, why do some PSU's connect 0V internally to earth ground, and others don't?

H if I know. If you've followed me this far, your guess is probably as good as mine.

I'd be inclined to connect 0V to earth ground within the PSU myself. I'd also probably leave the metal standoffs in place just for the extra static protection provided.

[RonnieG]

If the "AC ground" isn't connected to the "DC ground" in you NMB PSU then to what is the "AC ground" connected to?

[RoadWarrior]

I favour the ground everything approach, I've had PCs crash randomly every hour or so when powered from a non-grounded outlet. I've had boards fail to boot unless grounded to the case. I've seen printers randomly spew rubbish when they're connected to a groundless supply and the interface cable doesn't make a ground connection to the computer case. I've also been able to stabilise marginal overclocks by grounding heatsinks. I've seen lightning hit PCs through the modem, that had every single component fried that were ungrounded and have just the modem cook that were grounded. I've even seen some grounded machines have thier modem mostly survive a distant hit, because only an output transistor blew, which was a common type so was replaced. I've seen harddisks have permanent block and head damage when the power has pulsed on and off on ungrounded PCs and I've seen grounded PCs reboot with a scandisk and all is well.

In a carefully controlled environment with every parameter known, you can probably prove that a system with a split ground is less susceptible to artificially inserted noise than a system with tied grounds. However, carefully controlled environments bear little relation to the trials of the real world. So I go with my experience.

Thinking about loops though, surely they're only really a problem when the ground circuit has a significant amount of resistance, creating a potential difference that will incur a current flow, then if your PSU is grounded already, and you isolate the board from the case when you fit it, then because the only grounding point is through the mobo power wires, you'll have a nastier ground loop than if you screwed the thing down to every brass spacer. I think that's the point really, if you're going to ground, don't do it half heartedly.

You'll probably get caught out trying to isolate the motherboard from the case anyway, since a lot of expansion cards have their backplates connected to the groundplane of the card, so if you really want your system to pick it's own ground through the PCI bus, or maybe even through a drive cable, you're going to have more problems than you think you were curing.

Road Warrior

[silverthumb]

Interesting question Nihili. Makes me wonder if you have been making your OWN PWR supply or something! Lol! Or is it a case??? I'm no electronics wizard so I can't contribute much. I do work in a music studio and this question has come up a LOT as noise and problems. I know it's an important issue. I wonder if one wired up two identical systems each way how much noise differance there would be.

BTW, Paxmax, you're CRAZY! Just don't spill yer beer on your computer while reaching over to pick up that dropped cd case! :D

[bigben2k]

I think that the answer is there, somewhere...

I'm going to throw a monkey wrench in the works, and see what happens:

What if I static zap the PC case?

(Having done this before, it causes a reboot)

[Sniperboy]

When your computer reboots when you shock it, it's probably just the PSU protecting itself. It probably recognized that shock as a live wire from the PSU that touches the case. That would be bad for the PSU, so it shuts off. Don't touch a live wire to your case on purpose, it's scary, and I'm not sure whethere it damages your PSU or not.

[Voidz]

Looks like super beardninja went the way of the dodo on this thread.

So does anyone care to sum up? I felt like I was reading your posts at a 4th grade reading level. Extra grounding is good or useless?

Have you read the posts about grounding your heatsink to the inside of the case/ps via a copper wire? To eliminate EMI? I don't know, sounds alot like hooey to me.

[RoadWarrior]

You only notice the effect of grounding your heatsink on overclock when you've got your CPU way cool and it's not going further, since most OCs are stopped by heat first, then a lot of people won't see a difference. Myself I've only experienced it on decapped K6-2s, it's worth up 50mhz on those.

[Thingi]

I used to own an ATX case which only had plastic stand-off's. That case is now with one of my friends. Now from this discussion comes a silly question :o

When I go into his case should I leave the PS plugged into the wall and make sure the rocker switch on the PS is set to off (as I normally do). That way the case/mobo is grounded.

What about the mobo though.... Since the case has plastic standoff's surely I need a Power Supply that connect's 0v to AC outlet ground to make sure the mobo gets a ground connection.
Would it be true that if the PS didn't have 0v-to-AC ground then the mobo would not be grounded because of the plastic risers?

So is the reason why 0v-to ground is common because cases can have plastic stand-off's instead of metal risers?

(I usually use my grounding mat and bracelet if doing alot of work inside PC's, but some times I don't have it with me if i need to do a quick tinker inside my mates PC).

thingi

[oz2cpu]

Most ATX PSU cases are connected to GND of the outputs.
See here on the bottom side pcb pictures I have:
www.webx.dk/oz2cpu/radios/psu-pc400.htm
or even here:
www.webx.dk/oz2cpu/radios/psu-pc400-mod.htm
(I have reversed engineered several PSU types)

But that grounding is not made for big amps !
The ATX connector is designed to deliver all the amps in the GND path. A PC will work fine with mobo and psu placed on a table with no metal case at all.. the case is SHILDING not power gnd

[shane46and2]

One of the reasons that a mother board gnd and a case are tied together is to prevent eletromagnetic radiation from escaping the PC. This is a serious problem from a EMI pollution point of view. Keep in mind that GHz class computers run on simlilar frequencies as celular phones, cordless phones, wireless networks etc. The PC needs to stay quiet so these other devices can work properly. As ozz2cpu suggested the case is sheilding important.
Now as to wheather the DC gnd should be tied to the earth gound in your power outlet? There is a vast number of buss traces on the motherboard that are switching at high frequencies(100Mhz,133MHz,166MHz etc.) these traces are antennas to some degree. Each signal on these traces will emit some EMI, to protect the outside world from this EMI one could tie the case to earth ground. Ie, basic eletrmagnetic theary suggests that radiation cannot pass through a good conductor. Since this energy can't pass through the metal case(ie the case absorbs the energy) it needs a place to dissapate, earth ground is a good choice.
If one was to not tie the power supply DC ground to the case, the high frequency noise on that ground would have to take a longer path back to earth ground. This is a potential source of more EMI. This could also case some problems with some external devices that may be grounded to earth, ie serial devices in industrial apps that are hundreds of feet away.
shane

[bigben2k]

Originally posted by shane46and2
If one was to not tie the power supply DC ground to the case, the high frequency noise on that ground would have to take a longer path back to earth ground. This is a potential source of more EMI. This could also case some problems with some external devices that may be grounded to earth, ie serial devices in industrial apps that are hundreds of feet away.
shane
...but isn't the reverse true as well? Couldn't noise from an outside source affect a PC more easily?

[larrymoencurly]

Is it enough to couple the case to DC ground only through capacitors, to handle the high frequency interference?

If a direct connection between the two grounds is so desirable, then why are so many PSU circuit boards designed to allow them to be kept separate? My Delta, Antec, and even my cheapo Maxpower all have provisions for connecting the DC ground to chassis through just a single jumper wire.

[Dragonprince]

I think you folks are thinking to hard here.
The ground strap from a PSU to ground the case is just that. An extra ground incase your mounting a PSU in a case that might have plastic rails or some such that the PSU sits in.

The issue of the PSU and the internal earth ground (AC) and PSU ground is a different breed all together. First you need to understand that in modern electronics a path to ground is used as a trigger for logic gates and transistors and other assorted electronic parts. Its not just ground, but the path to ground, thats important here. For instance say you want to trigger a logic gate only when another component is energized. How will you know ? You can either tap the positive voltage feed side or tap the negative (ground side). If you tap the feed side your job could be much harder because now your adding inductance, capacitance and resistance to the feed side due to the extra wiring. Much easier to tap the output side of a DC circut, now your down stream from your component but you still know its in use right ?
Now you want to use that negative dc rail for signaling like we just explained but you also want to protect your equipment from stray voltage (noise) that might come in off that AC ground (who says your tv in the next room is good ? might be throwing out tremendous EMI )....how do you do it ? Isolation. Use some sort of resistive link between your "must be protected because its a part of the circut" negative rail in the PSU and the AC ground. A direct connection would let the noise in the "backdoor" of your circutry. Everyone on board now ?:D

[badgers]

it violates UL listing to take the black wire of the PSU to the green case ground.
the black wire can never be at case ground.

there is always some voltage between the neutral and ground due to voltage drop in the neutral.
there is also a forward diode drop between the line and the black rail of unregulated DC.

if you connect this to ground you will impress a current on the green ground wire.

National electrical code only allows the neutral and ground to be bonded at one location(the service entrance) by tying the black of the PSU to the ground you have(through diodes) tied the neutral to ground again.

also consider that this may trip a GFI circuit. But I doubt there will be enought current(5ma) to trip the GFI.

[larrymoencurly]

But what about the black ground wires on the low voltage side of the PSU? Most

[larrymoencurly]

But what about the black ground wires on the low voltage side of the PSU? Most UL approved PSUs have them grounded to the case, just like the green wire is.

[The Overclocker]

i belive the main reason for not using the 'safty' (external) ground is to prevent damage to components - when a fault occurs there would be a large current going down the ground which could damage components in the powersupply as well as where the fault occurs.

note that 'old' (some AT) powersupplies used real ground - they dont now - must be a valid reason - perhaps if someone could find the ATX psu spec it might explain why these changes have been made

[badgers]

who has a PSU that has the green ground tied to their PSU output?

I have never seen that larrymoencurly I am looking at one right now and the green wire is screwed to the metal case and that is it. I really know the electrical code and it violates the listing and the NEC to do what you are talking about.
ps. the neutral and ground are tied together when they enter your building(house, office ect) the green ground is NOT at zero volts. As long as there is a voltage drop on the neutral there is a voltage on the green ground. This is the source of "Stray voltage" that farmers have tried to deal with. the green ground has a voltage that animals standing in mud get low levels of shock from.

[Annoyingrob]

Correct me if I'm wrong, but wouldn't grounding the black lines to an actual ground, as well as grounding the case to the same ground cause some sort of ground loop? Also, the psu ground should be neutral realative to the other voltages to create a proper voltage differential.

[bigben2k]

Yes that's correct, but a "ground loop" as you describe it, refers to an effect by which the current travelling through it will stray all around, creating different voltage potentials, at any point. It's something a PCB designer tries to avoid. Maybe someone else can describe this more accurately.

There's also the possibility of creating inductance/capacitance, but that's usually not an issue in DC applications (but here it would be).

Maybe that's why some PSU manufacturers don't ground the case.