Hypothetical question about Isolated Supply, GFCI & Oscilloscope use

[bsodmike]

Hi all,

Here are the assumptions as play:

• We always float/isolate our DUT, never the other way around.
• The mains breaker has GFCI already installed; this is how it came.

Now, let's for a moment consider we have a mains supply, connected through an isolation transformer.  The HOT lead at 240VAC runs through 2x 60W bulbs in parallel, acting as current limit devices for safety reasons of course.

The oscilloscope is connected to mains-earth reference.  The grounding lead on the probe is tied to mains-earth ground through the its power lead - for this demo it's left to flap in the breeze (safely!).  We connect the probe tip-end onto the HOT (Live) wire coming off the Isolation supply.

• Would the flow of AC current from Live through the Oscilloscope to ground cause the GFCI to trip?
• Is it safe to use two probes, one connected to the Isolated HOT lead and the other on the Isolated Neutral lead; we can then use the MATH function to provide a computed signal of the mains voltage

Of course, isn't the best approach here to use isolated differential probes for taking mains AC readings on an oscilloscope?

Thanks!

[bdunham7]

You should be more clear in your terms.  Not that there are rigid definitions for terms like 'HOT', but you should be careful to be clear.

HOT would typically mean a mains circuit with a significant potential to ground, or perhaps in some areas isolated from ground.
NEUTRAL would mean a mains circuit current-carrying conductor with a potential near ground.  Think of it as a return to ground, because that is how it is often wired. 

Depending on your situation, one or both of your incoming mains wires could both be HOT.

ISOLATED means isolated and there is no hot or neutral unless you ground one side, either accidentally or deliberately.  If I understand you correctly, you are using  a normally connected scope to measure the outputs of an isolation transformer that is not otherwise connected--is that right?  In that case, your single probe measurement will only get you noise, not a reliable reading.  It could be anything, but you'll likely see your mains frequency, probably distorted at some voltage that I can't easily guess.  The mains GFCI will not trip, unless perhaps you use a 1X probe with a 50R scope input and your isolation transformer leaks very badly or is under water. 

Two probes and MATH is the same principal as a differential probe and will work fine, only limited by how well your scope does the math.  However, the point (one of them anyway) of the isolation transformer is that you now can somewhat arbitrarily select your ground point and measure from there.  So if you simply grounded one side of your isolated output, you can now measure the other side directly.  Of course, this defeats the touch safety feature of isolation.  You can't have both at the same time.

[David Hess]

Would the flow of AC current from Live through the Oscilloscope to ground cause the GFCI to trip?

No, because the isolation transformer on the DUT prevents any imbalance of hot and neutral currents.

Is it safe to use two probes, one connected to the Isolated HOT lead and the other on the Isolated Neutral lead; we can then use the MATH function to provide a computed signal of the mains voltage

Yes but ... since the DUT is isolated, its common mode voltage can drift around which may not provide the best results as it will subtract from the input voltage range of the probes.  Since the DUT is isolated, the probe grounds can be connected to any single point on the DUT.

The reason I use an isolation transformer is to that I can choose my own single point ground on the DUT, which is where the probe's ground clip goes.  Without the isolation transformer, this point is chosen for you where ground and neutral are tied together which is typically back at the circuit breaker box.

[nctnico]

Of course, isn't the best approach here to use isolated differential probes for taking mains AC readings on an oscilloscope?

Yes it is!

Don't think about using isolation transformers. In order to use these safely you need your entire bench to be setup properly and you'll need training as well.

With a differential probe you keep the potential danger area limited to the DUT itself so it is much easier to identify what is dangerous to touch and what is not.

[bsodmike]

You should be more clear in your terms.  Not that there are rigid definitions for terms like 'HOT', but you should be careful to be clear.

HOT would typically mean a mains circuit with a significant potential to ground, or perhaps in some areas isolated from ground.
NEUTRAL would mean a mains circuit current-carrying conductor with a potential near ground.  Think of it as a return to ground, because that is how it is often wired. 

Depending on your situation, one or both of your incoming mains wires could both be HOT.

Cheers for the heads up, I was using Live/Hot interchangeably, but that doesn't apply in the Isolated scenario.

ISOLATED means isolated and there is no hot or neutral unless you ground one side, either accidentally or deliberately.  If I understand you correctly, you are using  a normally connected scope to measure the outputs of an isolation transformer that is not otherwise connected--is that right?  In that case, your single probe measurement will only get you noise, not a reliable reading.  It could be anything, but you'll likely see your mains frequency, probably distorted at some voltage that I can't easily guess.  The mains GFCI will not trip, unless perhaps you use a 1X probe with a 50R scope input and your isolation transformer leaks very badly or is under water. 

Yes, exactly. Due to the galvanic isolation provided by the Isolation transformer, initially I thought that GFCI at the breaker box would not be affected, but later on I was like HMM... "What happens if I touch the probe's ground clip to one side of the Isolated output and the sensing-end of the probe to the other wire of the Isolated output"... of course at 10X probe impedance; thanks for the info on the 1X probe/50R scope input configuration.. interesting!!

Two probes and MATH is the same principal as a differential probe and will work fine, only limited by how well your scope does the math.  However, the point (one of them anyway) of the isolation transformer is that you now can somewhat arbitrarily select your ground point and measure from there.  So if you simply grounded one side of your isolated output, you can now measure the other side directly.  Of course, this defeats the touch safety feature of isolation.  You can't have both at the same time.

"Of course, this defeats the touch safety feature of isolation." -- Do you mean that the Oscilloscope frontend circuitry ground-points (especially the BNC connectors) are raised to a higher potential, and this could be a hazard or...

[bsodmike]

Don't think about using isolation transformers. In order to use these safely you need your entire bench to be setup properly and you'll need training as well.

All the devices on my bench are tied to mains-earth reference, through my APC UPS (which powers everything).

The isolation transformer I'm planning to use shortly (it's a DIY build), runs a Variac into an Isolation transformer.  The output is wired in series through 2x 60W lamps (in parallel) to act as current limiting devices.

Furthermore, none of the other bench devices (Oscilloscopes, Signal generator) will be floated, only the DUT will be floated by powering it via the Isolated supply.

You've mentioned a "proper" bench configuration and training; in terms of the bench setup, what else is needed, and also with regards to training, roughly what does this cover beyond the points I've raised?

[bdunham7]

of course at 10X probe impedance; thanks for the info on the 1X probe/50R scope input configuration.. interesting!!

Just so we're clear, that was intended as an example of improper setup.  What I was getting at is that your GFCI won't trip when you probe the circuit anywhere with a 10X probe because the probe impedance is too high.  With isolation, you have two separate reasons the GFCI won't trip.

"Of course, this defeats the touch safety feature of isolation." -- Do you mean that the Oscilloscope frontend circuitry ground-points (especially the BNC connectors) are raised to a higher potential, and this could be a hazard or...

No, sort of the opposite. Floating the scope can result in unsafe touch voltages on the scope. If the DUT is floating, then the scope stays grounded.  When people tell the cautionary tale of a skilled engineer being killed this way, apparently they are referring to a situation where this person ungrounded the scope and connected to some sort of high voltage, then was electrocuted while operating the floating scope.  You actually can float a scope or a signal generator b/t/w, but I'd only consider it when you are dealing with lower voltages, like the output of an audio amplifier. 

With your isolation transformer, you are theoretically safe if you touch either output (but not both) if it isn't connected to anything because there isn't any ground reference. The other side from where you touch has very high impedance to ground and very little current can flow through you.  However, if you ground one side, the other side can now zap you because there is a complete path.  So if you want to maintain touch safety and take a measurement, you have to use a method with at least two leads but without a low-impedance path to ground anywhere--and that could be a device with completely isolated, floating input section like a typical DMM or isolated scope, a handheld battery-operated device, etc, or you can use a system that is not truly isolated but has a path to ground that is of high enough impedance that the leakage current won't shock you.  This can be two separate scope probes without any ground leads, or it could be a differential probe.

[nctnico]

Don't think about using isolation transformers. In order to use these safely you need your entire bench to be setup properly and you'll need training as well.

All the devices on my bench are tied to mains-earth reference, through my APC UPS (which powers everything).

The isolation transformer I'm planning to use shortly (it's a DIY build), runs a Variac into an Isolation transformer.  The output is wired in series through 2x 60W lamps (in parallel) to act as current limiting devices.

Furthermore, none of the other bench devices (Oscilloscopes, Signal generator) will be floated, only the DUT will be floated by powering it via the Isolated supply.

You've mentioned a "proper" bench configuration and training; in terms of the bench setup, what else is needed, and also with regards to training, roughly what does this cover beyond the points I've raised?

Basically floating everything and constructing your bench and furniture from non-conductive materials.

If you connect a grounded scope to a floating DUT the DUT becomes grounded again at and it is uncertain at which points the DUT carries dangerous voltages. Worse; the GFCI will be defeated as well in this situation. Clip the ground leads of your probe to two points in the DUT; you might short something. Buying CAT rated differential probes will be cheaper and safer to work on mains voltages. Get rid of the isolation transformer if you love your life.

[bdunham7]

Basically floating everything and constructing your bench and furniture from non-conductive materials.

Seems a bit over the top, the sort of thing that happens when one lets the imagination run wild.  Why is it OK not to do this without an isolation transformer?

Get rid of the isolation transformer if you love your life.

Rather than debate the various technical arguments that have been thrashed out before, let me ask this.  Are you aware of any instance, ever, anywhere in the history of the entire world, where someone was killed while using an isolation transformer on a DUT?  I'll settle for anything, even if their bench was outside in the rain, or they caught a stray bullet, or even if the user was on the bomb squad and the DUT was a bomb.

[bsodmike]

If you connect a grounded scope to a floating DUT the DUT becomes grounded again at and it is uncertain at which points the DUT carries dangerous voltages. Worse; the GFCI will be defeated as well in this situation. Clip the ground leads of your probe to two points in the DUT; you might short something. Buying CAT rated differential probes will be cheaper and safer to work on mains voltages. Get rid of the isolation transformer if you love your life.

You've definitely got loads more experience than me, but I certainly would be uncomfortable plugging in a faulty DUT into mains directly without any Isolation.  I'd much rather have an Isolated supply and use that with caution (60W bulbs etc), when probing at the start with DMMs.  What I like about this approach is that any shorts cause the bulbs to turn on fully, and even if you intentionally (or otherwise) cause a short, they just go on fully with minimal sparking at all. 240VAC at max ~1A goes through (see YouTube link below):

https://youtu.be/51mjt9nFoeA?t=526


For any Oscilloscope work: I agree to the point that differential probes are the way to go, so I'm taking your advice one that front