Weird USB oscilloscope ground issue?

[Alex_Baker]

Today I ran into an odd problem when adjusting two old scopes. The tools I am using are a B&K 3026 signal generator and a picoscope 2204A for sanity checking the B&K. The problem is that when I connect the pico to the generator the trace is offset by about 240mv, regardless of if the generator is on or not, it only has to be plugged in. I put an ammeter between the grounds of the pico and the generator and get nearly 180mA of current flow. I can also get this current flow between the pico and a different grounded object, like a grounded power supply case.

So from what I can tell, there is a potential difference of 240mV between the GND (i.e. the black wires of the PSU) of my PC, and earth (the pin on the outlet). I need some help figuring out why this would happen. The power supply is fairly new, but I have been using the picoscope for years on this PC. Is this a failure of the PC or is it user error? 

[shakalnokturn]

180mA sounds an awful lot !
Is that AC or DC or a mix of ?

The Pico actually works (powers on) when it isn't connected to a DUT's ground ? (Thinking USB cable ground could be cut then it powers itself finding a ground path on the DUT.)
If this isn't the case I'd get really worried, start checking power outlet ground continuity and being weary of the various protagonist's power supplies safety (Y capacitors).

[coromonadalix]

can you plug  the pico on a lappy  battery powered ?        if this offset / current dissapear

the generator could be at problem,  something in the ac section,  ac filter  hoping its not those who get defectives

 or phases problems between your main computer and the generator ?

[Alex_Baker]

180mA sounds an awful lot !
Is that AC or DC or a mix of ?

The Pico actually works (powers on) when it isn't connected to a DUT's ground ? (Thinking USB cable ground could be cut then it powers itself finding a ground path on the DUT.)
If this isn't the case I'd get really worried, start checking power outlet ground continuity and being weary of the various protagonist's power supplies safety (Y capacitors).

It is purely DC as far as I can tell, at least the 240mV I measured is. The pico seems to work perfectly fine, always has.

can you plug  the pico on a lappy  battery powered ?        if this offset / current dissapear

the generator could be at problem,  something in the ac section,  ac filter  hoping its not those who get defectives

 or phases problems between your main computer and the generator ?

I don't believe the signal generator is the problem. I can get current to flow between the shield of the usb cable and a grounded surface, regardless of how the scope or the generator is hooked up. I did not test with a laptop, but I would assume that the issue would go away. I will test with a laptop tomorrow just so that I can finish what I was doing before I noticed this problem.

[Alex_Baker]

I tested with a laptop on battery, not issues at all. There is definitely a problem with my PC, power supply or otherwise. The power supply is a Corsair RM850e that I bought back in march, not super high end but not bottom of the barrel either.

[shakalnokturn]

It is purely DC as far as I can tell, at least the 240mV I measured is. The pico seems to work perfectly fine, always has.

All grounds being common, even with a good USB cable, it makes sense that the supply current to the Pico should divide out through the various ground paths offered. (The infamous ground loops that all audio designers must avoid.)
You can confirm this by using a beefier USB cable or doubling the ground link between PC and Pico, in both cases the Pico probe ground to DUT current (and thus offset voltage) should decrease.

Haven't looked around but guess this is a common known problem.

The solution would be an USB isolator unless you can get away with not tying the probe ground clip to the DUT.

[2N3055]

This is not ground loop question. Something is wrong.

240mV voltage drop with 180mA current suggest 1.3 Ω resistance.

With everything else disconnected, just PICO into PC with USB cable, measure voltage between Pico side USB ground and PC ground, on the cable ends.  You should not have more than 20-30mV.

With everything disconnected measure voltages between PC and siggen ground.
Measure around..

[wraper]

I tested with a laptop on battery, not issues at all. There is definitely a problem with my PC, power supply or otherwise. The power supply is a Corsair RM850e that I bought back in march, not super high end but not bottom of the barrel either.

Not necessarily based on that alone as you're simply make that side float so no current can possibly flow regardless of the reason. There are two things I can think about. Either current flows from laptop PSU from some reason, you can check that by measuring current between PSU GND and mains earth. Or there is an issue with USB GND connection between the laptop and the oscilloscope, so current from VBUS flows through the earth of other device as low resistance path and back to earthed laptop PSU GND through the mains wiring. I suggest measuring if there is a voltage drop between Oscilloscope probe GND clip and laptop GND, you can probe another laptop USB connector shell for that.

[m k]

So from what I can tell, there is a potential difference of 240mV between the GND (i.e. the black wires of the PSU) of my PC, and earth (the pin on the outlet)

Disconnect everything but keep cords connected to machines.
Short earth pins of cords.

Can you find a steady resistance when fiddling with variations?

[Jeroen3]

180mA sounds an awful lot !
Is that AC or DC or a mix of ?

The Pico actually works (powers on) when it isn't connected to a DUT's ground ? (Thinking USB cable ground could be cut then it powers itself finding a ground path on the DUT.)
If this isn't the case I'd get really worried, start checking power outlet ground continuity and being weary of the various protagonist's power supplies safety (Y capacitors).

It is purely DC as far as I can tell, at least the 240mV I measured is. The pico seems to work perfectly fine, always has.

I hope not! That would be very bad.

Can you also measure this current on the USB cable (clamp entire cable) when you touch the USB Chassis to a BNC of the signal generator?
If yes, that means you have a DC ground fault. Then be very cautions, your GFCI will not operate anymore.

[Alex_Baker]

I put a hvac style current clamp meter around the usb cable, nothing. I live in the US where GFCIs are not standard equipment, only bathrooms and outdoor outlets have them.

Does this mean that my PC's power supply is at fault? I have several other ATX PSUs sitting around and they all measure a couple of ohms between black and earth, I did the same test on my PC with it turned off and got 2 ish ohms. So it seems that whatever bonds black to earth is ok.

[wraper]

I put a hvac style current clamp meter around the usb cable, nothing.

Most of the clamp meters only measure AC current.

Does this mean that my PC's power supply is at fault?

No. As I said measure voltage between oscilloscope and computer GND. Also if PSU has GND and earth shorted, it cannot be at fault.

[Alex_Baker]

Most of the clamp meters only measure AC current.

This was in response to Jeroen3

No. As I said measure voltage between oscilloscope and computer GND. Also if PSU has GND and earth shorted, it cannot be at fault.

I missed your earlier response, I apologize. This issue only occurs with my PC tower, not the laptop. I measured 19mV between scope ground and the shield on another usb port on the front of the computer, the scope is on a very long cable. I measured 100mV between earth and scope ground today, which is interestingly not as bad as yesterday.

This ground "leakage" problem has nothing to do with the scope, I can replicate this stray current flow from any GND point on the computer, not just the scope.

[wraper]

I have another thought. Plug everything in the same mains extension cord and see what happens. It could be something funny is going on with mains earth wiring and you connect devices in question into different "earths".

[shakalnokturn]

This is not ground loop question. Something is wrong.

240mV voltage drop with 180mA current suggest 1.3 Ω resistance.

I don't often disagree with you but here I'll have to, not on the fact that the resistance is rounded down from 4/3  
There is a ground loop sketch it out, you'll see... Probe ground, USB ground and PC ground are all common now if you connect the probe ground to another length of wire leading to the PC's PSU inlet ground connection you create a loop which is also an additional return path for the supply current to the Picoscope. This alternate return path runs through the Pico's frontend ground tracks and probe ground wire which also have resistance (say approx. 1.3 ) so measuring anything is offset by the voltage drop across the probe's ground wire.

Edit: OTOH you weren't saying there wasn't a ground loop, rather it wasn't the problem. It still looks the most plausible problem to me considering it's DC.

[Alex_Baker]

There is a ground loop sketch it out, you'll see... Probe ground, USB ground and PC ground are all common now if you connect the probe ground to another length of wire leading to the PC's PSU inlet ground connection you create a loop which is also an additional return path for the supply current to the Picoscope. This alternate return path runs through the Pico's frontend ground tracks and probe ground wire which also have resistance (say approx. 1.3 ) so measuring anything is offset by the voltage drop across the probe's ground wire.

What puzzles me is that I have never ran into this problem before. I have used the pico dozens of times over the last several years and never noticed any funny business before.

So If I do create a ground loop when connecting the pico to an earthed BNC, is that normal? if so how do I avoid it in the future?

[shakalnokturn]

Creating a ground loop in this situation is normal and unavoidable without additional hardware, why it should only show effects now is another problem...
The most obvious to me would be higher resistance in the conductors or connections somewhere between probe ground clip and PC PSU. (Probe, BNC connector, Pico PCB, Pico USB connector, USB cable,PC USB connector, optional USB to mainboard connector, mainboard to PSU connector...)
Try hooking everything up so it shows the offset then wriggle things around to see effects on offset.

Out of curiosity with Pico connected to unplugged PC what resistance measurement do you get between probe ground clip and PC's PSU earth terminal?

I'd look for a high speed USB data isolator and a (ungrounded) linear regulated power supply for the Pico. Still I'd first try to understand why this issue is showing now if it hadn't previously...

[2N3055]

This is not ground loop question. Something is wrong.

240mV voltage drop with 180mA current suggest 1.3 Ω resistance.

I don't often disagree with you but here I'll have to, not on the fact that the resistance is rounded down from 4/3  
There is a ground loop sketch it out, you'll see... Probe ground, USB ground and PC ground are all common now if you connect the probe ground to another length of wire leading to the PC's PSU inlet ground connection you create a loop which is also an additional return path for the supply current to the Picoscope. This alternate return path runs through the Pico's frontend ground tracks and probe ground wire which also have resistance (say approx. 1.3 ) so measuring anything is offset by the voltage drop across the probe's ground wire.

Edit: OTOH you weren't saying there wasn't a ground loop, rather it wasn't the problem. It still looks the most plausible problem to me considering it's DC.

My English. Sorry.

There is a wire loop. Always is when you measure with any scope.

Problem is that voltage is DC and is huge. That is what I meant.
Ground loop is not source of the problem, it is what allows us to measure that something is injecting current to ground. And it should not. And even if it does, impedance of grounding should be lower.

10-30mV is normal.
250mv means something is wrong.

All the stuff on your desk should be connected to same grounding point.
Connected with short cables, there should be order of magnitude less potential difference in what essentially aspires to be equipotential net.
Something is not right with grounding and/or power distribution wiring.

In addition to DC current being injected into ground from somewhere. Also not right.

[shakalnokturn]

In addition to DC current being injected into ground from somewhere. Also not right.

Yes that is another possibility of course DC current in a shared ground wire could give similar symptoms and be potential safety hazard if ground continuity is lost at any time.
Once power mains (line) power distribution ground conductors have been confirmed to have sufficiently low resistance (which is one of the requirements for a safe power distribution) a couple of quick measurements should be able to reveal if cause is DC current flowing in the ground conductor.

The easiest to me still seems to be a comparative test:

Take a random remote ground point that shows DC voltage / current WRT to Picoscope probe ground as per OP's issue, take the same random point and repeat measurements WRT to PC's casing.

If the same voltages / currents exist in both cases 2N3055 and wraper's explanation fits here, if not mine applies.

It looks like I missed this along the way:

This ground "leakage" problem has nothing to do with the scope, I can replicate this stray current flow from any GND point on the computer, not just the scope.

So I could be wrong about the origins of the problem in the end.... Then OP should get very worried about the leakage source.

[m k]

This ground "leakage" problem has nothing to do with the scope, I can replicate this stray current flow from any GND point on the computer, not just the scope.

When measuring those resistances, did you include the power cord?

Move the tower to the bathroom.
Disconnecting mass storage will speed things up.

What say your GFCI?

If nothing check earth connection.

[Alex_Baker]

I did a bit more testing but came up inconclusive.

The resistance between GND and earth seems fine, a couple of ohms on my crappy meter that reads 1 ohm with the probes shorted. Something odd that I noticed with an analog ammeter is that the needle will drift cyclically up and down a few milliamps. As a point of comparison I checked three other ATX PSUs, all from different decades and none showed current from GND to earth. I did the same test to the one in my PC with everything disconnected and got no stray current, so this phenomenon only occurs with the whole system hooked up.

Another more concerning and puzzling thing that I found is that with the ammeter attached to GND in the PC, and earth, the needle jumps around in tune with a new squealing sound that the PSU is making. I don't know when this noise started, but I only noticed it when I went snooping around my computer. Remind me to never buy a corsair power supply again, because every one I have owned, new or otherwise, has squealed like a stuck pig.

[magic]

As much as it sucks, this is absolutely normal and not a sign of any mains wiring issue.

Your ATX PSU is internally grounded to mains protective earth and there is no current flowing through this connection until you connect your USB gadget. The gadget is grounded to the motherboard, which is grounded to the PSU through a bunch of black cables, as you know. Those cables carry many amperes of current, hence motherboard ground is some millivolts above true earth and its exact potential jumps around in sync with varying power consumption of the CPU. If you ground your USB gadget elsewhere, you provide a new path for the motherboard ground current to return to the PSU (through the USB cable and mains protective earth), so part of it diverts there.

About the only solution is to reduce resistance between the PSU and the motherboard, and particularly between the earth connection inside the PSU and the USB port ground pin.

[Alex_Baker]

As much as it sucks, this is absolutely normal and not a sign of any mains wiring issue.

Your ATX PSU is internally grounded to mains protective earth and there is no current flowing through this connection until you connect your USB gadget. The gadget is grounded to the motherboard, which is grounded to the PSU through a bunch of black cables, as you know. Those cables carry many amperes of current, hence motherboard ground is some millivolts above true earth and its exact potential jumps around in sync with varying power consumption of the CPU. If you ground your USB gadget elsewhere, you provide a new path for the motherboard ground current to return to the PSU (through the USB cable and mains protective earth), so part of it diverts there.

About the only solution is to reduce resistance between the PSU and the motherboard, and particularly between the earth connection inside the PSU and the USB port ground pin.

This makes sense, thanks!

I am still wondering how I have never noticed this before when using my scope, perhaps I had something hooked up weirdly that one time. It sounds like the proper way to avoid this issue is a USB isolator and external power supply(or some devices have built in isolated DC-DC).

[magic]

Yes, USB isolation will do it. A dedicated low power SBC for running the scope also should.

I don't know if USB has enough common mode noise immunity for this to work, but one could also try to star-ground the USB port in the PC. Say, buy one of those rear brackets with USB ports connected to pin headers on the motherboard, cut the GND cable and connect it to a free Molex plug going straight to the PSU.

I discovered this issue long ago with audio equipment, because analog outputs have the same offset and CPU noise problem. It is a widely known problem, although few people appear to know the reason.

[wraper]

As much as it sucks, this is absolutely normal and not a sign of any mains wiring issue.

Your ATX PSU is internally grounded to mains protective earth and there is no current flowing through this connection until you connect your USB gadget. The gadget is grounded to the motherboard, which is grounded to the PSU through a bunch of black cables, as you know. Those cables carry many amperes of current, hence motherboard ground is some millivolts above true earth and its exact potential jumps around in sync with varying power consumption of the CPU. If you ground your USB gadget elsewhere, you provide a new path for the motherboard ground current to return to the PSU (through the USB cable and mains protective earth), so part of it diverts there.

About the only solution is to reduce resistance between the PSU and the motherboard, and particularly between the earth connection inside the PSU and the USB port ground pin.

To the level of what he'd seen it's not normal.