Galvanic Isolators

[chris w]

With an isolation transformer, the BOAT EARTH is not connected to the SHORE EARTH. The SHORE Live & Neutral are connected to the input of the transformer while the SHORE Earth is connected to the iron core of the transformer over which the input and output (insulated) wire is wound but has no physical connection to the transformer output.

 

The BOAT derives its power from the output of the isolation transformer. The BOAT earth is connected to either one of the output connections. This means one of the outputs is now at the same voltage as the boat (we'll call that our zero) or Boat Earth and the other connection becomes the live.

 

This mimics the normal situation that occurs in normal household mains. The live and neutral are delivered to your house and, at the substation, the neutral is connected to planet earth. So only the live is live(!) if you follow me. Without one of them being connected to earth, both output leads would be "live" if you touched them.

 

To understand why, you need to understand that voltages are measured "relative" to something. When we say a mountain is 10,000 feet high, it is normally assumed we mean "10,000 feet above sea level". If we stood on the summit of the next highest mountain in the range, we might say the tallest mountain is 2,000 feet high meaning 2,000 feet higher than where we are standing.

 

By connecting the boat hull (boat earth) to the neutral output of the transformer, these two points must therefore be the same voltage. So when we say the live output is, for example, 240v we mean it is 240v higher than our reference point which is the boat hull.

 

Since there is now no connection between the hull and the shore earth, no galvanic action can take place as there is no physical connection to the shore anymore.

 

With regard to the point that "Idleness" raised, there is not 120v ac across your galvanic isolator. In simple terms, if you remove the earth and measure it, as Idleness suggested, it will appear to float to the mid-point because it isn't connected. Your electronic equipment earths WILL be connected (to the boat earth) and therefore they will be at the same voltage as the boat which is our reference zero.

 

Chris

[GUMPY]

 

 

With regard to the point that "Idleness" raised, there is not 120v ac across your galvanic isolator. In simple terms, if you remove the earth and measure it, as Idleness suggested, it will appear to float to the mid-point because it isn't connected. Your electronic equipment earths WILL be connected (to the boat earth) and therefore they will be at the same voltage as the boat which is our reference zero.

 

Totally correct, Im not good at explaining things. The "voltage" will still have to pass the GI to be bonded so that it will still cause the GI to conduct.

I have put a scope on my system and was amazed just how much RFI there was from the SM PSU causing I believe the GI to conduct.

What amazed me was thet I hadnt thought about RFI leakage before as we have mains inverters ( not DC to mains but 415v to 415v ) here at work that cause RCDs to trip because they have EMC filtering to comply with EU rules

BTW good explantion of the iso transformer.

 

 

J

Edited November 10, 2006 by idleness

[chris w]

The "voltage" will still have to pass the GI to be bonded so that it will still cause the GI to conduct.

 

The GI is (should be) fitted BETWEEN the shore earth and the boat earth so all equipment earths are bonded to the boat earth BEFORE they reach the GI. I'm not saying that stray AC couldn't override the GI (see my post above) but at the level of earth currents in most equipment it is extremely unlikely.

 

RCD's trip on current not voltage are are extremely sensitive to any imbalance between Live and Neutral (usually about 30mA difference). This difference is what is leaking to earth.

 

I notice there is one manufacturer who sells GI's that have 4 diodes either way meaning the galvanic (or AC leakage) voltage would have to be above about 2.4v to trigger it.

 

They do this (I presume) so that they can fit LED's across the unit (in opposite directions)in order to detect DC or AC currents. LEDs need about 2v across them before they conduct so a normal 1.2v GI wouldn't work with LEDs. If there is a DC leakage current one LED will light and if an AC leakage current both will light. (The LEDs will contain a series resistor in line with them to limit the LED current).

 

Seems like a neat way of keeping an eye on things and calming any worries (hopefully!).

 

Chris

[GUMPY]

The GI is (should be) fitted BETWEEN the shore earth and the boat earth so all equipment earths are bonded to the boat earth BEFORE they reach the GI. I'm not saying that stray AC couldn't override the GI (see my post above) but at the level of earth currents in most equipment it is extremely unlikely.

 

 

Of course that is correct but I was refering only to the return path to the supply mains as that is all that is important when discussing whether GIs work or not as I said I am not good at explaining things.

 

 

J

Edited November 10, 2006 by idleness

[Keeping Up]

With regard to the point that "Idleness" raised, there is not 120v ac across your galvanic isolator. In simple terms, if you remove the earth and measure it, as Idleness suggested, it will appear to float to the mid-point because it isn't connected. Your electronic equipment earths WILL be connected (to the boat earth) and therefore they will be at the same voltage as the boat which is our reference zero.

Surely the point is that the RFI is effectively 120 volts presented through a high internal resistance. This generates a leakage current to earth and it doesn't matter what voltage the GI needs to turn on (0.6, 1.2, 2.4 or even 4.8 volts) it will still conduct.

 

The added complication is that the GI (between boat hull and mains earth) is effectively paralleled by the path hull-water-shore. Now the question is whether the leakage current through the water will be large enough to develop (across the RF impedance of that path) enough voltage to turn the GI on (remember that the GI may be almost conducting already due to the galvanic DC voltage which it is protectingagainst). Practical measurements by people with access to the necessary equipment suggests that this voltage can indeed be enough to make the GI conduct.

 

Allan

[chris w]

the GI (between boat hull and mains earth) is effectively paralleled by the path hull-water-shore

 

The water path isn't a parallel path it is THE path. Galvanic corrosion occurs because 2 dissimilar metals (your boat steel and your neighbour's slightly different boat steel) are immersed in an electrolyte (a conducting medium, viz: canal water in this case). If a circuit is completed between the metals then galvanic corrosion will take place. The circuit IS completed the moment you and your neighbour both connect to the shore earth thus electrically joining your hulls together.

 

Depending on which steel is which, one of the boats will lose steel and one will gain it. So if your boat appears to be floating higher in the water these days, it's probably you that's losing metal

 

Galvanic corrosion voltages are typically around 0.8 volts so a pair of diodes (which need 1.2 volts to conduct) will block the corrosion current and effectively isolate you from your neighbour (or any other boats). This small "hill" will not affect any protection from large AC mains faults however, so AC protection is still viable.

 

If you have a GI fitted, simply measure the DC voltage across it. If it's less than say 1.1 volts the GI IS blocking any corrosion current (assuming there is no AC leakage).

 

Chris

[GUMPY]

Much as I hate to agree with John O, he has tried to point out that to be completely electrically safe and comply with present regulations everything should be bonded with no interruptions to the ground conductor.

 

The ONLY way of doing this is either having no galvanic protection or with an isolating transformer which is NOT on the boat but placed at the source of supply in a similar way that the yellow building site transformer is (or should be).

There are discussions going on at present for the next edition of BS7671:2001 to compile section 609 that deals with marinas and many are awaiting the outcome.

I would like to think that they will be pragmatic about this but I have my doubts and we will see it become mandatory to have continuous grounds from the supply with GIs not being acceptable due to the possible failure of the diodes under fault conditions.

 

BS7671:2001 is the IEE regulations.

 

J

[blackrose]

 

I notice there is one manufacturer who sells GI's that have 4 diodes either way meaning the galvanic (or AC leakage) voltage would have to be above about 2.4v to trigger it.

 

They do this (I presume) so that they can fit LED's across the unit (in opposite directions)in order to detect DC or AC currents. LEDs need about 2v across them before they conduct so a normal 1.2v GI wouldn't work with LEDs. If there is a DC leakage current one LED will light and if an AC leakage current both will light. (The LEDs will contain a series resistor in line with them to limit the LED current).

 

Seems like a neat way of keeping an eye on things and calming any worries (hopefully!).

 

Chris

 

That's the one I mentioned earlier when Gibbo was asking how it managed to light the leds.

 

Much as I hate to agree with John O, he has tried to point out that to be completely electrically safe and comply with present regulations everything should be bonded with no interruptions to the ground conductor.

 

The ONLY way of doing this is either having no galvanic protection or with an isolating transformer which is NOT on the boat but placed at the source of supply in a similar way that the yellow building site transformer is (or should be).

There are discussions going on at present for the next edition of BS7671:2001 to compile section 609 that deals with marinas and many are awaiting the outcome.

I would like to think that they will be pragmatic about this but I have my doubts and we will see it become mandatory to have continuous grounds from the supply with GIs not being acceptable due to the possible failure of the diodes under fault conditions.

 

BS7671:2001 is the IEE regulations.

 

J

 

I'm not up on how isolation transformers work but if they insist on no interuptions to the ground connector wouldn't they have to ban both GIs & ITs on boats?

Edited November 10, 2006 by blackrose

[chris w]

To overcome this problem (of possible failing diodes) in the US, the ABYC (American Yacht & Boat Council) insist that to pass spec the GI has to be self checking. It does this by sending an electronic signal up the earth wire and monitoring to check the signal returns down the neutral line. (Neutral and earth are joined together at the substation) This then confirms that the diodes and indeed the shore earth are OK.

 

You can get them in this country for about £500 (!) from here

 

Chris

Edited November 10, 2006 by chris w

[GUMPY]

 

I'm not up on how isolation transformers work but if they insist on no interuptions to the ground connector wouldn't they have to ban both GIs & ITs on boats?

 

Not unless they also ban isolating transformers on building sites as well.

 

The ground will have to be continuous back to the source of supply, the secondary of the isolation transformer is regarded as the source.

 

J

 

 

To overcome this problem (of possible failing diodes) in the US, the ABYC (American Yacht & Boat Council) insist that to pass spec the GI has to be self checking. It does this by sending an electronic signal up the earth wire and monitoring to check the signal returns down the neutral line. (Neutral and earth are joined together at the substation) This then confirms that the diodes and indeed the shore earth are OK.

 

 

 

Agreed, it appears that they do not disconnect the power in the event of a fault and that the monitoring is every few minutes rather than continuous so it still doesn’t allow for the fact that they could fail during the fault that is the one that kills you, You would never hear or see the alarm......

J

[Keeping Up]

The water path isn't a parallel path it is THE path. Galvanic corrosion occurs because 2 dissimilar metals (your boat steel and your neighbour's slightly different boat steel) are immersed in an electrolyte (a conducting medium, viz: canal water in this case). If a circuit is completed between the metals then galvanic corrosion will take place. The circuit IS completed the moment you and your neighbour both connect to the shore earth thus electrically joining your hulls together.

I think you either didn't read my post carefully enough Chris, or did not think about it enough.

 

We were talking about the return path to earth that is taken by RF currents from the on-board devices whose RFI filters pass it down the wiring earth to the boat's hull. These currents must ultimately find their way back to substation ground; this they can do from the hull via the GI and the shore-based earth connection, or via the parallel path from the hull through the water. In this way as far as the RFI currents are concerned, the paths ARE effectively in parallel. Being fed from a high impedance high-voltage source the RF current will generate a voltage across whatever impedance it finds; initially this will be the water path, unless the voltage so generated is enough to drive the GI diodes into conduction.

 

As I said, practical measurements (by Idleness and by Gibbo) seem to support the propsition that the voltage can indeed be sufficient to cause this effect, thus reducing the effectiveness of the GI.

 

Allan

[GUMPY]

To overcome this problem (of possible failing diodes) in the US, the ABYC (American Yacht & Boat Council) insist that to pass spec the GI has to be self checking. It does this by sending an electronic signal up the earth wire and monitoring to check the signal returns down the neutral line. (Neutral and earth are joined together at the substation) This then confirms that the diodes and indeed the shore earth are OK.

 

 

I have to assume that the signal is sent from the boat side of the diodes rather than the shore side because if its the shore side it tells you nothing exept that the bonding at the substation is good.

 

If its the boat side then how does it pass the diodes?

 

One of the problems of self checking GIs is that one then effectively ends up with a testing system that is connected to a potentially lethal earth system. If it wasn't potentially lethal then why is it being checked ? (think about it).

 

J

[chris w]

idleness

 

I have to assume that the signal is sent from the boat side of the diodes

of course

 

If its the boat side then how does it pass the diodes?

the signal is >1.2v

 

Allan

Being fed from a high impedance high-voltage source the RF current will generate a voltage across whatever impedance it finds

Not if it's fed from a HIGH-impedance source; the source impedance would drop the voltage internally. High impedance sources can only deliver minute currents. The RFI is not at high voltage anyway. The DC may be high voltage on the switched-mode charger but the AC ripple after rectification will be very tiny.

 

These currents must ultimately find their way back to substation ground

So what? Even if we accept a current from the hull to the jetty that is a different phenomenom than galvanic corrosion. Galvanic corrosion is specifically the corrosion caused by 2 dissimilar metals being in contact in an electrolyte (in this case, the canal water) and joined together externally (via the shore earth). Since the RFI currents you propose would be AC currents, there is no net electrolytic effect.

 

Chris

[Keeping Up]

Allan

Not if it's fed from a HIGH-impedance source; the source impedance would drop the voltage internally. High impedance sources can only deliver minute currents. The RFI is not at high voltage anyway. The DC may be high voltage on the switched-mode charger but the AC ripple after rectification will be very tiny.

 

So what? Even if we accept a current from the hull to the jetty that is a different phenomenom than galvanic corrosion. Galvanic corrosion is specifically the corrosion caused by 2 dissimilar metals being in contact in an electrolyte (in this case, the canal water) and joined together externally (via the shore earth). Since the RFI currents you propose would be AC currents, there is no net electrolytic effect.

 

Chris

The source impedance will be high but could still allow up to a few mA to flow, from a voltage that may effectively be anything up to 120v. These were the findings that generated the original postings (some months ago)

 

I agree that the RF ac current itself will probably not cause any corrosion. But the whole point of this disussion is the proposition that the it can drive the diodes into conduction so that they fail to protect against galvanic corosion. You could regard it as being that the sum of the DC bias and the RF ac, will cause more current to flow in one half-cycle of the RF than in the other, giving a net DC current that can cause corrosion.

[chris w]

Alan

 

But you were specifically referring to a current from the hull to the jetty. Since this would bypass the diodes, it can't drive them into conduction.

 

I of course agree that any AC current that did go through the diodes will bias them either into partial or full conduction depending on its magnitude thereby reducing or negating their effect on DC currents..

 

Chris

[Keeping Up]

Alan

 

But you were specifically referring to a current from the hull to the jetty. Since this would bypass the diodes, it can't drive them into conduction.

 

I of course agree that any AC current that did go through the diodes will bias them either into partial or full conduction depending on its magnitude thereby reducing or negating their effect on DC currents..

 

Chris

Indeed. The point I was trying to make (damned difficult translating these concepts into unambiguous words, but my skill at drawing diagrams on the PC is such that I'm staying with words for now) is that the current from the hull to the jetty MAY be enough to generate a voltage between hull and jetty that will for at least part of the cycle drive the diodes into conduction.

 

There are just to many variables including the fact that we don't know enough about the RF voltages/currents, we don't know about the impedance of the water path from hull to jetty, and we don't know how much DC bias already exists on the diodes. So we're really pretty ignorant, and we're all guessing a bit. But certainly there is a theoretical means by which the GI could be rendered ineffective. Whether it translates into practice is somewhat uncertain.

 

Allan

[chris w]

the current from the hull to the jetty MAY be enough to generate a voltage between hull and jetty that will for at least part of the cycle drive the diodes into conduction

 

I think you are getting a little confused Allan about currents and voltages. The hull voltage will not be raised by a current flowing to the jetty through the water. The jetty is clamped at zero (earth) so the hull voltage will be raised by the presence of the RF AC voltages at the hull, not because of current flowing through the water. This point (the hull) is also connected to one side of the diodes so if the magnitude of the AC voltage happens to be large enough to forward bias the diodes, it will do this regardless of whether there is a path to the jetty through the water. The alternative path (if it exists) is irrelevant to the diodes' conduction.

 

The value of any current through the water to the jetty will be determined by the magnitude of the AC voltages and not the other way round.

 

Chris

[Gary Peacock]

You know it would be far more simple to just unplug the land line.

 

If not you could always go for an isolation transformer or how about a timer and contactor to disconnect the shore power at times when it is not needed to minimise galvanic action?

[Keeping Up]

I think you are getting a little confused Allan about currents and voltages. The hull voltage will not be raised by a current flowing to the jetty through the water. The jetty is clamped at zero (earth) so the hull voltage will be raised by the presence of the RF AC voltages at the hull, not because of current flowing through the water. This point (the hull) is also connected to one side of the diodes so if the magnitude of the AC voltage happens to be large enough to forward bias the diodes, it will do this regardless of whether there is a path to the jetty through the water. The alternative path (if it exists) is irrelevant to the diodes' conduction.

 

The value of any current through the water to the jetty will be determined by the magnitude of the AC voltages and not the other way round.

 

Chris

Oh no Chris I'm not likely to be confused between currents and voltages, having myself been a qualified electronics engineer for 40 years. Your second sentence is simply incorrect, the hull voltage WILL be raised by a current flowing to the jetty through the water. It's just Ohm's law (or the ac equivalent V=IZ).

 

Any small current that flows from the live supply to the hull (whether chopped up into RF or not) and thence through the water to the jetty, will produce at the hull a potential relative to the jetty that is equal to the product of the current and the impedance of the watery path. This potential difference is effectively applied across the diodes, which may or may not be turned on as a result.

 

I suspect that other members of the forum may be losing the will to live after reading this level of detail.

 

Allan

[chris w]

Allan

 

[Next level of detail mode ON!]The impedance of the capacitors that decouple the supply will be very low at RF frequencies The resistance of the water will be many orders of magnitude greater. Therefore you have a potential divider consisting of the decoupling capacitors and the resistance of the water. To all intents and purposes therefore the hull voltage will be the same as the magnitude of the ac ripple voltages and this will drive a current through the water. The hull voltage will not be affected by this current due to the relative values of the effective potential divider. [Next level of detail mode OFF!]

 

Chris

[Gibbo]

 

Allan

Not if it's fed from a HIGH-impedance source; the source impedance would drop the voltage internally. High impedance sources can only deliver minute currents. The RFI is not at high voltage anyway.

 

 

Actually, if you do a practical test you will find that it *is* at a much higher voltage than you first assume. With the earth lifted, i.e. feeding into an infinite impedance, I measured anything between 40 and 300 volts of RFI depending upon the equipment and mode of operation.

 

The fact that high impedances can only deliver minute currents is irrelevant. Until the GI conducts there is *no* current drawn so the RFI will remain at it's unattenuated level. This is disregarding swamping effects from other connected equipment and also disregarding the parallel path through hull and water. For it is, despite your previous assertion to the contrary, a parallel path.

 

The DC may be high voltage on the switched-mode charger but the AC ripple after rectification will be very tiny.

 

Irrelevant. We are talking about the RFI thrown back down the mains earth.

 

So what? Even if we accept a current from the hull to the jetty that is a different phenomenom than galvanic corrosion. Galvanic corrosion is specifically the corrosion caused by 2 dissimilar metals being in contact in an electrolyte (in this case, the canal water) and joined together externally (via the shore earth). Since the RFI currents you propose would be AC currents, there is no net electrolytic effect.

 

Chris

 

I think you are missing the point. The theory shows that the RFI (and indeed 50Hz) presented to the earth conductor due to the potential divider of the filter caps can turn the GI on. Once those diodes are conducting in both directions they might aswell not be there for the purposes of galvanic corrosion.

 

That's the theory. The practicalities confirm the theory to be correct and that it *does* indeed happen. My GI, on my boat, showed almost flat topped white noise across the GI clipped at the clamping voltage of the GI.

 

This isn't wild speculation. It has been measured, seen and confirmed by several others.

 

Gibbo

[Gibbo]

Allan

 

[Next level of detail mode ON!]The impedance of the capacitors that decouple the supply will be very low at RF frequencies The resistance of the water will be many orders of magnitude greater. Therefore you have a potential divider consisting of the decoupling capacitors and the resistance of the water. To all intents and purposes therefore the hull voltage will be the same as the magnitude of the ac ripple voltages and this will drive a current through the water. The hull voltage will not be affected by this current due to the relative values of the effective potential divider. [Next level of detail mode OFF!]

 

Chris

 

Did you actually write what you meant to write here?

 

You seem to be saying that the hull will be at the same unattenuated AC voltage as dictated by the various filter caps. Which would, without any shadow of a doubt, turn on the GI.

 

Thankyou for agreeing with the rest of us

 

Gibbo

[chris w]

1. If the water presents a parallel path then the AC won't affect the diodes as it's bypassing them which would be a good thing from the galvanic corrosion standpoint!

 

2. You will NOT be getting 40-300v of RFI voltage. Lifting the earth is not a valid measurement as you have then just open-circuited the RFI decoupling capacitors but taken the measurement to earth!!!

 

3. Since so many devices now operate on a switched-mode principle, if you are correct then GI's would all be rendered useless. That's patently absurd. No one would be allowed to produce equipment that presented 300V of ac to earth! Stick a scope across your diodes when in normal operating mode and take a true measurement. If you then have a high ac voltage, you have some real equipment problem.

 

Chris

[tomsk]

1. If the water presents a parallel path then the AC won't affect the diodes as it's bypassing them which would be a good thing from the galvanic corrosion standpoint!

 

2. You will NOT be getting 40-300v of RFI voltage. Lifting the earth is not a valid measurement as you have then just open-circuited the RFI decoupling capacitors but taken the measurement to earth!!!

 

3. Since so many devices now operate on a switched-mode principle, if you are correct then GI's would all be rendered useless. That's patently absurd. No one would be allowed to produce equipment that presented 300V of ac to earth! Stick a scope across your diodes when in normal operating mode and take a true measurement. If you then have a high ac voltage, you have some real equipment problem.

 

Chris

 

 

Take it outside boys I love these full and frank exchanges of views on matters electrical, one of these days one may even come to a mutually agreeable conclusion, chill and drink beer, it's friday...

[Gibbo]

1. If the water presents a parallel path then the AC won't affect the diodes as it's bypassing them which would be a good thing from the galvanic corrosion standpoint!

 

That's making the rather blanket assumption that the water will have sufficient conductivity to swamp the RFI and 50Hz down below the clamping level of the GI. Rather a naive assumption. Especially when it's just an "idea" without any measurements to back it up. My measurements, and those of others, show this not to always be the case.

 

2. You will NOT be getting 40-300v of RFI voltage.

 

Really? I'll get all our equiment calibrated shall I? The measurements have been done!

 

Lifting the earth is not a valid measurement

 

Yes it is. It is a *perfectly* valid measurement.

 

as you have then just open-circuited the RFI decoupling capacitors but taken the measurement to earth!!!

 

Which is *exactly* the situation with a GI until it conducts.

 

 

3. Since so many devices now operate on a switched-mode principle, if you are correct then GI's would all be rendered useless.

 

And it is looking like that is exactly the situation. Watch this space. I am not alone in this viewpoint but it is quite recent and may have passed you by.

 

That's patently absurd. No one would be allowed to produce equipment that presented 300V of ac to earth!

 

Really? There is no limit in either CE or UL standards for voltage. Only maximum current limits. Which therefore naturally limits the size of the filter caps.

 

Stick a scope across your diodes when in normal operating mode and take a true measurement. If you then have a high ac voltage, you have some real equipment problem.

 

As precviously stated this is *exactly* what I, and others, did. You clearly didn't

 

Gibbo