Galvanic Isolators

[VicS]

Given that topic a good thrashing this time round. Always something to learn.

THIS is the diagram that Roger Shaw refers to.

It was posted with the question "Is fig 2 an acceptable alternative way of wiring a GI"

My thinking was that fault currents were unlikely to pass though it. At least not the full current if the shorepower earth connection is good. It would not therefore be subjected to the high transient currents so damaging to its diodes.

I'll leave you all to think about it

 

[nigelmercier]

My problem with Figure 2 is that there will be mains equipment that is also connected to the battery negative. If this has any leakage between the negative and earth, then the GI is bypassed. This is exactly the problem I had on my boat.

 

[VicS]

My problem with Figure 2 is that there will be mains equipment that is also connected to the battery negative. If this has any leakage between the negative and earth, then the GI is bypassed. This is exactly the problem I had on my boat.

Good point.

 

[HappyHunter]

Barbican - check PM

 

[laika]

As there's another thread on Galvanic Isolators running thought I'd add a coda to this. I bought a safeshore and it turned up very promptly. It sat there for week or so while I did Other Things until I overheard a couple of people talking about what a nightmare Brighton marina is for galvanic corrosion so it's now rigged up in a temporary location.

Temporary because I didn't really consider the size of it. It can handle more current than necessary and is probably physically bigger than I need and frankly finding somewhere to put it is hard. I could mount it in the void above the galley ceiling board where the shore power comes in but there's no airflow there. Or I could move everything around in the cockpit locker where the RCD is (the wall just under the combing is a bit "busy" with Stuff) but it can be damp in there (the RCD's got a water resistant housing). So when thinking about buying a galvanic isolator I'd say do look at the size of what's available and consider where you're going to mount it

 

[lpdsn]

My problem with Figure 2 is that there will be mains equipment that is also connected to the battery negative. If this has any leakage between the negative and earth, then the GI is bypassed. This is exactly the problem I had on my boat.

That's where the safeshore remote monitor comes into play. Nothing flash, just LEDs connected in parallel with the GI. If the GI diodes are conducting, so are the LEDs. There are two LEDs, so it will show DC leakage either way or AC leakage.

I went for a Fig 2 type installation. So the 230V Earth is directly connected to the shorepower earth, but the DC negative and boat's ground are connected through the GI. Assumption is that in any mains short circuit large currents will pass through the diodes. I believe they are generally only design to survive a quarter cycle, so about 5ms. Enough time for the RCBO/RCD to trip.

I'm happier that for the 230V circuits I don't have to rely on the GI, yet if somehow 230V AC gets onto the 12V circuits, then the GI will provide a safe route to Earth for a few ms anyway.

Only snag I've identified so far is what would happen if there was a fault with the 230V immersion heater. That might cause earth leakage and corrosion. Easy at the moment as I have no immersion heater. When I fit one I may even fit a second GI just for that circuit.

And the GI definitely has reduced the anode wastage. When I first bought the boat, two shaft anodes didn't last a year. Now I replace one each year and they last about three years.

 

[VicS]

Only snag I've identified so far is what would happen if there was a fault with the 230V immersion heater. That might cause earth leakage and corrosion. Easy at the moment as I have no immersion heater. When I fit one I may even fit a second GI just for that circuit.

The RCD should trip.

I am not quite sure what you have in mind when you say you might for a second GI for the immersion heater. I cant see where there will be any "galvanic " currents to block