Isolation Transformer

Transformer With my new hobby room almost complete it’s time to think about reinstalling my isolation transformer. I have used it in our previous house, so technically there is nothing new about it. Simply put it is a 550VA lump of iron and copper in the meter stand on the workbench, connect the wires to a plug, an on/off switch, a fuse and the output socket and we’re done. O, I almost forgot the front panel of course.
But I wanted to do a little bit more this time. I want to add an ananlog current meter. I’m not so much interested in adding a voltage meter though. The voltage can easily be measured with a multimeter should I want to know it precisely. Current is different. Adding a current meter is a bit more complicated because you’d have to break the circuit in order to put the meter in series. And current of course is always the great unknown, whereas the voltage is relatively constant. And I want to see when I’m at risk of blowing the fuse when I connect a load to my isolation transformer.

What’s The Purpose Of An Isolation Transformer?

First of all a small introduction to what an isolation transformer is and what it does. An isolation transformer is a normal transformer with a 1:1 input to output ratio, which means that the voltage on the secondary winding is equal to the voltage on the primary winding. So if you put 230V in, you'll get 230V out.

Working on mains powered devices can be dangerous because of the high voltages and relatively high power capabilities of the mains net. The isolation transformer doesn’t guard you against that. At best is lowers the maximum deliverable power, when an appropriate fuse is fitted. However it is a bit safer to work on mains powered devices when you run them from an isolation transformer though.

On most mains networks one of the output poles is connected to earth in the fuse box of your house. That pole is appropriately called Neutral. The other pole is called Live and carries the 230V, or whatever is available in your area.

No isolation transformer

With isolation transformer

You can safely touch the Neutral pole, if you know which one it is. But touching the Live pole can be quite dangerous, depending on how well you are connected to earth yourself. I remember getting a nasty shock in the distant past while working on a TV in somebody’s living room when I touched the TV chassis, while touching the central heating with my butt. Those were the days when the chassis of the TV was directly connected to the mains. I’m not sure RCDs were invented back then. At least I had never seen one.

Both output poles of an isolation transformer are floating. Neither is connected to earth. This makes it safer to touch any single output pole because no path to earth can exist through your body. Mind the word single in the previous line. Touching both poles at the same time is equally dangerous as touching the wires directly from the mains.
Also keep in mind that the RCD will not detect residual currents occurring at the output of the transformer. If you connect one output pole to earth and when you touch the other output pole you can still get a nasty shock and it will not trip your RCD! So in that sence the isolation transformer has made the safety worse.

Why would you connect one pole to earth of an isolation transformer? The whole idea of the thing is not to do that. But sometimes you’ll have to. Suppose you want to measure some waveforms with your oscilloscope on the hot side of a mains powered device. The ground terminals of your scope are internally connected to earth. Connecting the ground clip of your scope probe directly to the hot side of a mains powered device is a disaster waiting to happen. Not only will you create some nasty ground loops, which might ruin the results of your measurements. But it may also destroy much more than that, like the earth path on the PCBs of your scope, or the device under test, or your scope leads, or yourself.
Carefully finding out what the Neutral pole is before connecting your scope probe to that line is not always enough. Suppose you want to troubleshoot a switch mode power supply. You would want to connect the ground clip of your scope probe to the minus pole of the rectified mains voltage. That is a big problem, because there will be no Neutral side after the bridge rectifier. Connecting your scope probe to the minus side will short out your bridge rectifier. Boom, there goes your scope and your unit under test!
That’s a problem which is solved with an isolation transformer. However, don’t forget that this will make working on the device a bit more dangerous for you again, because of the connection to earth. And your RCD won’t save you if you do touch the hot pole this time.
My best advice is not to touch any of the poles of mains carrying wires, ever. Not even on the output side of an isolation transformer.

The Current Meter

Wiring up an isolation transformer is a non-event really. In our old house I had only wired an on/off switch and a fuse on the primary side. And the output was simply connected to a wall outlet. This time I wanted to add an analog current meter and a few banana output terminals, just to make the project a bit more interesting.

Circuit diagram Click the diagram to download it in PDF format

R2 is a 0.1 Ω resistor which is put in series with the output. The voltage drop across this resistor is proportional to the output current, drawn by the load. In my case the maximum output current of 2.5 A will create a voltage drop across the shunt resistor of 250 mV.
With these low resistor values it is important to sense this voltage as close to the resistor as possible. Otherwise the resistance of the extra wires will increase the total resistance considerably, which will increase the voltage drop which we want to measure. This is called a Kelvin connection.

Now this AC voltage drop needs to be rectified and amplified, before we can feed it to the analog meter movement. This can be done in different ways.
The first thing that would come to mind is to use a simple diode to rectify the voltage. This is not feasible because of the low voltage we want to measure and the relatively high voltage drop across the diode. Of course I could amplify the voltage first and then feed it to the diode, but that still would introduce some linearity problems at the bottom end of the scale.
A better solution would have been to use a precision rectifier circuit, in which case two diodes are used in the feedback loop of an Op-Amp. The advantage of this circuit is that the forward voltage drop of the diodes is compensated for by the feedback loop.
I have settled for an even more elegant solution. I’ve used a True RMS converter. They come in the form of a dedicated IC, an AD736 from Analog Devices, which needs only a few external components. The advantage of a True RMS converter is that it can accurately measure non sinusoidal signals, up to a point. The True RMS converter I have chosen can accurately measure signals with a crest factor of at least 3. The crest factor is the ratio between the peak value of the signal and the effective value of the signal. For instance the crest factor of a sine wave is 1.414. The crest factor of a square wave is 1, and for the output of a half bridge rectifier the crest factor is 2.

Anyway, the output of the True RMS converter is amplified and attenuated to create the appropriate full scale values for my analog meter movement.
I’ve used an old VU meter as analog meter. This meter has a resistance of 1100 Ω and requires some 1.6 mA to drive it to full scale. This means I need an output voltage of 1.76 V to drive the meter to full scale.
I wanted to have a bit of over range, so my scale runs from 0 to 3 A, which is 0.5 A more than the transformer can handle. This should be OK for short periods. Going any higher will blow the fuse. The scale between 2.5 A and 3 A is made red to indicate that the current has reached an unsafe value.
I don’t connect devices which draw 2.5 A very often. So it would be nice to have a lower scale as well. The easiest way for me was to divide the scale by 3.333, which results in a full scale of 0.9 A. Dividing the scale by 3.333 means that I have to amplify the signal by 3.333 to get a full scale deflection again. At this scale the analog graticule can be the same as for the 3 A range. A full scale of 1 A would have required a completely different graticule, which means a more cluttered scale and much more paint (Gimp) work for me.

I have added banana sockets on the front panel. The red and black ones are the 230 V output. I can connect a multimeter to those to measure the output voltage if I want to. I could also connect a load to them, instead of using the normal power outlet.
The shunt resistor is connected in between the black and blue ones. These can be used to measure the voltage drop more precisely with a digital multimeter. Or I could connect an oscilloscope across it to measure the exact current waveform.
The yellow/green socket is the earth connection. While the yellow socket is connected to earth via a 1 MΩ resistor. This connection can be used for ESD protection.

Powering The Current Meter

The current meter requires its own power supply. Any voltage pair between ±5 V to ± 16.5 V will do. Both voltages don’t even have to symmetrical. I’ve simply used a 10 V transformer which I had laying around, used two half bridge rectifiers and two tank capacitors to get ± 14V.
Voltage regulators are not necessary because the output of the True RMS converter and the gain of the Op-Ams are independant of the supply voltages.
The total circuit draws about 3 to 4 mA which, together with the ridiculously high tank capacitor values, results in a neglectable ripple voltage anyway, even with the use of only a half bridge rectifier.

Calibrating The Current Meter

Naturally all dimensions are tailored to my particular situation. My isolation transformer can deliver a maximum of 2.5 A, therefore the scale of the analog meter runs up to a maximum of 3A. If you want to build this circuit yourself with an isolation transformer with different specifications you may have to change some resistor values.

I’ve decided to set the high and low scale amplifiers to a gain which produces approximately a full scale value of 5 V on their outputs. Thus the high range will produce an output of 5V when the input voltage is 0.3 V, which is caused by a 3 A current flowing through a 0.1 Ω shunt resistor. This is a gain of 16.7. For the low range this means that 90 mV of input should create a 5V output, hence the gain would be 55.6 times.

Using Ohm's law it now should be easy to calculate the output resistors of both amplifiers, to get the maximum deflection of the meter movement correct. Two trimmer pots are used to calibrate the two ranges. Both ranges can be adjusted independently, so it doesn’t matter which one you calibrate first.

The trick is to measure the output current to an artificial load with an accurate enough Ampere meter. Suppose we adjust the upper range first. Connect some incandescent light bulbs in parallel to the output of the isolation transformer and monitor the output current with the reference meter. The trick is to load the circuit to almost the full load, which is just below the 2.5 A mark in my case. Getting it just right to 2.5 A can be tricky, but I have an easy trick in store to adjust the last part of the load to exactly 2.5 A. And once your reference meter shows 2.5 A it is only a matter of turning the trimmer pot to make the analog meter point at 2.5 A too.

Now comes the trick to accurately adjust the load to 2.5 A. Connect the input of a linear lab power supply to the output of the isolation transformer, in parallel to the already connected incandescent light bulbs. Then short the output of this power supply and crank up the current limiter knob to get a load from the isolation transformer of exactly the wanted value.

The same principle can be used for the 0.9 A scale, only with less incandescent light bulbs to start with.

Painting The Scale

It appears that meter scale programs are available on the internet. Perhaps some are even online tools to draw them. However I wanted total freedom on how to draw mine. One reason why I chose to do it myself is because I didn’t want to have my scale arc to have the same radius as that of the meter movement. I want my scale to have a radius of about 5 times the radius of the meter movement. This results a nicer, somewhat flattened scale arc.

Meter scale

I have used The Gimp to draw my scale. First I’ve started to draw the gradients at 1.5 ° intervals across a 45 ° scale coming from one single center point. This center point represents the center of the meter movement.
Then, on a new layer, I’ve drawn the arc of the scale, coming from a center point a lot lower than the previous center point. I can’t exactly remember how far down that new center point was. I simply experimented with it until I had it right.
On a next layer I’ve drawn some helper arcs from the same lower center point, which would be used to draw the different lengths of the graticule lines.
And finally I’ve used a copy of the first layer to erase all the parts of the graticule lines which I don’t want anymore.
The rest is as simple as adding some text (some of it under a slight angle to follow the arc of the scale), and colouring part of the scale red.

And then comes the challenge of printing the drawing exactly to scale on my inkjet printer. Once I’ve found the right scale I’ve printed the screen on a glossy piece of sticker paper, cut it to size and stuck it to the back of the original VU meter scale. And the result is like a bought one!

The Result

Front panel

I’m quite happy with the result, with its beautiful linear scale. Any imperfections which might exist are invisible by lack of resolution to which I can read the analog scale.
The meter movement I’ve used is a repurposed VU meter. And it shows. The movement is not damped in any way, which a real measurement unit would have been. This makes the movement very fast in responding to current changes. However it may also mean that it might fly off the scale every now and then due to inrush currents from loads. We’ll see how long it will last.