How to Kill Yourself with an Isolation Transformer

[calzap]

Isolation transformers can provide a degree of safety when working with high voltage equipment (device under test or DUT).   Here, I’m defining high voltage as a voltage as one that can give a harmful shock via dry skin contact.  However, the safety provided by isolation transformers can be easily defeated.  Worse, isolation transformers can nullify the benefits of  better protection … a GFCI/RCD-protected power source.

It’s easy to re-reference an isolation transformer output to ground via test equipment … with potentially lethal results.  For example, connect one side of the transformer output to the DUT and connect the ground lead of a mains-powered oscilloscope to a part of the DUT that has low resistance to where the transformer output is attached.  Attach the other output of the transformer to a part of the DUT that has a high resistance to where the first output is attached.   Pay your life insurance, and touch this second part of the DUT with one hand and something grounded with the other.

The diagram and pictures below illustrate the scenario.  In the pictures, there is a light bulb instead of a body.  I do not recommend doing the setup in the pictures unless you are very familiar with working with mains power.  In the pictures, you are the light bulb.  In the first picture, the DUT, a plug cord in this case, is connected to the 120 VAC output of an isolation transformer.  Your left hand (the white lead of the bulb socket) is touching something conductive connected to one side of the isolation transformer.   Your right hand (black lead) is touching a part that is grounded … the red clips, green and then black wire go to the ground pin of the orange plug in the receptacle.  What could go wrong   … you’re isolated   … right?  Wrong in this case!  The other side of the transformer secondary is also grounded via the oscilloscope.

The isolation transformer in the picture was plugged into a GFCI-protected socket.  The GFCI didn’t trip because there was no ground fault that it could detect … it was isolated from the fault by the transformer.  The GFCI receptacle on the isolation transformer secondary provides little extra protection.  Something would have to go seriously wrong inside the transformer for it to trip.  Its ground socket is not connected to anything.

The scope was plugged into a the same GFCI-protected receptacle as the transformer.  So why didn’t that trip the GFCI?  It didn’t because current was passing through the scope ground chassis to the ground line of the cord and pin.  From there, it passed into the ground contact of the receptacle and into mains wiring ground.  A GFCI monitors hot and neutral currents and trips if they aren’t the same; it does not monitor current in ground wires directly.

In the second picture, the DUT is plugged directly into an outlet protected by a GFCI breaker.  The bulb is not lit.  Why?  Because the GFCI tripped.  No shock to the body here!

If the DUT was plugged directly into a functioning GFCI/RCD, a real body might feel a brief tingle before power was removed.  I experienced this when a garage door installer put a lag screw into a mains cable.  I’ve seen GFCI receptacles fail live.  Rare, but it happens.  That’s why they have test buttons.  When’s the last time you tested yours?  I thought so … you’re like me.     But yeah, a DUT plugged directly into a functioning mains-powered GFCI is shock-safer with regard to grounding than using an isolation transformer.  All my shop receptacles are GFCI-protected.
 
So, will GFCI protect test equipment against a ground fault?  With regard to starting a fire or shocking the operator … yes.  However, GFCI’s operate very quickly, but not instantly.  It might be possible for a ground fault to cook some semiconductors before power is removed.  Battery powered test gear is best for high voltage work.  However, some battery powered instruments are not designed for high voltage use, especially some oscilloscopes.  For mains-powered oscilloscopes, differential probes or probe isolators are recommended.  And plug any mains-powered test gear into GFCI/RCD-protected receptacles.

Now, all you have to do to get shocked is touch two hands to high voltage parts of a DUT.

Mike

[TrickyNekro]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

Most test equipment anyhow is isolated, apart from the ground reference they have for the signals.
The only thing that one would lose with isolating the test equipment, is maybe the power supply
fault detection.

The load is almost always higher power than the test equipment. To have the RCD where the energy and
a mistake could kill you, is most probably the better thing to do. In the end we have isolated test equipment
and they are called multimeters. 

Or use isolated probes. Not the Differential ones, with the 10MΩs. Isolated.

[Benta]

I think there's a Darwin award lurking here.
Why on earth (pun intended) would you want to touch the high voltage secondary end of an isolation transformer when the other end is grounded?
I mean, if you have a 500 V battery, you don't touch both ends at once (unless you're a moron).

[Benta]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

NEVER!
That will get you killed.

[CaptDon]

I think most experienced technicians were aware of this type of fatal flaw just as we / they are aware that even with an isolation transformer if you happen to be working on a Carver PM-2.0T that incorporates both a + and - 130VDC power rail with a kilowatt of power available you can still be killed and D.C. doesn't like to let go. I guess the real problem comes through the rumor mill that connecting your unit under test to an isolation transformer protects you in every imaginable situation. It seems many noobs get their information 3rd or 4th hand from folks who weren't very electrically savvy in the first place. It is amazing on Iceland folks working with 240vac may tend to take chances working a hot circuit but when faced with 240vdc which exists in some areas they take insane amounts of precautions!!! A 240vdc arc flash will sustain forever and the plasma ball can get huge when copper vapor is involved. An incredibly bigger fireball than a 240vac arc flash.

[tggzzz]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

No. Never Never Never.

Read and understand the safety references and anecdote at https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/

Consider why someone did this...

Or use isolated probes. Not the Differential ones, with the 10MΩs. Isolated.

Using an HV differential probe is a much better starting point.

[coppercone2]

man I renember playing blue shift (half life security guard episode) a long time ago thinking it was a sick joke where it tells you that protecting the black mesa facility equipment is first or second priority objective and personal safety is tertiary.

"Low Priority : Personal Safety"

 I guess its true. That 200$ clapped out ebay oscilloscope is that important. 

what happened to it being cool to destroy malevolent machines, like captain kirk?

Like even at work near the 200k oscilloscope, its just a box.... no one really cares about it. literally like a flashlight will get more attention because it does useful immediate work for humans. Everyone knows its marked up as hell because its cutting edge and it bearly does anything better then the other thing and its only good for advertising that the tests are done 'diligently'. May be useful for precision on extreme manufacturing scales to give people better 'projections' for making decisions that involve millions of units  being made with theory based design changes to meet deadlines. If you need it, its considered cheap for you. if you don't need it, its just breaking some expensive luxury good. luxury. aka that thing you might play with if you have free time. not necessary.


get a cheaper scope that you can afford to break ....... its common sense stuff

[TrickyNekro]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

No. Never Never Never.

Read and understand the safety references and anecdote at https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/

Consider why someone did this...

I do get why someone did this. Every metal of the oscilloscope will be referenced to whatever point one is probing with the ground lead. But you would be essentially "protected" by the same RCD as the load.
And you can definitely get some very very sensitive RCDs, that you can use extra for your specific test bench. There are also 6mA RCDs.

If you feel you need even more protection, then in comparison to the year old tek scope there, you can remote control most scopes through ethernet nowadays.
The 1500V isolation should be enough for most mains powered applications, one would guess. Probe with the power down, control remotely.

And you probably should be wearing gloves or similar if you have anything in your load that can dump some serious energy on you. No other protection will save you if you are probing capacitor banks.

And there are isolated scopes also, those seperate the ground at the inputs. But cost a pretty penny.

So unless you are working with motors, which tend to leak some residual current and you anyhow should be very careful and use differential / isolated probes, well you can get away with a lot.
And you are going to need differential / isolated probes, if you are trying to probe different parts at different potentials anyhow. ( for example you are probing the driver of a Motor ).

Edit: As a side note, if you use ethernet, take extra care that the cable is unshielded.

[TrickyNekro]

Or use isolated probes. Not the Differential ones, with the 10MΩs. Isolated.

Using an HV differential probe is a much better starting point.

Depends on what you are probing. If you are probing sensors along side with power points then the ~60 - 30uA could potentially throw your measurements off, or your system big time.

[TrickyNekro]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

NEVER!
That will get you killed.

Then why is Tektronix offering battery powered oscilloscopes for "floating oscilloscope measurements"?

I do get that every exposed oscilloscope metal part will be out to kill you, but this is actually practiced.

Edit: No Seriously, unless I am missing something big here... They say in their webpage that using isolation transformers
is dangerous and can kill you and all the nice stuff, but then tell you that using a TDS3000 series oscilloscope is ok, just use
with caution... just a bit of caution, it´s all ok, it uses batteries and everyone is happy... just a bit of caution...

Damned if I ever seen some legal speech and marketing speech in my life combined...

And again, unless I am missing something big here, I can not really take all this seriously. Use caution with our exposed metal
but dare you not buy our certified product, you will die a horrible death... I mean come on.

Either understand what you are doing and use generally caution, because a mistake could seriously cost your life, or this is all marketing for handheld oscilloscopes.
I do seriously stand to be corrected here.

Edit No.2: And I do understand that isolation transformes leak themselves and in some scenarios that could again be dangerous or messy for the readings. As always understand what you are doing
and use much more than needed caution. It is the "marketing double speak" that drives me a bit nuts here.

[coppercone2]

that is a qualified oscilloscope

but its still not as good as active probes. because channels are common ground. differential active probes = full isolation on every channel.

[TrickyNekro]

but its still not as good as active probes. because channels are common ground. differential active probes = full isolation on every channel.

Differential probes are NOT isolated probes. They might be good enough for some application probes, but they are not isolated.

Try to probe different differential sensors with them and you´ll see ( or won´t see ). Depends on the system. But they are definitely NOT isolated.

Maybe there are some models that are, but normally they have a 5MΩ path to oscilloscope ground.

[coppercone2]

that is pretty good isolation. You probobly wont blow up anything with 28 microamps at line voltage

[johansen]

How many of you all drive home from work slightly tired?

Now, all you have to do to get shocked is touch two hands to high voltage parts of a DUT.

no shit sherlock.

all you have to do is nod off at the wheel.

happens pretty often.

[ejeffrey]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

NEVER!
That will get you killed.

Then why is Tektronix offering battery powered oscilloscopes for "floating oscilloscope measurements"?

I do get that every exposed oscilloscope metal part will be out to kill you, but this is actually practiced.

Edit: No Seriously, unless I am missing something big here... They say in their webpage that using isolation transformers
is dangerous and can kill you and all the nice stuff, but then tell you that using a TDS3000 series oscilloscope is ok, just use
with caution... just a bit of caution, it´s all ok, it uses batteries and everyone is happy... just a bit of caution...

Damned if I ever seen some legal speech and marketing speech in my life combined...

And again, unless I am missing something big here, I can not really take all this seriously. Use caution with our exposed metal
but dare you not buy our certified product, you will die a horrible death... I mean come on.

Either understand what you are doing and use generally caution, because a mistake could seriously cost your life, or this is all marketing for handheld oscilloscopes.
I do seriously stand to be corrected here.

A battery powered oscilloscope is designed for that use.  That's a different story than floating a scope designed to be grounded.

Also, many battery powered oscilloscopes are designed for automotive applications, and they are intended to be grounded to the vehicle chassis.  They aren't necessarily intended to be floated at a high voltage.

[coppercone2]

How many of you all drive home from work slightly tired?

Now, all you have to do to get shocked is touch two hands to high voltage parts of a DUT.

no shit sherlock.

all you have to do is nod off at the wheel.

happens pretty often.

these people should join the army it pays better then trying to protect a scope if you don't mind the risk


this way the government pays for a new scope and you don't have rent. risk management. put the risk else where

[TrickyNekro]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

NEVER!
That will get you killed.

Then why is Tektronix offering battery powered oscilloscopes for "floating oscilloscope measurements"?

I do get that every exposed oscilloscope metal part will be out to kill you, but this is actually practiced.

Edit: No Seriously, unless I am missing something big here... They say in their webpage that using isolation transformers
is dangerous and can kill you and all the nice stuff, but then tell you that using a TDS3000 series oscilloscope is ok, just use
with caution... just a bit of caution, it´s all ok, it uses batteries and everyone is happy... just a bit of caution...

Damned if I ever seen some legal speech and marketing speech in my life combined...

And again, unless I am missing something big here, I can not really take all this seriously. Use caution with our exposed metal
but dare you not buy our certified product, you will die a horrible death... I mean come on.

Either understand what you are doing and use generally caution, because a mistake could seriously cost your life, or this is all marketing for handheld oscilloscopes.
I do seriously stand to be corrected here.

A battery powered oscilloscope is designed for that use.  That's a different story than floating a scope designed to be grounded.

Also, many battery powered oscilloscopes are designed for automotive applications, and they are intended to be grounded to the vehicle chassis.  They aren't necessarily intended to be floated at a high voltage.

Then I´ll go ahead and mention all the above sources:

Here is the Tektronix blog / post / website:
https://www.tek.com/en/documents/technical-brief/floating-oscilloscope-measurements-and-operator-protection

Here is the teardown by Dave:


There is absolutely nothing special about that scope, as far as I can tell, in comparison to any other low / mid range scope in the market today, apart that it can use a battery.
Exposed metal back and front referenced to the ground of the BNCs.

Somehow it´s ok to float that scope and not others. Again, I do get that most isolation transformers are leaky and that might be a concern.

But, in practice, I absolutely do not see how that differs from floating any other oscilloscope, or equipment in that case, with a low to no leak isolation transformer ( or use two, at which point just buy probes ).

Sure, don´t touch exposed metal when connected, that much is clear. The rest is legal speech.

About the automotive case:

Most of the time, you need the isolation because you need to probe different signals and buses that are either differential, like the CAN bus, or because of the long wiring of the car and different loading conditions might
sit in a bit different ground point. This is another use case scenario and has to do with measurement and not with safety ( unless I guess you are probing the battery of an EV ).


The major concern is not to touch any exposed metal when the instrument is floated and connected to a high potential. And if the potential is that high you might also not want to touch the intrument in general, at which
point you´ll have to remote control it. And you might have to remote control it even wirelessly via an accesspoint if the common mode is that high. And at which point you might already need a high voltage probe.

But the question of the OP was about household mains, as far as I understood it. Not 11kV lines...
It is much much much safer, for someone who wants to experiment with these relatively low voltages to have a very sensitive RCD with the load and have someone around to kill the power and call for help,
if anything goes wrong, than anything else ( plus some rated gloves ). And that´s because, touching any naked metal will trip the RCD and in case you touch two wires, you will not let the transform keep dumping
its energy on you. Edit: This is assuming you won´t touch two wire at the absolute exact same moment and that there is a path to earth through your body.

Isolating the load is EXACTLY the reason why so many people got injured or killed by trying to make Lichtenberg Figures on wood!!!!

Just screaming you´ll die, if you do it, is for the most part unproductive, offers no reasoning and no solution other than throwning money at a problem, while still not teaching you where exactly the problem lies.
And you better understand what you are doing, because one day the time will come, when you´ll have to play dirty to make something work.

[TrickyNekro]

that is pretty good isolation. You probobly wont blow up anything with 28 microamps at line voltage

Blow up no, but it could potentially mess up measurements of sensitive signals. Talking from experience.

And you´ll get it relatively fast when you see ac on your readings. But if it´s between signals, it will give you
a very good head scratcher...

[tggzzz]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

NEVER!
That will get you killed.

Then why is Tektronix offering battery powered oscilloscopes for "floating oscilloscope measurements"?

They aren't.

They are selling scopes that are completely designed from the ground up (pun intended) to be operated when floated, and those scopes happen to be battery powered.

"Little" things like the BNC socket locks being made from plastic, not metal.

[tggzzz]

Or use isolated probes. Not the Differential ones, with the 10MΩs. Isolated.

Using an HV differential probe is a much better starting point.

Depends on what you are probing. If you are probing sensors along side with power points then the ~60 - 30uA could potentially throw your measurements off, or your system big time.

You appear to have missed the phrase I have emphasised.

There is a reason there are many different types of probe: there are many difference measurement use cases which have different requirements and tradeoffs.

Your use case is too vague for it to be possible to comment.

[tggzzz]

but its still not as good as active probes. because channels are common ground. differential active probes = full isolation on every channel.

Differential probes are NOT isolated probes. They might be good enough for some application probes, but they are not isolated.

Try to probe different differential sensors with them and you´ll see ( or won´t see ). Depends on the system. But they are definitely NOT isolated.

Maybe there are some models that are, but normally they have a 5MΩ path to oscilloscope ground.

Why are you fixated on floating a scope and not using HV differential probes?

Your contentions are strawman arguments: nobody has suggested that one (of many) type of probe is suitable for every use case.

In that vein, I suggest that the probe-earth capacitance of a floated scope will cause problems when "probing differential sensors".

[xvr]

Isolation transformer is not a safety device. Output voltage from it the same deadly as input. It is not a silver bullet, there is ONLY usage area for them is debugging DUT that directly Mains connected (without any internal isolation).

Debugging such devices always is a dangerous business, with transformer or without. Engineer that works with such device should have appropriate qualification and experience. Isolation transformer will not help him to be more safe, it only allow not to fry up test equipment, but instead it could help him to fry up himself 

IMHO beginners should not deal with Mains connected devices at all (with Isolation Transformer or without - doesn't matter).
Unfortunately we can't stop him. As I can see in this conference they sometimes try to build some derivation of electric chair  An this is much worse than any Isolation Transformer 

[TrickyNekro]

but its still not as good as active probes. because channels are common ground. differential active probes = full isolation on every channel.

Differential probes are NOT isolated probes. They might be good enough for some application probes, but they are not isolated.

Try to probe different differential sensors with them and you´ll see ( or won´t see ). Depends on the system. But they are definitely NOT isolated.

Maybe there are some models that are, but normally they have a 5MΩ path to oscilloscope ground.

Why are you fixated on floating a scope and not using HV differential probes?

Your contentions are strawman arguments: nobody has suggested that one (of many) type of probe is suitable for every use case.

In that vein, I suggest that the probe-earth capacitance of a floated scope will cause problems when "probing differential sensors".

Dunno, that was just an example and a side discussion.

I don´t have a problem against diff probs. One should use whatever is appropriate. But sometimes, when one only has an
isolation transformer, then an RCD is cheaper than a diff prob. I would have thought.

Anyhow, the question is not whether diff probs are a problem. The question is whether or not it is ok to float equipment.

[TrickyNekro]

Maybe the best way to avoid all these problems is to simply isolate the test equipment.

NEVER!
That will get you killed.

Then why is Tektronix offering battery powered oscilloscopes for "floating oscilloscope measurements"?

They aren't.

They are selling scopes that are completely designed from the ground up (pun intended) to be operated when floated, and those scopes happen to be battery powered.

"Little" things like the BNC socket locks being made from plastic, not metal.

I pointed exactly to the sources I was referring to in a post above and the reasoning of my comments.

[joeqsmith]

I've yet to have a need for an isolation transformer at home.  As for safety, I guess the only 100% sure safe way is pay someone who is competent to do the work.