Strange ground loop or something similar in my home

[hydrogen18]

Over a year ago, I installed this wooden cabinet on the wall that I can put equipment in. Recently I added in a DC power supply that is used to run 19 Volts DC around my house. It's fused and stuff and meets code from what I understand here in the US.



In the wooden cabinet I put a small steel grounding strip. I ran a 14 awg wire over to the edge of my house(through the attic, etc.) but forgot to ground it. So my grounding strip is just a "pretend ground", I will call it that from now on. It's got a whole bunch of surge suppressors tied to it but it is not tied to the mains ground.

I was troubleshooting some unrelated issues today and put a multimeter between the mains ground and this "pretend ground". It read 26 VAC. So I found a AC current meter and it showed 200 ma between the same points. At this point I figured all my meters were bad. So I rigged up this.



It is exactly what it looks like. An LED between mains ground and "pretend ground". It has a 100 ohm resistor as well. It lights up, so there is real current here. Remember that the "pretend ground" is 100% isolated from earth.

I eventually discovered unplugging my 19 VDC power supply would "fix" this current. I can't comprehend why. This power supply runs over a 3 wire cord about 20 ft long. Two wires being the isolated DC output and another being the "pretend ground". The "pretend ground" is connected to a cabinet mounted on a wooden wall in my garage. So it's sort of earthed but 2 x 4s aren't really great conductors. So there are 3 states I found

1. Power supply unplugged - no current, LED not lit
2. Power supply plugged into multi outlet strip that is off - very little current, LED lit dimly
3. Power supply plugged into multi outlet strip & turned on - 200+ma current indicated, LED lit brighter

Next I dug out the scope and hooked that across the LED + resistor. I captured the following traces from the cases above

Case 1:



Case 2:


Case 3:



So in case 1 it looks like some switch mode power supply is coupling into the long unterminated "pretend ground" wire. It shows approx 1 volt amplitude. Case 2 is just weird, I don't understand that. But case 3 has 30 volts of amplitude and some sort of 60 Hz component. So mains is getting coupled in here! Enough to light the LED.

I tried a transformer based linear regulated power supply and did not see anything on the scope nor did the LED light.

I tried a 24 VDC wall wart (300 mA) rating that came with some Harbor Freight drill. The scope shows this waveform and the LED is really dim. Still seems to have a 60 Hz component.



Can someone give me an idea what is going on here? I've shorted the grounding strip to mains ground for now. But I seem to have 200 mA of mystery current floating around that I just don't understand at all. I don't think you can get this kind of capacitive coupling between stuff at 60Hz. I have tested the outlets  in my house and they all appear to be wired properly. The electrical system of my house is grounded with 4 ft long copper clad steel rod connected to the main breaker panel on the outside of the house

[wraper]

Sounds like Gateway PSU is faulty. Probably a short between primary and secondary within a transformer. That's why you need a real ground when using devices which have ground connection. Devices which require ground usually have weaker insulation compared to those which do not (reinforced/double insulated). Also that's why having GFCI is important. What you described is a life-threatening issue. Of course there is possibility that PSU injects high frequency noise into earth connection and your meter just goes nuts, no real 200 mA present there. Small leakage current like 1 mA is acceptable and is completely enough to light a led. Also it would be way more informative if you measured waveform just across resistor, without LED present. So you could see a real current flowing through it.

BTW, you should learn how to use/not use autoset. 500 ns/div and 2.5 us/div timebase is useless for displaying this issue. Other are too low as well but not that bad to be completely useless.

[T3sl4co1l]

SMPS normally have line-to-ground caps for filtering.

You shouldn't see much waveform though, on account of the filtering, so it is quite possible something has gone bad.

Tim

[f4eru]

You're measuring normal EMC with the scope. That's normal when you have a gigantic antenna connected to your scope, and a SMPS nearby.

The 200mA is much too high, it should be much lower through the typical Y-caps -> it could be a crappy meter overreacting to the EMC, or the PSU being faulty.

Your wiring is dangerous. Fix the Ground connection before somebody gets hurt or dies.

[hydrogen18]

When you say EMC do you mean Electromagnetic compatibility? This is a new concept to me.

[jmelson]

When you say EMC do you mean Electromagnetic compatibility? This is a new concept to me.

Yes.  Almost all electronic gear today has some sort of line filter in it to prevent it from sending switching noise out the power cord (conducted emissions) or being affected by noise coming in the power cord (electromagnetic compatibility).  These typically involve capacitors from some power cord pin to another.  If the appliance has a 2-prong plug, then it can't conduct any current to the safety ground pin, but if it has a 3-prong plug, it often will conduct current to the safety ground pin.  This current can be surprisingly high for some equipment like big test gear (computers, logic analyzers, scopes, spectrum analyzers).
If you have a bunch of these all plugged into the same power strip, the current can sum to "shocking" levels (pardon the pun).  My logic analyzer knocked me onto the floor when we had a bad safety ground in the wall outlet.  Typically, on all this type of gear, the line filter is connected to the mains at all times, regardless of the power switch position.

Jon

[hydrogen18]

Good to know, I'm starting to think this may somehow be related to the GFCI in the garage that keeps tripping.

[f4eru]

Yes.
A typical 1nF Y cap, if connected L-G will drive approx 60 uA to Earth (G) on 120V/60Hz, and should not trip a 30mA RCD.

These caps are also the cause of the metal parts of unearthed devices (class II) to feel "a little electric" when lightly touched.

A leaky device will, and often it is multiple leaky devices adding that trip the RCD. -> you can use a good and filtered multimeter to measure the earth leakage to check each leakage and find the offender.

EMC is electromagnetic compatibility, unwanted radio transmissions in fact. Every non linear device will generate some noise, and some other device (like I presume your multimeter) are affected.

[hydrogen18]

Thanks everyone who replied.

I took more scope traces today as suggested: Here are some shots of the waveform measured with the scope







So it has an obvious 60 Hz component, nearly 27 volts peak to peak. So I'm leaning towards the theory around bypass caps between ground and the line in power supplies.

With a 2200 ohm resistor connecting earth and the "pretend ground" I get this



So I guess there is no real current here since it dropped to almost nothing.

I can measure this waveform at a much higher frequency and lower amplitude.




But this just looks like some SMPS noise getting coupled into either the earth wiring or the scope leads. Also with an amplitude of ~1 volt it is not as concerning to me.

So I think I will just finish installing the earth rod I already have and just never put in, as well as binding together this ground strip in the cabinet with the mains. Not really sure how to go about doing that in a code compliant manner here in the states. I guess I could just make a run of #14 awg THHN to the grounding rod near the main panel on the house.

[wraper]

If this was measured over 100 Ohm resistor as before, this power brick is deadly dangerous, around 100mA RMS. If just across open circuit, then nothing exceptional.
 

[wraper]

Again, with such fast timebase, this measurement is useless.