workbench safety: mains isolation

[scrat]

Modern 'scopes are double insulated which means the outputs are isolated from earth anyway so it gains you nothing.

My Rigol DS1052E is definitely grounded at the BNC inputs, I just checked it.
So unfortunately you cannot make such wide sweeping statements.

It MUST be earthed,since otherwise, connecting the probe's ground to mains or high voltage would be normally harmful.

[Zero999]

It MUST be earthed,since otherwise, connecting the probe's ground to mains or high voltage would be normally harmful.

What about battery powered 'sopces? Rigol 'scopes have battery powered option.

You won't get a shock providing the device under test is connected to an isolation transformer.

Although that won't protect you if you connect other pieces of floating test equipment to the same circuit, which is why you should make sure the chassis of each piece of test equipment are connected together so they can't float at different voltages.

[alm]

It MUST be earthed,since otherwise, connecting the probe's ground to mains or high voltage would be normally harmful.

What about battery powered 'sopces? Rigol 'scopes have battery powered option.

Not sure about the Rigol. The manual of all IEC class I scopes with battery option I've seen explicitly states that the scope should be grounded by a separate ground connection when used battery powered (my scope has a separate ground screw on the back for this purpose), or only used for potentials below 30V or so.

You won't get a shock providing the device under test is connected to an isolation transformer.

Now it only takes a single fault to create a dangerous situation, without disconnecting the ground lead of the scope, it would take a double fault (broken ground and broken isolation transformer / connection of person to second conductor).

What safety do you gain by connecting a piece of class I test equipment to an isolation transformer? Connecting the ground clip to a dangerous potential will just blow the fuse, inconvenient, but not a safety issue. Why do feel the need to ignore safety advice by scope manufacturers (I quoted Tek, but I'm sure Agilent and Lecroy will tell you the same)?

Although that won't protect you if you connect other pieces of floating test equipment to the same circuit, which is why you should make sure the chassis of each piece of test equipment are connected together so they can't float at different voltages.

This sounds like a bodge job, can you refer to instructions from the past thirty years or so by any reputable test equipment manufacturer that tell you to do this? I referred you to instructions by Tek that tell you to never, ever do it (with normal scopes).

[jahonen]

You won't get a shock providing the device under test is connected to an isolation transformer.

Now it only takes a single fault to create a dangerous situation, without disconnecting the ground lead of the scope, it would take a double fault (broken ground and broken isolation transformer / connection of person to second conductor).

What safety do you gain by connecting a piece of class I test equipment to an isolation transformer? Connecting the ground clip to a dangerous potential will just blow the fuse, inconvenient, but not a safety issue. Why do feel the need to ignore safety advice by scope manufacturers (I quoted Tek, but I'm sure Agilent and Lecroy will tell you the same)?

Why it would be any more dangerous if you have both DUT and scope behind their own isolations? Of course, measuring mains connected equipment would be quite unsafe with isolated scope.

And yes, user manual for my Agilent MSO6034 says that

Maintain oscilloscope ground connection. Do not negate the protective action of the ground connection to the oscilloscope. If the circuit under test has voltages greater than 30 Vrms, the oscilloscope must be grounded through its AC power cord (or its ground post on Option BAT models) to prevent electric shock to the user.

I have always thought that spirit of that advice is that some people will try to measure potentials galvanically connected to the mains with battery powered scope (or, generally power supplies galvanically connected to surrounding environment), thus creating the hazard with the surrounding environment since everything connected to the scope (like a computer etc.) is now live respect to the environment.
However, I see no big problem if I always have both scope and DUT isolated from mains and each other. Of course, that requires care and discipline to always use isolated powering for mains powered DUT.

Regards,
Janne

[Zero999]

Now it only takes a single fault to create a dangerous situation, without disconnecting the ground lead of the scope, it would take a double fault (broken ground and broken isolation transformer / connection of person to second conductor).

A good isolation transformer should have class 2 insulation between the primary and secondary so the chances of that happening are the same as any other piece of class 2 equipment.

What safety do you gain by connecting a piece of class I test equipment to an isolation transformer? Connecting the ground clip to a dangerous potential will just blow the fuse, inconvenient, but not a safety issue. Why do feel the need to ignore safety advice by scope manufacturers (I quoted Tek, but I'm sure Agilent and Lecroy will tell you the same)?

Most 'scope manuals are written with the assumption that you're not running it off an isolation transformer and bonding its chassis to that of the device under test.

This sounds like a bodge job,

No bodge job, it's recommended that the chassis of all equipment running off an isolation transformer be connected together so there's no risk of them floating at different potentials. It's written down in the IEE regulations as protection against electrical shock using a separated supply. The idea is that if all metal parts are bonded together and the power supply is isolated from earth, there's no chance of receiving a shock. It's pretty safe because there would need to be a triple fault in order for you to be shocked: the insulation would have to fail on both appliances as well as the bonding between them.  Using an earth bonded 'scope run directly from the mains is more risky because there only needs to be two faults to receive a deadly shock: insulation and earth bond failure.

Setting up such a system is easy, just connect a ganged mains socket to the isolation transformer secondary and connect all the devices under test and test equipment to it. All the chassis will be now connected together because even though there's no earth connection, there's still the earth bus inside the gang socket connecting the chassis together.

[tyblu]

Why it would be any more dangerous if you have both DUT and scope behind their own isolations? Of course, measuring mains connected equipment would be quite unsafe with isolated scope.
...
However, I see no big problem if I always have both scope and DUT isolated from mains and each other. Of course, that requires care and discipline to always use isolated powering for mains powered DUT.

When the 'scope is isolated and has a normal ground (class 1 something-or-other), its chassis and reference floats. The danger is with respect to your voltage. If you are 'Earth' (mains) grounded -- for eg. touching your grounded workbench -- then the oscilloscope may have a "hot chassis", floating its reference high above yours. This is why alm mentions this...

The only exception would be something like a Tek A6901 ground isolation monitor, which continuously monitors the potential between safety ground and scope ground, and will protect the user if it reaches dangerous levels (above 30V or so) by turning off the scope and grounding it.

I stick a banana plug into the Earth ground in a North American socket and keep a voltmeter between the chassis and Earth ground. Using a cheap meter, I think this actually 'bleeds' any high voltage chassis capacitive charge, but I may be blowing smoke -- regardless, it hasn't been a big deal for me yet, but I'll keep doing it.

[PetrosA]

It sounds like there's some misconceptions going around about how floating voltages are or aren't dangerous. Here's my 2 cents

If you have a separate or isolated voltage source that doesn't share a ground connection anywhere in the system with the mains, then floating voltages are safe. Try this experiment - take a UPS, unplug it from the mains so it starts producing voltage on its own and check voltage to system ground. It won't exist. If the UPS switches on while plugged in because power fails, you will get voltage to system ground. Edit to add: If the disconnected UPS does make a second, accidental contact with mains phases, neutrals or ground, especially through your body, there would be a dangerous situation because there would then be a path for current to flow.

I don't pretend to know how isolation transformers outside the US are wired, but here they must have their ground connected to the system ground. The only way to know this for sure is to take resistance readings between the ground and neutral on the isolated side, and isolated ground to system ground, and isolated neutral to system neutral.

[RayJones]

I agree, I think more is being made out of isolation transformers than really exists.

The #1 problem with normal mains is that Neutral is earthed somewhere.
This means touching live is very likely to induce a possibly deadly current flow down through your boots to good ol mother earth.

The sole purpose of the isolation transformer is to float your neutral (and live) away from earth, and indeed the earth connection is still carried through to the end equipment.

So if you grab live with your hand brushing against the metal chassis you won't havethat killer current path.

As for test equipment needing to be isolated - total utter bullshit.
You still have the protective earth connected through, and that's what the BNC outers on the CRO attach to!

[tyblu]

It sounds like there's some misconceptions going around about how floating voltages are or aren't dangerous. Here's my 2 cents

If you have a separate or isolated voltage source that doesn't share a ground connection anywhere in the system with the mains, then floating voltages are safe. Try this experiment - take a UPS, unplug it from the mains so it starts producing voltage on its own and check voltage to system ground. It won't exist. If the UPS switches on while plugged in because power fails, you will get voltage to system ground. Edit to add: If the disconnected UPS does make a second, accidental contact with mains phases, neutrals or ground, especially through your body, there would be a dangerous situation because there would then be a path for current to flow.

I don't pretend to know how isolation transformers outside the US are wired, but here they must have their ground connected to the system ground. The only way to know this for sure is to take resistance readings between the ground and neutral on the isolated side, and isolated ground to system ground, and isolated neutral to system neutral.

You are quite right, and I feel silly. This happens more often than is comfortable. 

[Zero999]

As for test equipment needing to be isolated - total utter bullshit.

I agree, you don't have to power your test equipment from an isolation transformer

You still have the protective earth connected through, and that's what the BNC outers on the CRO attach to!

That's not necessary providing the test equipment is connected to an isolation transformer and its chassis is connected to the chassis of the other equipment connected to it.

I'm talking about regulation 471-12-01
http://www.theiet.org/publishing/wiring-regulations/mag/2006/18-portablesupply.cfm?type=pdf
http://books.google.co.uk/books?id=Z432pJcac6QC&pg=PA80&lpg=PA80&dq=Regulation+471-12-01&source=bl&ots=bbTjcRH3vG&sig=eo84q5YnT6LuFqPNxMLT62HYrBo&hl=en&ei=Dgw_TevyA9-ShAf37aSLCg&sa=X&oi=book_result&ct=result&resnum=4&ved=0CDQQ6AEwAw#v=onepage&q=Regulation%20471-12-01&f=false

All it means it that if more than one appliance is running from an isolation transformer, the exposed metal parts need to be bonded together to prevent them floating at dangerous potentials.

Of course if you're not using an isolation transformer to power the 'scope you need to connect the chassis to protective earth.

[Chasm]

All it means it that if more than one appliance is running from an isolation transformer, the exposed metal parts need to be bonded together to prevent them floating at dangerous potentials.

Using more than one isolation transformer is widely seen as both unnecessary and a recipe for disaster.
Such a setup is rarely warranted.

When working on HV circuits apply the KISS principle at all times.
And try to think up front.

[Zero999]

Using more than one isolation transformer is widely seen as both unnecessary and a recipe for disaster.
Such a setup is rarely warranted.

I agree, which is why I've never said anything about using more than one isolation transformer, read my posts.

When working on HV circuits apply the KISS principle at all times.
And try to think up front.

It's not a high voltage circuit, mains is low voltage.

[RayJones]

As for test equipment needing to be isolated - total utter bullshit.

I agree, you don't have to power your test equipment from an isolation transformer

You still have the protective earth connected through, and that's what the BNC outers on the CRO attach to!

That's not necessary providing the test equipment is connected to an isolation transformer and its chassis is connected to the chassis of the other equipment connected to it.

I'm talking about regulation 471-12-01
http://www.theiet.org/publishing/wiring-regulations/mag/2006/18-portablesupply.cfm?type=pdf
http://books.google.co.uk/books?id=Z432pJcac6QC&pg=PA80&lpg=PA80&dq=Regulation+471-12-01&source=bl&ots=bbTjcRH3vG&sig=eo84q5YnT6LuFqPNxMLT62HYrBo&hl=en&ei=Dgw_TevyA9-ShAf37aSLCg&sa=X&oi=book_result&ct=result&resnum=4&ved=0CDQQ6AEwAw#v=onepage&q=Regulation%20471-12-01&f=false

All it means it that if more than one appliance is running from an isolation transformer, the exposed metal parts need to be bonded together to prevent them floating at dangerous potentials.

Of course if you're not using an isolation transformer to power the 'scope you need to connect the chassis to protective earth.

So your CRO develops an internal fault and active mains connects to the internal chassis which should be grounded.
Guess what?
Your "Bonded to exposed metal" has suddenly become live, connected to the mains, and no amount of isolation transformers will protect you from the inevitable shock.

I also suggest we refer to this document on pg4 under " "Ed" and "Ing" " http://www.apcmedia.com/salestools/SADE-5TNQYQ_R0_EN.pdf

This clearly states that the isolation transformer ONLY OPENS ACTIVE/NEUTRAL - ground must be carried through.

So by bonding together your exposed metal, you are doing what you should always be doing, connecting that to earth.
If you are not, why oh why would you deliberately open the protective ground?

Connect all your exposed metals to real earth.
You still need to have your wits about you.
Stupidity will still be rewarded with a jolt.

The isolation transformer allows you connect your test equipment without adversely shorting sections out, like the input full wave bridge of a SMPS.
And this goes back to what I stated earlier, turn it off, and allow large caps to discharge before playing about with your test leads, especially the earthy end.

[Zero999]

Thgis is hard work. . .

So your CRO develops an internal fault and active mains connects to the internal chassis which should be grounded.
Guess what?
Your "Bonded to exposed metal" has suddenly become live, connected to the mains, and no amount of isolation transformers will protect you from the inevitable shock.

No because if it's powered by an isolation transformer (which is what I said) there would be no risk of a shock.

If a device is powered by an isolation transformer, connecting its chassis to earth provides no extra protection against electric shock.

[RayJones]

Yes it is hard work.

So what are you connecting your exposed metal work to?

[alm]

I'm talking about regulation 471-12-01
http://www.theiet.org/publishing/wiring-regulations/mag/2006/18-portablesupply.cfm?type=pdf
http://books.google.co.uk/books?id=Z432pJcac6QC&pg=PA80&lpg=PA80&dq=Regulation+471-12-01&source=bl&ots=bbTjcRH3vG&sig=eo84q5YnT6LuFqPNxMLT62HYrBo&hl=en&ei=Dgw_TevyA9-ShAf37aSLCg&sa=X&oi=book_result&ct=result&resnum=4&ved=0CDQQ6AEwAw#v=onepage&q=Regulation%20471-12-01&f=false

All it means it that if more than one appliance is running from an isolation transformer, the exposed metal parts need to be bonded together to prevent them floating at dangerous potentials.

Google Books link doesn't work, but the PDF states that 'there should be no connection between the separated circuit and any other circuit, or to earth'. I believe this would include connecting a scope connected to an isolation transformer to a DUT on a different isolation transformer. So it doesn't appear to be that the proposed isolated workbench would comply with this.

No bodge job, it's recommended that the chassis of all equipment running off an isolation transformer be connected together so there's no risk of them floating at different potentials. It's written down in the IEE regulations as protection against electrical shock using a separated supply. The idea is that if all metal parts are bonded together and the power supply is isolated from earth, there's no chance of receiving a shock. It's pretty safe because there would need to be a triple fault in order for you to be shocked: the insulation would have to fail on both appliances as well as the bonding between them.  Using an earth bonded 'scope run directly from the mains is more risky because there only needs to be two faults to receive a deadly shock: insulation and earth bond failure.

Triple fault? You connect the scope ground to some potentially dangerous potential in your DUT (eg. live wire), this is safe, because everything is floating. Now a fault occurs and the neutral wire in the isolated DUT gets connected to something you touch (eg. case): you get shocked, since you're likely to be touching the scope ground somehow (all exposed metal on most test equipment is connected to ground). If the scope had been grounded, there would have been sparks and a blown fuse as soon as you clipped it to the live wire, and the ground conductor would have carried all current and the GFI/RCD would trip. Isolation transformer does not make the voltages less dangerous, it just makes it safe to touch one of the wires at a time. If you somehow get connected to the second wire, you can still get shocked.

The issue with ground is that there tends to be lots of grounded stuff around, for example piping for heating and water, some floors, workbenches, almost all large fixed metal objects, so it's quite likely that you connect your floating ground to earth ground somehow. From the signal integrity front, there is usually something like 1000pF of capacitance from your scope to ground (in your average lab, obviously less if you suspend it in mid-air somehow), so if there is any connection between your DUT and ground, this can give some nasty coupling.

Regarding isolation transformers and ground: some do connect the ground conductor, some don't. Connecting the scope to the former won't do any harm, but it won't do much good either. It takes a double fault for class I equipment to shock someone, so the effort is probably better spent in improving safety some other way. Connecting equipment like scopes to the latter can be dangerous, and is recommended against by the manufacturers. I still don't see how cutting the ground connection would improve safety, the only advantage is that it can be convenient for measurements to non-isolated circuits without proper tools like differential probes, and that's precisely the most dangerous application.

[Zero999]

Google Books link doesn't work, but the PDF states that 'there should be no connection between the separated circuit and any other circuit, or to earth'. I believe this would include connecting a scope connected to an isolation transformer to a DUT on a different isolation transformer. So it doesn't appear to be that the proposed isolated workbench would comply with this.

The PDF is a shortened form of the full standard.

The PDF is talking about powering one device off an isolation transformer. The full standard goes on to talk about powering multiple devices off the same transformer which is perfectly safe as long as their chassis are connected together.

Triple fault? You connect the scope ground to some potentially dangerous potential in your DUT (eg. live wire), this is safe, because everything is floating.  Now a fault occurs and the neutral wire in the isolated DUT gets connected to something you touch (eg. case): you get shocked, since you're likely to be touching the scope ground somehow (all exposed metal on most test equipment is connected to ground).

No because the DUT is also powered by an isolation transformer and its chassis bonded to the 'scope.

The issue with ground is that there tends to be lots of grounded stuff around, for example piping for heating and water, some floors, workbenches, almost all large fixed metal objects, so it's quite likely that you connect your floating ground to earth ground somehow. From the signal integrity front, there is usually something like 1000pF of capacitance from your scope to ground (in your average lab, obviously less if you suspend it in mid-air somehow), so if there is any connection between your DUT and ground, this can give some nasty coupling.

It doesn't matter if there's a ground around or not. Just don't power anything on your test bench with a grounded power supply and bond all the chassis together and you'll be fine.

I think a drawing is in order to get across what I'm trying to say.

[alm]

The PDF is talking about powering one device off an isolation transformer. The full standard goes on to talk about powering multiple devices off the same transformer which is perfectly safe as long as their chassis are connected together.

But multiple isolation transformers are not permitted, and that's the solution that was suggested by multiple people in this thread. What's the point of the isolation transformer if the DUT doesn't float relative to your scope? I can see why this might be useful in an outdoor environment without any ground nearby (eg. powered from battery/generator), which is what I believe was the intention of the regulations you quote, but what do you gain on the bench except a false sense of security?

No because the DUT is also powered by an isolation transformer and its chassis bonded to the 'scope.

If you make sure there are no grounded objects nearby and never measure anything not connected to the same isolation transformer / ground, this setup is probably no worse than no isolation transformer at all, since you're basically creating your local circuit with local ground, but what's the advantage over the usual setup of grounded test equipment and DUT on an isolation transformer? The latter allows the DUT to float relative to the user and the scope, something that your proposed setup does not (since the scope ground in connected to DUT ground, and the user is likely to be connected to the scope ground). With your proposed setup, there is always the risk of connecting the scope to something not on the same isolation transformer, which can be dangerous.

[Zero999]

But multiple isolation transformers are not permitted, and that's the solution that was suggested by multiple people in this thread.

Well I agree that multiple isolation transformers are a bad idea, even though bonding the chassis together should eliminate mos of the shock risk there's still the possibility of voltages adding together.

What's the point of the isolation transformer if the DUT doesn't float relative to your scope? I can see why this might be useful in an outdoor environment without any ground nearby (eg. powered from battery/generator), which is what I believe was the intention of the regulations you quote, but what do you gain on the bench except a false sense of security?

Isolation transformers are routinely used indoors in the UK for shock protection, for example the only sockets allowed in the bathroom are for shavers and are powered by an isolation transformer.

If you make sure there are no grounded objects nearby and never measure anything not connected to the same isolation transformer / ground, this setup is probably no worse than no isolation transformer at all, since you're basically creating your local circuit with local ground, but what's the advantage over the usual setup of grounded test equipment and DUT on an isolation transformer? The latter allows the DUT to float relative to the user and the scope, something that your proposed setup does not (since the scope ground in connected to DUT ground, and the user is likely to be connected to the scope ground). With your proposed setup, there is always the risk of connecting the scope to something not on the same isolation transformer, which can be dangerous.

Again, as I said above, nothing bad should happen if the chassis are connected together but I wouldn't recommend using more than one isolation transformer ever.

Did you look at the image I posted? Before making this post it had zero views, the first was me clicking on it to see if it worked, there again it could be the server playing up.

[alm]

Isolation transformers are routinely used indoors in the UK for shock protection, for example the only sockets allowed in the bathroom are for shavers and are powered by an isolation transformer.

This would apply if you just connected the DUT to the isolation transformer, touching both ground (likely in a bathroom) and one of the conductors won't result in you getting fried in that case. Touching 'local ground' and any ground-referenced potential will get you shocked, however, with one isolation transformer and the ground connections together. With just the DUT on an isolation transformer, ground could be floating, so touching scope ground and the now floating potential would be safe. And GFI/RCD would still be functional even if you touch the scope ground and a non-floating potential. I guess especially small isolation transformers (like shaver sockets) will probably limit the current, but this is hardly the case for large isolation transformers for the whole bench, and I'd probably consider a GFI/RCD more important.

Did you look at the image I posted? Before making this post it had zero views, the first was me clicking on it to see if it worked, there again it could be the server playing up.

I did, but I didn't feel the need to click on it to enlarge it, since it just confirmed that we had the same set up in mind.

[Zero999]

Yes shaver sockets are normally low power, typically 25VA.

I suppose all I'm saying is I feel safer working at a bench which doesn't have an earthed power supply.

One of my 'scopes has to be run off a transformer as it's the US model and we have 230V in the UK. Naturally I used an isolation transformer rather than an autotransformer, although the chassis is earthed because the transformer is in a metal box so is not class 2: one of the connections on the secondary could easily make contact with the case, still even then it'd need the insulation on my 'scope to fail too but I decided it wasn't worth the risk. I do use another isolation transformer for the DUT so I suppose having to isolation transformers is a necessary evil, not something I'd do by choice.