How to Kill Yourself with an Isolation Transformer

[watchmaker]

I listened to those who put the utility of the CFCI over an isolation transformer.  And I grew up with isolation transformers.

One hand rule still applies.  GFCIs did not exist in the 1960s.  Polarized plugs are now standard. Differential probes are available.

I was troubleshooting a device under power and shorted a power transistor collector to ground.  The device was two prong power into a transformer.
GFCI tripped immediately and saved not only me but the device.

I am a reformed believer.

I do not know the use of iso xformers in vintage equipment today.  But I can see that they can invoke a false sense of security and I do not see the sense  in circumventing the GFCI as I think some have suggested.

[madires]

If you rely on a GFCI/RCD for your workbench make sure to get a type B or B+ to cover most fault current situations.

[pdenisowski]

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

If you mean the THS3000, it was discontinued some time ago.  We (R&S) do still manufacture and sell a battery powered, isolated oscilloscope (the RTH) - up to 600 V at CAT IV and 1000 V at CAT III

https://www.rohde-schwarz.com/us/products/test-and-measurement/oscilloscopes/rs-scope-rider-handheld-oscilloscope_63493-156160.html

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...

The Tek app note on floating and isolated measurements is, in my opinion, very well written.  I have often been asked to do a video on floating and isolated measurements, and if I ever were to do one, it would start with a huge warning / disclaimer as well (in red, all caps, bold, and underlined).

I knew someone who used to joke that the reason he studied electrical engineering instead of civil engineering is that people wouldn't die if he made a mistake.  Floating / isolated scope measurements is a good example of where that (tongue-in-cheek) statement clearly isn't always true.

[Vovk_Z]

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

Ethernet twisted pairs are isolated at both ends with high frequency isolating transformers. So, unshielded twisted pair is more safe then shielded with grounding connectors. Basically, if you don't have a shield - you don't have a problem.

[madires]

In TP terminology 'shielded' refers to a single pair and 'screened' to the whole cable with multiple pairs.

[DavidAlfa]

WARNING: Don't put your dick inside bulb sockets, you might die!!

Same as you don't die while driving, your gas oven doesn't blow your home up, you don't suffocate with plastic bags...
People learn their skills so they don't die using isolation tranformers.

Five people might die using isolation transformers every year, but hundreds from electric shock and thousands due car crashes.
Accidents do always happen, same as the 1849813th time you cut chicken, but that day the knife slipped.
Let's ban chicken and knifes! And cars, electricity...

Calm down your overthinking!

[tggzzz]

Quite right.

Also: stop wearing seatbelts, and stop telling inexperienced (little) people not to run into the road without looking.

[DavidAlfa]

Don't run downstairs in darkness with a knife in your mouth, hands in the pockets! 

[tggzzz]

Don't run downstairs in darkness with a knife in your mouth, hands in the pockets! 

Some risks are worth taking.

Is that risk (or the subject of this thread) one of those?

[bdunham7]

I was troubleshooting a device under power and shorted a power transistor collector to ground.  The device was two prong power into a transformer.
GFCI tripped immediately and saved not only me but the device.

Hmmm.  Was this power transistor on the primary or secondary side of that transformer?  Assuming secondary--since a BJT on mains would be rather unusual--it seems to me that your GFCI likely tripped due to a surge and not an actual ground fault.  (No input fuse??)  In that case, while the GFCI tripping may have been fortuitous, your bacon and perhaps the device was actually saved by the isolation transformer.  Stepdown transformers are typically isolating as well and have the same protective characteristics--breaking ground reference, limiting surge voltage and limiting fault current.  Typically if you short something on the mains side, you'll get quite a spark and physical damage to your probes even with a GFCI.  No GFCI or MCB is fast enough to protect any normal-sized semiconductor from direct mains fault currents.  The bond wires would simply be vaporized.

[watchmaker]

I have to digest this.  Thanks. I am sure I will have followup questions.

This was a power transistor regulating the sweep voltage in a BK 501A.  It was on the secondary side and I knocked it into the frame (fucking essential tremors; what a joke).  It was the GFCI that tripped, not breakers.  Fuse on unit was not blown

It sounds like I do not understand what happened.

As far as risk; most of us are not risk averse.  We would not be doing what we do.  But we DO value risk mitigation.  I refuse to work with "Pennsylvania Mountain Men" who use a 24 inch bar with no chaps or hat and wear sandals while using a chainsaw.  But I have no problem approaching a 28 inch oak that needs to be made horizontal; always with two others (one to get help, the other to take care of injuries).

It is pretty easy to tell a pro from a wannabe.  The pro wants to come home with all their fingers.  The wannabe wants to prove he is fearless.

It this offends anyone, too bad.   At 71, I have to earned the right to speak my mind.

[tggzzz]

If that offends anybody, it says lots about them and nothing about you

It is very difficult to accurately assess all possible risk/failure mechanisms. That has to be factored into plans and procedures. Unfortunately some of the reasons behind some of the more obscure procedures is.easily omitted and forgotten.

Too many people think (I use that word loosely) that if they can't recognise/understand a risk, then it doesn't exist.

Beginners think about how things work. Professionals think about how things fail.

[TrickyNekro]

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

Ethernet twisted pairs are isolated at both ends with high frequency isolating transformers. So, unshielded twisted pair is more safe then shielded with grounding connectors. Basically, if you don't have a shield - you don't have a problem.

I do think that´s what I said 
Basically it´s a reference to NOT using CAT 7 or similar cables where the outside "sleeve" of the connectors is metal
connected to each other and usually grounded, or directed to ground through a snubber.

[TrickyNekro]

Floating / isolated scope measurements is a good example of where that (tongue-in-cheek) statement clearly isn't always true.

Indeed. Generally HV can be an expensive hobby... one way or the other...

[ironcurtain]

The one corner case: what about folks doing EMC testing with a LISN for AC? What do you think is the sensible approach to isolate mains from DUT to measure EMC characteristics?

I have good filtering in place for the lab (two inline filters plus a workbench mounted 20db + 60db filter bank where all the test equipment is plugged). Recently got an isolation transformer for AC LISN use primarily. Say you want to do EMC testing of a connected ATX PSU, what would be your suggested approach?

I think this is interesting for hamradio folks too. I was recently testing an old HF transceiver that got noise coming in through a benchtop PSU, the harmonics of the switcher peppered pretty much the entire HF spectrum. Because of the cycling/spacing it was obvious, and I changed the PSU and was able to do proper tests and measure the noise floor, but someone else (or myself on a rainy day!) could easily forget and think the DUT is busted.

[thm_w]

When repairing SMPS in consumer equipment I power it from a UPS which is not plugged into the AC supply.  I keep the scope grounded.  Doesn't obviate the need to be careful.

Interesting but no different than using an isolation transformer. I don't see any advantage to it.
With isolation transformer you can add a variac to ramp up the voltage.

[m k]

With SMPS variac can be a pain.

First nothing happens and then *something* happens.

[wasedadoc]

When repairing SMPS in consumer equipment I power it from a UPS which is not plugged into the AC supply.  I keep the scope grounded.  Doesn't obviate the need to be careful.

Interesting but no different than using an isolation transformer. I don't see any advantage to it.
With isolation transformer you can add a variac to ramp up the voltage.

The advantage for me is that I have the UPS but don't have an isolating transformer. 

[johansen]

Ground the lisn to your test equipment ground.. as in actual ground.

Kinda defeats the point of an isolation transformer. Suppose you need an isolation transforner to make your dc - or + your ground point for measurements. Emc measurements in that case may be useless because the equipment is not functioning as it would in the wild

[wobbly]

I'd like to ask if someone could review the safety advice given in this video posted on this exact topic just today (1 hour ago at time of writing)...



I'm not saying it's right or wrong, I'm just trying to learn.  I like this guy's channel, his methodical approach has been useful on many occasions.

Mains power makes me very nervous which is why I want to learn more about the safety aspects.  I'd value some of your opinions on this video.

Many thanks.

[madires]

When Rich grounds the output of the isloation transformer via his lab PSU (lab PSUs usually have isolated outputs) and/or scope (probe's ground clip) the output isn't floating anymore. While probing with the scope he needs to be careful (one hand in his pocket). A better approach would be to keep the DUT floating and to use differential probes.

[Phil1977]

Can anyone explain why these discussions are still active?

HV differential probes are easily available today - no need to ground your DUT with the scope. Measurements with HV differential probe are much more trustworthy than without because of the symmetrical input of defined impedance. So if available, I´d always use this probe for everything >100V.

Isolation transformers can be an experts tool for risk mitigation but should never be seen as something that makes lethal voltages safe to touch. And at least in countries with serious mains voltage you also should never see a RCD as a personal safety device. It´s intended to switch off defect appliances, and also not to make live wires touchable. Especially if you are working with larger SMPS then the charge in the main buffer capacitor alone can really badly harm you. 

If you want to work with these devices:
- Earth all equipment that has an PE-connection
- Fix your test leads before power up so that you don't need to take your hands into the danger zone
- Solder a small neon to the mains cap - just as a warning indicator
- Use high impedance probes without any ground wiring leading to the DUT
- Use an isolation transformer to prevent ground loops, as an EMI-filter or whatever. But never as a single point of operator insulation!

What also many people miss in my opinion: The isolation transformer only isolates very low frequencies like DC and mains power. Its impedance for SMPS-like frequencies is much lower. That means, if you insulate your DUT with such a transformer, and connect scope GND to SMPS primary ground then you not only have a very serious safety risk, then you also have an additional ground loop that´s not present in normal operation. And this also spoils your measurement results. Things get even much worse if you connect scope GND not only to SMS primary ground but even e.g. to some point in the snubber network.

Sorry if I repeat things anyone in this field should know. But experience shows many people believe it´s worth to measure with a risky setup. And in 99.9% of the cases it´s just pain no gain.

[tggzzz]

Can anyone explain why these discussions are still active?

Firstly because there is no shortage of new inexperienced people. They, just like children, will make old mistakes. While everybody was young and inexperienced at one time, the sane ones try to avoid making old mistakes.

Secondly because there is no shortage of people[1] confidently stating the equivalent of "I walk into the road without looking and haven't come to any harm, so that means you can too".

There's an English proverb "A little knowledge is a dangerous thing". I expect there is the equivalent in most languages.

[1] especially on social media and doubly so on yooootooob.

[Phil1977]

It´s of no shame to have little experience - but many of the stories and social media contributions demonstrate that it would be of any necessity or advantage to measure that way. And that´s what is indeed annoying, that people show off by doing something stupid while they are claiming to spread knowledge.

[magic]

HV differential probes are easily available today - no need to ground your DUT with the scope. Measurements with HV differential probe are much more trustworthy than without because of the symmetrical input of defined impedance. So if available, I´d always use this probe for everything >100V.

This only works if source impedances are matched, or at least close enough.

Example: connect the negative to a hot ground and the positive to a 10kΩ voltage divider. Positive rise time for common mode signals increases by 100ns assuming 5pF probe input capacitance, negative remains unaffected.