My isolation transformer has its protective earth connected to the output neutral. How should I rewire it for my home lab for "DUT probing"?

Question

I have a rather sturdy isolation transformer in my home lab and I just realized that the primary side earth is not only connected to the output side earth but also to the output side neutral. I don't understand why, but it explains some odd measurement results I've had when I assumed everything was isolated from earth.

^{Inside of the whole device, toroid is potted inside and halfway up outside. Fuse is in the rightmost switch.}

^{Closeup of the connections corresponding to the schematic.}

^{simulate this circuit – Schematic created using CircuitLab}

The simplified schematic (without the filter and protection) shows where the input earth is connected to the chassis but also to the respective output earth and to the output neutral.

I could not find documentation for it. The label is old and faded, judging from the markings on the capacitors and power switch it was made no earlier than 1989.

PROD NO :   60065
LOW IMP TRANS ISOL
SERIAL NUMBER :
 0750-UK  9003048

MADE IN SWEDEN

Why is it like that?

When is this ever useful? Since I got this from an electronics lab surplus auction I don't have the manual, but the label in the back says "LOW IMP TRANS ISOL". Have they just used it as a large line filter? It seems to me that with the current configuration it is as deadly as a direct connection to the outlet.

I thought that the point of an isolation transformer was to remove the earth ground potential or tie on the neutral wire. Even the venerable Olin was taken by surprise by a device such as mine and blew a fuse.

How should I modify it?

My primary goal is to isolate devices I want to measure with my oscilloscopes (DUT).

First of all I should probably keep the transformer chassis connected to the input earth and remove the link from the chassis to the output connectors.

But then I have at least two or three options:

1. Remove every earth wire on the output connectors, including the link C to the output neutral. This is how I actually expected the transformer to be wired.

2. Connect the output earths to each other but remove the link C. This seems pointless to me. The only thing it does is to make sure two or more different devices will have their earths connected, something you may or may not want but probably don't care about in this context.

3. Leave the outputs as they are, just remove the connection between input and output. The output neutral will then still be connected to the output earth, but will be floating with respect to the input side earth (the "real" earth).

If I understand some other answers here, option 1 is really the way to go, but I wonder if option 3 would give me any kind of benefits?

Answer 1

Your device has an isolation transformer in it, but the device itself isn't called an isolation transformer. Mostly because it doesn't isolate.

The device with your description is a "line conditioner" or "power conditioner".

Their point is not to isolate in the sense where you want to use an isolation transformer, but their function is to be a passive mains supply filtering device, which helps to make a clean mains supply for sensitive equipment, perhaps in a heavy industrial environment where the raw mains is polluted by noise from high power loads and their transients.

I don't know how your device specifically works, but the transformers used may be ferroresonant. Even if your transformer isn't, it still has filter caps on the output.

It basically gives you a local noise free earth-referenced mains output, which could be achieved only by wiring a separate outlet back to your main junction box where Neutral and PE are bonded together, although it might pick up interference along the way, so this is better in that sense.

It might be useful as it is in some cases, so there might be no reason to modify it into an isolation transformer.

Isolation trasformers generally have only two prongs for output, so I would be a bit wary when connecting equipment that needs a functional earth like a desktop PC to a isolating tranformer. I would still earth the case externally, as the mains filter caps are from PE to N and from PE to L, so there is no guarantee at what potential the case floats, and you also need other equipmen to connect to the device that may be earthed or floating.

Answer 2

An isolation transformer makes things both safer and more dangerous:

• If you touch one of the output wires, no current will flow through you and to Earth, so you won't get zapped.

• However, because no current can flow from mains to Earth through the transformer, a Residual Current Device (RCD) cannot detect faults. So if you touch one of the wires, it won't trip, it will just wait until you touch both wires at the same time to zap you.

As it is wired currently, it offers no isolation from Earth, but will still prevent the RCD from tripping, as the RCD measures fault current on the primary side, and this current is blocked by the transformer.

Remove every earth wire on the output connectors, including the link C to the output neutral. This is how I actually expected the transformer to be wired.

That would result in a standard isolation transformer, with safety advantages and drawbacks as outlined above.

If you use it with a load that requires Earth, and the Earth pin is disconnected, the load's earthed chassis will be isolated from Earth. This means any fault to chassis will go undetected. To get zapped, you need a fault to chassis from one wire, then touch both the chassis and the the other wire.

Leave the outputs as they are, just remove the connection between input and output. The output neutral will then still be connected to the output earth, but will be floating with respect to the input side earth (the "real" earth).

This makes it more dangerous, as the load's "earthed" chassis is now connected to the neutral, creating a permanently fault to chassis. If you work on the device, you will most likely touch the chassis and rest your hands on it, so all you need to do to get zapped is touch live, and the RCD on the primary side won't trip.

To use this mode you'd need a RCD on the secondary side, which would trip if current flowed through the output live wire, but returned through the output "earth" wire instead of output neutral. While output "earth" and output neutral are connected, this is done upstream of the RCD, so it can work.

I'm not too fond of this option. Using an isolation transformer on Class I devices (which require Earth) is a bit problematic.

If the end goal is to probe mains voltage with the standard earthed lab equipment (ie, scope) then the scope should be earthed, the device under test should be isolated, and for safety purposes you should behave as if it was not isolated.

Floating a scope with an isolation transformer and probing mains is a Really Bad Idea, especially if it has a touchscreen, or worse, is connected to a PC via USB...