Isolation transformer ground question

[james_s]

The relation to earth is at the panel, neutral is bonded to earth,

maybe thats always the case in america,but in many parts of the world the neutral - earth link could be anywhere on the network,and making your own link at the panel would break local reguations

Yes, the OP is in America, I'm in America, that's the location I'm discussing here. Other parts of the world bond neutral to earth in other ways, but as far as I know it is always done somewhere, and it is never done downstream of the panel anywhere.

[themadhippy]

but the origin of the the post above yours,that you apeared to be replying to was spain,

and it is never done downstream of the panel anywhere

apart from were concentric earthing on the consumers side is used,ok it maybe never made it to the states,and its an extremly rare beast in the uk  nowerdays,but they do exist.

[gf]

It is of course possible to turn the output of the tranformer into a private TN system in order to make a RCD work, by declaring one of the two conductor as Neutral and tying it to PE. But this defeats the purpose of an isolation transformer which is rather supposed to provide an IT system at the secondary side.

[tatel]

Yeah, in Spain I don't know any home that has those devices in, say, the bathroom. Just one at the panel, period. They seem to be 30 mA, not 5. Not really sure about it, but earthing seems to be done at least at the electric company transforming sub-station *and* somewhere into the building, the last being absent in many old buildings, I think. And, of course, the plugs can be inserted into the sockets both ways. I guess these are EU rules, not just spanish rules.

When I read abouth earthing, three-prong plugs, sockets with fuses and GFCIs, etc, in UK-US-Australia, etc, I wonder why EU citizens aren't dying by electrocution by the dozens daily... I asked about that to some friends of me, pro electricians, and got answers that (I guess) english-speaking people wouldn't like. Not that I understand technical reasons behind those answers, either.

So I'm going to catch (I'm being facetious) one of these friends by the neck and get my home electrical installation renewed and re-certified (I guess somebody had it certified somewhere in time or I wouldn't be getting any service from the electricity provider). I live in an old building fully owned by Ebenezer Scrooge, so I will pay for it, then I will try to get my money back from the landlord (it will be like mission impossible). The only thing I know for sure is that earth line is non-existant on the entire building, which is some 150 years old. This has made me afraid of things like having an electric water heater or using the oscilloscope I bought last year. I can solve this problem by just throwing money at it, so let's go for it. That said, neither me nor any neighbour got electrocuted in more than 20 years I'm living there.

But, for the life of me, I can't understand why a GFCI wouldn't work after an isolation transformer or, for the case, why would an isolation transformer would spare my life from electrocution. So I'm guessing I don't understand how the earthing works. Your answers seem to imply that a circuit is closed between two earthing points, one of them being me and the other at the sub-station-panel-wathever?

I was under the idea that all the earth is more or less at the same potential everywhere and that it is enough to be the path between two points at a different enough potential -say, line and earth- to get pretty darn good toasted. This misunderstanding can't be really solved by just throwing money at it. Could you guys give me any pointers?

[gf]

Touching two points with different potential only toasts you if the loop impedance of the closed circuit is low enough so that enough current can flow through your body. This is not granted if you touch only one conductor of the secondary side of an isolation tranformer, since (due to the isolation) there is a very high impedance between the other conductor and earth. If you touch both conductors you still get toasted, of course.

Edit: Not the voltage toasts you, but the current flowing through your body.

[james_s]

But, for the life of me, I can't understand why a GFCI wouldn't work after an isolation transformer or, for the case, why would an isolation transformer would spare my life from electrocution. So I'm guessing I don't understand how the earthing works. Your answers seem to imply that a circuit is closed between two earthing points, one of them being me and the other at the sub-station-panel-wathever?

I was under the idea that all the earth is more or less at the same potential everywhere and that it is enough to be the path between two points at a different enough potential -say, line and earth- to get pretty darn good toasted. This misunderstanding can't be really solved by just throwing money at it. Could you guys give me any pointers?

I think what you're missing is that electrical potentials are always relative to each other. The potential of the earth is not an absolute value, it is a value relative to some other node that is referenced to it. For example if you take a 12V battery and you tie the negative terminal to a rod driven into the earth then you can call the earth "ground" and you will measure 12V between it and the battery. If on the other hand the battery is just floating without either terminal connected to the earth, you will measure 0V between either terminal and the earth, the battery is floating, it has no connection to the earth so there is no voltage relative to the earth.

An isolation transformer is exactly the same, the secondary is floating, just like a battery. If you are standing on the earth and your body is at the same potential as the earth you can touch either side of the isolation transformer secondary and you won't get shocked because there is not a complete circuit and no current can flow. To pass a current you need a circuit, a complete, circular path connected between two nodes that have a potential relative to each other.

[tatel]

OK, got it. So the reason an isolation transformer makes it safe to touch just one exposed part of the secondary side, is that the non-galvanic connection to the primary side, wich is the related to earth one, has high impedance. This means the relation to earth has been broken on the secondary side. To get electrocuted by touching just one exposed part of the secondary, a current must flow from that secondary to earth throug me, but it can't happen since their potentials aren't related (=0V=no current can flow). No electrocution then.
 
Should I touch two exposed parts at different enough potential on secondary side, a current would flow trough me, without going to earth, so the GFCI couldn't see any difference between  L and N windings, so it wouldn't break.

Thank you very much, guys.

[Ian.M]

What's worse is that as soon as your test equipment introduces a ground (e.g. you start probing the primary side of the PSU with an ordinary scope probe), that isolation and the limited safety it provides goes away, leaving you with ground referenced mains voltage that can electrocute you with a single touch, but due to the action of the transformer, wont trip any upstream RCD/GFCI protection.

RCDs/GFCIs have saved a *LOT* of idiots from receiving Darwin awards, and a lesser number of smart careful people from unexpected electrocution. 
Are you certain you would fall in the 'smart careful people' category, as you need to be one if you use an isolation transformer on your test bench!

[Zero999]

What's worse is that as soon as your test equipment introduces a ground (e.g. you start probing the primary side of the PSU with an ordinary scope probe), that isolation and the limited safety it provides goes away, leaving you with ground referenced mains voltage that can electrocute you with a single touch, but due to the action of the transformer, wont trip any upstream RCD/GFCI protection.

RCDs/GFCIs have saved a *LOT* of idiots from receiving Darwin awards, and a lesser number of smart careful people from unexpected electrocution. 
Are you certain you would fall in the 'smart careful people' category, as you need to be one if you use an isolation transformer on your test bench!

An insulation monitoring device can provide a warning when that happens.

There are off the shelf devices available, or if it's just for hobby use, build your own from a couple of neon lamps and high voltage resistors.
https://docs.rs-online.com/e9c5/0900766b812cdc34.pdf



If an audible, as well as visual warning is required, the neon can be substituted for an AC input opto-coupler connected to a BJT to activate a small piezo buzzer and LED.

[tatel]

Are you certain you would fall in the 'smart careful people' category, as you need to be one if you use an isolation transformer on your test bench!

I'm careful enough not to use my benchtop scope unless the DUT is a batery powered one. I'm careful enough to scratch my not-so-deep pockets to do what the landlord should do if *he* were smart. I'm careful enough not to put an electric water heater in a house where electrical installation has no earth wire. I'm careful enough to watch Dave's video about "How not to blow your scope" and to follow related threads on the forum for more than a year to know what "not to do unless you really know what are you doing".

But, above all, I'm stubborn enough to continue asking again and again until being sure I'm getting right any important concept that crosses my way. Those who ask may look stupid once. Those who don't ask, afraid to look stupid once, remain stupid forever.

Now that I get how an isolation transformer and a GFCI work, I can begin to think about what should I choose to do. I'm very grateful to both gf and james_s for telling me what I was missing.

That said, thank you for the "advice". Do you realize that giving real knowledge, like gf and james_s do, is much better than just calling names and saying things without giving any knowledge?

[Ian.M]

If you feel offended by my comment, think about those two categories again. 

I'm assuming you are smart and careful, because you *ARE* asking these questions, not just going ahead and putting yourself into danger, but even smart careful people can make mistakes, so when you are confident you have the necessary knowledge and understanding, *PLEASE* don't let yourself get complacent about the hazards of working on devices under test fed by an isolation transformer.  If you've *EVER* tripped a RCD (GFCI) while holding or touching something wet or metallic, you are probably one of the lucky guys who's been saved by one, but downstream of  an isolation tranformer you don't have that extra layer of protection.

[tatel]

An insulation monitoring device can provide a warning when that happens.

There are off the shelf devices available, or if it's just for hobby use, build your own from a couple of neon lamps and high voltage resistors.
https://docs.rs-online.com/e9c5/0900766b812cdc34.pdf

Thank you for that information. However for use after an isolation transformer, those devices in all cases seem to have in some way connection to earth. This would be defeating the isolation transformer, wouldn't it? As said before, I now have to think about isolation transformer being worth to have or not.

I have read somewhere that to properly fully check an SMPS, I would need a dual channel scope. So, no matter if there is an isolation transformer, I could easily blow something. In that case, having an isolation transformer could be useless.

All in the room is wood, so I don't really fear so much becoming the path for a current to flow to earth. I think there it's a much higher probability of becoming the lesser impedance path between line and neutral. In that case, not just the isolation transformer but also any GFCI, no matter where, could be useless.

Then I would like not just me, but also my equipement to be as safe as possible.

Last but not least, the lesser the bulk, the more clarity.

So I'm beginning to think that having a couple of micsig DP1007 probes could be the simpliest way to go. Not the cheapest one, however, but that's secondary. Any thoughts?

[tatel]

If you feel offended by my comment, think about those two categories again. 

Understood. Excuse me if you think I've been rude in any way. I'm on this forum to learn and I really appreciate any answers, unless I feel really insulted. And, after seeing how bullets fly between our primma donnas in the threads like the zoom quirk, or how much some developers could despise reports about regressions, I'm not sure I'm thick-skinned enough to be here. So I've probably been so touchy that I feel compelled to apologize again.

[Ian.M]

Apology accepted.  It was *NEVER* my intention to personally insult you or your intelligence, and I unreservedly apologise for giving you the impression that I was.   You are still here and talking sense, so I assure you, you are thick skinned enough to benefit from the technical content here without getting drawn too far into personality conflicts.  Engineers and good technicians frequently tend to be excessively blunt and a significant number fall short in social skills, because the profession self-selects for people who are more comfortable with dealing with complicated things than with complicated people.   However that doesn't justify deliberate targeted rudeness on our part.

Personally,  for working on SMPSUs, I would prefer at least one differential probe with HV isolation and also an isolation transformer.  That combo decreases your risk, as a RCD (GFCI), and no isolation, wont protect you effectively from high energy DC shocks e.g. from the SMPSU's primary side DC bus reservoir capacitor.    If you can afford it, get a second isolated probe.   The isolation transformer also increases your safety when working on devices where the PSU is exposed and maybe even on the same board as the circuit you are working on, even if the setup has the PSU's secondary side grounded. 

One thing to be aware of if you are in an older building with whole house RCD protection (rather than per circuit protection) is you should have at least *some* bench lighting with a maintained supply, or enough natural daylight at your bench, as if you trip the RCD, and all the lights go out while holding a test probe in place, you can easily cut yourself or shock yourself on a charged capacitor, or accidentally knock stuff off the bench because there is no longer enough light to see where your hand is.  If you do have per circuit protection, make sure that at least one light is on a different circuit and is switched on whenever you are working.

[bdunham7]

Thank you for that information. However for use after an isolation transformer, those devices in all cases seem to have in some way connection to earth. This would be defeating the isolation transformer, wouldn't it? As said before, I now have to think about isolation transformer being worth to have or not.

I have read somewhere that to properly fully check an SMPS, I would need a dual channel scope. So, no matter if there is an isolation transformer, I could easily blow something. In that case, having an isolation transformer could be useless.

So I'm beginning to think that having a couple of micsig DP1007 probes could be the simpliest way to go. Not the cheapest one, however, but that's secondary. Any thoughts?

A while back I posted a simple explanation of how to use an isolation transformer to probe an SMPS.  My post and photos appears to have gotten a bit scrambled, but it should be understandable.  A lively discussion follows which reveals some disagreement.  I"m not persuaded and I would still do it this way if I didn't have an isolated scope.  The bottom line is that this type of work should be done carefully and mostly hands-off, meaning you make all your connections on an unpowered DUT and then don't touch it while making measurements. 

https://www.eevblog.com/forum/testgear/probing-smps-with-oscilloscope/msg3152256/#msg3152256

The trick in using a multi-channel scope in this context is to remove the ground clips from all but one of your probes.  One and only one grounding point, although you can vary its location in the circuit--which is the primary reason for doing all this in the first place.  Some may protest that this results in some signal integrity issue of one sort or another, but then they tell you to use a cheap differential probe which is typically much worse in this regard.  I find that with a bit of thought and experimentation, it works just fine.

As far as 'blowing up' something, an isolation transformer usually provides a much lower fault current if you do make a serious mistake.  Mine allows me to vary the input voltage (variac) and measure leakage current as well as limiting the fault current to a few amperes.  I'm unlikely to 'blow up' my oscilloscope, I'll just pop a breaker.

Differential probes have CMRR issues, the Micsig ones have big CMRR issues.  I have one, the DP10013, and it works and is handy for some situations.  It probably would be OK for a lot of SMPS issues and it could be handy for adding a channel where you just can't manage to set up the measurements you want with any one particular ground point.  However, I really don't see how using such a probe without an isolation transformer improves safety, since the DUT is then live with full mains power and a large fault current.  As I stated in the earlier post, RCD/GFI is a nice additional safety feature--like airbags in a car--but they should not be your primary plan.  There are cases where they won't trip and in any case there is still plenty of stored energy in some SMPS primary capacitors.  If you use differential probes with an isolation transformer, then you can avoid introducing a low-impedance ground (differential probes are not galvanically isolated and will have leakage current) to the DUT and thus keep it 'touch-safe', at least in theory and with respect to a single point.

[Zero999]

An insulation monitoring device can provide a warning when that happens.

There are off the shelf devices available, or if it's just for hobby use, build your own from a couple of neon lamps and high voltage resistors.
https://docs.rs-online.com/e9c5/0900766b812cdc34.pdf

Thank you for that information. However for use after an isolation transformer, those devices in all cases seem to have in some way connection to earth. This would be defeating the isolation transformer, wouldn't it? As said before, I now have to think about isolation transformer being worth to have or not.

Some connection to earth is necessary to monitor if either side is connected to earth.

No, it doesn't defeat the isolation. The secondary is connected to earth via two 150k resistors which limit the current to a safe level if there's a fault. In this case it'll be under 1.5mA. I think I originally designed the circuit for 120V or made an error, because the current should be limited to under 1mA, so change R1 & R2 to 270k for 240V.

[tatel]

Apology accepted.  It was *NEVER* my intention to personally insult you or your intelligence, and I unreservedly apologise for giving you the impression that I was.

No worries, man. I really appreciate your words. Let's go ahead 

Personally,  for working on SMPSUs, I would prefer at least one differential probe with HV isolation and also an isolation transformer. 

I'm not in a hurry to work with SMPS. It's true that one died after just brushing the dust from inside. It bothers me because I'm used to clean them anytime I got a salvaged PC and never had any problems. So I'm willing to do a complete forensic work and even to try and repair it, just to know what happened, to learn something, and to have fun. But, I do have spares and other, more urgent projects, so this is going directly to the backburner for the time being.

A while back I posted a simple explanation of how to use an isolation transformer to probe an SMPS.  My post and photos appears to have gotten a bit scrambled, but it should be understandable.  A lively discussion follows which reveals some disagreement.  I"m not persuaded and I would still do it this way if I didn't have an isolated scope.  The bottom line is that this type of work should be done carefully and mostly hands-off, meaning you make all your connections on an unpowered DUT and then don't touch it while making measurements. 

https://www.eevblog.com/forum/testgear/probing-smps-with-oscilloscope/msg3152256/#msg3152256

This post from you and the one in that thread (and that whole thread indeed), are really interesting. Somehow it got under my radar. I have to read carefully the whole thread, for the time being, I've just skimped through it.

You have reminded me that I could get an isolated Owon VDS1022i which I do really like. I have old, small laptops available that can run WinXp/Win7/Linux. I use them for OBD work on my car. That, and the Owon, could be a nice thing for automotive work, while also being good enough to check what's the matter on a PSU with it's 5 mV/div IIRC. It would be about the price of a MicSig DP1007 probe, and I would get much more use of that equipement, I think.

Fact is, I'm going to buy a chinese benchtop lathe (not the infamous mini-lathe). A couple of MicSig probes + isolation transformer (I would think about 400VA) could easily be half the price of that lathe. Let alone the variac. I have to choose, and the need for that lathe is quite a bit more pressing.

I'm not planning to do much work with SMPSUs, I'm much more focused about DUTs that could be powered from a battery. So that approach with no differential probes looks quite atractive to me. Then I could get the differential probes later if really needed.

Now I'm going to read carefully that thread. But, what you guys would do if having that isolated scope? I'm trying to have the different options clear. I'll quite probably post more questions after reading it.

Thank you all.

[tatel]

No, it doesn't defeat the isolation. The secondary is connected to earth via two 150k resistors which limit the current to a safe level if there's a fault. In this case it'll be under 1.5mA. I think I originally designed the circuit for 120V or made an error, because the current should be limited to under 1mA, so change R1 & R2 to 270k for 240V.

Now that's really interesting, because it seems to do something like the incandescent bulb method, or better, increasing safety. Incandescent bulbs are pretty rare around here lately. I guess I'll have to look at it closely. Thank you very much for the tip

[Zero999]

No, it doesn't defeat the isolation. The secondary is connected to earth via two 150k resistors which limit the current to a safe level if there's a fault. In this case it'll be under 1.5mA. I think I originally designed the circuit for 120V or made an error, because the current should be limited to under 1mA, so change R1 & R2 to 270k for 240V.

Now that's really interesting, because it seems to do something like the incandescent bulb method, or better, increasing safety. Incandescent bulbs are pretty rare around here lately. I guess I'll have to look at it closely. Thank you very much for the tip

The lamp in my schematic is neon, not incandescent. A neon will light with a tiny current, whilst an incandescent takes a relatively large current. The neon could be replaced with a high efficiency green or white LED and a small bridge rectifier.

Incandescent lamps are used to limit the current through a power supply, so it doesn't blow up if there's a fault.

[tatel]

Yeah. I meant, it provides visual warning, perhaps somewhat like the incandescent bulb method, where you can know capacitors have been charged. And it would limit the current more. I know it's not exactly the same, sorry. I still need to look at it more closely.

The earthing could be to some chassis wich is also earthed from mains, but the secondary side would still have high impedance thanks to that couple of high value resistors, it isn't?

[tatel]

Well, it turns out that I was wrong about prices. Eleshop.eu has the MicSig probes cheaper than I thought.

I'm leaning towards DP750-100 since it has the most CMRR. 600V CAT II seems to be good enough. However output voltage is the lowest at 1.5V. But I have seen reports about it working with low voltage differential busses:

https://www.eevblog.com/forum/testgear/differential-probe-for-offline-pfcd-psu/msg4293502/#msg4293502

https://www.eevblog.com/forum/testgear/micsig-dp10007-or-dp750-100-differential-probe/msg4378921/#msg4378921

I'm sure there are better differential probes but €177+VAT is already in the upper range of what I'm willing to pay.

For isolation transformer, I'm thinking this one should be good enough to power that 300W ATX PSU, and probably other SMPSUs in the future, as long as they get just the load needed to regulate properly (I'm using a 10W 10R resistor right now).

https://www.tme.eu/en/details/pfs250_230_230v/safety-transformers/breve-tufvassons/

Or perhaps this one, wich is somewhat cheaper and could be combined in the future with a variac into a case. Not sure about that. All AC things here are 220V unless they are tree-phase. Most probably I'll go ahead with just the isolation transformer and a 60W incandescent bulb in series (the only one I have been able to find). Any thoughts?

https://www.tme.eu/en/details/pvs250_230_230v/safety-transformers/breve-tufvassons/

With this, I think the way recommended by Ian.M gets covered.

Now, should I follow bdunham7's recommendation, same isolation transformer. By the way, I got quite a few clues from your posts in that thread

Not sure about the probes, however. My GW Instek 1054B accepts 300V CAT 1 max input voltage IIRC and I assume this would be with 10x probes. It's said to be RMS voltage. So I guess this would be enough to work on the secondary side? Anyway I looked for the HV probes you recommended:

https://www.ebay.es/itm/332835001398.

In the pictures you posted, I see various different probes at same time, i.e, a PP510, the linked High voltage one and still another that I can't identify, which seems to have an alligator adapter.

I see you are using the high voltage probe on the primary side and the PP510 on the secondary, but could you elaborate some more about the probes? One or two HV probes? 100x atenuation low voltage ones? Just 10 x attenuation would perhaps be enough on the secondary?

Last, the other way -getting an isolated scope (Owon VDS1022i)

https://eleshop.eu/owon-vds1022i-usb-oscilloscope.html

I like this way because it would be the safest one for my benchtop scope, and being so portable I could use it for automotive work, I think. The specs seem to be good enough for most PSU work. Max input voltage is 400V, but p-p this time. Attenuation goes to 500x. Channel isolation would be 100:1 at 50 Hz and 40:1 at 10 MHz.

However I'm not sure if differential probes can be used, I looked for Testec TT-SI 7002 probes and I can see the warning "Only connect to a grounded measuring device!". Which wouldn't be the case with this isolated scope. Any comments about probes to use with this device?

[wasedadoc]

When working on SMPS I isolate it by running from a UPS whose mains plug is not plugged in. No problem to attach grounded scope probe anywhere.  I have a 100x probe for looking at transformer primary side.

[Zero999]

Yeah. I meant, it provides visual warning, perhaps somewhat like the incandescent bulb method, where you can know capacitors have been charged. And it would limit the current more. I know it's not exactly the same, sorry. I still need to look at it more closely.

The earthing could be to some chassis wich is also earthed from mains, but the secondary side would still have high impedance thanks to that couple of high value resistors, it isn't?

That's the idea. The resistors will limit the current to a safe value, which will reduce the shock and the lamp will provide an indication when either live conductor is earthed.