Confusion around probing isolated circuits. Is this a safe practice?

[KungFuJosh]

Exactly. Some idiot on youtube recommended using an isolation transformer so that a standard probe could be used as a diff probe by connecting the ground lead to a different part of the circuit instead of ground.

What's wrong with that?

One of the reasons isolation transformers are dangerous is because inexperienced (and sometimes experienced) people will do something stupid with their test setup, at which point, the user may become the path to ground. Ooops.

[EEVblog]

uA of current is not going to damage the scope, if you ground 120V with your scope probe, its going to blow something up though. Possibly damage the scope, definitely the probe. Probably y cap leakage as you say.

Yes, scopes are very rigidly grounded to mains earth, just look at any oscilloscope teardown video. The BNC is often bolted the metal chassis, so it's crazy low impedance. The scope probe ground lead is the thing that vaporise first.

[EEVblog]

- The scope is and remains grounded.
- And differential probes are used for measurements.

Exactly. Some idiot on youtube recommended using an isolation transformer so that a standard probe could be used as a diff probe by connecting the ground lead to a different part of the circuit instead of ground.

What's wrong with that?

Becaause it can be dangerous, especially with say measuring mains circuits (a common reason for doing this). It means that the entire metal chassis of your scope will be at a live potential relative to ground, and if you touch it, ouchy.
Oscilloscopes are mains earth referenced for safety reasons. It's why USB powered scopes like the new Rigols for example, come with a safety earth clip so you can ground it.

[AmeliaBuns]

uA of current is not going to damage the scope, if you ground 120V with your scope probe, its going to blow something up though. Possibly damage the scope, definitely the probe. Probably y cap leakage as you say.

Yes, scopes are very rigidly grounded to mains earth, just look at any oscilloscope teardown video. The BNC is often bolted the metal chassis, so it's crazy low impedance. The scope probe ground lead is the thing that vaporise first.

Huh that's cool! I didn't realize those scopes are USB powered! It can act as a portable one in a pinch I guess Good for very small workspaces like mine. Although in the case of USB powered scopes, I am slightly confused why grounding would help with safety (Although I can see how it'd help with noise/measurement). Unless the USB adapter isn't isolated, the other side is low voltage anyways. I thought transformers are dangerous because the other side is also still high voltage. Most USB charger's I've seen do have horrendous earth leakage tho.

But as for my original question:

The confusion for me has been two things so far. the strange spark I saw once, and what the range or "safe practice" is here. Is it okay to do? I'm confused as why arcs would form at only 100uA.

I keep hearing online about how you can isolate your DUT to avoid spending money on a DIFF probe that costs a ton and has a lot of noise and low bandwidth. Maybe I can get away with it with another layer of isolation but I also love my scope so much and 600-800CAD is a LOT of money to me. Don't want to see my precious investment smoking! Or worse, become inaccurate/malfunction without me knowing! I assume the device has protection built-in but I rather not rely on that :sweat:. I'm kind of a semi-newbie and I don't wanna do anything stupid...

[shapirus]

- The scope is and remains grounded.
- And differential probes are used for measurements.

Exactly. Some idiot on youtube recommended using an isolation transformer so that a standard probe could be used as a diff probe by connecting the ground lead to a different part of the circuit instead of ground.

What's wrong with that?

Becaause it can be dangerous, especially with say measuring mains circuits (a common reason for doing this). It means that the entire metal chassis of your scope will be at a live potential relative to ground, and if you touch it, ouchy.
Oscilloscopes are mains earth referenced for safety reasons. It's why USB powered scopes like the new Rigols for example, come with a safety earth clip so you can ground it.

Well no, we're talking about a different scenario here, if I understand correctly:

- the scope is grounded, meaning, chassis connected to mains PE;
- DUT is powered via an isolation transformer;
- the probe's ground clip is attached to an arbitrary node in DUT.

Apart from seeing a tiny spark produced from the capacitance between the transformer's windings when the ground clip is attached, what else can go wrong in this setup? Obviously, assuming that we don't try to connect another probe's ground to a node with a different potential in DUT.

[AVGresponding]

Looking at the PSU you have, I would check the continuity between the earth pin on the power plug, and the green "GND" post on the front. If it's good (low ohms, less than 1), I would short the -ve (black) and GND (green) binding posts together. If it's not good (more than a couple of ohms), I would get a different PSU, assuming the internal GND connections are not loose (I have seen this).

If you have a capacitance meter, you could remove the Y capacitor and check it's not failing, or swap it for a smaller capacitance Y cap. A smaller capacitance value would mean a lower leakage current. There's also a non-zero chance that the "Y" cap in the PSU isn't Y rated, in which case, replace it with one that is.


EDIT: You might stalk the Buy/Sell/Wanted section of this forum, to see if you can find a decent used PSU in your area.

huh there's a buy sell section?
Is this even dangerous? as I was mentioning to another poster, the currents are only 96uA. tho they can add up.

The ground pin is probably good, but it comes as isolated by default? I'm assuming that they wanted to give you the option to earth it or not on your end? Siglent SPD1168X is also floating(you can just short the green and black). If using an isolated power supply as a substitute for a dif probe is an okay practice, I'd much rather keep my lab bench PSU isolated. I'm actually wondering if those siglent ones are even isolated or not.

Yep:  https://www.eevblog.com/forum/buysellwanted/

Although the leakage current (probably) isn't dangerous to you (I have no idea if you have a medical condition that increases your sensitivity), it will add noise to your measurements, and every time you connect to your scope, the Y capacitor will dump all its charge through your scope ground, and every time you disconnect, you'll have a tiny arc, which has the potential to damage your scope input. Connecting the PSU -VE to GND on the front panel literally short-circuits this problem, protecting your scope front end, and reducing the noise of your measurements.

In the context of what you are doing, your PSU stops being isolated as soon as you connect your scope probe ground, so you may as well do it at the source, and save the extra stress to the scope.

All PSUs that are even half decent will have the option to connect to ground; my Siglent SPD3303X does, my HP and Agilent ones do, my TTi EX354D does, and the various EP PSUs I use for HV stuff come with -VE linked to GND by default.

oh nooo! I didn't know it's dangerous. hope I didn't ruin my expensive precious scope D: I'm always scared of that. I spent so much on it. (And it doesn't even have 50 ohm inputs!)
Being honest. I'm more worried about my scope. Wounds heal, scopes won't

Why does the arc happen btw? would a TVS diode trying the negative/positive terminal to earth help prevent this issue?
I've been trying to use this to my advantage to probe across a resistor as a shunt or to measure something relative to VCC not GND. I guess i'll stop doing that? I'm debating DIY-ing one of those differential probes. although I wonder how well they really work....
I also wish I had a power rail probe. those are so much more expensive, more so than even differential probes

The arc happens because as you disconnect the scope ground lead, the point it was connected to is no longer shorted to ground, and the AC current flowing through* the Y capacitor allows the potential to rise to a level that an arc occurs.


*technically the current doesn't flow through the capacitor, but that's a different question.

- The scope is and remains grounded.
- And differential probes are used for measurements.

Exactly. Some idiot on youtube recommended using an isolation transformer so that a standard probe could be used as a diff probe by connecting the ground lead to a different part of the circuit instead of ground.

What's wrong with that?

Becaause it can be dangerous, especially with say measuring mains circuits (a common reason for doing this). It means that the entire metal chassis of your scope will be at a live potential relative to ground, and if you touch it, ouchy.
Oscilloscopes are mains earth referenced for safety reasons. It's why USB powered scopes like the new Rigols for example, come with a safety earth clip so you can ground it.

Well no, we're talking about a different scenario here, if I understand correctly:

- the scope is grounded, meaning, chassis connected to mains PE;
- DUT is powered via an isolation transformer;
- the probe's ground clip is attached to an arbitrary node in DUT.

Apart from seeing a tiny spark produced from the capacitance between the transformer's windings when the ground clip is attached, what else can go wrong in this setup? Obviously, assuming that we don't try to connect another probe's ground to a node with a different potential in DUT.

The spark isn't going to be from capacitive coupling in the transformer, that's going to be far too small to be noticeable in this way at the voltages we're talking about. It's going to be from the Y capacitor, connected between a live mains conductor (should be the neutral when using a 3-pin plug; with a 2-pin, all bets are off) and the 0V rail of the secondary circuit. It's not unusual to see 1-2mA via this route, though currents in the few hundred uA range are more common.

It's not a great idea to probe this way, because it's far too easy to make a mistake and connect another ground at a different potential.

[EEVblog]

- The scope is and remains grounded.
- And differential probes are used for measurements.

Exactly. Some idiot on youtube recommended using an isolation transformer so that a standard probe could be used as a diff probe by connecting the ground lead to a different part of the circuit instead of ground.

What's wrong with that?

Becaause it can be dangerous, especially with say measuring mains circuits (a common reason for doing this). It means that the entire metal chassis of your scope will be at a live potential relative to ground, and if you touch it, ouchy.
Oscilloscopes are mains earth referenced for safety reasons. It's why USB powered scopes like the new Rigols for example, come with a safety earth clip so you can ground it.

Well no, we're talking about a different scenario here, if I understand correctly:

Correct, I was making a generic point about scope safety because KuFuJosh was also talking generally.  "floating the scope" is an age old technique that can be dangerous, even deadly.
If someone doesn't understand it, they they certainly should not be encouraged to use that technique.

Proper battery powered scopes avoid this safety issue because they are supplied with special insulated BNC connectors and ground clip leads, so you can't accidently touch the ground system.

[rf-messkopf]

The confusion for me has been two things so far. the strange spark I saw once, and what the range or "safe practice" is here. Is it okay to do? I'm confused as why arcs would form at only 100uA.

I keep hearing online about how you can isolate your DUT to avoid spending money on a DIFF probe that costs a ton and has a lot of noise and low bandwidth. Maybe I can get away with it with another layer of isolation but I also love my scope so much and 600-800CAD is a LOT of money to me. Don't want to see my precious investment smoking! Or worse, become inaccurate/malfunction without me knowing! I assume the device has protection built-in but I rather not rely on that :sweat:. I'm kind of a semi-newbie and I don't wanna do anything stupid...

As others have said, a couple of hundreds of µA leakage current into your scope ground are not going to hurt it, so you are doing okay here.

But: If you measure an isolated low-voltage DC powered DUT fed from your power supply, the leakage current from your power supply will flow through your probe's ground lead to the scope, and though its PE connection back to mains earth. Since the ground lead has a finite impedance, the scope and DUT ground potential can become slightly different. This may affect sensitive measurements and you may see a mains ripple superimposed on the scope trace (but we are talking microamps and milliohms here, so this is generally not a big problem). To avoid this you can try the following:
- Add another low impedance connection between your DUT and scope ground that carries most of the leakage current.
- Ground one output of your power supply, usually the negative one (this assumes that the negative output is you DUT ground).

Another issue, not related to mains leakage, is that the ground lead of your probe has quite a large inductance which can affect the measurement of fast signals with a large bandwidth, such as fast rising edges. For these signals you will want the impedance between the ground ring of your probe and the DUT ground as low as possible, e.g. by using a ground spring.

And again: never isolate your safety class I test equipment (scope and power supply) with an isolation transformer (unless you know exactly what you are doing). Their safety relies on their cases being connected to the PE of your installation.

[watchmaker]

- The scope is and remains grounded.
- And differential probes are used for measurements.

Exactly. Some idiot on youtube recommended using an isolation transformer so that a standard probe could be used as a diff probe by connecting the ground lead to a different part of the circuit instead of ground.

What's wrong with that?

Becaause it can be dangerous, especially with say measuring mains circuits (a common reason for doing this). It means that the entire metal chassis of your scope will be at a live potential relative to ground, and if you touch it, ouchy.
Oscilloscopes are mains earth referenced for safety reasons. It's why USB powered scopes like the new Rigols for example, come with a safety earth clip so you can ground it.

Well no, we're talking about a different scenario here, if I understand correctly:

Correct, I was making a generic point about scope safety because KuFuJosh was also talking generally.  "floating the scope" is an age old technique that can be dangerous, even deadly.
If someone doesn't understand it, they they certainly should not be encouraged to use that technique.

Proper battery powered scopes avoid this safety issue because they are supplied with special insulated BNC connectors and ground clip leads, so you can't accidently touch the ground system.

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued; likely because of the issue she found) and mains, LEAVING her scope properly grounded.

In this context, user naivete is a canard.  Even a soldering iron is dangerous if the user does not know how to use it. Why is RTFM such a popular acronym in this forum?  What equipment IS safe if the user does not know how to use it?

For AmeliaBurns, I have a simple, cost effective, safe solution if she is working on circuits below 20V.  Buy a Digient Analog Discovery for $300 which provides all the TE she probably needs.  In her situation, buying a lab grade PSU may well be shooting a flea with an elephant gun.  She is not a professional.

I am surprised no one brought up ensuring the PSU is meets UL or EU safety standards.

[EEVblog]

I think KungFuJosh was the only one suggesting floating a scope. 

You're not aware we tend to go off on tangents on here?

[rf-messkopf]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued; likely because of the issue she found) and mains, LEAVING her scope properly grounded.

Floating a safety class I power supply is as dangerous as floating a scope. See reply #15. And her lab power supply has isolated outputs, the isolation is provided by the transformers you see in the pictures. The case also seems to be properly connected to PE. Of course we don't know if it is compliant to all applicable safety standards. The leakage current she has seen does not, by itself, call its safety into question.

And in Josh's defense, I don't think he said that, but I'll leave it to him to clarify. 

[shapirus]

BTW, is there a "go-to" design for a DIY HV differential probe with a reasonable bandwidth and CMRR high enough to be usable for probing generic SMPS'es?

I remember there have been quite a few threads to that effect, is there some design that would stand out in being not overly complex and good enough at the same time? Surely someone tried to build one?

[watchmaker]

I think KungFuJosh was the only one suggesting floating a scope. 

You're not aware we tend to go off on tangents on here?

I can be soooo naive. 

But seriously, especially when we are dealing with amateurs like myself, we should avoid confusion.  I remember how flummoxed I was by the correct, but confusing advice I got when I first started here.  I appreciate and support making threads a "matter of record" for future readers, but sometimes we make it so complicated.

What do people think of the Analog Discovery for people working on low voltage circuits who have not yet learned about mains voltage; and may not for a pretty long time.

[EEVblog]

BTW, is there a "go-to" design for a DIY HV differential probe with a reasonable bandwidth and CMRR high enough to be usable for probing generic SMPS'es?

I remember there have been quite a few threads to that effect, is there some design that would stand out in being not overly complex and good enough at the same time? Surely someone tried to build one?

https://www.eevblog.com/forum/projects/cheap-hv-differential-oscilloscope-probe/
https://www.eevblog.com/forum/projects/diy-mains-voltage-differential-probe-for-smps-design/
https://www.eevblog.com/forum/projects/hv-50mhz-differential-probe-(diy-proyect)/
https://www.eevblog.com/forum/projects/my-high-voltage-differential-probe-design/
https://www.eevblog.com/forum/beginners/low-cost-diy-differential-probe/

[watchmaker]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued; likely because of the issue she found) and mains, LEAVING her scope properly grounded.

Floating a safety class I power supply is as dangerous as floating a scope. See reply #15. And her lab power supply has isolated outputs, the isolation is provided by the transformers you see in the pictures. The case also seems to be properly connected to PE. Of course we don't know if it is compliant to all applicable safety standards. The leakage current she has seen does not, by itself, call its safety into question.

And in Josh's defense, I don't think he said that, but I'll leave it to him to clarify. 

Thanks for this.  I was going to offer to send her a Tenma plug and play iso until you posted this.  Thank you for the caution.

[shapirus]

BTW, is there a "go-to" design for a DIY HV differential probe with a reasonable bandwidth and CMRR high enough to be usable for probing generic SMPS'es?

I remember there have been quite a few threads to that effect, is there some design that would stand out in being not overly complex and good enough at the same time? Surely someone tried to build one?

https://www.eevblog.com/forum/projects/cheap-hv-differential-oscilloscope-probe/
https://www.eevblog.com/forum/projects/diy-mains-voltage-differential-probe-for-smps-design/
https://www.eevblog.com/forum/projects/hv-50mhz-differential-probe-(diy-proyect)/
https://www.eevblog.com/forum/projects/my-high-voltage-differential-probe-design/
https://www.eevblog.com/forum/beginners/low-cost-diy-differential-probe/

Yes, thanks, I can use the search function .

The question was whether there's a certain one that's more attractive than the rest. Will still review all of them though.

[KungFuJosh]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued

No! I stated floating the scope should NOT be done. I was referring to an idiot on youtube who suggested doing that. My primary point was to not mess around dangerously and/or ignorantly.

And in Josh's defense, I don't think he said that, but I'll leave it to him to clarify. 

Indeed. Thank you.

[AmeliaBuns]

The arc happens because as you disconnect the scope ground lead, the point it was connected to is no longer shorted to ground, and the AC current flowing through* the Y capacitor allows the potential to rise to a level that an arc occurs.



*technically the current doesn't flow through the capacitor, but that's a different question.

Wait, but unconnected, the voltage on the negative/positive terminal barely goes above 4.8Vp. That's barely enough to create an arc. Why would that happen during a disconnect? I thought it's caused by "parasitic inductance" in most devices, but 96ua is so low. Is there a way to avoid this?

But the main reasons seems to be connecting two different ground clips in places you shouldn't? or the sparks.

I usually avoid using more than one probe at a time when doing this. once another probe is connected to my scope, I make sure everything's earthed and only connected to ground.

[AmeliaBuns]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued; likely because of the issue she found) and mains, LEAVING her scope properly grounded.

In this context, user naivete is a canard.  Even a soldering iron is dangerous if the user does not know how to use it. Why is RTFM such a popular acronym in this forum?  What equipment IS safe if the user does not know how to use it?

For AmeliaBurns, I have a simple, cost effective, safe solution if she is working on circuits below 20V.  Buy a Digient Analog Discovery for $300 which provides all the TE she probably needs.  In her situation, buying a lab grade PSU may well be shooting a flea with an elephant gun.  She is not a professional.

I am surprised no one brought up ensuring the PSU is meets UL or EU safety standards.

Wait my PSU isn't discontinued, it's still up for sale! Plus, don't even the fancy power supplies have some leakage in a similar range to my my unit? How much should  I expect out of those? I didn't realize/think the leakage would be all that different. 90-100uA seems to be considered very good for leakage? What is "non-xfornered"? Sorry for all the questions I'd be tempted to see if I can modify the components to reduce this etc, but that'd be not a great idea for a beginner honestly. I do feel like I'm a bit too paranoid about mains.

I'm technically a software engineer I do embedded too. but I do electronics for fun. Although I have this terrible habit of trying to be "professional" at every hobby I do. so I read all about impedance matching and all the rules even tho I'm just designing a crappy LED Controller and worry about EMI  :-D I forced myself to learn all that in just 4 days so my head hurts a bit. Might be why I'm getting so confused, I might need a break. This crazy attitude did get me my last job, but I did get laid of so...

I might design my own active power rail probe/differential probe if I'm good enough to trust myself

I also am still surprised nobody suggested the DC/DC isolation circuit. It's used in medical equipment apparently where strong isolated is needed for vulnerable patients where apparently even 100uA can hurt them(?). That'd have to be DIY tho.

That test device is interesting, a scope, power supply and logic analyzer all in one? Still too expensive for me at this moment. I also doubt any of those functions are decent for such a compact and relatively cheap device. Although I'm a bit confused how it'd help. Is it because it contains differential inputs already?

Also sadly my current circuit is 24v, although that's only for the MSOFET and switching side, the rest is 3.3v. I do a lot of different things tho so it does vary. Most of my projects are more embedded style, but I do want to eventually do RF and maybe some analogue etc.

[AmeliaBuns]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued; likely because of the issue she found) and mains, LEAVING her scope properly grounded.

Floating a safety class I power supply is as dangerous as floating a scope. See reply #15. And her lab power supply has isolated outputs, the isolation is provided by the transformers you see in the pictures. The case also seems to be properly connected to PE. Of course we don't know if it is compliant to all applicable safety standards. The leakage current she has seen does not, by itself, call its safety into question.

And in Josh's defense, I don't think he said that, but I'll leave it to him to clarify. 

Yeah this is what confuses me. it's not a GREAT power supply, the output is a bit noisy (slightly less than I anticipated for the price/type. although my measuring was vey crappy as it was hard to use a ground spring). but it's not a god awful unit and for this specific purpose. isolation is the main thing that matters!
From my understanding, the sparks are the main thing I should remove/worry about in terms of damage as long as I don't attach multiple probes, but I'm not sure what is the best aproch. I still wonder if a TVS at slightly above my voltage (30v) would do much to help? and i'm a bit confused by how such a low voltage and current can do that (only 100uA max, 4.8Vpp AC at 60hz)

[AmeliaBuns]

BTW, is there a "go-to" design for a DIY HV differential probe with a reasonable bandwidth and CMRR high enough to be usable for probing generic SMPS'es?

I remember there have been quite a few threads to that effect, is there some design that would stand out in being not overly complex and good enough at the same time? Surely someone tried to build one?

I've seen quiet a few! Although not many of them are properly tested or succeeded well. Some of them don't do some things I've heard is good to do, but it's designed by people who surely know a lot more than I do.

https://resources.altium.com/p/differential-oscilloscope-probe-project
This one doesn't look... extremely good? https://www.instructables.com/DIY-1GHz-Active-Probe-for-Under-20/ it's an active fet tho not differential.
https://xellers.wordpress.com/electronics/1ghz-active-differential-probe/ this one didn't really work well up to 1ghz, but it still worked at lower freq? also these days 4 layer PCBs with specific layouts barely cost a few dollars. I can do 50x50mm for 2$ at
https://hackaday.io/project/167197-poor-mans-1ghz-differential-probe
JLCPCB!! Rogers is 96$ for a 2 layer 100x50mm tho, 48 for teflon.


I thought about building one of these!

[watchmaker]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued; likely because of the issue she found) and mains, LEAVING her scope properly grounded.

In this context, user naivete is a canard.  Even a soldering iron is dangerous if the user does not know how to use it. Why is RTFM such a popular acronym in this forum?  What equipment IS safe if the user does not know how to use it?

For AmeliaBurns, I have a simple, cost effective, safe solution if she is working on circuits below 20V.  Buy a Digient Analog Discovery for $300 which provides all the TE she probably needs.  In her situation, buying a lab grade PSU may well be shooting a flea with an elephant gun.  She is not a professional.

I am surprised no one brought up ensuring the PSU is meets UL or EU safety standards.

Wait my PSU isn't discontinued, it's still up for sale! Plus, don't even the fancy power supplies have some leakage in a similar range to my my unit? How much should  I expect out of those? I didn't realize/think the leakage would be all that different. 90-100uA seems to be considered very good for leakage? What is "non-xfornered"? Sorry for all the questions I'd be tempted to see if I can modify the components to reduce this etc, but that'd be not a great idea for a beginner honestly. I do feel like I'm a bit too paranoid about mains.

Here is what I get when I go the link you provided..  I never saw an item restocked when it is marked like this.

[watchmaker]

I think KungFuJosh was the only one suggesting floating a scope.  I explicitly described how to use it between her non-xformered SWMPS (which has been discontinued

No! I stated floating the scope should NOT be done. I was referring to an idiot on youtube who suggested doing that. My primary point was to not mess around dangerously and/or ignorantly.

And in Josh's defense, I don't think he said that, but I'll leave it to him to clarify. 

Indeed. Thank you.

Note the "I THINK".  This was not a definite statement.  Apologies for the ambiguity.

[KungFuJosh]

Note the "I THINK".  This was not a definite statement.  Apologies for the ambiguity.

No worries. I just don't want anybody else thinking I suggested that.

[Martin72]

Hmm so what I'm hearing overall is that even if you're circuit is isolated, just don't probe it as is without a differential probe? and that I should ground my circuits unless for some reason, it must not be grounded:?  I'm a bit confused. the 60-200ua of current going into the ground of my scope is apparently just fine, the tiny spark is the spooky part. (not sure how that's happening)

I drew a picture to illustrate what the main problem is:


As mentioned, all inputs of a scope are connected to each other (unless you have a scope where the channels are individually isolated).
However, this also means that if you want to measure with several channels, for example, but at different (separate) potential circuits, their galvanic isolation from each other is thereby eliminated.
The consequences can range from malfunctions to destruction of the circuits.
An isolation transformer is used to isolate the potential from the DUT to the mains.
A scope should not be connected via an isolation transformer.
Why use an isolating transformer on the DUT?
If the device is not grounded because it belongs to protection class II.
These devices usually have a 2-pole power plug that can be plugged into any outlet.
However, this also means that the line can then be reversed with the phase.
If you then connect the negative terminal of a probe to the phase, you short-circuit the phase to ground – not good...