Suggestions Needed for Monitoring Line Voltage Glitches

Well, being lazy, I built a "power line to sound card" interface to analyse the waveform offline to try and find off-peak mains signalling (https://goughlui.com/2014/02/02/project-analyze-mains-power-flicker-issue-ripple-signalling/ ). It's basically just an isolated transformer and a voltage divider. I didn't develop much in the way of software applicable for your situation, but I suppose that's one approach you might think about by defining a "mask" and comparing samples for each cycle by looking at zero-crossings. It's a lot cheaper than a cycle-by-cycle capable power analyser or using an oscilloscope and trying to capture every cycle with it, but perhaps may need a little "calibration".

- Gough

Hi Gough,

Thank you for the link and the extremely interesting exploration of the noise you had on your power line. You always do such an awesome thorough job with every presentation.

John

Decided to write a quick follow-up - thanks for the kind words jw0752. Perhaps if you aren't bothered to write some software, you might build it anyway and just use a sound recording app to record the waveform over a long period. Then you could try various filters on a copy of the file to see where things are suddenly different - e.g. perhaps you can try a notch at 50/60Hz then normalize to 100% (or even above) and then see if there are any "spikes" that remain. Transients usually have high frequency components as they have a sudden onset - so they may leave a tell-tale spike after filtering. Then you note down the time, go back to the original file and study the waveform carefully.

Another way which I forgot to mention, assuming you have a decent oscilloscope that can ride through the power interruptions, is to use that with a high voltage differential probe (for safety) - something like these which I reviewed earlier: https://goughlui.com/2018/12/16/review-eevblog-hvp-70-70mhz-micsig-dp10013-100mhz-high-voltage-differential-probes/

The idea would be to get the screen to capture a few cycles, triggered across the zero crossing, under a "mask test" set-up where you may capture a good set of waves from the mains as a reference, "increase" the mask around it say by a few percent, and set the oscilloscope to beep and capture screenshots/waveform files to USB when it violates this mask. Of course, this is not a feature all oscilloscopes have.

If you have the scope but not the probe, assuming you get the polarity of live/neutral right, you could get away with the 10x probe but you would be violating the safety CAT ratings of the oscilloscope and potentially put things in danger. If you get it wrong, you'd be shorting live to earth which is a terrible thing to do in general. Instead, you could take the hybrid approach and just use a linear brick transformer like I did before and scope the output of that. It should be noted that the transformers themselves may have resonance/distortion due to the core materials and inductance properties, so I would be less inclined to trust the absolute reading/waveform shape after it goes through a magnetic transformer compared to a high-voltage differential probe.

What you might be suffering from is some loose connection or intermittent shorts somewhere which is causing some short periods of "notching" on the waveform, or a short dip which is not going to easily be caught by a DMM that might only sample at about 4Hz. In the very catastrophic case, it could even be a sign of arcing somewhere in the network which happens transiently (e.g. bats hanging across phases, trees clashing with wires in the wind occasionally, failed ceramic insulator on a pole, high-resistance cut-out contact, a minor partial discharge inside a transformer perhaps?). Other causes may be more mundane - a weak/failing fuse, breaker contact, perhaps even a surge protective device going a bit funny and "clamping" during higher-voltage peaks (rather than surges) as the MOV wears out.

- Gough

jw0752  Please forgive the slight side track.  I'd recommend that you also check your neutral to earth voltage. Infrequently, the neutral connection can become loose or corroded and what should be zero isn't.  It exhibits itself as flicker, dim and/or bright, in the early stages.  If the neutral floats... stuff burns up.  I know that you checked all the connections in the breaker box.  I hope that included the white and green - especially the main bonding jumper - the one connection between the neutral buss and the ground/earthing buss.  At work, we see it every so often.  Oh yeah... happened to me, also.

Great point kmikemoo! Definitely worth checking especially for split-phase 120/240V installations, perhaps depending on where the earth-neutral bond is, it could also be on the transformers on poles and have broken off there (I think I remember videos from Bobsdecline's YouTube channel mentioning this for Canada).

Things are different down-under where I am, where 3-phase distribution and 1-phase/3-phase supply is the norm. Our main earth-neutral bond is in the meterbox usually, losing it just means potential safety issues (e.g. tingles when touching electrical devices and taps) but rarely results in voltage issues for the device itself as houses usually get a neutral wire and don't depend on earth connection except for safety.

- Gough

John,

I know how frustrating it can be when you see glitches in the lights, but you can't determine the source or quantify the severity of the events.

Many years ago, almost in another life, I worked for a company that built very high power inverter systems.  One of our clients wanted to use our inverter in a grid power quality system.  The idea of the system was to stand in between the grid and a remote factory.  The system would monitor the grid voltages on a cycle-by-cycle basis, measuring the amplitude, frequency and phase of the grid.  From these readings we would generate a signal that was a mirror of the grid voltages, that we could use to create a envelope (very similar to the mask test that that Gough Lui describes, except active through the entire cycle) we could compare (instantaneously) against the incoming readings, giving us a go/nogo decision on the incoming signal.  If the grid dipped (brown out), we would fire up the inverter to support the grid voltages.  The cool part of this system is that the power needed to support the grid through a brown out came from a massive superconducting magnet.  We could stand in from the grid and support a load of 1.2MW for about a minute.   Btw, the client for this system was an outfit in Madison, WI.

This was task required quite a bit of computing power (and custom electronics), so it is not too likely that you would want to try this at home, although I guess most PC and such today could likely do as much as the floating point DSP (TI) that we were using back then.

Good luck with you search with glitches!

Gene

Thanks Gough for the follow up ideas and suggestions. I will be experimenting and following up.

John

Hi Mike,

this is a good insight and I will check it out. I do have one of my meters set to Min Max set between Neutral and one Phase just so I could see if I was getting glitches there and not across the two phases at the same time. So far it looks like a glitch on a phase to neutral happens at the same percentage drop as the drop across the 240 volts.

John

Hi Gene,

thanks for sharing this interesting story from your past. I had never heard of using a super conducting magnet as practical storage device. The Power company guy with the chart recorder was suppose to come today but he didn't make it. Hopefully tomorrow. In the mean time I am gathering as much data as I can on the problem.

John

Yep, something similar happened to me only last week. Air gap in old fuse (more likely) or possibly loose connection had caused flickering but could smell some ozone too (usually the sign). Failure triggered due to higher than normal current as had a couple of electric heaters on for a change (one was also my fan heater) and then the wife put the kettle on...