How stable is the mains voltage?

[soldar]

Well, from a test and measurement point of view, the AC mains voltage provided by a "standard" wall outlet is often (at least in my experience) not stable enough for some types of testing.

For example, if you want to run standby power tests as per EIC 62301, the voltage and frequency supplying the device under test must be within 1% of the nominal values and there is also a limit on the amount of harmonic distortion -- I think THD must be < 2% for the first 13 harmonics. 

It's not uncommon that running these kinds of measurements in the United States requires the use a "lab grade" AC power source in order to meet these requirements: you can't just plug the power analyzer into the wall.  The usual culprit is ... surprise! ... THD, probably as a result of all the non-linear, non-PFC'ed devices being plugged into the grid these days.

(Coincidentally, I just finished a video on this very topic -- standby power measurements -- which should go live on YouTube in the next couple of weeks)

I think pretty much in any country if you look at the mains on a 'scope you are going to see a very flattened top of the wave and not even close to a perfect sine.  Lots of distortion there.

Not to mention other types of distortion like spikes here and there.

[soldar]

Here in the UK, the frequency does vary a little, but has long term correction. Restored a couple of the old 1930's (Temco) synchronous motor clocks, one running in the lab for months, that is still within a few seconds of the radio clock next to it.  ~Ñinety years between them. Short term variations, but corrected in the long term

The old clocks with sync motors kept time much better than the newer digital ones which used the grid as reference which, in many cases could run fast due to glitches and spikes in the signal being counted.

[dietert1]

Should be easy to build an opamp bandpath filter that does something similar to the synchronous motor. There will be more parts yet no moving parts..

Regards, Dieter

[dietert1]

The device on my bench has a comparator looking at mains voltage via a two stage clipping plus low pass filter, so here is the requested measurement. This is with 50 Hz mains and of course this measurement may not be representative.
The scope gets triggered on the comparator output positive edge and shows the next negative edge (10 msec delay). One can see the jitter.
The blue curve is a trend plot that shows 1000 past delay measurements. The exact 10 msec would be a constant line in the middle. So the average frequency was a little higher than ideal.
The delay histogram has its peak shifted to lower delays, too. Jitter is about +/- 10 usec. There are about 10 or 20 outliers (not visible on screen).

Regards, Dieter

[tridac]

The accuracy of frequency of the mains is quite amazing, considering that the frequency depends on the rotational speed of thousands of tons of turbo alternators, all in sync. Variiations is load, input variations from renewables, the rotational inertia time constant of the system, probably measured in minutes, and the control system for that must be a pretty cool piece of design. The frequency does drop a little at peak load, but must be advanced a litlte later to ensure the second accuracy of clocks in the long term...

[coppice]

The accuracy of frequency of the mains is quite amazing, considering that the frequency depends on the rotational speed of thousands of tons of turbo alternators, all in sync. Variiations is load, input variations from renewables, the rotational inertia time constant of the system, probably measured in minutes, and the control system for that must be a pretty cool piece of design. The frequency does drop a little at peak load, but must be advanced a litlte later to ensure the second accuracy of clocks in the long term...

The frequency has to be kept accurate. If the generating sets start to advance or retard more than a very small amount from the network mean you can start to get instabilities in the whole network. Even when they let the frequency fall during periods of excessive load, the frequency has to ramp down in a slow and smooth manner.