Isolating Transformer (UK)

[Ian.M]

Unfortunately, its common practice to connect grounded testgear to a D.U.T. fed from an isolating transformer.   As soon as you introduce an earth downstream of the isolation, you are totally unprotected against shock, even if the isolating transformer is fed from a RCD.  Also although a floating isolated supply is safe under single fault conditions, it is unlikely to have automatic earth fault detection so there is a high risk of continued use until a second fault occurs.  The fact that you can touch one (LF) node of a circuit fed from a fully floating  supply with no immediate consequences promotes carelessness and unsafe working practices.   It is much much harder to intentionally or accidentally defeat the protection offered by a RCD and their use does not encourage unsafe working practices.

[SteveyG]

Unfortunately, its common practice to connect grounded testgear to a D.U.T. fed from an isolating transformer.

That simply comes down to stupidity.

[madires]

Don't rely on a single protection method! It's not isolation transformer vs. RCD, it's the question how we can add several layers of protection:
- a RCD is a good idea anyway (though being defeated by an isolation transformer)
- an isolation transformer (more reliable than a RCD and less painful)
- use isolated T&M (or differential probes for your scope)
- rubber gloves?
- procedures (as described in another post)
- check the protection devices regularly

[cprobertson1]

So should I be forking out for an isolation transformer: or will a dedicated 10mA RCD be enough, bearing in mind that it's fairly rare that I have to actually deal with a mains potential circuit and that I obviously follow general safety procedures* when doing so.

*Which I'll probably revise with a qualified electrician in the near future, just to be certain I'm not doing anything unsafely and not realising it - better safe than sorry!

[MrSlack]

From experience, a qualified electrician in the UK is the last person you want to ask about electrical safety.

[cprobertson1]

From experience, a qualified electrician in the UK is the last person you want to ask about electrical safety.

HA!

Good point 

In fact, to quote myself,

Though I've met a surprisingly large number of electricians who say they've had a shock before.......

'Nuff said

I was thinking along the lines of the HV technicians at my place of work who regularly deal with the 450V/600V CNC machines - but then again... they're probably not the best folks to ask either... I've seen some of them do.....things...

Take it the HSE manuals are the way to go?

[nctnico]

Most of the personal safety benefits of an isolating transformer are moot assuming your bench supply is effectively RCD protected.

RCDs are not suitable as primary protection against electric shock. Sure they help reduce the risk of death, but people have still gone into fibrillation from contact with an RCD protected circuit. A 'working' RCD only guarantees a maximum disconnection time not any kind of current limit.

Isolation transformers however are a very useful tool providing you understand what you are doing. I would never substitute working on some live gear with an RCD instead of an isolation transformer - they just aren't reliable enough.

The problem is in the understand what you are doing part. An RCD/GFI at least prevents a lethal/prolonged current to flow. An isolation transformer can easely be defeated by grounded test equipment or give a false sense of safety. The only safe way to work on mains (or any high voltage circuits for that matter) is to use isolated probes and never ever touch the circuit.

[Ian.M]

Personally, I'd have both isolating transformer and high sensitivity  RCD, and also an insulating floor mat (under an ESD mat if you need that) + a bench that does NOT have a metal frame.  All sockets on the bench apart from the isolating transformer output should be RCD protected.

Most of the time in the TV shop I used to work in, a 250VA isolation transformer was sufficient.  A 500VA one will certainly suffice for everything except heating/cooking appliances and those with large motors.  An isolating transformer, used intelligently, improves your safety when working on the low voltage side of equipment with an open frame mains PSU or one integrated on the same board.  However you must always be aware of anything that introduces a ground - which would trip the RCD without the isolation transformer - as that instantly removes the protection offered by both.

[Mark Hennessy]

In an electronics workshop - which might be slightly different to what the OP is doing - then there is only one use of an isolation transformer: to allow the safe connection of test equipment to a DUT. The classic example has already been given; the primary side of a switched-mode PSU (where the -ve output of the input bridge is a negative-going half-wave waveform wrt mains earth with a peak value of -325V). In days gone by, prior to baseband video/audio inputs, TV sets had a live chassis, so also needed to be isolated for the same reason.

Note that "allowing safe connection of test gear to the DUT" is nothing to do with the probability or magnitude of electric shocks - the transformer is not intended to change that. Naturally, as soon as you've connected test gear, the DUT is no longer floating, so it remains hazardous; but the transformer has allowed you to tie it to ground at the point you need to make a measurement...

A really important rule - not yet mentioned, I think - is this: only one DUT may be connected to the transformer.

Also, the mains earth should not be carried through to the outlet. If you have a DUT with a N-E swap, the case becomes live.

The default working practice it to power a DUT from an RCD, and only use the isolation transformer for those occasions described above. Naturally, when using the isolation transformer, any RCD in line becomes redundant, but doesn't hurt.

Away from all that, if you wish to use an isolation transformer to reduce the magnitude of an electric shock, then you need a centre-tapped secondary, and you ground the centre-tap. This gives you two lines at 115V wrt mains earth. And you should be able to use an RCD at the output (but check it works properly in that mode). But naturally, this supply is no longer floating, so can't be used to safely connect test gear to a DUT as described above.

Personally, I doubt the utility of this arrangement in an electronics workshop environment, and have never used it myself, but it might be appropriate for the electrical repairs that the OP does? Of course this scheme is used on building sites - those yellow-cased "site transformers" produce 110V from a transformer with a grounded centre-tap, so the maximum voltage wrt mains earth is 50V.

I've attached a BBC EGN which might be of interest.

All the best,

Mark

[nctnico]

Another problem I see is distance to the distribution/switch board comes into play with ground protection and that can add delay. Making an additional ground/earth is a no according to some wiring regulations (depends where you live of course), they are also a proximity hazard as well.

Now you are not making sense. The delay is in the order of nanoseconds where the speed of impulses through your nerves is in the several tens of meters per second ballpark so it takes a pulse to go from your brain to your toe tens of milliseconds.

@Mark Hennessy: TVs with a hot chassis are something from the 70's maybe 80's which shows how antiquated using an isolation transformer is. Nowadays just use a differential probe and be done with it. With multiple differential probes you can look at several points in the primary part of a PSU if you like.

[SteveyG]

The problem is in the understand what you are doing part. An RCD/GFI at least prevents a lethal/prolonged current to flow. An isolation transformer can easely be defeated by grounded test equipment or give a false sense of safety. The only safe way to work on mains (or any high voltage circuits for that matter) is to use isolated probes and never ever touch the circuit.

No it doesn't. Conditions can be such that you sustain an electric shock with a current that is not sufficient to trip the RCD, yet be greater than the threshold to cause muscles to contract or be enough to cause cardiac arrest. An RCD won't trip at 20mA, but you'll have a hard time letting go and will lead to death.

[nctnico]

That is easely solved by using a more sensitve RCD and not a reason to disqualify any RCD.

[cprobertson1]

Now you are not making sense. The delay is in the order of nanoseconds where the speed of impulses through your nerves is in the several tens of meters per second ballpark so it takes a pulse to go from your brain to your toe tens of milliseconds.

Sorry, I'm not understanding why you brought up nerve impulses?

Looking at this from a biological perspective, an electric shock usually passes through your lovely salty electrolytic fluids that are between your cells rather than charging through the oily cells (cells have lipid bimembranes which are surprisingly good insulators: so much so that many cells use them to generate a potential difference by separating +ve and -ve ions onto opposite sides of it - similar to charges on a disc capacitor!) - nerves don't provide a particularly good pathway for electricity to flow, and are certainly no more conductive than regular tissues - your circulatory system is the best candidate for passing current from one pace to another in your body (as blood contains lots of salt and water - though I've never actually measured the resistance of a droplet of blood... I might try that next time I cut myself, and perhaps observe it under a microscope while I pass current through it to see what happens....)

Please note that this assertion that neurons are bad conductors in the case of electrocution is speculative (based on my being a qualified biochemist) - but believe it or not, most biochemists only care about the macro-effects of being electrocuted - not what's actually happening at the microscopic level (i.e where the current is actually flowing and how and why)


(Did you know that nerve impulses are actually mostly chemical? An electrical signal (or more specifically, as each cell is discrete, a potential difference across the membrane of the initiating cell) - which triggers voltage-gated calcium channels which then open up - allowing for the uptake of calcium from the synaptic cleft (the gap between two synapse) - the high Calcium concentration within the cell triggers the release of neurotransmitters into the synpatic cleft, which then binds to chemical receptors on the neighbouring neuron, which can then trigger one of several responses - the traditional nervous impulse involves the uptake and release of certain ions within that cell, which triggers that cell to repeat the process with it's neighbour, and so on and so forth (and at the end of the impulse there is generally a chemical feedback loop which amplifies the (chemical) signal into an actual affect - this is obviously an undetailed summary - there's a lot more to it of course! (my degree was in Biochemistry, specialising in Immunology - that was before I became a tooling engineer though!))

[Delta]

No it doesn't. Conditions can be such that you sustain an electric shock with a current that is not sufficient to trip the RCD, yet be greater than the threshold to cause muscles to contract or be enough to cause cardiac arrest. An RCD won't trip at 20mA, but you'll have a hard time letting go and will lead to death.

I assume you mean a 30mA RCD won't trip at 20mA?

[nctnico]

Now you are not making sense. The delay is in the order of nanoseconds where the speed of impulses through your nerves is in the several tens of meters per second ballpark so it takes a pulse to go from your brain to your toe tens of milliseconds.

Sorry, I'm not understanding why you brought up nerve impulses?

Just to outline that the 'systems' in your body that may be affected by a shock are extremely slow compared to the speed at which electricity flows through a conductor. The delay caused by wiring to detect a leakage by an RCD is totally insignificant.

[cprobertson1]

Ah, got you! I misinterpreted it at first - sorry!

The speed of conduction through a human is still pretty high though - the extracellular matrix and circulatory system in particular are moderately good conductors (compared to other tissue that is - they're terrible conductors compared to "normal" conductors!) - but the current speed is still pretty fast.

As you mentioned though - the time delay caused by longer wiring should be fairly small? Might Shock mind clarifying the effect he mentioned (i.e how it works ?)

[Ian.M]

I suspect Shock has confused different types of breakers.
For MCBs, there is an issue with wiring length (actually total loop impedance of the L conductor from the distribution transformer and E back to) when clearing a hard 'bolted' fault to P.E.  Within their max clearing current rating, a lower loop impedance results in a faster trip.

However modern RCDs almost invariably use a balanced L-N current transformer for leakage current detection, and are unaffected by any wiring impedance that doesn't interfere with normal operation of even low wattage loads.

The other issue of relying on a single mechanical device for safety isn't strictly accurate.  At least in the UK, there are now very few ring circuits with 13A outlets that aren't RCD protected at the consumer unit, usually by a whole house 30mA RCD or a 32A/30mA RCBO just for that ring.  So, assuming the bench has local RCD protection, both it and the consumer unit RCD would have to fail, + the user would have to fail to follow safe working practices before there would be a significant risk of shock with serious consequences.

[Delta]

Shock is getting confused and thinking of the operation of Overcurrent Protective Devices (MCBs, fuses...).  The speed at which these will disconnect a hard earth fault and thus protect against indirect contact depends upon the fault loop impedance, and thus the length of the cable run (and conductor CSA) will affect the speed of operation.  all of which is completely irrelevant when talking about RCDs.  If current flows at 0.7 * the speed of light in copper and 0.6 * the speed of light in human tissue; so fecking what! 


EDIT - Beaten to it by less than a minute!

[MrSlack]

To quote my engineering manager about 18 years ago:

"technology (GFI/RCD etc) doesn't solve a whole lot of electrical safety problems. It just stops the next person who comes along who stupidly tries to drag your dead body off the bench from getting toasted too"

[nctnico]

Well, don't take my word for it:
www.tek.com/dl/51W_10640_1.pdf

Oh, and I do take safety very seriously. When I was a teenager a good friend of mine electrocuted himself while tinkering with electricity.

[Delta]

Well, don't take my word for it:
www.tek.com/dl/51W_10640_1.pdf

Oh, and I do take safety very seriously. When I was a teenager a good friend of mine electrocuted himself while tinkering with electricity.

That says not to use an isolation transformer to float the instrument, it absolutely does not say that an isolation transformer should not be used to supply the DUT!

VERY important distinction!

[cprobertson1]

VERY important distinction!

Lol, even I know that one

We really need a sticky post outlining the facts and issues of bench safety in the first post, this comes up all too often and we get people like the "I touch the live mains all the time and I don't die" guys who just distort reality to new players.

That would be a very good idea - especially given how easy it is to get seriously injured (and/or killed) if you mistreat it =/

[nctnico]

Well, don't take my word for it:
www.tek.com/dl/51W_10640_1.pdf

Oh, and I do take safety very seriously. When I was a teenager a good friend of mine electrocuted himself while tinkering with electricity.

That says not to use an isolation transformer to float the instrument, it absolutely does not say that an isolation transformer should not be used to supply the DUT!

VERY important distinction!

You didn't read far enough! In the table on the next page it says not to use an isolation transformer to make floating measurements -period-.

[Mark Hennessy]

Well, don't take my word for it:
www.tek.com/dl/51W_10640_1.pdf

Oh, and I do take safety very seriously. When I was a teenager a good friend of mine electrocuted himself while tinkering with electricity.

That says not to use an isolation transformer to float the instrument, it absolutely does not say that an isolation transformer should not be used to supply the DUT!

VERY important distinction!

You didn't read far enough! In the table on the next page it says not to use an isolation transformer to make floating measurements -period-.

Says the people who sell differential 'scope probes

I think the advice in the attachment I provided is perhaps more balanced, especially as it was written by someone who didn't want their employees to come to harm, rather than a marketing department...

The problem is not the transformer - the problem is education.

[nctnico]

I've read that BBC document but it is basically trying to regulate stupidity (also note it refers to 'rules' from 1989 which is more than 25 years ago). What is more simple than: don't touch it when powered/charged and use a differential CAT rated probe to measure? I really don't see why people still try to defend the use of isolation transformers with more do's & don'ts than the 10 commandments which people also don't follow.