Marine Galvanic isolator help

[MadScientist]

Hi guys. I’m looking at building a galvanic isolator for my boat , the commercial ones all have poor monitoring solutions

Here’s the basic idea ( sometimes 4x diode drops are used )


This device is inserted into the PE line leaving the boat and prevents small DC currents from flowing , the capacitor is soley as a fail safe system if the diodes go open circuit.

What I want to do is monitor the diodes for two things

(A) a diode  drop voltage appearing , which indicates there are currents flowing

(B) a detection system that detects the diodes have failed either open or shorted

The monitoring needs to be isolated for input into a micro

(a) is straightforward , (b) is where I could do with some ideas. I could inject a small dc current occasionally to test diode operation etc.

The solutions have to withstand the full fault mains voltage and current.

Any other thoughts on how to approach it.

[Ian.M]

(A) is far from straightforward as a 'diode drop voltage' does not indicate significant currents are flowing unless it becomes excessive, it simply indicates the diodes are doing their job.   There will inevitably be a DC voltage offset across the diodes due to the boat's anode(s) impressing an offset voltage on all grounded metalwork, with the possibility of a further offset of the marina TT supply PE (ground) if the marina uses cathodic protection systems for sheet-piling or other structures.

You are going to run into trouble with AC ground leakage current from any equipment with either a grounded output and a SMPS, or with a 3 wire mains lead and an EMI filter on the supply input with Y capacitors to ground.  To avoid the diodes being biased into conduction nearly half the time by half-waves of the AC leakage current aiding the DC offset, you should probably put a high value bipolar capacitor across the diode stack to bypass AC leakage currents.

(B) is trickier.  If you apply your test pulse to the middle of each diode string via a capacitor and carefully control the voltage swing and thus the charge pumped by each pulse, you can get the test pulse charge transfer to balance out to a high degree of accuracy, provided there is no diode fault, yet still get a high enough test current to unambiguously determine the state of the diodes by monitoring their Vf drop during the test pulse.  Some simulation and experimentation may be required . . .

[CaptDon]

I should think that all National Unified Electrical Codes (no matter what nation) would prevent any such circuit breaking the intentionally low d.c. resistance path back to P.E. earth ground. There are a lot of extra rules that apply to marinas here in the U.S. effecting grounding and power distribution and proposed and accepted by both the N.F.P.A. and N.E.C.C. Some local code rules may go above and beyond, like the 'green wire nut for ground, no other color wire nut acceptable. Or 'the ground wire closest to a home run to the breaker panel shall be the one protruding through the green wire nut and shall be the one anchored to the metallic enclosure.' I feel for your situation. The boat I run is fed by 208 3ph and we have a ground/neutral amperage indicator. Due to the age of the boat its ground and neutral are bonded everywhere and typically I can measure about .25 to .5vac between the hull of our boat and the metal pier. We had zincs on the boat which are useless in fresh water. I think they were just replaced during our 5 year hull inspection with magnesium. Strange that a small metallic boat (say 27ft or less) in fresh water can accidentally have the metallic hull connected directly to 120vac hot line and not trip any breakers but still surely electrocute any swimmers within about 5 to 10 feet of the hull. Salt water on the other hand acts like a dead short and will trip the breaker. I think you will be stuck with some kind of a current transformer detection circuit on your low resistance/impedance protective earth. Cheers mate!!
 

[CaptDon]

So after investigating the commercial units I wonder if one of the A.C. lines powered digital panel meters which have their metering inputs isolated from the A.C. lines would be of any help? You could perhaps use the readily available 0-2vac and perhaps a 0-2vdc panel meter pair to look 'across' your isolator?. Because we are in the 3 phase 100+ amps per leg category a huge dockside weatherproofed housing isolation transformer would be our only option so we are basically financially bound to sacrificial anodes. As said in my other post we do in fact see .25 to .5vac between pier and hull. This is mainly the imbalanced phase current across the d.c. resistance of our shore cord (I X R = V). I was unaware the smaller non-commercial boats were using isolators. None of my friends have them on their pleasure boats, maybe they should? We are trending away from zincs in fresh water and I think they are even starting to rule against them like they did with the copper element hull paint. If I recall correctly we now use magnesium. Cheers!!

[MadScientist]

Let’s not go down NEC ABYC debates. NEC rules are US based and I’m not

GIs are common on small boats in both US and Europe , many are fitted in our marina.

Couple of things

1. There is normally no current flowing in the PE circuit due to galvanic action particularly as European boats do not have PE grounded in the boat ( nor any boat side connection to neutral ) and such boats not have underwater metals bonded together , this is a far better situation then the US practice. Far less galvanic action as a result. Of course there are always inadvertent paths from the boats PE to seawater but that’s the whole point of the GI.

AC currents will of course be shunted by the capacitor

Hence with two diodes or four there will normally be NO conduction and hence no voltage drop. This is not up for debate I’ve measured it

Hence ( A) is quite easy determined by sensing across the diodes , a 2x or 4x diode drop is a fault and indicated as so. No drop no fault , assuming functioning diodes , anytime the diodes conduct its an indication of a GI fault ( minor spikes aside )

( b) is the issue.

My thoughts were to inject every 30 minutes a small DC pulse in each direction with a voltage greater then the combined voltage drop , of the diodes conduct they are good. If not I have a fault

[Ian.M]

Sorry, I missed you already had a bypass capacitor across the diode strings.  However it is *NOT* a failsafe as it will be incapable of carrying the hundreds of amps possible during a bolted fault so may prevent upstream breakers tripping.  Of course upstream GFCIs (RCDs) will still trip as they trip on current imbalance in the live conductors, not on current in the Line conductor alone.

DC ground offsets do exist simply due to the electrode potentials of ground electrodes of different metals or alloys.  The magnitude of the DC offset isn't a good proxy for the current through the diodes as their Vf vs If characteristic is highly nonlinear and both temperature dependent and dependent on the die structure and doping.

Never the less, if you decide on a maximum DC current you can tolerate (based on the rate of galvanic corrosion of your exposed anode area being small enough to not excessively reduce anode life), you can find a worst case minimum Vf for the string at that current and use that as a threshold for fault detection, you can detect some electrical faults, both on-board and with the shorepower supply.

[floobydust]

Lifting ground with 15A branch feed, I use a 25A bridge rectifier with additional diode(s) across the (+) and (-) if higher blocking voltage is required. You can also stick LED's (plus dropping resistor) across it to monitor surprises. And there's the EMI cap say 0.1uF across it all.
I would check your galvanic potentials, aluminum and iron salt-water is greater than 2 diodes I believe.

[Zero999]

The idea of the circuit is to block the DC due to the small voltage generated by the electrochemical cell formed between the boat's hull and the earth electrode on land. Low DC voltages will be completely blocked by the diodes and the capacitor will pass the AC, with a low voltage drop.

The capacitor just passes high frequencies to eliminate EMI and can be a low value. It doesn't need to have a high voltage rating.

Another thing to not is, an RCD capable of DC operation should be used, that way it'll still trip if one of the diodes fails. A hall effect sensor should be used, rather than a current transformer, in your circuit, so it works on DC.

[floobydust]

I notice large ships (docked) are hard-connected to PE? but cathodic protection involves local sacrificial anodes on the hull, which are driven with DC in the "impressed current" scheme.

[MadScientist]

Lifting ground with 15A branch feed, I use a 25A bridge rectifier with additional diode(s) across the (+) and (-) if higher blocking voltage is required. You can also stick LED's (plus dropping resistor) across it to monitor surprises. And there's the EMI cap say 0.1uF across it all.
I would check your galvanic potentials, aluminum and iron salt-water is greater than 2 diodes I believe.

Yes I’m of the view after measuring things that 4 diode drops are needed. This is typically what’s in commercial GIs , ie two bridge rectifiers , on EU boats PE is not connected to any boat metals unlike US boats so the potential should be far less especially on dry GRP hulls . My PE goes to a battery charger , water heater and sockets , that’s it. The water heater is the biggest risk of unintended leakage paths, but the plumbing is all plastic , on ABYC compliant boats underwater metals are bonded together and ABYC requires PE to be linked to DC negative. This opens up far greater issues around impressed corrosion , and is the result of US boats fixation on sea cock failures for example. The ISO / RCD Eu standards based around the assumption of an RCBO are a better compromise between impressed corrosion and mains safety.

The cap is there to allow AC fault currents if the diodes fail , and hence the conduction through the cap trips the RCBO

yes monitoring the voltage drop is easy enough , though I want an isolated detection , as I have a boat wide  LIN  bus for slow speed monitoring systems.

The issue is a practical circuit to detect failed diodes. Commercial systems don’t bother as it’s complex but the problem is simple voltage drop monitoring , cant differentiate between a system with no drop and open or shorted diodes , ie if the diodes short the voltage monitor looks ok

[MadScientist]

The idea of the circuit is to block the DC due to the small voltage generated by the electrochemical cell formed between the boat's hull and the earth electrode on land. Low DC voltages will be completely blocked by the diodes and the capacitor will pass the AC, with a low voltage drop.

The capacitor just passes high frequencies to eliminate EMI and can be a low value. It doesn't need to have a high voltage rating.

Another thing to not is, an RCD capable of DC operation should be used, that way it'll still trip if one of the diodes fails. A hall effect sensor should be used, rather than a current transformer, in your circuit, so it works on DC.

No the cap was added by ABYC as a “ fail safe “ ie that in the event of a diode failure an AC fault path still exists.

Good point re dc RCBO , I’m redoing my AC panel and will be upgrading the breakers , I’ll spec the appropriate RCBO

[CaptDon]

Floobydust, your circuit diagram is very similar to our commercial 140 passenger vessel. Four of my anodes are directly attached to the hull and are easily replaceable. One larger anode has a controllable bias applied to it and in my wheelhouse I have a meter with a 'green' sweet spot in the middle and 'red' on either end. One end of the meter indicates 'Under Protected' the other end indicates 'Over Protected'. I don't think once I did the initial adjustment (front panel knob) I have ever had to readjust it except this year when we went from zincs to magnesium. One adjustment and it seems stable again. As mentioned before, I do have .25 to .5vac between the metal hull and the metal pier. The implementation of our neutral being bonded to ground at both ends of our 208 3ph shore cord really sucks. If we ever lose both neutral and PE ground we'll be in big trouble, so will swimmers!!!

[Zero999]

The idea of the circuit is to block the DC due to the small voltage generated by the electrochemical cell formed between the boat's hull and the earth electrode on land. Low DC voltages will be completely blocked by the diodes and the capacitor will pass the AC, with a low voltage drop.

The capacitor just passes high frequencies to eliminate EMI and can be a low value. It doesn't need to have a high voltage rating.

Another thing to not is, an RCD capable of DC operation should be used, that way it'll still trip if one of the diodes fails. A hall effect sensor should be used, rather than a current transformer, in your circuit, so it works on DC.

No the cap was added by ABYC as a “ fail safe “ ie that in the event of a diode failure an AC fault path still exists.

Good point re dc RCBO , I’m redoing my AC panel and will be upgrading the breakers , I’ll spec the appropriate RCBO

Then the capacitor will need to be much larger. Two large  electrolytic capacitors, with a high ripple current rating, connected back-to-back to form a non-polarised capacitor will do.