Understanding the risks/trade-offs of a non-grounded Isolation Transformer

[shaunakde]

I was watching a video about Isolation Transformers and understand there are two types of isolation transformer devices on the market: 1. medical and 2. tech/service/lab grade. The difference seems to be that in the medical grade, the Neutral and Ground are bonded to one another, while on the other they are not.

If my understanding of this device is correct, as long as the ground and neutral are connected, probing a live circuit with say an oscilloscope still has the same risks as doing so directly from mains voltage. In this case, the transformer is more to break ground loops and remove noise from the supply lines / ensure all devices are bonded to the same ground point.

Where as, if the ground is not bonded, the secondary side of the transformer is floating, and it should be okay to measure any live circuit with an oscilloscope.

A comment on the video mentioned that there are different trade-offs and risks of doing this as well. As far as I can think, the main danger is that there is no grounding protection anymore, so touching exposed metal parts of the scope (like the BNC connector, or exposed parts of the case) could be dangerous. Am I right in my assessment? Are there more risks to this kind of a setup?

Additionally, will using a GFCI socket (ground fault detecting socket) reduce the risks involved in this kind of isolation?

Sorry if these questions make no sense, because I completely misunderstood the topic (and please feel free to indicate this! Its all helpful in my learning process!)

Shaunak

[bob91343]

Let's reduce this to fundamentals.

1. If there is a conductive path from the secondary to the primary side, it's not an isolation transformer.

2.  The ground lead should be connected to the load chassis.  This means the ground lead of the transformer, the DUT, and the test equipment.

The purpose here is to ba able to place the accessory test equipment ground or chassis at mains ground.  When measuring places in DUT, its chassis can now be connected to test equipment ground without hazard.  What happens inside the load/DUT no longer matters.

[bdunham7]

I was watching a video about Isolation Transformers and understand there are two types of isolation transformer devices on the market: 1. medical and 2. tech/service/lab grade. The difference seems to be that in the medical grade, the Neutral and Ground are bonded to one another, while on the other they are not.

If my understanding of this device is correct, as long as the ground and neutral are connected, probing a live circuit with say an oscilloscope still has the same risks as doing so directly from mains voltage. In this case, the transformer is more to break ground loops and remove noise from the supply lines / ensure all devices are bonded to the same ground point.

Don't worry so much about the 'type' or 'grade' (medical grade means many things...) just know that for electronics service, any proper isolation transformer will have the secondary completely isolated from the primary and not connected to ground on either side.  If you choose to reference one side to ground for some specific reason, that's your option, but the transformer should not be made that way.  Typically, if there is a ground, it is passed through directly without being bonded to either primary or secondary.

Where as, if the ground is not bonded, the secondary side of the transformer is floating, and it should be okay to measure any live circuit with an oscilloscope.

Bzzzzzzt!! There's a lot of hazards there and some aren't obvious. 

A comment on the video mentioned that there are different trade-offs and risks of doing this as well. As far as I can think, the main danger is that there is no grounding protection anymore, so touching exposed metal parts of the scope (like the BNC connector, or exposed parts of the case) could be dangerous. Am I right in my assessment? Are there more risks to this kind of a setup?

Additionally, will using a GFCI socket (ground fault detecting socket) reduce the risks involved in this kind of isolation?

There are many risks.  First, an isolation transformer will completely defeat any GFCI protection on the primary side.  Secondly, if you are referring to isolating the oscilloscope, that's a very bad idea in most cases because there are many things that can go wrong.  First, yes, the scope, it's case (if metal) and anything it is connected to may float up to some high voltage.  Second, unless you use a cheater (2-prong) adapter or cut the ground pin off the cord, your scope will still be grounded with most isolation transformers.  Third, even if the scope is floating, the channel grounds are typically not isolated so if you forget and connect any of the probe grounds in different spots, you may blow up your oscilloscope anyway.

How you use an isolation transformer is very specific to the DUT, your scope (although most are similar in the relevant ways) and what type of testing you are doing.  It is not possible to give simple, generic advice on this matter, IMO.

[helius]

How you use an isolation transformer is very specific to the DUT, your scope (although most are similar in the relevant ways) and what type of testing you are doing.  It is not possible to give simple, generic advice on this matter, IMO.

I mostly agree, although there are some points that bear repeating:
1. Do *NOT* isolate or float (i.e. operate without its Protective Earth connection) a scope unless it was designed to be used that way. Examples of isolated scopes are the Tek 212 and THS700, which operate from batteries and have no exposed metal parts. All other scopes depend on the Protective Earth to protect the user from shock. This also applies to other mains-powered test equipment.

2. Whenever this topic comes up, which is disturbingly often, there is confusion about what to do with the Protective Earth of the DUT (should it be grounded or floating?). In my opinion, the only good reason to use an isolation transformer is for servicing live-chassis DUT. By definition, these appliances have no earth connection and must be isolated to service safely. The simple fact that this question arises points to a dangerous lack of knowledge about the reasons for isolation.

3. It is bad and unsafe practice to use isolation transformers in order to make measurements more convenient (probe on bridge rectifier, ground clip on neutral, etc). While this will work for a single probe, you must be extremely careful in multiple-probe scenarios, or when signals are injected. The use of differential probes, or lacking that, removing the probe ground clips and using two probes in A-minus-B mode is safer than isolating the DUT. Why? Because isolating the DUT removes the protection afforded by the Protective Earth lead for wiring faults. If the DUT is isolated, a wiring fault can occur with no indication of trouble. It doesn't matter if the DUT is earthed or not, isolation makes the ground useless as a safety device.

It can be argued that with the DUT isolated, you can safely touch a live terminal with one hand (!). Apart from live-chassis appliances (where isolation is mandatory), this is not much of a safety improvement. You should use appropriate insulated grabber accessories for your probes and wear standard rubber lab gloves which will eliminate the possibility of contacting live terminals with the skin.

[VooDust]

How you use an isolation transformer is very specific to the DUT, your scope (although most are similar in the relevant ways) and what type of testing you are doing.  It is not possible to give simple, generic advice on this matter, IMO.

I mostly agree, although there are some points that bear repeating:
1. Do *NOT* isolate or float (i.e. operate without its Protective Earth connection) a scope unless it was designed to be used that way. Examples of isolated scopes are the Tek 212 and THS700, which operate from batteries and have no exposed metal parts. All other scopes depend on the Protective Earth to protect the user from shock. This also applies to other mains-powered test equipment.

2. Whenever this topic comes up, which is disturbingly often, there is confusion about what to do with the Protective Earth of the DUT (should it be grounded or floating?). In my opinion, the only good reason to use an isolation transformer is for servicing live-chassis DUT. By definition, these appliances have no earth connection and must be isolated to service safely. The simple fact that this question arises points to a dangerous lack of knowledge about the reasons for isolation.

3. It is bad and unsafe practice to use isolation transformers in order to make measurements more convenient (probe on bridge rectifier, ground clip on neutral, etc). While this will work for a single probe, you must be extremely careful in multiple-probe scenarios, or when signals are injected. The use of differential probes, or lacking that, removing the probe ground clips and using two probes in A-minus-B mode is safer than isolating the DUT. Why? Because isolating the DUT removes the protection afforded by the Protective Earth lead for wiring faults. If the DUT is isolated, a wiring fault can occur with no indication of trouble. It doesn't matter if the DUT is earthed or not, isolation makes the ground useless as a safety device.

It can be argued that with the DUT isolated, you can safely touch a live terminal with one hand (!). Apart from live-chassis appliances (where isolation is mandatory), this is not much of a safety improvement. You should use appropriate insulated grabber accessories for your probes and wear standard rubber lab gloves which will eliminate the possibility of contacting live terminals with the skin.

Thanks, that was very enlightening.

Is my understanding correct that even with an isolation transformer, I get zapped if I touched Plus and Minus of the DUT? (at 220V I mean).

[shaunakde]

Thank you VooDust, helius, bdunham7, bob91343 for your replies. It really helped me understand a few things about Isolation transformers and electricity in general. Unfortunately, I have more questions now 

1. If there is a conductive path from the secondary ...

Interesting. Am I right in assuming that the purpose of the devices that have their ground bonded are less for electrical safety, and more for signal integrity?
 

First, an isolation transformer will completely defeat any GFCI protection on the primary side

I did not think of this at all. After digging a bit more, I now understand that GFCI protection works by checking the difference in current between L and N, and thus the Isolation Transformer would completely disable this protection since all current in the primary side would be returned, independent of any leakage in the secondary side. I hope I am correct in this line of reasoning.

It is not possible to give simple, generic advice on this matter, IMO.

I do agree. It does feel like one of those things where if you have to ask, you probably are going to seriously put yourself at risk. From what I understand, it is far better to use a differential probe than to "float" the scope (unless you really know what you are doing).

Whenever this topic comes up, which is disturbingly often... isolation makes the ground useless as a safety device

Before I made this question, I also came across various threads (mostly from people working on valve amplifies etc.) that seemed to give the floating scope advice. I was sceptical and that's why I decided to go down this rabbit hole. I think my takeaways from your answer are - just don't. Pay for a differential probe.

I do have a question around what would be a common scenario where you would need the device to be galvanically isolated from the mains. (And what does that even mean - intuitively). I looked at Wikipedia and this is what they had to say:

Isolation transformers provide galvanic isolation and are used to protect against electric shock, to suppress electrical noise in sensitive devices, or to transfer power between two circuits which must not be connected.

The part I am confused about is how they protect against electric shock. The one path of current flow it does reduce is between live and earth ground (through one's feet), but other than that, it doesn't seem to reduce any risks.

The other application I can think of for an Isolation transformer is impedance matching, and again, google seems to spit out several commercial products in the audio space for this, but I don't fully understand why you would need over an Opto-isolator+mosfet/BJT for example. Ofcourse I do understand that in very high power circuits the transformer if probably the way to go, but for things I would be able to do on a home supply (110v 10A) - I can't really think of much that couldn't be isolated by other means.

Thank you for your patience and responses.

Regards,
Shaunak

[ArthurDent]

Here is a good description of medical/commercial grade isolation transformers and leakage currents that may clear things up. A shield between primary and secondary can make a big difference in leakage between windings and interference. I have a power supply for a specialized tube-type interference receiver where the A.C. goes first through an hermetically sealed isolation transformer with a grounded shield between windings, then to the regular power transformer for the receiver. They were obviously concerned about getting clean power to the receiver.

https://canadatransformers.com/faraday-shield/

[bdunham7]

I do agree. It does feel like one of those things where if you have to ask, you probably are going to seriously put yourself at risk. From what I understand, it is far better to use a differential probe than to "float" the scope (unless you really know what you are doing).

Before I made this question, I also came across various threads (mostly from people working on valve amplifies etc.) that seemed to give the floating scope advice. I was sceptical and that's why I decided to go down this rabbit hole. I think my takeaways from your answer are - just don't. Pay for a differential probe.

I do have a question around what would be a common scenario where you would need the device to be galvanically isolated from the mains. (And what does that even mean - intuitively). I looked at Wikipedia and this is what they had to say:

Isolation transformers provide galvanic isolation and are used to protect against electric shock, to suppress electrical noise in sensitive devices, or to transfer power between two circuits which must not be connected.

The part I am confused about is how they protect against electric shock. The one path of current flow it does reduce is between live and earth ground (through one's feet), but other than that, it doesn't seem to reduce any risks.

The origin of using isolation transformer for service and repair was older, generally vacuum tube era, devices such as TVs that had a metal chassis that was connected to either one side of the line (in which case it might be 'hot' depending on which way the plug was) or to the negative side of a mains bridge rectifier (in which case it was always 'hot').  In use, the chassis was enclosed, often in wood!  But during service, the risk to the repairman of touching the chassis was very high.  So companies like BK sold isolation transformers that would prevent electric shocks via a ground reference.  Of course  you can still get fried if you touch two points in the circuit with high enough potential differences.

You keep mentioning 'floating the scope', but the phrase "unless you know what you are doing" is misleading--it usually implies doing something you shouldn't but thinking you'll get away with it because you are careful.  The real choices are DUT isolation, differential probes or isolated (inherently floating, often battery powered), along with hands-off operation and probing (Turn it off, connect probes, turn it on, measure, turn it off, repeat as needed).  In addition to the dangers, floating a scope may result in unpredictable, noisy readings.  I commonly use my iso-variac for the DUT and an isolated scope where needed, simply because it happens to be easy for me and the iso-variac has other features (like a much lower current limit if I short something) that are handy.  The isolation also just might save me a good buzz if I get sloppy--but I give up GFCI protection. 

The other application I can think of for an Isolation transformer is impedance matching

Not relevant to mains power isolation.  Audio outputs of vacuum tube amplifiers will also generally be isolated by transformers that also provide impedance matching but that's a different subject entirely.

[iteratee]

There are many risks.  First, an isolation transformer will completely defeat any GFCI protection on the primary side.

Not necessarily. You want a primary-side GFCI. My isolation transformer has a grounded core and shield with an ordinary GFCI.

The extra secret sauce is the nanosecond secondary-side fault detector and high-z isolated secondary-side ground circut. I detect faults and lock the door before the slow-ass pixies can escape. Not an entirely trivial system though. E.g. the fastest off-fhe-shelf solid-state switches are way too slow.

Secondly, if you are referring to isolating the oscilloscope, that's a very bad idea in most cases

Meh yeah that's true (in the ubiquitously retarded cases anyway.) My scope is pretty heavily modded with its board on polyimide standoffs, numerous "effective" grounds that may or may not be "isolated" in some real or virtual manner, etc.  Not recommended for beginners and/or pussies. 

[helius]

There are some legitimate ways to use isolation, but you must be very careful. A popular (and still very expensive on ebay) tool was the Sencore PR57 iso-variac, which includes an electrical safety analyzer to measure ground bonding and leakage. An indication of high leakage may be sufficient warning that a wiring fault exists in the DUT.

[bdunham7]

There are some legitimate ways to use isolation, but you must be very careful. A popular (and still very expensive on ebay) tool was the Sencore PR57 iso-variac, which includes an electrical safety analyzer to measure ground bonding and leakage. An indication of high leakage may be sufficient warning that a wiring fault exists in the DUT.

The prices on Sencore stuff in general is batshit crazy.  The BK 1655/A is the same show, but you can still buy them new for under $500.

https://www.tequipment.net/BK1655A.html