USB, leakage currents and earth referenced and non-ER equipment.

[paulca]

A lot of the electronics I work with these days are USB powered.  Either via an earth referenced computer PSU or a floating w/ leaking current USB power adapter/charger.

As per other threads I am developing stuff with USB audio, ADCs, DACs etc.  This involves connecting a whole bunch of different equipment together.

So far I've been a little hesitant when connecting a breadboard to two different PCs USB ports, this obviously isn't that bad as both PCs have the same (incredibly noisy) ground reference.

However, if I then connect my oscilliscope to that bread board, it's inductors all POP and whine as soon as the ground is connected and it immedaitely starts showing noise of about 1 volt peak to peak with a whole range of frequenies.  Even scoping a pulled low MCU pin, has the same noise.  The scope is completely floating, DC powered from a battery and why I wasn't really worried hooking it up to the breadboard.

I would like to use a signal generator however to test filters and spot check audio quality etc with the scope.  The signal gen has however got a SMPS (that I never did replace).  It has the usually 100V+ ground leakage current of uAs.  It's still 100V+ though.

I would imagine connecting the sig-gen and PC USB to the same breadboard will be fine.  The breadboard can take it and as both USB/PC and SigGen are on the same earth/neutral reference and both have SMPS there won't be much potential across the board?  The uA of current will not bother either end as they are used to it anyway.

The scenario I would be more worried about would be connecting the scope to the sig-gen, and then the USB powered breadboard, without first connecting the two grounds together.  Would this not lead me into the scenarios that my scope grounds become the conduit for the 100V+ ground current?  If the scope is a 50-100V chinese scope... will that result in smoke?

[m k]

Since I live in schuko world I may have a different approach.

Back in the day when regular dry area home socket was two contact and totally round 0-class thingy the usual first level interference elimination was flipping wall plugs around.
Everybody were also mainly very indifferent how possible leakage current flew in ground wires.
Ground loops were also generally much less probable since connected mains trafos were generally also much less.

But if your scope is floating and you use 10x probe it's still floating and seeing 10V+.

Earth grounding everything galvanically isolated is also a good practice.
Every now and then it becomes clear that different parts of the globe have different gut feelings.
One good example is USA and its center tapping and 208V use.
We have only one ground and correctly connected neutral is the same.
https://www.thermalcorporation.com/208-volts-is-a-weird-voltage-where-did-it-come-from/

[Terry Bites]

It depends on the wiring scheme of the ER. 
In the UK you will have a TN-C-S or TN-S or TT. In all cases way all earths must be bonded together. [Even in Finland shuko world]
If not you've got a big problem in your lab and home. Check your bonding and/ or get an electrican to test and certify your wiring- reccomended.
If you have an old consumer unit get rid of it ASAP. Be safe. Also see man shed connection schemes- www.voltimum.co.uk/articles/supply-detached-outbuilding-technical

If you cant trust your earths to be at the same potential then you need to float the scope inputs via an isolation barrier.
This will get you away from dodgy earthyness kill off plenty of common mode noise. The downisde is that most off he shelf analog iso-amps will have poor frequency response and poor distortion figures.
You may well get away with an IL300 based circuit (320volt.com/en/il300-optical-audio-signal-separator-low-frequency/)

You need a decent iso-supply fror the floaty side- this needs to be low noise. Batteries are best, whack in some LiPos.



I'm trying to visualise your set up - are the scope gnd and the PC gnds not all the same?

[paulca]

The main issue is using USB power originating on a desktop PC.  With that comes a plug earth connection.  The same earth the PC is notorious for dumping noise onto from it's high power SMPS.  And the USB ground is also subject to all the noise from the PC motherboard/cpu and display ports.

My "proper" 100Mhz scope arrived yesterday and I did spent about 15 minutes browsing around my general noise with the probe looped back on itself.  It has all manor of stuff in there.  From 50Mhz stuff to big inductor ring pulses from SMPSs.

Interesting, what the "proper" scope does not show is any 50Hz artefacts because it's earth referenced I assume.  In fact the different between connecting the ground to the USB and not connecting it is a tiny little bit of noise.  Again I assume because, technically they are already grounded(shared ground), just the long way around via the plug socket.

It's MUCH more alarming looking at it from the floating scopes point of view.  For example.  If I connect the ground to the breadboard, the (cheap) scope literally, audibly pops and whines.  If I then hold the probe tip in my finger it shows me a noise modulated 50Hz triangle wave with a P2P of over 20V!  (probe set correctly at 10x).

Part of me is wondering if the issue is not the ground and the USB noise, but the floating solar system powering the "floating" parts of the lab.  Anything that runs off 12V or less runs off that.  It's not earthed.  It's floating in the breeze including 10 meters of 6mm cable outside the house and a solar panel sitting outside just asking for electrical storm activity.  I probably shouldn't be connecting "that" to a breadboard!

[paulca]

Confirmed with the multi meter.  Me and the solar panel system are floating apart by 8V RMS at 50Hz. 

It can't have much charge though as even trying to measure it directly with the meter between Solar GND and House Earth the voltage rapidly drops to 0, even though it look like it might spike a bit, it quickly zeros.  It immediately floats back up to the same relative "to me" voltage though.

Maybe I should just go now and connect the - pin of the battery to an Earth pin in the plug socket beside it.  Just to keep it referenced somewhere sane.  I somewhat worry that it might not be 8V it's floating at all the time.  An electrical storm could make it float a lot lower/higher.

Is the downside of doing so not going to be a wealth of noise on my lovely quiet battery?

I always get confused when multimeters and "field effects" are involved.  Sometimes you can just be reading an anomaly caused by the meter itself or your own bodies electric fields ... or your body being an antenna for the mains 50Hz induced field.

I never know if it's lying to me of if I'm measuring correctly.

I am NOT going to check any of this with my brand new scope though.  I figure I "could" run a current test and scope the response from a resistor put between Earth and the Solar panel battery "-".

Looked at another way, I have a whole bunch of metal, including an entire cable up the side of the house.  It is obviously subject to the electrical field from all the mains wires in the house and neighbourhood.  Just like most things.  As it's not grounded on any end however it just sits and bounces along to whatever that field is currently doing.  The voltage on the end relative to ground is probably subject to change with the weather and how much power is being run through the house cables. 

If that is the case, then it's unlikely to float very far and probably only has as much current as you might expect between neutral and a local earth.  Next to none. 

The fact that me and the panel are floating differently and it's forming a triangle wave is most likely after it's been subject to all manor of phase delays along the cables compared to my relatively discrete location in said field.

I'm less sure about what happens when the local 50Hz mains field is no longer the prevalent field, such as if a thunderstorm rolls in overhead or even the static in a snow storm.  Just how far will a floating DC setup drift from earth?  It's not grounded.  Panel is sitting on a wooden table, so is the battery and charge controller.  Said wooden table is of course soaking wet.  It can't float far surely.  Not far enough to be dangerous?  Unless it's literally got a lightening streamer off it or takes a direct hit it should be safe, right?

[David Hess]

Even if the grounds all come from a single outlet, or even better a single power strip, capacitive coupling through the Y class interference suppression capacitors will couple high frequency currents into the ground affecting all devices.  That is what the oscilloscope is seeing.

Isolation transformers can help with this, but they have their own capacitive coupling from input to output and may be difficult to use properly.  A simpler but safer solution, at least as far as an oscilloscope goes, is to use differential probing so that the high frequency noise between grounds can be rejected.  This means putting up with the inherently higher noise of a differential probing solution.

An oscilloscope with galvanically isolated inputs, like those intended for off-line measurements, is a more specialized solution and preserves lower noise if that matters, but they are not very common.  Tektronix discontinued their last one and replaced it with $10k isolated probes.