How to wire and test an isolation transformer

[s/v Jedi]

I have been offline for a while due to travel, so here’s a bit delayed response to the grounding issue.

First, I did not create a new ground return path. The ground wires that go to the isolation transformer are not interconnected inside the transformer.

Second, an isolation transformer is not for creating a floating power network. It is designed to create complete galvanic isolation of the boat from the shore. So all three conductors: hot, neutral and ground are all completely isolated from shore power.

Third, there are two kinds of appliances: grounded and double isolated. A double isolated appliance does not have a ground pin at the plug and is completely safe, but a grounded appliance has exposed metal surfaces and when a fault develops, one of the power carrying conductors can short to the case.

As we have eliminated ground as a return path, we can safely touch that metal surface, even when standing on a wet floor. However, in the very unlikely even that a second appliance creates a fault and shorts the other power carrying conductor to the metal case, we now have two metal surfaces with a lethal voltage between them and the risk of electrocution exists.

This risk is eliminated when, like I recommend, ground conductors are installed to the outlets. Even though this ground is no longer bonded to earth, it still is a chassis ground and when the above mentioned faults occur, it will complete a short and the breaker will trip, preventing electrocution.

Someone, I think GBN, commented that he was against installing this chassis ground safety feature, claiming that it is less safe than not having it. Unfortunately he failed to provide reasoning for that claim and he will never do so because it simply isn’t true. I described the safety provided, there is no negative other than the cost and weight of the conductor, which is negligible.

Furthermore, he claimed that a RCD type breaker at the isolation transformer would provide protection against this. This however is INCORRECT as well. You will be electrocuted all the way down to ashes and this RCD will not trigger.

Here is why: a RCD (ELCI, GFCI, RCBO) device triggers when the current through one of the conductors becomes different than the current through the other conductor. This difference must be more than a certain value to trigger the device. For a GFCI outlet this is 5mA, for a system wide ELCI, RCBO etc. it is 30mA.
These devices are fantastic but they have no function behind an isolation transformer. It’s just craziness to claim this, because the only way that the current between the two conductors becomes different is when another return path, like ground, exists. As we have removed that path, the current is the same and the device will not trigger.

To be sure this is understood, here is how the electrocution would happen: from the first current carrying conductor at the transformer, it goes through the output breaker, to the outlet, to the metal surface of the appliance, to your hand and through your body to your other hand that touches the metal ofa second appliance that has a fault to the OTHER current carrying conductor, back to the transformer output breaker and into the transformer.

As you can see, current in and out of the transformer is equal and changing this breaker to a RCBO does not change the outcome: you are electrocuted.

Now let’s assume one needs compliance with ABYC and has installed a jumper between one of the current carrying conductors of the transformer output and ships ground. We call this creating a polarized network and the grounded conductor is called the Neutral and the other the Phase or Hot or Live. So now the chassis ground has become an alternative return path for neutral.

And as the jumper is located at the transformer, before the output breaker, this return path via ground can bypass the breaker and a RCBO would trigger at 30mA, which will still electrocute you but probably not kill you.

But in reality this doesn’t happen, because you have ground conductors installed and as soon as an appliance develops a fault between the Phase and it’s exposed metal, this creates a short via the ground conductor and the breaker trips.

So even when you have grounded the neutral, a RCD type breaker at the transformer output is pure nonsense. This is not a matter of “it’s so cheap so why not just do it” because it is nonsense. It would be crazier than installing sprinklers under water because those would at least still spray water under water, while the RCD type device simply never gets triggered.

So here it is. While I agree that a floating network (floating neutral) behind an isolation transformer is good, it does require double pole breakers and grounding the neutral behind the transformer does not reduce protection by the transformer in any way. Only when you connect your ships ground to shore ground you make a potentially fatal mistake. Even using a galvanic isolator instead of a transformer increases danger significantly, nothing beats the protection provided by the isolation transformer.

[s/v Jedi]

Originally Posted by sailing_jack

Since the isolation transformer is the power source in the boat. Any ground fault current must find its way back to the source. Having a common ground providing this path is a good idea IMHO

Exactly. I hope my last post makes this clear for everyone

[team karst]

Jedi; any thought given for a simple monitor system to check for main faults in the boat?
Being iso driven, it is single fault tolerant; meaning a single line fault to GND will not cause a GFCI trip (no current flow). Its the second fault to the other high line that will trip the GFCI (or even your main DP breaker), as the current exceeds 5 mA.
A couple of low current LED based lamps might do the trick.

[s/v Jedi]

Originally Posted by team karst

Jedi; any thought given for a simple monitor system to check for main faults in the boat?
Being iso driven, it is single fault tolerant; meaning a single line fault to GND will not cause a GFCI trip (no current flow). Its the second fault to the other high line that will trip the GFCI (or even your main DP breaker), as the current exceeds 5 mA.
A couple of low current LED based lamps might do the trick.

Yes, this isn’t hard to do but remember it does not improve safety: these monitors function is to improve power availability. When you have an alarm on the first fault, you can fix that (remove appliance) before a second fault develops that may trigger a breaker with loss of power as the result. This is a really bad thing during a medical procedure or when you are a war ship and missiles are incoming, but on a cruising sailboat, losing AC power for a bit is mostly just a nuisance.

Another point to consider is that when you ground one conductor to make it the Neutral and conform to ABYC or to allow single pole breakers (questionable as well) then you create the first fault on purpose.

All this doesn’t negate the point that installing a monitor enables one of the strong points of a floating network

[HeywoodJ]

An RCBO at the transformer or panel will not catch the fault (death) in your scenario, but a GFCI at each point of use (PoU) would. This is why many model codes require point-of-use GFCI protection in damp or wet areas, or other areas at risk of alternate return paths developing. While a PE return path may be a good thing in general, it will only clear a low-resistance short, higher resistance faults can still present a risk under your scenario, and a PoU GFCI can clear those faults when a breaker doesn't.

The risk/nuisance/cost equation is up to you to decide.

[Bowdrie]

Originally Posted by HeywoodJ

An RCBO at the transformer or panel will not catch the fault (death) in your scenario, but a GFCI at each point of use (PoU) would. This is why many model codes require point-of-use GFCI protection in damp or wet areas, or other areas at risk of alternate return paths developing. While a PE return path may be a good thing in general, it will only clear a low-resistance short, higher resistance faults can still present a risk under your scenario, and a PoU GFCI can clear those faults when a breaker doesn't.

This is one of the key features in my boat, ALL onboard AC devices are using a GFCI as a "feed thru" device as well as the normal outlets.
This is in addition to dockside and boatside RCDs.
The individual breakers feed the Line on the GFCI, and the device is hardwired to the Load side.
All of the GFCIs are installed in weatherproof boxes up high in the boat, (there is NO AC wiring below waist/chest height,) and utilize the gasketed hinged covers.
Nuisance trips are virtually non-existent, and it's easy to go round and push the test buttons every month or so.

[s/v Jedi]

Originally Posted by HeywoodJ

An RCBO at the transformer or panel will not catch the fault (death) in your scenario, but a GFCI at each point of use (PoU) would. This is why many model codes require point-of-use GFCI protection in damp or wet areas, or other areas at risk of alternate return paths developing. While a PE return path may be a good thing in general, it will only clear a low-resistance short, higher resistance faults can still present a risk under your scenario, and a PoU GFCI can clear those faults when a breaker doesn't.

The risk/nuisance/cost equation is up to you to decide.

The codes you talk about are for shore side installations without isolation transformer, so out of the scope of this thread.

If both appliances are on the same GFCI outlet, which is a high possibility in case you are to touch both simultaneously, then the GFCI will not protect you.

Also, please describe the potentially dangerous scenario that you claim exists for my diagram in case of a higher resistance fault but are cleared by a GFCI outlet.

And last but not least, please describe how the test button of a GFCI outlet works for installation behind an isolation transformer

[team karst]

I suspect that very few Master Electricians have really dealt with galvanically isolated power systems, and even fewer casual users. It might help to understand why the vast majority of power systems are polarized, ie; one side earthed.

First; most all land based systems, including marinas, are fed from higher voltage feeders. It may even start from 238kV and go thru several transformers before 240V is achieved. So, one risk is already in place. A transformer fault from primary to secondary. If the secondary is "floating", then you may well have 14kV instead of 240V at your home receptacle! So, to prevent that catastrophic issue, one leg of the secondary is earthed, to provide fault current to pop a primary side fuse. Another problem is a bare primary wire falling on a secondary wire. And, what would happen if the grounded conductor opens just upstream of your step-down transformer??

Second; Our friend lightning. One mitigation strategy is to provide a low impedance path for lightning thu-out the distribution system. Again, done with bonding to earth at many points in the system. If this is not done, a direct hit will allow much higher voltages to develop in the system.

So, you see that it it quite difficult to engineer a large scale system with truly isolated conductors from earth. In a "small scale" system, the story changes. Perhaps there is no threat from a high V primary to secondary fault. Perhaps the risk of lightning is low at the secondary system. And, you can enjoy the benefit of isolation. Meaning, you can touch either (but not both) ungrounded conductor and not receive a shock. You can't do that with a polarized system.

There are some second order effects in isolated systems that would be a bit much to explore here. They involve Common Mode rejection due to the high Z to Earth of the distribution over large areas. Basically, noise may be more of an issue on larger systems. This is not limited to ac systems. Power plants and substations use isolated DC systems very frequently, and not as a shock prevention strategy.

hope this helps.

Question for the group. What ac V reading do you get if you meter s/v Jedi ac leg(s) to Earth/seawater?

[s/v Jedi]

Originally Posted by team karst

I suspect that very few Master Electricians have really dealt with galvanically isolated power systems, and even fewer casual users. It might help to understand why the vast majority of power systems are polarized, ie; one side earthed.

First; most all land based systems, including marinas, are fed from higher voltage feeders. It may even start from 238kV and go thru several transformers before 240V is achieved. So, one risk is already in place. A transformer fault from primary to secondary. If the secondary is "floating", then you may well have 14kV instead of 240V at your home receptacle! So, to prevent that catastrophic issue, one leg of the secondary is earthed, to provide fault current to pop a primary side fuse. Another problem is a bare primary wire falling on a secondary wire. And, what would happen if the grounded conductor opens just upstream of your step-down transformer??

Second; Our friend lightning. One mitigation strategy is to provide a low impedance path for lightning thu-out the distribution system. Again, done with bonding to earth at many points in the system. If this is not done, a direct hit will allow much higher voltages to develop in the system.

So, you see that it it quite difficult to engineer a large scale system with truly isolated conductors from earth. In a "small scale" system, the story changes. Perhaps there is no threat from a high V primary to secondary fault. Perhaps the risk of lightning is low at the secondary system. And, you can enjoy the benefit of isolation. Meaning, you can touch either (but not both) ungrounded conductor and not receive a shock. You can't do that with a polarized system.

There are some second order effects in isolated systems that would be a bit much to explore here. They involve Common Mode rejection due to the high Z to Earth of the distribution over large areas. Basically, noise may be more of an issue on larger systems. This is not limited to ac systems. Power plants and substations use isolated DC systems very frequently, and not as a shock prevention strategy.

hope this helps.

Question for the group. What ac V reading do you get if you meter s/v Jedi ac leg(s) to Earth/seawater?

Great post

I can measure that from our bonding plate to the AC legs but unfortunately we are on the hard at the moment…

[goboatingnow]

Originally Posted by sailing_jack

Since the isolation transformer is the power source in the boat. Any ground fault current must find its way back to the source. Having a common ground providing this path is a good idea IMHO

This illustrates a fundamental ms understanding the presence of earth in a fsukth return path is a historical artifact to ensure fuses blew on hot failures

The whole point of an isolating , is to remove esrtb as a fault return. Rcbos can be used tomprotected against inadvertent or unexpected return paths

GIVEN YOUR ISOLATING transformer removes ground as a return fault path it’s the height of sillyness to then Bypass the sillyness and reground the the transformer to seawater or ships ground

This grounding of the isolating Trafford introduces a potential Fatal return path thst didn’t exist.

Do any so called proper “ reference diagram “ should (a) terminate shore ground at the shore power receptacle , (b) preferably house traffo in a plastic case , with no connection to earth , no other appliances should be case earthed but will be protected by the output rcbos on the Traffo

Hence Jedi diagram is fine as except for the common earth busbar

[s/v Jedi]

Originally Posted by goboatingnow

GIVEN YOUR ISOLATING transformer removes ground as a return fault path it’s the height of sillyness to then Bypass the sillyness and reground the the transformer to seawater or ships ground

This grounding of the isolating Trafford introduces a potential Fatal return path thst didn’t exist.

Do any so called proper “ reference diagram “ should (a) terminate shore ground at the shore power receptacle , (b) preferably house traffo in a plastic case , with no connection to earth , no other appliances should be case earthed but will be protected by the output rcbos on the Traffo

Hence Jedi diagram is fine as except for the common earth busbar

Aha! Now it becomes clear, you don’t understand “ships ground” Ships ground is a chassis ground and not earthed!

Yes, if you would earth ships ground (with a plate under water) then you reintroduced a return path. If you check my diagram then you will see there is no earthing done for ships ground: it’s just a conductor interconnecting all the ground pins of outlets, metal housings of transformers, inverter/charger etc.

[goboatingnow]

Originally Posted by s/v Jedi

Aha! Now it becomes clear, you don’t understand “ships ground” Ships ground is a chassis ground and not earthed!



Yes, if you would earth ships ground (with a plate under water) then you reintroduced a return path. If you check my diagram then you will see there is no earthing done for ships ground: it’s just a conductor interconnecting all the ground pins of outlets, metal housings of transformers, inverter/charger etc.

Explain to me the electrical purpose of a ships groin point in a properly wired isolating transformer with emphasis of gaily circuit paths.

[s/v Jedi]

Originally Posted by goboatingnow

Explain to me the electrical purpose of a ships groin point in a properly wired isolating transformer with emphasis of gaily circuit paths.

Me and others already did: protect against electrocution after second isolation fault in an appliance.

[goboatingnow]

Originally Posted by s/v Jedi

Me and others already did: protect against electrocution after second isolation fault in an appliance.

Fault in an appliance does not pose a shock risk with an isolate. Supply. Postulating two simultaneous fault is rather “ reaching “ and RCBOs protraction will cover thst anyway

[flee27]

Following what you are saying about not creating an earth ground because of no grounding plate. Is is correct to say that one could not have metal through hulls or other under water metal bonded like my boat has. This basically creates an earth ground? Correct?

Thanks for lessons.

Foster