240V mains measuring and sampling.

1. I need to "see" the 240VAC mains voltage on the oscilloscope screen.
Is there a way to isolate the oscilloscope so that all the combinations of L+N, L+Gnd, N+Gnd be safely seen and measured?

2. The 240 mains is contaminated with power harmonics, RF, noise, brown-outs, surges, spikes etc.
I need to record via the PC sound card a scaled down copy of the 240V, eg. 1/100th of the 240, a 2.4 V.
How to isolate the desktop PC from the mains so that the 2.4V can be safely recorded?

I guess that a 240/240 isolation transformer is needed, but is it enough? am I missing something?


2012-08-06 12:26 pm
The probe's ground is connected to earth (verify this with a multimeter!). If the scope and the probe can handle it (most likely with the x10 divider enabled) you can then measure L->Gnd and N->Gnd. If the scope has two channels and can subtract them you can then get L->N as well.

If you want to measure it on the PC get a step down transformer, like even a 5V AC wallwart and a couple of resistor, sure there might be some you might some distortion on the harmonic content but the other events will be visible (heres someone that did this to measure the frequency to check how good it was for timekeeping Accuracy and stability of the 50 Hz mains frequency )
referring back to the accuracy link in post2.
The graph shows the "time clock" error.
It seems that the slope of that "error" plot gives an indication of the deviation from correct frequency.
If the slope is zero then the clock frequency is not changing.
If the slope is slightly up or slightly down then the clock frequency is higher of lower than the target frequency. I am assuming the target frequency is 50Hz and that all the clocks were correct when the test was started at error = 0.

@ the period 17.5days to 19.5days the graph is steeply upwards.
The gain is ~30seconds over that 2days. But some parts of that 2day period are obviously very much steeper than the 2day average.
If I take 15seconds per day from the average and make a further assumption that this is actually a "clock correction" being applied, then the average correction can be assumed to be no greater than the 17seconds/day (10mHz error from target frequency).

It is clear that the hourly fluctuation is at least double the 10mHz and more likely from the shape of the graph to be at least three times this and possibly around five times the 10mHz daily correction.

That would amount to 50Hz +-50mHz (+-0.1%) as maximum fluctuation that I can resolve from the data in the way it is displayed.

I would expect the minute by minute frequency fluctuation to be even higher than the hourly average.
Could the minute by minute variation be as much as +-1%?
That would be +-500mHz on a 50Hz target.

I wish I had seen that plot when this was being discussed some years ago.
I was being accused of all sorts of misinterpretations of the data at that time.
Last edited:
Minute by minute fluctuations are not much larger than long term changes. 100mHz or perhaps 200mHz at the extreme end of 'normal' conditions. Various things start taking automatic actions to correct frequency for excursions much beyond this.

The generators have quite high rotational inertia so they can't change speed quickly. The biggest changes in frequency occur in two situations:
1. unplanned generator trip (or, rather rare, line trip)
2. TV pickup - everyone puts the kettle on or visits the loo at the end of a very popular programme.
The producers cannot "turn on" generators just at the instant the load increases.
The Inertia of all the generating equipment is what initially meets a demand change.
For an increase in demand the first thing that happens is that the voltage falls.
The second thing is that the generators slow down. It's that slowing down that releases the energy to try to maintain voltage.
This is a second to second as well as a minute to minute variation.
That variation in frequency, of the whole generating grid, is minimised by the inertia of the grid. A bit like Gootee's capacitor analysis. Increase the demand and the capacitor discharges to a lower voltage. It's exactly the same for the generating grid. Increase the demand and the generators slow down. The reduced rotational speed releases energy to meet the changed demand.
As soon as the supply generation capacity matches the new demand the slowing down stops. When the supply exceeds the demand the grid speeds up.

It's the slowing down and speeding up that changes the supply capacity to meet the changing demand.


2012-08-06 12:26 pm
Or everybody goes to fetch a beer during the commerical break and causing all the fridge compressors to turn on at the same time :D

Mismatch of active power between loads and generation causes frequency shifts, apart from the high inertia that was mentioned, its all automated these days so it catches up quickly. They're also pretty good at forecasting usage ELES - Load and Generation
Mismatches of reactive power cause voltage variations (transmission lines and transformers have a very inductive character).

I've got no idea if harmonic content changes throughout the day. I've tested it a couple of times and it was always awful, more like a trapezoid than a sinewave.


2006-09-08 2:04 pm
2. The 240 mains is contaminated with power harmonics, RF, noise, brown-outs, surges, spikes etc.
I need to record via the PC sound card a scaled down copy of the 240V, eg. 1/100th of the 240, a 2.4 V.
How to isolate the desktop PC from the mains so that the 2.4V can be safely recorded?

I guess that a 240/240 isolation transformer is needed, but is it enough? am I missing something?
Getting everything at the same time will be difficult: ordinary 50Hz transformers are pretty unpredictable above 10KHz or so, and their common mode behavior will be even more so.

If you want to make seroius harmonics measurements, ordinary transformers will also be a problem, particularly the small ones.
I would recommend you get two identical transformers, not too small like 10VA, and you wire the primaries in series the secondaries in parallel.
That way, you will get rid of some of the distortions.
Another possibility is to use a transformer for industrial applications with a 400V primary. If it is used on 230V it will remain far from the saturation.
Toroidals have a high capacitance, and will couple transients in an unpredictable way, especially the common mode ones.
Tektronix publish a nice short guide - 'Fundamentals of Floating Measurements and Isolated Input Oscilloscopes'. (copy/paste that into google...)

NB Tek DO NOT recommend using a mains isolation transformer to 'float' a scope, ever.

Meanwhile that's an elegant technique, Elvee. I've only used small toroids but found with the sig gen that with a nominal loading - say 1K across the secondary - regular smallish mains step-down toroids can be quite good enough to present a flat passband beyond at least ~30Khz to a soundcard. More than enough for what I was interested in, and cheap to try.

Edit to add - last toroids I measured had under 500pF stray C pri:sec (650VA 35-0-35VAC in my power amps, actually). That's the real reason for a nominal secondary load on top of any two-resistor potential divider to set your signal input level.
Thank you all for your help and prompt replies.

@sim0n #2:
Yes, probe's ground is connected to chassis GND.
I think I can't use the scope to measure the mains voltage directly because the RCD will trip due to the internal connection chassis GND to mains earth. TT or TN system, must look at it.
My intention was to use a 240/240 isolation transformer to power the oscilloscope so that it 'floats', thus permitting me to connect scope's chassis GND to Neutral and the probe to Live. Of course EXTREME care is mandatory when dealing with mains voltages.

The same isolation transformer could be used to power the desktop PC. Then a 100:1 resistive attenuator will bring the 240VAC within soundcard's range. I feel that a laptop running off its batteries could also be used and no transformer is required.

sim0n and DF96: I think the transformer will filter out some of the components that corrupt the mains power. My intention is to expose and record the contamination of mains power.

@DF96 #3: very clever the idea of a resistor to load the secondary in order to avoid saturation. I had forgotten that, thanks.

@AndrewT #4: Yes, to investigate the components of mains power.

@Elvee #9: indeed, transformers can do a lot of things. I had read somewhere that transformers can be used as power conditioners, up to a point ofc.

EDIT_1: @ martin clark #11, the .pdf is invaluable, thank you.
Last edited:


2012-08-06 12:26 pm
The input impedance of scopes (typically 1Megaoogm) isn't high enough to trip the RCD (typically 30mA @ 240V -> 8000ohms trips it)

Also, if you're going to be monitoring this all day do you need the exact trace of each individual disturbance or are you just interested in the frequency of how often they occur (i.e. every event is bad anyway)
Last edited:

Right, the input impedance of the scope is >1MOhm. It's doubtful whether the oscilloscope allows the user to disconnect the mains earth from the probe GND. I will look at it anyway. Perhaps some other scope has a switch that connects/disconnects the mains earth to chassis.

The soundcard will do the monitoring and the recording, I will use the oscilloscope occasionally to take pictures of the disturbances. (if I am around and one occurs)


2006-09-08 2:04 pm
You could use any regular, say, 220:12V transformer, resistively load its secondary to put magnetizing current and low level hysteresis out of the way (or to be more precise, to make conditions more realistic)
Resistive loading won't change anything to magnetizing current or hysteresis.
It could slightly alter the magnetizing current if it is unduly heavy, but anyway the magnetizing current per se is not a problem, the saturation is.
On the other hand, loading will significantly degrade the reproduction of harmonics. It will also slightly magnify the effect of hysteresis.