DC current leakage problem

[nb Innisfree]

6 hours ago, BruceinSanity said:

My experience is that a polarity warning light is more common, a neon across neutral and earth which lights if there is voltage across them. I think the Victron provides connections for that.

I fitted one on Innisfree, after many false clues we eventually found it to be the cause of our shoreline tripping occasionally, leakage must have hovered around 30 milliamp, sometimes more but mostly less. 

[nicknorman]

2 hours ago, Dr Bob said:

Thanks Nick...but if the current is only very low,  then the GI will be isolating the earth, so no problem with the boat dissolving away. That would mean I didnt need to minimise the charging - which I would like to leave on when in the marina (2-3 days at a time)?

Thanks Cuthound, just off out for a few hours so will look at video later. Never done a diode test on a multimeter. I assume my meter has one!

Well as I’ve just said, if the LED is glowing it’s passing current. But not much. I suppose our set up is similar, we have a mechanical voltmeter across the GI. When I plug the shore power in, the meter deflects very slightly so it is passing current between the hull and the shore. But not very much.

[Sea Dog]

19 minutes ago, nicknorman said:

If the LED is glowing, it’s passing current! Current between the boat earth and shore earth. But not much.

Ah yes, the led is indeed using something, however small, to produce a glow. The protective device itself ought not to be though - assuming it's not defective, it ought to be isolating the boat earth from the shore earth until such time as a fault over the 1.4V required to overcome the diodes occurs. Right?

[nicknorman]

17 minutes ago, Sea Dog said:

Ah yes, the led is indeed using something, however small, to produce a glow. The protective device itself ought not to be though - assuming it's not defective, it ought to be isolating the boat earth from the shore earth until such time as a fault over the 1.4V required to overcome the diodes occurs. Right?

Yes.

[Dr Bob]

46 minutes ago, nicknorman said:

Well as I’ve just said, if the LED is glowing it’s passing current. But not much. I suppose our set up is similar, we have a mechanical voltmeter across the GI. When I plug the shore power in, the meter deflects very slightly so it is passing current between the hull and the shore. But not very much.

Nick. I am thinking along the same lines as SeaDog. As long as the Voltage is less than 1V (1.2, 1.4V whatever?) then the GI isolates the earth wire and nothing flows. If you have a volt meter that goes from the GI inlet to GI outlet then that effectively bypasses the GI and current will flow round the voltmeter circuit allowing the current to go between the boat and shore. I was working on the principle that ANY current on the earth wire from boat to shore is bad as it potentially sets up a galvanic cell with other boats etc and mine can disappear.

What I think I have learnt here is that the diode in the GI can light up without the earth contact being made in the GI to connect boat to shore. If that is not true then yes some current must be passing when the LED comes on. Which is correct?

Edit.....I think you just answered the question!

Edited November 16, 2017 by Dr Bob

[nicknorman]

26 minutes ago, Dr Bob said:

Nick. I am thinking along the same lines as SeaDog. As long as the Voltage is less than 1V (1.2, 1.4V whatever?) then the GI isolates the earth wire and nothing flows. If you have a volt meter that goes from the GI inlet to GI outlet then that effectively bypasses the GI and current will flow round the voltmeter circuit allowing the current to go between the boat and shore. I was working on the principle that ANY current on the earth wire from boat to shore is bad as it potentially sets up a galvanic cell with other boats etc and mine can disappear.

What I think I have learnt here is that the diode in the GI can light up without the earth contact being made in the GI to connect boat to shore. If that is not true then yes some current must be passing when the LED comes on. Which is correct?

Edit.....I think you just answered the question!

The GI is not powered so the only way the LED can light is with the shore earth connected and a voltage difference between hull and shore earth. What’s not clear is the voltage the LED will start to conduct at (and thus glow) vs the main diodes conduction voltage. Hopefully the latter is more than the former. LED forward voltage depends on the chemistry ie the LED colour. What colour is your LED?

[rowland al]

2 hours ago, nicknorman said:

If the LED is glowing, it’s passing current! Current between the boat earth and shore earth. But not much.

Yes. but does it depend where the LED is in the circuit? 

I read this on their web site. Does this mean other boaters in the marina  (if they know what they are doing), end up improving the thickness of their hulls?  - 

"Believe me when I tell you that this is no fairy tale – my own hull has lost around half a tonne in weight through corrosion that could have been prevented for just a few pounds with a simple galvanic isolator (sometimes known as a Zinc Saver)."

Edited November 16, 2017 by rowland al

[nicknorman]

10 minutes ago, rowland al said:

Yes. but does it depend where the LED is in the circuit? 

Well not really. The LED must be powered by something, and the only possible thing is a voltage between hull and shore earth. The LED must be across the diodes (I think there are normally two LEDs, one each way)

[MtB]

On 16/11/2017 at 23:53, rowland al said:

"Believe me when I tell you that this is no fairy tale – my own hull has lost around half a tonne in weight through corrosion that could have been prevented for just a few pounds with a simple galvanic isolator (sometimes known as a Zinc Saver)."

 

Anyone starting a sentence with "believe me", is probably best not believed.

[smileypete]

18 hours ago, Dr Bob said:

Anyone help with a DC current leakage problem. We have recently started to use our marina berth a bit more and connected to shore power via a galvanic isolator - a cheap one, see photo below. Now the pram cover is up, I can see there is a faint glow in one of the LED's indicating dc current leakage (never saw it in the summer on the few occasions we used it). The battery to earth connections seem fine and the earth connector on the Victron Multi (charger/inverter) 12V 2000W - probably 10 years old - I havent checked as not sure how to get to it.

Wonder if it's an issue with the hull earth bond or the earthing relay in the combi.

Can you easily and safely access the earth connection from the galvanic isolator on the combi side? (This is why I don't like 'inline' galvanic isolators as it makes this harder to do)

If so try connecting earth the connection from the galvanic isolator on the combi side (not the shoreline side) securely to the metalwork of the hull and see if the LED stops lighting.

[WotEver]

9 hours ago, nicknorman said:

Well not really. The LED must be powered by something, and the only possible thing is a voltage between hull and shore earth. The LED must be across the diodes (I think there are normally two LEDs, one each way)

Any idea how these LEDs are connected? If they’re directly across the protection diodes then surely any current at a voltage above 0.6V ish will be passing through the LEDs instead of the protection diodes. If they’re in series with another diode and then across the protection diodes then the voltage would have to be pretty high (over 1V) to get them glowing. Besides, any major fault would blow the LEDs instantly, making them pretty pointless. There’s not enough voltage for a little Op-Amp circuit to drive the LEDs so I’m a bit puzzled as to how they can be wired without compromising the operation of the GI. 

Have I mentioned yet in this thread that I much prefer an IT?

[forsberg]

31 minutes ago, WotEver said:

Any idea how these LEDs are connected? If they’re directly across the protection diodes then surely any current at a voltage above 0.6V ish will be passing through the LEDs instead of the protection diodes. If they’re in series with another diode and then across the protection diodes then the voltage would have to be pretty high (over 1V) to get them glowing. Besides, any major fault would blow the LEDs instantly, making them pretty pointless. There’s not enough voltage for a little Op-Amp circuit to drive the LEDs so I’m a bit puzzled as to how they can be wired without compromising the operation of the GI. 

Have I mentioned yet in this thread that I much prefer an IT?

They are not connected across the diodes as they would clamp the voltage at 2*0.6V so the Leds would never lit. There is however 220V available to drive any circuitry you like, or you can use a current transformer. With Three diodes it would be possible i assume but I have yet to find a GI with 3 diodes. I would prefer a IT aswell as long as it's wired not to disable the RCD.

But more to the OT, all current leaks add up so it would be intresting to know if you got anything else connected besides the Victron ?

The Led might allready be very close to trigger and the Victron is just the tip of the Iceberg that sets it off. Calorifier's are a very common cause.

[Flyboy]

1 hour ago, WotEver said:

Any idea how these LEDs are connected? If they’re directly across the protection diodes then surely any current at a voltage above 0.6V ish will be passing through the LEDs instead of the protection diodes. If they’re in series with another diode and then across the protection diodes then the voltage would have to be pretty high (over 1V) to get them glowing. Besides, any major fault would blow the LEDs instantly, making them pretty pointless. There’s not enough voltage for a little Op-Amp circuit to drive the LEDs so I’m a bit puzzled as to how they can be wired without compromising the operation of the GI. 

Have I mentioned yet in this thread that I much prefer an IT?

This image implies that the LED's are connected across the diodes.     

[nicknorman]

1 hour ago, WotEver said:

Any idea how these LEDs are connected? If they’re directly across the protection diodes then surely any current at a voltage above 0.6V ish will be passing through the LEDs instead of the protection diodes. If they’re in series with another diode and then across the protection diodes then the voltage would have to be pretty high (over 1V) to get them glowing. Besides, any major fault would blow the LEDs instantly, making them pretty pointless. There’s not enough voltage for a little Op-Amp circuit to drive the LEDs so I’m a bit puzzled as to how they can be wired without compromising the operation of the GI. 

Have I mentioned yet in this thread that I much prefer an IT?

They can’t be in series as they can’t take the fault current, so they must be across the diodes. Do we know how many diodes are in series? Maybe 3? A red LED starts to glow at around the same voltage as 2 diodes so that would be a bit close!

7 minutes ago, forsberg said:

They are not connected across the diodes as they would clamp the voltage at 2*0.6V so the Leds would never lit. There is however 220V available to drive any circuitry you like, or you can use a current transformer. With Three diodes it would be possible i assume but I have yet to find a GI with 3 diodes. I would prefer a IT aswell as long as it's wired not to disable the RCD.

But more to the OT, all current leaks add up so it would be intresting to know if you got anything else connected besides the Victron ?

The Led might allready be very close to trigger and the Victron is just the tip of the Iceberg that sets it off. Calorifier's are a very common cause.

There is no 220v available on GIs. They just have 2 terminals, for shore earth and boat earth. Perhaps the diodes are doped in some way to increase the Vf?

[Sea Dog]

32 minutes ago, nicknorman said:

They can’t be in series as they can’t take the fault current, so they must be across the diodes. Do we know how many diodes are in series? Maybe 3? A red LED starts to glow at around the same voltage as 2 diodes so that would be a bit close!

 

And if the main isolating diodes were 'switched on' by a fault to earth on the boat, wouldn't they be shorting out the indicator leds and therefore cause a normal indication from that diagram above?

 

Edited November 17, 2017 by Sea Dog
Sp

[forsberg]

Think I stand corrected , forgot that the forward voltage drop  increase with current so that would probably work very well to let the Led's go from glow to lit.

As there probably are a resistor in series with a led, any excess current will go thrue the diodes.

 

[dmr]

The LEDs will be across the diodes led but they will have a resistor (or something cleverer) in series with them, otherwise they would be instantly destroyed the first time the GI did any real work.

Visualising the LED as indicating the voltage across the main diode is probably more useful than thinking about the LEDs handling a leakage current.

This is a case where a good old simple little mechanical voltmeter would have been a better approach ?????

.................Dave

[Sea Dog]

9 minutes ago, dmr said:

 

This is a case where a good old simple little mechanical voltmeter would have been a better approach ?????

Mine has that arrangement and, from an earlier post, I believe Nicknorman also has one. Mine does occasionally show a slight deflection, but is zoned green to red. Since its green, I'm reading that as the device protecting the boat against the pd the meter is indicating. If it was in the red zone I take it that would indicate a pd sufficient to switch on the diodes was present. The documentation for it is sketchy to say the least!

[dmr]

The more I think about this, and looking at Nicks VI graph of LEDS (I never knew that different colours where so different) then LEDS are really of limited use in this application. They will indicate a fault (GI working) but are not much good at indicating approaching trouble.

Anyway have a look at this: 

http://galvanic-isolator.co.uk/galvanic-isolator-gi5016sa-instructions/

It specifically mentions issues with battery chargers.

................Dave

[nicknorman]

43 minutes ago, dmr said:

The LEDs will be across the diodes led but they will have a resistor (or something cleverer) in series with them, otherwise they would be instantly destroyed the first time the GI did any real work.

Visualising the LED as indicating the voltage across the main diode is probably more useful than thinking about the LEDs handling a leakage current.

This is a case where a good old simple little mechanical voltmeter would have been a better approach ?????

.................Dave

The problem with a meter, which is what we have, is that it doesn’t respond to ripple. If you mix a small fixed dc offset with ripple, one half of the ripple cycle can cause the diodes to conduct whilst the other half cycle doesn’t. Thus the diodes are conducting DC whilst the meter still shows a safe dc offset. LEDs give an instantaneous reaction and thus show such ripple, as per the OP.

[WotEver]

1 hour ago, Sea Dog said:

And if the main isolating diodes were 'switched on' by a fault to earth on the boat, wouldn't they be shorting out the indicator leds and therefore cause a normal indication from that diagram above?

Nope. Diodes don’t get ‘switched on’, they simply pass a current in one direction only. When passing that current they drop around 0.6V across them. Two in series will drop 1.2v etc. Another two in parallel with the first two but pointing the other way allows current to flow in the opposite direction. So a GI works by isolating the boat’s hull from shore earth unless the voltage difference exceeds 1.2V. 

14 minutes ago, nicknorman said:

The problem with a meter, which is what we have, is that it doesn’t respond to ripple. If you mix a small fixed dc offset with ripple, one half of the ripple cycle can cause the diodes to conduct whilst the other half cycle doesn’t. Thus the diodes are conducting DC whilst the meter still shows a safe dc offset. LEDs give an instantaneous reaction and thus show such ripple, as per the OP.

Yeah, but that’s the purpose of the capacitor in the better units, to pass that ripple. So an LED in theory still won’t show it. 

[nicknorman]

5 minutes ago, WotEver said:

Nope. Diodes don’t get ‘switched on’, they simply pass a current in one direction only. When passing that current they drop around 0.6V across them. Two in series will drop 1.2v etc. Another two in parallel with the first two but pointing the other way allows current to flow in the opposite direction. So a GI works by isolating the boat’s hull from shore earth unless the voltage difference exceeds 1.2V. 

Yeah, but that’s the purpose of the capacitor in the better units, to pass that ripple. So an LED in theory still won’t show it. 

If the LED doesn’t show it, it’s because there’s nothing to show.

[dmr]

9 minutes ago, WotEver said:

Nope. Diodes don’t get ‘switched on’, they simply pass a current in one direction only. When passing that current they drop around 0.6V across them. Two in series will drop 1.2v etc. Another two in parallel with the first two but pointing the other way allows current to flow in the opposite direction. So a GI works by isolating the boat’s hull from shore earth unless the voltage difference exceeds 1.2V. 

Yeah, but that’s the purpose of the capacitor in the better units, to pass that ripple. So an LED in theory still won’t show it. 

Switched on is not a bad analogy for a non electronics person though. I 'spose "start to conduct" might be a bit more correct, but in a GI we are not using the one way conduction characteristic (its actually a disadvantage) but are exploiting the fact that diodes do not start to conduct (switch on) till they have about 0.6 volts across them.

I think the capacitor is to remove high frequency switching stuff, whilst Nick is talking about ripple at mains frequency?

.............Dave

Edited November 17, 2017 by dmr
added more stuff

[WotEver]

4 minutes ago, nicknorman said:

If the LED doesn’t show it, it’s because there’s nothing to show.

Because it’s being bypassed by the capacitor straight to the hull. 

Have we ever decided if AC ripple causes stray current erosion? 

4 minutes ago, dmr said:

Switched on is not a bad analogy for a non electronics person though. I 'spose "start to conduct" might be a bit more correct, but in a GI we are not using the one way conduction characteristic (its actually a disadvantage) but are exploiting the fact that diodes do not start to conduct (switch on) till they have about 0.6 volts across them.

But the disadvantage of considering them to ‘Switch on’ is that the reader can imagine a dead short like a switch, exactly as SeaDog did. Analogies rarely work well in electrical descriptions. 

[dmr]

24 minutes ago, nicknorman said:

The problem with a meter, which is what we have, is that it doesn’t respond to ripple. If you mix a small fixed dc offset with ripple, one half of the ripple cycle can cause the diodes to conduct whilst the other half cycle doesn’t. Thus the diodes are conducting DC whilst the meter still shows a safe dc offset. LEDs give an instantaneous reaction and thus show such ripple, as per the OP.

Right then, a tiny little built in oscilloscope with a little LCD display, I will add it to my list of interesting projects that I will never get around to.

What about a meter connected via a bridge rectifier made from Schottky diodes?????

.............Dave