Isolator Transformer Query

[G6Tanuki]

Originally Posted by GMB

That's why you don't do it like that. Safety transformers have separate bobbins so we just need to add enough screening so that the secondary cannot "see" the mains and you can get a good result. Actually it may be enough to earth the laminations if the bobbins are far enough apart.

That 'separate bobbins' construction style is pretty much the standard design for old-style copper-and-iron wall-wart transformers.

[It's great for us technical-frobber types because it means you can rewind the secondary without disrupting the primary-windings].

The windings-on-top-of-each-other-with-a-grounded-interwinding-screen only works in situations where you have a ground available, making the design rather hard to implement in countries like the UK and loads of European countries where 2-pin mains sockets are the norm.

[kalee20]

Originally Posted by GMB

That's why you don't do it like that. Safety transformers have separate bobbins so we just need to add enough screening so that the secondary cannot "see" the mains and you can get a good result. Actually it may be enough to earth the laminations if the bobbins are far enough apart.

Good point, and yes it would give better isolation and lower capacitance of the floating secondary to everything else.

Although, most of the isolating transformers that I've seen, have been as I've described. The twin-section bobbin type is the exception rather than the rule. It also has slightly less-good regulation, due to higher leakage inductance.

[Lucien Nunes]

around 7uA, still enough to give a tingle!

Not sure I can feel 7μA as a tingle. To get a response on such low currents, I stroke the conductor with the back of the fingers, which causes a muscular response that can be perceived by the resulting vibration, instead of feeling the current itself.

If the primary winding was powered by a floating earth free supply, then would that stop any capacitance induced voltage [referenced to earth] in the secondary winding ?

No, unfortunately. The problem as mentioned above is not just that the secondary is coupled to the primary, but that it is always coupled to earth by its stray capacitance whatever form it takes. Even a hand-cranked AC generator that has no electrical input, will have stray capacitance and hence be able to produce earth leakage despite perfect insulation.

The question is, what is an acceptable leakage current for something that is considered to be electrically separated from earth? It depends what you are trying to do with it. For example, 200μA is quite harmless, much lower than the limit for touch-leakage on Class II appliances, but will still light an indicator. Imposing a leakage limit in turn imposes a limt for the stray capacitance (or its unbalance) which imposes a limit on size. The bigger your transformer, the higher the capacitance so the higher the leakage is likely to be.

[woodchips]

Doesn't seem to have been mentioned but don't leave the secondary of a mains isolating transformer floating. Why? if you earth one side then you can use an RCD before you start holding the wires, get the considerable protection these afford if you do dob your finger where it shouldn't.

On a similar point about capacitive coupling I have recently had an interesting example of this. My last readout I designed 20+ years ago was battery powered using a 1/2AA lithium cell. The readout, being used on machine tools and similar is usually in a fairly dark environment, to the display used were LED, I finf LCD to be rubbish. Whilst you can wave a DVM around to get the reflection to work, not quite so easy with 3 tonnes of drilling machine. The readout had power saving to blank the display but many customers found that irritating so I offered a mains version. This used a 9V wall wart with regulator to drop it to 3.6V.

No problems for years, then customer said the readout was jumping positions, not really much use. In the end I discovered that it was the capacitive coupling through the PSU that was upsetting the readout.

The opto sensor uses power saving, turning the LEDs off, also the display is software scanned so there is no digits displayed if a low reading. This meant that the readout current consumption was zero for a considerable time as the readout was used. This allowed the capacitive coupling to blatt the ADC input and give the false reading.

Testing produced an about 80V pk-pk 50Hz sinewave on the power connector, plenty to overload the ADC input. This was at minimal current, testing suggested 10uA or less. Further testing showed that a 1k resistor across the 9V, so about 1mA, was enough to completely stop the 50Hz noise voltage and solved the jumping problem.

As an aside, when I started selling these I supplied a 9V 300mA PSU, but transformer based, these never had the capacitive problem, so no unhappy customers. Then transformers were out of fashion and switched mode PSUs were in, so it took many years to use up my stock of PSUs.

Something to be aware of? It was the time based change in current drawn, and being zero for blanked digits that caused the head scratching, you see a digit and automatically assume they are on 100% of the time, nope!

[GMB]

Originally Posted by woodchips

Doesn't seem to have been mentioned but don't leave the secondary of a mains isolating transformer floating. Why? if you earth one side then you can use an RCD before you start holding the wires, get the considerable protection these afford if you do dob your finger where it shouldn't.

Doing that means it is no longer an isolating transformer!
If floating then touching either secondary wire, or indeed any part of the connected circuit, has little effect because there is no solid path to earth - something that earthing it totally ruins! It only kills you if you touch both wires - something that is also lethal with the mains and is the situation where the RCD just sits and watches you die.

[ms660]

Out of interest....It can happen.

Lawrence

[Lucien Nunes]

if you earth one side then you can use an RCD

But the RCD requires you to get a shock to trip, whereas if it's floating you don't get the shock in the first place.

I just did some tests with my own bench isolating transformer. This is a 1650VA Carroll and Meynell (3kVA 30 mins) 230-230V rated. It has an earthed inter-winding screen and earthed core.

At the time of testing, the mains was 242V and the off-load secondary voltage 247V. Our mains intake at the workshop is TN-C-S and my bench is near it, so the N-E voltage at the bench is always trivial, in this case 0.15V.

Insulation test at 1kV DC. Insulation Pri-Sec, Pri-E and Sec-E >1GΩ. Using Megger MIT310.

Open-circuit AC voltage from each end of Sec to earth: L1-E 64V, L2-E 118V. Using Fluke 289.

Short-circuit AC current from each end of Sec to earth: L1-E 15μA, L2-E 29μA. Using Fluke 289.

I then attempted to measure the capacitances with the Fluke but the readings jumped around wildly. I don't know how the capacitance range works on that but it clearly didn't like the capacitance being distributed over the inductance, so I switched to a simple LCR meter on its 2nF range, which uses a measurement frequency of 1kHz. This read all three capacitances as lumped values with no change according to which end (or both) of the winding was tested.

Capacitance Pri-E 1.0nF, Sec-E 0.53nF, Pri-Sec 0.352nF.

The theoretical series value of Pri-E and Sec-E would be 0.347nF, so the very close measured value of 0.352nF, and the fact that that neither reading changes when the other winding is shorted to earth, suggests that the two windings are effectively screened from each other. Leakage current measured from Sec-E is therefore confined within the secondary, and exists only due to its stray capacitance and not to the presence of a TN supply on the primary.

I was about to calculate the equivalent source impedance of the secondary earth leakage current but I can't immediately confirm the input impedance of the 289 on AC volts which is needed as it's in the same order of magnitude as the source impedance (it's not resistive IIRC). Will return to the subject later.

[Lucien Nunes]

Out of interest....It can happen.

The old 'crossed mains lead E&N' ploy. That's one reason not to connect the earth terminal in the socket of the isolated output. I recall a discussion of the pros and cons of this a while back. Mine is disconnected - if I want to earth bits of the DUT I will take an earth there myself.

[G6fylneil]

I recall a 13A socket at an industrial unit with a crossed L&N. It went unnoticed for over 30 years, until an extension lead with a crossed N&E was plugged into it.
I've got a 500/750VA RS isolation transformer (looks like a Caroll & Meynell), with nothing plugged into it I get 74V N to E and 46V L to E measured with a Fluke 89IV. It's input impedance is quoted as 10M < 100pf. (Luciens' 289 is probably the same). The earth of my transformer output socket is connected to the transformer supply lead earth, but there is provision to disconnect it.

[Roger Ramjet]

I struggle to understand how one can get a L&N reference on the floating output of an isolation transformer i.e. if the secondary was tied down to earth at one end that would indeed become the NEUTRAL reference and the opposite end would become the LIVE reference but without a tied down earth both wires have no relationship to earth (other than the capacitive effect)

Rog

[Lucien Nunes]

I think that's a matter of terminology, in that we might call the two ends of the secondary connected to the left and right-hand contacts of the output socket 'N' and 'L' respectively, just to identify them, but as you say neither is a neutral. They are just two wires of a floating circuit. In my post #27 I called them L1 and L2.

[G6fylneil]

I used the L, N, and E designations because my transformer is fitted with a BS 1363 socket, its terminals are labeled L, N, and E. I think in the UK these days both Line and Neutral are considered "Live" as they are "connected to the source of supply". Internally the transformer has two red wires coming out of the potting, They are probably connected "the other way round" from Lucien's, and I get a lower voltages because the transformer has a lower VA. (or my meter has a higher input capacitance)
My 89IV measured 320pF from secondary to E, 277pF secondary to primary, and 540pF from primary to E, but that's also got 1.3M of 3-core flex on it.

[Lucien Nunes]

I think that in the UK both Line and Neutral are considered to be "live" these days as they are "connected to the source of supply".

Yes indeed and this distinction is often glossed over on the forum, because of traditional usage and old habits. The word 'Live' has multiple meanings and the new definitions try to clarify that, by avoiding such ambiguities as a 'neutral conductor that is live.'

In 2022 any conductor that is energised in normal use, e.g. both the brown and blue wires in a mains lead, but not the green/yellow one, is termed a 'live conductor'. Of these, the one that is held at or near earth potential is the 'neutral' and the other one is the 'line.' With the floating output of an isolating transformer, there are two live conductors, but they are not identified as line and neutral because neither is held at or near earth potential.

Obviously, the key difference is that what we now call line used to be called live. In the vintage equipment context, this presents a continual conflict between old and new usage. We connect the line conductor to the terminal marked live etc. I try to use the terms in a way that is clear in the context. Talking to an electrician, I will always use the modern definitions, but discussing the history of fuse boxes with a collector, I will call the two poles live and neutral because that is what they are physically marked in the box.

There are certain areas of electrical definitions such as the distinction between lines and phases, TN-C-S and PME, 2-phase / split phase / two phases of 3-phase; etc where people do tend to be a bit careless even though precise definitions exist.

[Roger Ramjet]

Originally Posted by G6fylneil

I used the L, N, & E designations simply because the transformer output is fitted with a BS1363 socket, it's terminals are labeled L, N, and E. I think that in the UK both Line and Neutral are considered to be "live" these days as they are "connected to the source of supply". The measured voltage was different for the two terminals, and my readings were lower than Lucien's, presumably because my transformer has a smaller VA.
I measure 320pF secondary to earth, 280pF secondary to primary, and 540pF primary to earth, but that's got 1.3M of 3-core flex on it.

One could argue that an isolating transformer should be fitted with a old fashioned two pin socket especially when connecting vintage "live chassis"'' equipment which originally would have been supplied with a two pin plug.

Conversely, is it best practice to plug an appliance with an earth wire and extraneous metal parts into a BS1363 3 pin socket of an isolating transformer?

Rog

[Lucien Nunes]

My isolating transformer is modern, and is often used with modern equipment that already has a BS1363 plug on. I can't see any advantage in fitting a non-standard socket.

is it best practice to plug an appliance with an earth wire and extraneous metal parts into a BS1363 3 pin socket of an isolating transformer?

Under the conditions where I need the live functional parts of the equipment to be electrically separated, yes. E.g. when scoping the line side of an SMPSU. Under other conditions, no, it should be connected to a TN supply with earthed neutral.

I think you mean 'exposed' rather than 'extraneous' in this case, as extraneous means introducing an uncontrolled potential into an equipotential zone (as with a metal water pipe etc). An appliance would only have extraneous conductive parts if it was sat on a patch of damp earth with its cable run indoors.

[Lucien Nunes]

I'm too late to edit the previous post but I would add a comment about the relative safety of being connected to an isolating transformer or not, when not technically necessary such as undergoing maintenance.

A live-chassis set is always likely to be safer run from an isolating transformer, as this overcomes one of the intrinsic hazards of most such sets, namely poor insulation of live functional parts and poor protection against direct contact.

A class II appliance in good condition won't really be affected safety-wise by the transformer, as none of its protection relies on the supply having an earthed neutral.

A class I device, again in good condition, is also unlikely to be much more or less safe when isolated (especially if in the non-isolated condition there is additional protection in the form of an RCD) although the mode of protection is changed.

Situations where an isolating transformer does make things less safe are:

a) Where an appliance wiring error, in conjunction with an earth present at the outlet of the transformer, leads to the specific situation illustrated by Laurence in post #26 where the secondary becomes connected between earth and the casing of the appliance.

b) Where multiple devices are connected to one isolating transformer, and a fault in one or the wiring (first fault) defeats the protection by electrical separation at another, yet remains undetected due to the lack of ADS.

c) To a minor extent, where a large and complicated Class I device can have multiple faults in separate sections, as though they were separate appliances as with b) above. This might include a projection outfit comprising separate projector, transformer and amplifier units, each Class I but with increased possibility of broken protective conductors in the interconnecting cables preventing their casings from being held at the same potential.

The main one is b) and this is why an entire shack or bench should never be supplied from one transformer, and why public mains supplies are not floating (you can't have one customer's safety depending on another customer's installation being free of faults.) The same situation can arise with an IT (isolated from earth) portable generator, which should never be used to power the installation of a building that is designed for a TN supply without an N-E link being made. The problem can be controlled up to a point by separately protecting each device with an RCD. These will not trip on a first fault as they would with a TN supply, but if two loads are faulted to opposite potentials, both RCDs should trip in the event of a shock between them. However, an IT supply should not really be used in such situations.

[woodchips]

An isolating transformer removes you from the infinite busbar incoming mains.

The RCD will detect and trip a leakage current before you even notice.

It is like having a generator outside, it is much better to know what voltage is around, like an isolating transformer, and the RCD is a useful safety device, either 230V or 400V 3 phase.

The isolating transformer with neutral earthed means there is only one wire that has electrocution possibilities, and the RCD cancels that.

[stuarth]

Regrettably, the argument that you should earth one side of an isolation transformer secondary and add an RCD has been discussed on here many times. It’s an appealing idea.

See GMB’s post #25 or the first line of Lucien’s post #27 on this thread for an explanation of why this is a bad idea.

Yes, if you earth one sided the secondary there is only one wire with electrocution possibilities. Just like the mains. But if you don’t earth anything on the secondary, there there are no (single) wires with electrocution possibilities.

Of course if you connect yourself across two wires of the secondary you could die, and if you had an earth connection and an RCD, then, as GMB says, the RCD will watch you die.

Leave the secondary with no earth connection, I.e. isolated!

Stuart

[TonyDuell]

I can see no point at all in having an isolating transformer and then connecting one side of the secondary to earth all the time. Since mains earth and neutral are bonded together on the incoming feed to my house (PME installation), the isolating transformer would seem to add nothing.

Of course there will be times when one side of the secondary does end up earthed A common example is if I am working on a typical AC/DC radio, powered from said transformer and I clip the earth lead of my 'scope probe onto the chassis. There's a connection from one of isolating transformer output wires via the 'neutral' wire of the radio's mains cable, radio chassis, 'scope probe earth lead, 'scope chassis to mains earth. But that's why I have an isolating transformer, so I can make that connection without tripping the RCD in my consumer unit.

And if I have a radio where there's a bias resistor between the neutral mains lead and chassis, I can earth the chassis without removing the bias from the valves. Or more typically I'll be working on some kind of switch-mode PSU which starts by bridge rectifying the mains. With an isolating transformer I can earth the -ve output of the rectifier without effectively connecting a diode across the mains (== blown fuses and diode hitting the ceiling)

High voltages are dangerous if mishandled and I do not believe for an instant that an isolating transformer, or anything else for that matter, makes them safe under all conditions. The transformer just lets me make measurements I couldn't do without it.

[G6fylneil]

Tony,

"I can see no point at all in having an isolating transformer and then connecting one side of the secondary to earth..."

At work we had testing machines which were supplied via isolation transformers to stop the necessarily high leakage currents from the filter capacitors for the motor drives etc. tripping the mandated RCDs in the supply. Earthing the secondary is a sensible thing to do.