Galvanic Isolators

[Gibbo]

 

Now, there may or may not be a problem with the earthing on "Blackrose's" pontoon (we're investigating) but with correct earthing on any boat, I now think there will never be a problem with earth leakage voltages if the analysis above is correct. This would also mean that adding a capacitor (while making a huge difference on the home set-up) would actually be superfluous in reality on a boat, which is in total contradiction to my original thoughts,

 

Gibbo/Allan/anyone - can you do a sanity check on this train of thought please.

 

Chris

 

I've just got back to this thread following a long session of debauchery.

 

Your conclusion is correct. The voltage on the hull is swamped by the path through the water. So the amount of voltage you will see on the hull depends upon the resistance path through the water and the source impedance of the offending equipment.

 

I did give the details in an earlier post about my findings on this on various boats. I can't be bothered looking it up now. In total I think I checked 11 boats (including my own). About 4 showed the GI foced into conduction. A couple showed the GI as marginal. The rest showed none (or at least very little) AC on the hull.

 

Gibbo

[Keeping Up]

I need to think of a way of replicating this in a valid, scaled-down version which I can do at home.

To get a fully definitive answer we should connect everything up, then measure the variations in voltage as the boat is slowly craned in and out. I'm not doing that with Keeping Up - but you could do it as your new boat is first launched.

 

Allan

[chris w]

Gibbo

 

In total I think I checked 11 boats

 

Were the boats checked against the same piece of equipment (eg: a specific charger or whatever?). If the equipment was different each time, it's less easy to be conclusive surely as the various equipments' parameters would probably be very different.

 

Chris

[Gibbo]

Gibbo

Were the boats checked against the same piece of equipment (eg: a specific charger or whatever?). If the equipment was different each time, it's less easy to be conclusive surely as the various equipments' parameters would probably be very different.

 

Chris

 

No each boat was checked as it was. I simply went onto some contacts'/friends' boats with a scope and meter.

 

I had a quick look round at what equipment they had. The ones with the GI forced into conduction tended to have tons of equipment on board (battery chargers, PCs, loads of Hi Fi equipment and TVs etc). The ones with no problem at all tended to have almost nothing (maybe a charger and some lights).

 

The idea of this rough and ready test wasn't to be conclusive about what would and wouldn't cause the problem. It was simply to see if the theorised problem did indeed exist in the real world.

 

Gibbo

[blackrose]

I've just has to explain the basics of galvanic corrosion & isolation to Ray Farrow of BW.

 

He's going to look into whether the marina is already isolated.

[blackrose]

Just had a phonecall from BW. The maina is not isolated and there is no special setup.

 

He reckoned it might be due to other boats nearby which are not earth bonded producing stray currents. I think he may have got the wong end of the stick.

[Yoda]

This subject would make a great paper for a BSc.

[chris w]

Here's a schematic of what, IMHO, is happening:

 

 

Although some of these “resistances” will be capacitively derived, I have for this simple analysis taken them to be resistances at a fixed frequency of 50Hz.

 

Let the input voltage on the GI be Vg, then

 

Vg = Vin [R3/(R3+Rs)] where Rs = the equivalent resistance of the source (nominally R1 in parallel with R2)

 

We don’t know the value of R3 but we do know what the ratio R3/Rs would have to be to NOT affect the GI. Let’s say Vg has to be below 0.3v peak. This is so Vg plus any galvanic voltage (0.8v) would still be below the diodes’ conduction voltage (1.2v). In practice, I saw much lower voltages than this on “Blackrose” of the order of a few tens of millivolts.

 

Thus, if the earth leakage voltage on the charger’s earth lead would be clamped at 1.2v, then

 

R3/Rs = 0.3/1.2 = .25 or, in other words, R3 would have to be at least 4 times lower in value than Rs. My charger will not deliver more than 1mA rms down the earth lead which implies that Rs for that particular charger is of the order of 120K. This would mean the hull/water/shore earth resistance would need to be no higher than 120K/4 = 30Kohms. In fact to get just millivolts, the hull/water/shore earth resistance would need to be an order of magnitude lower than this, so let's say around 3K ohms.

 

What we need to know is whether this is a plausible value for the hull/water/shore earth reistance. If it is, then it would explain what we are seeing in practice.

 

Chris

[Gibbo]

Here's a schematic of what, IMHO, is happening:

 

 

Although some of these “resistances” will be capacitively derived, I have for this simple analysis taken them to be resistances at a fixed frequency of 50Hz.

 

Let the input voltage on the GI be Vg, then

 

Vg = Vin [R3/(R3+Rs)] where Rs = the equivalent resistance of the source (nominally R1 in parallel with R2)

 

We don’t know the value of R3 but we do know what the ratio R3/Rs would have to be to NOT affect the GI. Let’s say Vg has to be below 0.3v peak. This is so Vg plus any galvanic voltage (0.8v) would still be below the diodes’ conduction voltage (1.2v). In practice, I saw much lower voltages than this on “Blackrose” of the order of a few tens of millivolts.

 

Thus, if the earth leakage voltage on the charger’s earth lead would be clamped at 1.2v, then

 

R3/Rs = 0.3/1.2 = .25 or, in other words, R3 would have to be at least 4 times lower in value than Rs. My charger will not deliver more than 1mA rms down the earth lead which implies that Rs for that particular charger is of the order of 120K. This would mean the hull/water/shore earth resistance would need to be no higher than 120K/4 = 30Kohms. In fact to get just millivolts, the hull/water/shore earth resistance would need to be an order of magnitude lower than this, so let's say around 3K ohms.

 

What we need to know is whether this is a plausible value for the hull/water/shore earth reistance. If it is, then it would explain what we are seeing in practice.

 

Chris

 

Your diagram is respresentative. Your maths is completely screwed up.

 

With infinite load resistance, the earth lead from the charger will be at 115 volts. To a first approximation, R3 in parralel with R2 will have to be 1/100th R1 to get this down to 1.15 volts. So more like 1/300th to get this down to 0.3v

 

Gibbo

Edited December 1, 2006 by Gibbo

[chris w]

As a follow-up (in case anyone out there is interested!) I measured the output series resistance of my charger's earth lead by monitoring potential divider voltages. By wiring various resistances across the GI, I could plot the way this affected the volts across the GI.

 

The result turned out to be about 450 ohms. So the charger's earth lead looks like a 1mA current source with a 450 ohm series resistance. In order to drop the volts down to 0.3v ac rms across the GI, I had to wire 300 ohms in parallel.

 

The issue I now can't get my brain around is that I just don't find it plausible to believe that the hull/water/shore earth resistance would be this low. A few Kohms maybe but of the order of 300 ohms? Allan Jones and I are going to do some real measurements of this on his boat in about 10 days time, so hopefully we'll solve the mystery then.

 

Chris

 

Gibbo

 

With infinite load resistance, the earth lead from the charger will be at 115 volts.

 

That's true, but I haven't got infinite load resistance - I've got a GI in the diagram clamping the earth lead volts at around 1.2v peak.

 

Chris

Edited December 1, 2006 by chris w

[Gibbo]

As a follow-up (in case anyone out there is interested!) I measured the output series resistance of my charger's earth lead by monitoring potential divider voltages. By wiring various resistances across the GI, I could plot the way this affected the volts across the GI.

 

The result turned out to be about 450 ohms. So the charger's earth lead looks like a 1mA current source with a 450 ohm series resistance. In order to drop the volts down to 0.3v ac rms across the GI, I had to wire 300 ohms in parallel.

 

The issue I now can't get my brain around is that I just don't find it plausible to believe that the hull/water/shore earth resistance would be this low. A few Kohms maybe but of the order of 300 ohms? Allan Jones and I are going to do some real measurements of this on his boat in about 10 days time, so hopefully we'll solve the mystery then.

 

Chris

 

Gibbo

That's true, but I haven't got infinite load resistance - I've got a GI in the diagram clamping the earth lead volts at around 1.2v peak.

 

Chris

 

You still have to do the calculations for the resistances required based on open circuit GIs.

 

Gibbo

[chris w]

Gibbo

 

Whatever the rights or wrongs of the theoretical calculation, in practice I had to put 300 ohms across the GI to drop the volts across it to 0.3v rms. (Actually, one would expect this value from the fact that the charger's earth lead is current limited at 1mA ie: the short circuit current in that lead is 1mA rms))

 

What's your view on this as compared to the likelihood of the hull/water/shore earth's path being this low.

 

Chris

[Space_Cowboy]

Mike on "Blackrose" and I spent the afternoon experimenting with his GI on his boat, so a real-life environment.

 

Mike had 2 Sterling chargers running plus a TV, fridge and a laptop. In addition we connected up my Sterling Charger, the one which gives 120v ac on the earth lead.

 

The curious thing is that the scope revealed no AC present at all on the earth lead (just a few millivolts of noise). Yet, when I tried it again at home this evening, sure enough there's the 120v ac on my charger.

 

This has led me to believe, with hindsite, that there may be an issue with Mike's shore earth on his pontoon.

 

I separated the hull earth, shore earth and ringmain earth and measured the voltage between them with the shore power ON on both a scope and a multimeter

 

Results

 

Shore earth to hull earth = 0v

 

Shore earth to ringmain earth = 28v ac

 

Hull earth to ringmain earth = 28v ac

 

Where is the 28v coming from? That's why I suspect there just may be a problem with the pontoon earthing. Mike is going to get more details from BW - maybe they are using an IT. Why the difference when I measure it at home - there should be 120vac as measured at home on the chargers, yet on Mike's boat there's zippo.

 

Any thoughts Gibbo?

 

Chris

 

 

I was just reading this thread and just was a little confused reading the above....especially reagrding the 120V on you charger earth when connected at home. Did you mean you removed the earth from your charger and measured with reference to your home ring main earth ? - this I could understand......

 

Did you take any neutral - earth measurements with every thing connected ? - just to see if they are at the same potential.

 

If the ring main circuit earth is securely bonded to the hull then then I would have thought there should be no potential present across any part of the earthing system on the boat.

[chris w]

I was just reading this thread and just was a little confused reading the above....especially reagrding the 120V on you charger earth when connected at home. Did you mean you removed the earth from your charger and measured with reference to your home ring main earth ? - this I could understand......

Yes that's correct - the "earth" lead is derived from two 4.7nF caps placed across the live and neutral

 

Did you take any neutral - earth measurements with every thing connected ? - just to see if they are at the same potential.

yes - there was 28v ac between them - not normal - one would expect maybe 2v

 

If the ring main circuit earth is securely bonded to the hull then then I would have thought there should be no potential present across any part of the earthing system on the boat.

That's true of course. The measurements I took were with the 3 "earths" disconnected (ie: shore earth, hull earth and ringmain earth). I wanted to try to find out from where the 28v ac was originating.

 

Chris

Edited December 1, 2006 by chris w

[Gibbo]

Gibbo

 

Whatever the rights or wrongs of the theoretical calculation, in practice I had to put 300 ohms across the GI to drop the volts across it to 0.3v rms. (Actually, one would expect this value from the fact that the charger's earth lead is current limited at 1mA ie: the short circuit current in that lead is 1mA rms))

 

What's your view on this as compared to the likelihood of the hull/water/shore earth's path being this low.

 

Chris

 

Ok 300 Ohms, 1mA = 300mV.

 

This is in parallel with the bottom Y cap. Do the maths and I bet you'll find the caps come out to about 10nF giving a source impedance of each at about 320K at 50Hz

 

Which agrees with what I said about the magnitude of the resistances as opposed to the 1/4 you suggested.

 

The resistance of the water is of course highly variable but this value doesn't seem too far removed from reality in a well polluted canal.

 

Gibbo

 

Results

 

Shore earth to hull earth = 0v

 

Shore earth to ringmain earth = 28v ac

 

Hull earth to ringmain earth = 28v ac

 

Where is the 28v coming from? That's why I suspect there just may be a problem with the pontoon earthing. Mike is going to get more details from BW - maybe they are using an IT. Why the difference when I measure it at home - there should be 120vac as measured at home on the chargers, yet on Mike's boat there's zippo.

 

Any thoughts Gibbo?

 

Chris

 

These are what I would expect. In fact it is often much higher. The 28 volts between hull to ringmain earth and between shore earth to ringmain earth is the AC from the various filter caps in whatever equipment happened to be connected at the time.

 

This gets swamped by the path through the water when everything is connected back up again. Unless there is lots of equipment on board in which case the water path resistance is insufficiently low to reduce it below the GI conduction voltage.

 

Gibbo

[chris w]

These are what I would expect. In fact it is often much higher.

That's the point - they should be higher. Floating at around 120v ac. The filter caps on mains filters in equipment are wired between the live and neutral with "earth" taken from the junction. Hence half of 240v.

 

The caps in my charger are 4.7nF each giving a nominal impedance at 50Hz of around 700K.

 

The 28 volts between hull to ringmain earth and between shore earth to ringmain earth is the AC from the various filter caps in whatever equipment happened to be connected at the time.

The 28v was still present with the RCD off so no power was reaching the ringmain (in theory). Just a thought, I wonder whether the neon polarity light was contributing to this as it's wired across Live and shore earth before the RCD and thus establishing some kind of path.

 

Chris

Edited December 1, 2006 by chris w

[Gibbo]

That's the point - they should be higher. Floating at around 120v ac. The filter caps on mains filters in equipment are wired between the live and neutral with "earth" taken from the junction. Hence half of 240v.

 

The Y caps are not always equal value. Sometimes the one between neutral and earth is a higher value (never seen it the other way round though). Some equipment only has a cap between neutral and earth. Without looking at every piece of equipment in isolation it's complete guesswork what the voltage will be.

 

Gibbo

[chris w]

Well mine DEFINITELY has 2 x 4.7nf caps cos I looked inside and mine was plugged in on "Blackrose"

 

Chris

[Gibbo]

Well mine DEFINITELY has 2 x 4.7nf caps cos I looked inside and mine was plugged in on "Blackrose"

 

Chris

 

I suspect you only saw part of the filter. Usually there will be 2 Y caps as you state. Then a differential mode choke (which will be practically a short circuit at 50Hz), then 2 more Y caps of the same value.

 

If you do the maths assuming 2 Y caps each of 4n7 you will find the figures come out about right to produce 300mV across your 300 Ohm resistor.

 

With the resistance at 300 Ohms we can safely ignore the bottom Y cap (i.e. the one between N and E)

 

2 X 4n7s in parallel gives 9n4. Impedance at 50Hz = 338K. 230 volts across this = 670uA which through 300 Ohms = 200mV. Not a million miles away from the measured figures.

 

Obviously this is a bit rough arse!

 

Gibbo

Edited December 1, 2006 by Gibbo

[Yoda]

Bet Jon doesn't know about this thread.

[Gibbo]

Bet Jon doesn't know about this thread.

 

Bet he's watching it like a hawk.

 

Gibbo

[DHutch]

I can assure you that the site crew are quite aware of this thread existance, and are montering closely.

- Im not going to pretend that I have the time to analysis in detail all the post, but i skim over it serveral times a day.

 

 

Daniel

[tomsk]

I can assure you that the site crew are quite aware of this thread existance, and are montering closely.

- Im not going to pretend that I have the time to analysis in detail all the post, but i skim over it serveral times a day.

Daniel

 

I have a traditional five berth Welsh coracle, replete with flume rebates and knurdling spars. My question is, would a GI enhance my boating experience and do you know where I can replenish my prescription of Largactil (AKA The Chemical Cosh) as I have this recurring hallucination since I dropped my last 'script overboard whilst scraping freshwater mussels of rocks, where threads keep recurring and the points therein keep recurring and recurring and recur.......(off stage) Nurse, the screens if you will.

[dor]

I can assure you that the site crew are quite aware of this thread existance, and are montering closely.

- Im not going to pretend that I have the time to analysis in detail all the post, but i skim over it serveral times a day.

 

Well I know I am just back from the pub, but am I missing something here?

 

I have a rudimentary knowledge of electronics, but the esoteric content of this thread is way past the current limit of my knowledge. However the main participants are clearly very knowledgable on the subject and it, to me, actually makes quite interesting reading.

 

Better anyway than someone saying "I'm right" and totally ignoring any alternative point of view.

[ChrisPy]

there used to be a paint called Galvafroid. would that help?