Sign in with
Sign up | Sign in
Your question

Shocking case?

Last response: in Systems
Share
October 1, 2009 7:01:14 AM

I assembled my first homebuilt system on Friday (after 25+ years of Compaq, Northgate, Gateway, Dell, Toshiba and IBM store-bought PCs) and it all went pretty smoothly, thanks to the research I did here and on similar sites. I really appreciate the effort this community makes to assist users at all levels of experience and interests.

My question is this: when the power supply is unplugged from the 120v source , the metal PC case is hot with 20v (twenty) AC coming from the ethernet cable. When the grounded supply is hooked-up again, the case is no longer hot.

If the switched power strip is switched off with the power cable plugged in - maintaining the ground connection -, there's no voltage detected on the case.

Just for fun, I pulled the power cord on an old Dell that's hooked up to the same router and there's no AC on that case.

Any thoughts on why this is happening? Does this put any components at risk? Is there a remedy besides keeping the PC plugged in to grounded supply when not in use?

(for the record, it's a Lancool K7 case and Gigabyte p55-UD4p board.)

Thanks for any suggestions or comments.

More about : shocking case

a c 113 B Homebuilt system
October 1, 2009 8:20:21 AM

Wow, you threw me for a loop there. Saying your PC case is "hot" is probably not the best choice of words.

You are detecting a 20V A/C current in your case when the PSU is unplugged? And this apparently goes away when you unplug the Cat 5 cable?

I would say you have some bare cable somewhere, or something is badly wrong. I can think of no reason your ethernet cable would be carrying an alternating current.

Did you mean DC? And yes, any current running through your chassis is a very bad sign.
m
0
l
October 1, 2009 9:07:03 AM

Thanx for the reply. I stand corrected.

Yes, with the case unplugged from 120V, but connected by ethernet cable to router, there's a 20V AC signal on the case. I can feel the small shock and it also triggers the little AC detector used for checking hot... I mean, "voltageful" 120v outlets. I also briefly hooked it up to an O-scope and saw a funky alternating signal. I'll check again to make sure there wasnt a DC component, too.

I will look more closely for possible shorts, but wouldnt a faulty path off the mobo that picks up the network voltage also interfere with some operation of the board, starting with the network? I did have some intermittent losses of signal that took some time to reconnect, so maybe it didnt work as well as I initially thought...

I appreciate your confirmation that it aint what it's supposed to be. Didnt seem right, but now I know....
m
0
l
Related resources
a c 122 B Homebuilt system
October 1, 2009 9:27:21 AM

You have what is known in electronics as a "floating" ground. If you check, I will bet that you have no DC component present.

Google "floating ground".

m
0
l
a c 113 B Homebuilt system
October 1, 2009 9:32:55 AM

Your router is DC. Your PSU is DC. Now, DC will have a fluctuation to some degree. It's not a flat line, because the PSU is not quite that good at converting AC to DC... neither would the wall wart powering your router be that good.

Did you get a crossover cable by chance? Some Ehternet connections require that some of the wires be crossed... that they go into a different pin than they left. This is for connecting one hub to another or one PC to another... if you got hold of a crossed cable, perhaps that could cause the problem?

I'm not a network guy, but that seems possible.

m
0
l
a c 113 B Homebuilt system
October 1, 2009 9:36:07 AM

Wouldn't a floating ground require both ends to be grounded?
m
0
l
October 1, 2009 6:20:08 PM

Here's a screen shot of the signal on the case. It's taken with everything disconnected from the system except the ethernet cable (which is not a crossover, Proximon). The positive lead of the scope is attached to the case and the negative lead is just hanging in the air. No apparent DC component to this.

http://inlinethumb16.webshots.com/42959/268810543010564...

Please correct me if Im wrong, but here's my diagnostic approach going forward:

1) the router works just fine with several interchangeable cables on two other PCs and mostly fine with the new p55, so I'm inclined to rule out the router or the cables.

2)The board has two LAN connectors and the same conditions occur in each, so I'm ruling out the connectors, too.

3) I will pull the mobo, and make sure there are no trace bridges or bits of wire or screws floating around (yikes!). then I'll hook it up on the bench making sure there are no shorted cables or other cabling issues.

If there's a signal on the ground of the board while it's running on the bench, does it not suggest a problem with the board that only be remedied by replacing the board?

I appreciate your previous replies and future ones, too.

jsc - if it's a "floating ground", as you say, what do you recommend I do about it?
m
0
l
October 1, 2009 9:18:36 PM

Update: the bare board is on the bench. Only the RAM sticks and CPU, sans cooling fan, are left attached. With only the RJ45 ethernet cable connected, ground shows the same 20v AC signal I got when it was all connected in the case.

I inspected the board as carefully as I could and turned it upside down. No change.

This would appear to warrant an RMA on the board. All in favor say "aye", all opposed say "nay"

Thanx all, for your assistance.

m
0
l
a b B Homebuilt system
October 1, 2009 10:48:35 PM

Nay - sounds like it the router, or if the router use a brick, it could be the brick.
Have you checked the Ground on the router, with it disconnected from the computer.

If the router uses AC from the outlet, then it is the ground wire on that ac cord OHMs check the ground wire on the AC Power cord, or the router DC ground is floating.
Rational: With the computer AC cord disconnect, you have ac on your case. With both the ethernet cable and the Computer Power cord disconnected - No AC. When you use a different AC cord to the router, -No Ac. Looks like the power cord.

IFyour router use a brick with two wire output to the router.
Also is the output of the brick AC, or DC. I think most are DC. O'scopes are handy (Have a digital one). I suspect that the ground in your router is floating (No direct path to ground) and if the Neg/return line coming from the brick is not tied to ground in the brick then the brick is probably the culprit. Easy to verify with meter, and/or Scope.
m
0
l
October 2, 2009 12:13:54 AM

Chief - thanks for the "no" vote! I really would prefer that this not be a mobo problem. I think you're on to something.

Random responses to your observations/questions:

1) Measuring the output of the router power supply cord, o-scope shows a 20v peak to peak funky sine-ish wave trying to sorta be a square wave. There also seems to be a 70mv DC component in there that doesnt show up on the case.

2) The router has a two prong brick. I'm not sure how I would check the ground on the router. What would you suggest? I have digital scope and multimeter.

3) just for fun, I confirmed with multimeter that the grounds on both outlets appear to be the same. (power strip for PC is close to but not the same as power strip for the router)

4) I replaced the existing router with a backup (also 2 prong brick, different brand) and got the same results.

5) the disconnected RJ45 connector triggers the AC detector when the other end of the cable is plugged into the router.

6) I plugged the cable modem RJ45 directly into the mobo and there was no AC signal on the mobo ground

7) your reference "When you use a different AC cord to the router, -No Ac. Looks like the power cord." may have been a speculative comment. That particular results, tho, is not something I have encountered. I dont have a spare AC brick/cord for the router. But, as noted, I did swap in a completely different router.

8) I think your last line is key: "I suspect that the ground in your router is floating (No direct path to ground) and if the Neg/return line coming from the brick is not tied to ground in the brick then the brick is probably the culprit." That makes a lot of sense. (I think one of the previous commenters said a similar thing in a terse form and I didnt appreciate the "connection" to the problem, if you'll pardon the weak word play...)

Now, perhaps my question should now be, how do I establish a good ground with a 2 prong brick in a 3 prong power strip?

Just a couple other points: I reversed the orientation of the brick in the power strip to no effect. (and now I'm thinking that really should have established the ground, if the brick is good, shouldn't it?) And, for what it's worth, I plugged the router into another PC, pulled the PC's power cord, and detected no AC signal on the PC's case.

A quick reality check: I have the bare mobo on the bench and Im plugging in only the ethernet cable connected to the router. And Im consistently reading that AC signal on the ground. But when the board is in the case and everything is grounded thru the PSU 3-way plug, there is no AC signal on the case. Is it correct to assume that the bare, ungrounded board should NOT have a signal on its ground coming from the router?

Even if you have no further suggestions, I think you've put me on the right track. And if you happen to have the "Magic Answer" to get this all humming along happily without too much trouble, I'll be amazed AND eternally grateful.


RetiredChief said:
Nay - sounds like it the router, or if the router use a brick, it could be the brick.
Have you checked the Ground on the router, with it disconnected from the computer.

If the router uses AC from the outlet, then it is the ground wire on that ac cord OHMs check the ground wire on the AC Power cord, or the router DC ground is floating.

Some rational: With the computer AC cord disconnect, you have ac on your case. With both the ethernet cable and the Computer Power cord disconnected - No AC
When you use a different AC cord to the router, -No Ac. Looks like the power cord.

IFyour router use a brick with two wire output to the router.
Also is the output of the brick AC, or DC. I think most are DC. O'scopes are handy (Have a digital one). I suspect that the ground in your router is floating (No direct path to ground) and if the Neg/return line coming from the brick is not tied to ground in the brick then the brick is probably the culprit. Easy to verify with meter, and/or Scope.
m
0
l
a b B Homebuilt system
October 2, 2009 1:13:26 AM

A coulpe of comments:
If your brick uses a two-prone non polorized AC plug, then they do not tie the output to the "Return" AC as that could put the HOT AC on the return. That is why revearsing the plug had no effect. If it was a polorized plug you could simply tie "Earth" ground to the Neg Output. Don't you hate it.

My brick has a +12 V DC 1 Amp output. Check data plate on yours. If it is DC then I suspect that only reading 70 millivolts DC is not to good (What does the meter give. To get an accurate reading on most bricks you need to load them otherwize they will be high and possiblly excessive ripple. It seams coincidental that the output AC from the brick is also the AC you read on computer case.

Myself I would simply plug in a known good +12 VDC (But then I have regulated DC PSs at home). You can use any Brick as long as it has the same Output Voltage, current Rating can be a little lower.

Still leaning toward the brick with the cavet that it may have messed something up in the router. The twenty volts didn,t really go away when you had the computer plugged in, It was just loaded down.

Missunderstood your comment
"Just for fun, I pulled the power cord on an old Dell that's hooked up to the same router and there's no AC on that case."
Your comment
"I plugged the router into another PC, pulled the PC's power cord, and detected no AC signal on the PC's case. " - Did you also swap the ethernet cable.

Added:
On your waveform measurement - Is it truely AC, that is going + and - , or is it all Positive with the "Bottem" of the waveform going to Zero. Depends on whether you use DC, or AC coupling. Also a voltmeter in AC position uses a Capacitor to block DC component (ie a sinewave going from 0 -> 20 will read the same as a 20V AC even though it is not AC.
Me and spelling do not get along, plus its getting late (Good excuse)
m
0
l
a c 113 B Homebuilt system
October 2, 2009 3:04:21 AM

I have to go back to the cable here. I think you have the normal DC current present in the Ethernet cable running into your computer is going into your ground because of a mis-wired plug. It's the only thing that makes any kind of sense.

You say that you have another computer on the network that is not exhibiting these symptoms. Move the cable to the other computer to test.
m
0
l
October 2, 2009 8:02:35 AM

I'm working on the suggestions you guys have made, especially from Proximon and RetiredChief. I'm pretty much convinced there's nothing wrong with the p55 build. Unfortunately it's hard to believe that two routers and multiple cables are all defective, too. Grounding issues do seem the most plausible. I'm surprised I havent seen others with similar reports but then again I wouldnt have noticed the voltage leak if I had kept the PC connected to the grounded power strip.

It's not over yet, but I really appreciate the help so far! I'll post results soon.
m
0
l
a c 113 B Homebuilt system
October 2, 2009 6:44:55 PM

I put out a request for some further expertise. Maybe someone will have seen this before.
m
0
l
October 2, 2009 8:19:10 PM

I think the problem has been solved.

First, big, big thanks to everyone who responded to my question, especially RetiredChief & Proximon. Literally, I couldnt have done it without you.

No surprise: the answer in one word: "Grounding Issue"

I've pretty much come to the conclusion that nothing is wrong with any of the equipment - including multiple cables that I meticulously checked for integrity - involved except to the extent that the cost-challenged router power supply is ungrounded. I suspect that most users would discover an unwanted voltage somewhere on their PC if they have the system ungrounded (PC power supply disconnected) yet connected to a powered, yet ungrounded router.

I took RetiredChief's - hereafter known as The Man - implied suggestion to hook up an orphaned, 3 prong PC power supply and just tap off the 5V needed by the router and let the rampant happiness ensue.

In following up, I actually would be curious to know if others can measure a "leak" at ground on their systems when hooked up (or unhooked, as the case may be) as I've described.

Three rousing cheers for the Power of the Crowd to solve problems, great and small: Huzzah! Huzzah! Huzzah!
m
0
l
October 2, 2009 8:21:03 PM

Proximon said:
I put out a request for some further expertise. Maybe someone will have seen this before.


Thank you, Proximon, your assistance was very helpful. I posted the outcome of this in separate post.
m
0
l
October 2, 2009 8:26:38 PM

Thank you, RetiredChief, this had me baffled and now it's not only fixed - hopefully - but I learned me up a bunch o'new stuff to boot. Detailed explanation in separate post.


RetiredChief said:
A coulpe of comments:
If your brick uses a two-prone non polorized AC plug, then they do not tie the output to the "Return" AC as that could put the HOT AC on the return. That is why revearsing the plug had no effect. If it was a polorized plug you could simply tie "Earth" ground to the Neg Output. Don't you hate it.

My brick has a +12 V DC 1 Amp output. Check data plate on yours. If it is DC then I suspect that only reading 70 millivolts DC is not to good (What does the meter give. To get an accurate reading on most bricks you need to load them otherwize they will be high and possiblly excessive ripple. It seams coincidental that the output AC from the brick is also the AC you read on computer case.

Myself I would simply plug in a known good +12 VDC (But then I have regulated DC PSs at home). You can use any Brick as long as it has the same Output Voltage, current Rating can be a little lower.

Still leaning toward the brick with the cavet that it may have messed something up in the router. The twenty volts didn,t really go away when you had the computer plugged in, It was just loaded down.

Missunderstood your comment
"Just for fun, I pulled the power cord on an old Dell that's hooked up to the same router and there's no AC on that case."
Your comment
"I plugged the router into another PC, pulled the PC's power cord, and detected no AC signal on the PC's case. " - Did you also swap the ethernet cable.

Added:
On your waveform measurement - Is it truely AC, that is going + and - , or is it all Positive with the "Bottem" of the waveform going to Zero. Depends on whether you use DC, or AC coupling. Also a voltmeter in AC position uses a Capacitor to block DC component (ie a sinewave going from 0 -> 20 will read the same as a 20V AC even though it is not AC.
Me and spelling do not get along, plus its getting late (Good excuse)

m
0
l
!