Cheap galvanic isolator

[gary955]

Ignoring for the time being (please) the very pursuasive argument for an isolation transformer. why shouldn't I make a Galvanic isolator from a couple of bridge rectifiers and a heat sink?

35a and 50a bridge rectifiers are available very cheaply on Ebay. The spec however rarely quotes the foreward voltage drop, usually stating "low foreward voltage drop". I understand that its usually about 0.6v, giving 1.2v combined, but a good one might be as low as 0.2v. Ironically I would want bad ones!! Ideally not conducting until 2 volts or more (combined)

How would I find such a thing.......and should I?

[Chalky]

Liability. I considered this as an option - why pay £70+ for about £5 worth of bits? Because when it goes wrong and someone gets electrocuted not only have you got it on your conscience, but you'd also be liable. Suddenly £70 becomes more reasonable - most of it is paying for someone elses liability insurance.

[gary955]

Liability. I considered this as an option - why pay £70+ for about £5 worth of bits? Because when it goes wrong and someone gets electrocuted not only have you got it on your conscience, but you'd also be liable. Suddenly £70 becomes more reasonable - most of it is paying for someone elses liability insurance.

Yes as well as my conscience being troubled, it might be ME that gets electrocuted!!!

But the only risk of electrocution is if it fails open curcuit with a massive current going to earth, and then only if the RCD fails as well. Its a passive piece of equiptment and if using similar or better components to a commercial one the risk is no greater. the wiring diagram as you know couldn't be simpler

[Gibbo]

You'd need to test it to make sure it will actually do its job. That's nowhere near as simple as it sounds. It's only simple to someone who doesn't understand the issues (a common scenario). Also, in order to be sure you know it will do its job you'd have to build several of them and test each one to destruction. By the time you've done the job properly you could have bought several.

 

Unless, of course, you can find someone who has the plans for one where someone has already done this. Doubtful... the plans I've seen online have been laughable.

 

I'll give you a quick clue to the problem. The device would have to survive a complete short circuit from a (say) 32 amp supply using a normal fuse. Have you any idea how much current that will force through it? I guess there will be one or two extra zeros in the actual figure from what you'd expect.

 

Then suppose it's on a 32 amp supply with no RCD and gets a partial short that passes 32 amps. How warm is it going to get?

 

These are just the beginnings of the issues.

 

That's why a commercial one is 70 quid for 5 quidsworth of bits.

[gary955]

You'd need to test it to make sure it will actually do its job. That's nowhere near as simple as it sounds. It's only simple to someone who doesn't understand the issues (a common scenario). Also, in order to be sure you know it will do its job you'd have to build several of them and test each one to destruction. By the time you've done the job properly you could have bought several.

 

Unless, of course, you can find someone who has the plans for one where someone has already done this. Doubtful... the plans I've seen online have been laughable.

 

I'll give you a quick clue to the problem. The device would have to survive a complete short circuit from a (say) 32 amp supply using a normal fuse. Have you any idea how much current that will force through it? I guess there will be one or two extra zeros in the actual figure from what you'd expect.

 

Then suppose it's on a 32 amp supply with no RCD and gets a partial short that passes 32 amps. How warm is it going to get?

 

These are just the beginnings of the issues.

 

That's why a commercial one is 70 quid for 5 quidsworth of bits.

You're right Gibbo, i dont really have a clue, but am willing to learn

Someone I came across on the net reverse engineered a commercial one and it literally was just a pair of bridge rectifiers going in opposite directions across the earth wire and mounted on a heat sink.

This is intended for me on my boat only, on a 16A shoreline with a regularly tested RCD.

Surely if a short of thousands of amps was possible it would burn out the earth wire anyway so the diodes failing open curcuit would be immaterial and there must be a fuse somewhere.

I may be wrong here but it does sound a little like arguing that you should get a qualified electrical engineer to wire a plug for you because the consequenses of getting it wrong might kill you or burn down your boat and that makes his £100 fee good value!

[mrsmelly]

Ignoring for the time being (please) the very pursuasive argument for an isolation transformer. why shouldn't I make a Galvanic isolator from a couple of bridge rectifiers and a heat sink?

35a and 50a bridge rectifiers are available very cheaply on Ebay. The spec however rarely quotes the foreward voltage drop, usually stating "low foreward voltage drop". I understand that its usually about 0.6v, giving 1.2v combined, but a good one might be as low as 0.2v. Ironically I would want bad ones!! Ideally not conducting until 2 volts or more (combined)

How would I find such a thing.......and should I?

 

Short answer is if you cant afford 70 odd squid for a peice of safety equipment you simply cannot afford to go boating...........

[gary955]

Short answer is if you cant afford 70 odd squid for a peice of safety equipment you simply cannot afford to go boating...........

Oh Please!!!!

I'm very willing to accept an argument that tells me I'm wrong.....but an argument that only states "I bought it in a shop and it was expensive so it must be better" is simply a silly argument.

Quality, or safety is not measured by cost.

[Chalky]

Surely if a short of thousands of amps was possible it would burn out the earth wire anyway so the diodes failing open curcuit would be immaterial and there must be a fuse somewhere.

 

The diodes will fail long before the wire shows signs of distress.

 

I may be wrong here but it does sound a little like arguing that you should get a qualified electrical engineer to wire a plug for you because the consequenses of getting it wrong might kill you or burn down your boat and that makes his £100 fee good value!

 

Where I work we have quite a few electrical engineers, the minimum qualification is one degree, many have second or 3rd ones. We still have an electrician come round once a year and test all of the electrical equipment. We're not allowed to put plugs on mains equipment (but we can play with high voltage high current DC which can be significantly more lethal ).

[mrsmelly]

Oh Please!!!!

I'm very willing to accept an argument that tells me I'm wrong.....but an argument that only states "I bought it in a shop and it was expensive so it must be better" is simply a silly argument.

Quality, or safety is not measured by cost.

 

Sorry not what I meant, the same arguement could be made about say a lifejacket you could buy some foam and material and other bits and make one yourself after all they are about fifty quid ! but some things in life have to be bought, I dont buy liquified rubber and make my own tyres for the car either, same arguement.

[WotEver]

You're right Gibbo, i dont really have a clue, but am willing to learn

Let me fill in a couple of the gaps that Gibbo left, then.

 

Someone I came across on the net reverse engineered a commercial one and it literally was just a pair of bridge rectifiers going in opposite directions across the earth wire and mounted on a heat sink.

Possibly he's telling the truth. If he is, what are the part numbers of the rectifiers? What is their specification? Did that particular GI meet any of the required legislation?

 

Surely if a short of thousands of amps was possible it would burn out the earth wire anyway so the diodes failing open curcuit would be immaterial and there must be a fuse somewhere.

Yes, there will be a fuse or breaker in the circuit, and if it's been correctly specified it will trip/blow well before any damage is done to the wiring. Can you say the same for the GI?

 

How long does a fuse take to blow? How long does an MCB take to trip? Until the power is disconnected the GI has to carry the full fault current without catching fire or blowing itself to pieces.

 

Part of the ABYC's specification for GIs (okay, that's not relevant in the UK, but it's still a good standard) states that a 70A GI must be able to carry a fault current of 5,000A (no, that's not too many zeros), applied three times, for a duration of at least one full cycle (1/60th sec in US, 1/50 sec in the UK) without any change to the electrical and mechanical characteristics of the isolator.

 

Could two arbitrary bridge rectifiers on a piece of aluminium do that? Can you be sure? Because if they can't, and they blow open circuit, you now have no earth protection whatsoever on your boat.

 

Tony

 

edited to add a comma

Edited May 11, 2011 by WotEver

[Proper Job]

Oh Please!!!!

I'm very willing to accept an argument that tells me I'm wrong.....but an argument that only states "I bought it in a shop and it was expensive so it must be better" is simply a silly argument.

Quality, or safety is not measured by cost.

Don't buy a cheap MPPT solar controller

[WotEver]

Don't buy a cheap MPPT solar controller

Oh no, you shouldn't - they're crap

[Gibbo]

Surely if a short of thousands of amps was possible it would burn out the earth wire anyway...

 

No. This is one of the bits that people get wrong.

 

A lot of this depends upon the length and size of the cable between the sub station and the "problem", but close to the substation, a direct short would result in a current of several thousand amps. Possibly (believe it or not, but under certain odd circumstances it is indeed possible) well into the tens of thousands of amps... for, say, 100uS until the fuse blew. This is nowhere near long enough to cause even the remotest problem to the cable. But the diodes will be toast within microseconds.

 

This is why the ABYC specs state the GI has to withstand repeated overcurrent tests of either 5,000 or 7,000 amps (depending upon the spec).

 

It isn't easy to find a diode that will handle that. An off the shelf bridge rectifier certainly won't even come close.

[gary955]

Sorry not what I meant, the same arguement could be made about say a lifejacket you could buy some foam and material and other bits and make one yourself after all they are about fifty quid ! but some things in life have to be bought, I dont buy liquified rubber and make my own tyres for the car either, same arguement.

Dont want to get bogged down with this because i appreciate the input of gibbo and tony but its not the same argument at all!!

You couldn't possibly make a tyre for your car from liquid rubber but a lifejacket made from some suitably bouyant foam and material might be every bit as effective as a shop bought one. It might not be as elegant but it should work.

THATS my point if a £70 or £100 GI contains a fivers worth of parts bolted to a lump of Alloy, then why not assemble the fivers worth of parts myself?

 

 

Gibbo

Are you saying that a high fault current would pop the diodes before blowing the fuse (or tripping the RCD?) leaving me unknowingly without an earth AFTER the fault has been rectified. Or that the failure of the diodes presents a very real danger DURING the milliseconds it might take for the fuse to blow?

If the former then could I not check the GI after a known fault, after all there would be some evidence of the fault having occured.

[Gibbo]

Are you saying that a high fault current would pop the diodes before blowing the fuse (or tripping the RCD?) leaving me unknowingly without an earth AFTER the fault has been rectified. Or that the failure of the diodes presents a very real danger DURING the milliseconds it might take for the fuse to blow?

If the former then could I not check the GI after a known fault, after all there would be some evidence of the fault having occured.

 

The diodes would pop way before the fuse.

 

The diodes could also pop without tripping the fuse. Imagine a very brief short circuit. Diodes pop. Fuse doesn't. You now have no earth but everything works. You're not even aware of the fact that there's been a problem.

 

Everyone knows that, to all intents and purposes, electronic components act instantly. What they often fail to understand (or believe) is that under gross overload conditions they also break instantly.

[gary955]

The diodes would pop way before the fuse.

 

The diodes could also pop without tripping the fuse. Imagine a very brief short circuit. Diodes pop. Fuse doesn't. You now have no earth but everything works. You're not even aware of the fact that there's been a problem.

 

Everyone knows that, to all intents and purposes, electronic components act instantly. What they often fail to understand (or believe) is that under gross overload conditions they also break instantly.

Hmmm now that does sound more worrying........but would the same be true of the commercially bought GI? for £70 or so, are they able to fit diodes that would be tolerant of such a fault?

I dont think the american spec is applicable here so wonder if UK models conform. What about the RCD. will that save me?

[Chalky]

Many commercially bought GIs in the American market have built in fault detection circuitry to check that the diodes are OK. Next question how do I know that the detection circuitry is working OK and has not failed as well or is showing the diodes as being OK when they've failed? The level of detailed design work and validation to answer this question isn't cheap... Put in a microprocessor as part of the design and you've entered a whole new ball game!

[Gibbo]

Hmmm now that does sound more worrying........but would the same be true of the commercially bought GI? for £70 or so, are they able to fit diodes that would be tolerant of such a fault?

 

Such diodes are available. Such diodes are indeed fitted in the "proper" GIs. But a cheap bridge rectifier is not going to come even close to meeting those requirements.

 

I dont think the american spec is applicable here so wonder if UK models conform. What about the RCD. will that save me?

 

This is actually a bit of a problem at present. I believe specs/standards are currently being drawn up for Europe regarding GIs (they may already be finished, it's been a while since I checked). But as there are US specs/standards, good engineering practice would dictate that they have to be used. If only because there aren't any others at present.

 

With regard to the RCD, a brief short circuit could pop a diode about 1000 times faster than an RCD could ever trip.

[RLWP]

Are we up to £70 of your professional development time yet Gibbo? It's important to include all the costs when assessing if something is "worth it"

 

Richard

[gary955]

Many commercially bought GIs in the American market have built in fault detection circuitry to check that the diodes are OK. Next question how do I know that the detection circuitry is working OK and has not failed as well or is showing the diodes as being OK when they've failed? The level of detailed design work and validation to answer this question isn't cheap... Put in a microprocessor as part of the design and you've entered a whole new ball game!

Hmmm not going to get into microprocessors! spanners and a soldering iron was more what I was thinking.

[Gibbo]

Are we up to £70 of your professional development time yet Gibbo?

 

Yes it took me longer than 5 minutes to type that

[RLWP]

So, the components are going to cost more than a fiver, and we've all spent hours on this thread.

 

That off the shelf item sounds a complete bargain now

 

Richard

 

[magpie patrick]

I find having a fibreglass hull is pretty effective...

Edited May 11, 2011 by magpie patrick

[RLWP]

I find having a fibreglass hull is pretty effective...

 

Edit to remove foolish reply...

 

Richard

 

Edited May 11, 2011 by RLWP

[gary955]

Such diodes are available. Such diodes are indeed fitted in the "proper" GIs. But a cheap bridge rectifier is not going to come even close to meeting those requirements.

 

 

 

This is actually a bit of a problem at present. I believe specs/standards are currently being drawn up for Europe regarding GIs (they may already be finished, it's been a while since I checked). But as there are US specs/standards, good engineering practice would dictate that they have to be used. If only because there aren't any others at present.

 

With regard to the RCD, a brief short circuit could pop a diode about 1000 times faster than an RCD could ever trip.

OK that seems fairly pursuasive but how do I ensure that my chosen commercially produced GI is not just a couple of bridge rectifiers back to back, like the reverse engineered one mentioned earlier?

It was supposidly from a well known manufacturer which wasn't named!