DIY Galvanic isolator.

[NorthUp]

Would someone in PBO care to buy a commercial version, and reverse engineer it-
i.e. whats inside the chandlers version to justfy the price tag, apart from a public liability insurance policy?

 

[jfkal]

With respect, 'these things' are there to prevent stray galvanic currents.

There are safety implications if they fail or don't work properly, but the prime use of them is to reduce stray galvanic currents. They are not built to 'save life'.

Rant over

Right. But they are also not supposed to kill if they fail. They do what they do at the expense of a proper earth connection in case of a fault.
1. Low voltage (stray galvanic), it is supposed to isolate
2. High AC current (short to earth), it is supposed to conduct and trip the fuse, RCD or whatever. Hence it needs that high current capability otherwise you die when you touch bare metal. So why don't some of the know-it-alls do the following:
Built the damn thing as shown in the magazine.
Short live to ground 10 times and then see whether it is still working.
AND THEN LET US ALL KNOW.

 

[john_morris_uk]

Right. But they are also not supposed to kill if they fail. They do what they do at the expense of a proper earth connection in case of a fault.
1. Low voltage (stray galvanic), it is supposed to isolate
2. High AC current (short to earth), it is supposed to conduct and trip the fuse, RCD or whatever. Hence it needs that high current capability otherwise you die when you touch bare metal. So why don't some of the know-it-alls do the following:
Built the damn thing as shown in the magazine.
Short live to ground 10 times and then see whether it is still working.
AND THEN LET US ALL KNOW.

I am not sure that the professionally built ones are designed to be shorted direct to ground ten times in a row, however, the current in the circuit will be limited by the wiring resistance and earth resistance, so I suppose that they should survive. If you want to blow the fuse ten times, or trip your over current breaker ten times, then carry on.

All the galvanic isolator has to do is to withstand the several hundred amps of a short long enough to trip the RCD or blow the fuse. The real problem is that most silicon devices are capable of thermal runaway and blowing far faster than conventional fuses. This is the reason that the diode bridges in the design need to be massively overrated.

If the home-built galvanic isolator is made with the correctly rated diode bridges, there is no reason why it shouldn't be as safe as a commercially made one.

 

[halcyon]

If the home-built galvanic isolator is made with the correctly rated diode bridges, there is no reason why it shouldn't be as safe as a commercially made one.

I admit I have not read all the thread, but has anyone looked inside a commercial galvanic isolator ?

I did, looked inside on well known one, about £80 / £90, a couple of years back, one heatsink, two 32 amp bridge rectifiers, and about 1 foot of wire.

So yes you can make one as good a commercial one, and yes just about under £10 if you do not count a box to isolate the terminals. I if fails, who's insurance cover the compensation claim ?

Am I happy with GI, not at the moment, I need to dig out my notes from 1990, when we first looked into it, I think we may have lost the plot.

What worries me the way the whole marine trade is going, we are selling more on what marketing can claim, and buying on which box makes the best claim. We are going away from making good solid products, to a marketing image.

Brian

 

[fireball]

Ahh .... common sense ... found it at last!!

Home Grown vs Commercial Build - it just comes down to what components you buy and put in - it is perfectly feasible to home build a GI that far outweighs the capability of the commercial build ...

Everything in life these days seems to be a safety item ... next we'll be told we can't splice our own rope because it is a safety feature ... FFS some of us are capable of DIY ... the rest of you will just have to pay over the odds for someone else to make it for you!

 

[bedouin]

What is surprising is the blind faith some people put in a black-box over a known circuit. If these things are selling for £50 at a chandlery then they have at most £10 worth of components inside (at "retail" prices rather than wholesale). What on earth do these people expect to find?

I would much rather trust a design where I had chosen the components and know their ratings than an off-the-shelf black box probably built with the cheapest components they could find.

A conventional 35A bridge rectifier in the recommended layout (parallelled) is rated to withstand at least 800A for half a cycle (10ms) given that you have two of these devices it is rated to withstand at least 800A for 20ms - and may do considerably better. That is long enough to blow pretty much any RCD (or MCB of reasonable rating).

 

[andythilo]

Just my 2p, if you are a competant electronics engineer who understands bridge rectifiers and how they work as a GI, then build one and be happy that you saved £50+

If however you have ANY concerns about it then just buy one and be happy that you KNOW it's doing what it should.

 

[goboatingnow]

Galvanic isolators shall be designed to withstand the application of power from a short-circuit test from a source capable of delivering 5,000 A r.m.s. symmetrically to its output test terminals for the time required for the circuit-breaker in the test circuit to trip. After
three applications of the short-circuit test, the electrical and mechanical characteristics of the isolator shall be unchanged."[\QUOTE]

This is teh ABYC standard and as far as I know EU manufacturers do not have to comply with it.

I opened my commercial GI ( from a well known EU company ) and its two cheapo bridge rectifiers.

As a previous poster says, if you know what you are doing , you will actually build a better product yourself, but then i;m an electronics engineer.

 

[billcole]

a few points, in no particular order...

As others have said, some of the huge fault currents being mooted are unlikely owing to the impedances of the supplies found in marinas.Anyway, if 800 amps were to flow through a piece of copper earth wire it too would get hot and not last long

If a full mains voltage to earth fault occurs then the size of the heatsink they might be bolted to will make no difference to the length of time the diodes survive, the p-n junction will be destroyed before the heat is even conducted to the diode/bridge rectifier casing, let alone having time to spread through a heatsink and be dissipated

Causing a fire is unlikely, the components are not made of flamable materials. They may go bang and smell nasty but a sustained fire is unlikely. If they are destroyed by a fault then they'll almost certainly go to either a short circuit, in which you've got the same situation as if you'd stuck with an uninterrupted earth wire, or an open circuit, in which case they will no longer be dissipating power, and so will not get hot.

It is true that if all the diodes went open circuit before the RCD tripped then the boat would be left momentarily without an earth connection, BUT...RCDs are tripped by an imbalance between the current flowing in the live and neutral conductors, not by current flowing in the earth connector ( Thats why they're not called Earth Leakage Circuit Breakers ) so the system would still have RCD protection.

A refinement of the two BR galvanic isolator design is to add a capacitor across the two rectifiers, thus providing some path for a.c. (a.c shouldn't cause galvanic corrosion as the two halves of the cycle cancel out, although there are circumstances where the biasing of the diodes can give rise to a dc component...see Nigel Calder's book for an explanation) A 1 microfarad capacitor will have an impedance of about 3kOhms at 50Hz, so will conduct about 60mA to ground if 220V 50Hz appears across it.

 

[nimbusgb]

The real point to consider is if you buy an expensive swindlery supplied unit if it kills you you have someone to sue

I'll bet you don't when the time comes.

 

[andythilo]

You wont be sueing the Chandlery anyway, you'll be sueing the Manufacturer.

 

[headrowe]

From beyond the grave????????

 

[misterg]

A conventional 35A bridge rectifier in the recommended layout (parallelled) is rated to withstand at least 800A for half a cycle (10ms)

*That* is what I set out to verify when I looked into it, and failed - would you mind pointing me in the right direction?

Since this discussion, I've found that Farnell let you search by peak surge current (IfSM), and there are certainly discrete diodes there that would address all my fusing / tripping concerns - say 4 of these @ £1.90 each (get 'em quick because the lead free ones are £5.82 each).

I'm not convinced that all 'cheapy' 30 amp bridge rectifiers are created equal, or that all commercial GIs are built with fault currents in mind.

(Incidentally, my 1000A estimate comes from 330V (peak of 220V AC) into ~0.3 ohm earth resistance (Zs) - say a socket outlet near the main distribution board - I don't know what typical Zs values are in practice, but believe that for a 32A outlet, it can't be more than ~1.4 ohm at the most distant outlet. This is just my 'order of magnitude' estimation of the potential currents involved. The IEE (IET?) regs should be your guide.)

I was reluctant to openly suggest testing a GI between live and earth, but would do that if I had any doubts. If it's any good, it should survive that on repeated occasions. If it's no good, you'll get a bang, and some bits of hot diode flying about.

Anybody got any good sources for isolation transformers ?

Andy

 

[bedouin]

*That* is what I set out to verify when I looked into it, and failed - would you mind pointing me in the right direction?
Andy

I always use RS (rswww.com) - they have an excellent selection of devices and good datasheets. If you check the data sheets for the standard 35A bridge rectifiers you will find loads of figures, including some about maximum current. They are also a good site to check the trip characteristics of RCDs / MCBs.

Comparing those leaves me in no doubt that the simple bridge rectifier design is safe - with the proviso that an RCD is fitted.

 

[Billjratt]

Just one final query if I may.
the G.I.s I have seen diagrams for, have had two diodes in series, opposed by another pair. this gives a 1.2 volt drop (0.6 each) in both directions of flow, as already described.
If I then introduce (effectively) another G.I. in parallel, as this project proposes, have I not cut the effective voltage back to 0.6v?
Or do diodes in parallel not act like resistors in parallel?

If diodes in parallel ad infinitum will still drop 0.6v, we could make isolators out of multiples of the cheapest available units. - Just keep adding them side by side until you are satisfied the wiring or (more likely) a connection, will blow before the unit.


How many of you have a neon lamp in the ground circuit to monitor correct Line/Neutral orientation? - Get one and use it to monitor the ground path status as well.( Which is effectively what they do-)
That way you will know if you've blown your GI and have no sense of smell.

 

[misterg]

... two diodes in series, opposed by another pair. this gives a 1.2 volt drop (0.6 each) in both directions of flow, as already described.
If I then introduce (effectively) another G.I. in parallel, as this project proposes, have I not cut the effective voltage back to 0.6v?
Or do diodes in parallel not act like resistors in parallel?

No, they don't act like resistors - the voltage drop will be the same. Incidentally, the PBO project uses all the diodes in each bridge in the same direction, so they don't act as 2 GIs in parallel (although you could wire it that way).

If diodes in parallel ad infinitum will still drop 0.6v, we could make isolators out of multiples of the cheapest available units. - Just keep adding them side by side until you are satisfied the wiring or (more likely) a connection, will blow before the unit.

Yep, you've got it. I don't think you need to do this, though, if you're slightly discerning in your choice of diode. (Some of the 35A bridges in RS are good for 475A surge current. I guess that by using all the diodes in the bridge, you should be good for double this (because you create a parallel path) - Job done.)

How many of you have a neon lamp in the ground circuit to monitor correct Line/Neutral orientation? - Get one and use it to monitor the ground path status as well.( Which is effectively what they do-)
That way you will know if you've blown your GI and have no sense of smell.

Good idea!

Andy

 

[Allan]

Although I have not read the whole thread I think I grasp the two sides put here. I have worried in the past about the next owner of a boat with lots of this type of DIY projects. My main reason being that owner could be me!
Allan

 

[goboatingnow]

Get one and use it to monitor the ground path status as well.( Which is effectively what they do-)
That way you will know if you've blown your GI and have no sense of smell. [\quote]

Now would you use a neon to monitor ground path status???

 

[halcyon]

*That* is what I set out to verify when I looked into it, and failed - would you mind pointing me in the right direction?

Anybody got any good sources for isolation transformers ?

Andy

Something has been bothering me since the start of this thread, and it has come to me.

Why are we fitting a GI in the main supply line ?

When GI's first came in, back in the 1990's, they were fitted between the mains earth, and the DC bonding system.

The original problem was from variation in mains pontoon earth, and seabead earth, I known as it was first seen on a Sealine 450, and I got blamed for a faulty battery charger.

We were getting a voltage between mains pontoon earth, and the skin fitting / props, and they were dissapearing in a month, by fitting a pair of apposed diodes between mains earth and the bonding system you broke the circuit for low voltages.

Except if you had a mains earth fault, at the same time lost the mains shore earth link, you still had a high voltage safety link to saebead completing the fault circuit via skin fittings.

So the circuit never normally went through the GI, so why have we added a potentional safety issue into the equation.

I need to dig out my old GI files, and take more interest.

Brian

 

[jfkal]

I am not sure that the professionally built ones are designed to be shorted direct to ground ten times in a row, however, the current in the circuit will be limited by the wiring resistance and earth resistance, so I suppose that they should survive. If you want to blow the fuse ten times, or trip your over current breaker ten times, then carry on.

All the galvanic isolator has to do is to withstand the several hundred amps of a short long enough to trip the RCD or blow the fuse. The real problem is that most silicon devices are capable of thermal runaway and blowing far faster than conventional fuses. This is the reason that the diode bridges in the design need to be massively overrated.

If the home-built galvanic isolator is made with the correctly rated diode bridges, there is no reason why it shouldn't be as safe as a commercially made one.

I did it to mine and it survived. So I feel pretty safe