Transcript
Introduction
Hello, and thank you for joining today’s webinar. My name is Wes Houck, and I’m an electrical engineer here at PMI. Today, we’ll be covering our white paper on detecting loose neutrals with Merlin.
And let me tell you up front why this one matters. A loose neutral is one of the most dangerous faults a single-phase service can develop. It’s also one of the least likely to announce itself. It doesn’t fail all at once. It degrades. The return path breaks or remakes itself over and over.
And meanwhile, the customer calls in and tells you that the lights flicker, and there’s an outlet in the back bedroom that quit working. No breaker trips. There’s no overcurrent to trip on. Nothing goes dark. And that’s exactly why these things survive for years, long enough to damage the customer’s equipment, and in the worst cases, long enough to start a fire.
We’ve been writing about this fault at PMI for over a decade. White Paper sixty-six was our first one, so I’m not here to tell you about a new fault. The question this paper asks is a different one. Who has time to go look?
Figure One: The Physics of a Loose Neutral
If you got the paper in front of you, pull it up because I’m going to walk it front to back, seven figures in order. If you don’t have it, that’s fine. I’ll tell you what you’re looking at as we go along.
So let’s start with figure one. This is on page one. Start with the physics because the physics is what makes this fault so quiet. In hundred and twenty, two forty volt split-phase service, the neutral has exactly one job: it carries the difference. When the load on the two legs are balanced, that neutral carries almost nothing at all.
Now let that connection go bad. A corroded lug up at the weatherhead, a set screw that’s worked itself loose over a few hundred hot/cold cycles. That return path picks up resistance. And here’s what happens. These two hundred and twenty-volt loads aren’t referenced to a solid center point anymore. They turn into a series circuit sitting across the full two hundred and forty volts, and the voltage divides by impedance instead of by design.
That center point becomes an artificial neutral, and it shifts every time somebody in the house turns something on. The arithmetic is simple, and it’s a little unsettling. Follow along figure one with me. One leg feeds the computer pulling six amps, so that’s twenty ohms. The other leg feeds a light pulling four amps, so that’s thirty ohms. With a good neutral, both of them see a hundred and twenty volts, and the neutral quietly carries back the two-amp difference, and everything’s fine.
Now open the neutral. The same two loads, except now they’re in series across two hundred and forty volts, so the current is two hundred and forty divided by fifty ohms. That’s four point eight amps through both of them. The computer drops to ninety-six volts, and the light gets driven at a hundred and forty-four volts. That’s a hundred and twenty percent of what it’s rated for.
Notice which one got hurt. The lightly loaded leg always loses. Whatever’s drawing the least current in the house is the equipment taking the overvoltage, which is generally why the damage claim comes from the room nobody was even using.
Figure Two: RMS Capture of a Real Event
Figure two, bottom of page one. This is an RMS capture of a single event on a real service, and I want you to find channel one and channel two in there. One of them climbs up towards a hundred and thirty volts. The other one falls towards a hundred and eleven volts, and they do it at the same instant. Mirror images right around the two hundred and forty volt band. Hold onto this figure. We’re coming back to this exact recording later, and it’s going to matter that it’s the same one.
The Diagnostic Rule: Mirrored Leg Movement
That mirrored movement is the fingerprint, and it is the diagnostic rule that holds up really well in the field. So let me give it to you plainly. If the two leg voltages move in opposite directions, one rising while the other falls, event for event, suspect the neutral. If both legs sag together or both swell together, your cause is upstream of the service: the transformer, the secondary, or the feeder.
There just aren’t many other faults that can push one leg above a hundred and twenty-six volts while dragging the other one below a hundred and fourteen volts in the same instance. And look at what that does to your compliance numbers. During an event, both legs are outside the ANSI C84.1 range A band. That’s a hundred and fourteen to a hundred and twenty-six volts at the same time in opposite directions. In between events, the same service averages a perfectly respectable hundred and twenty volts. So your steady-state metrics miss this thing completely. The averages are doing their job. The fault just doesn’t live in the averages.
Why This Is a Monitoring Problem
The other thing about this fault is the timing. It’s a mechanical failure, so it’s intermittent. It makes and breaks with temperature, with load, sometimes with nothing more than the wind. Early on, the events are short, and they’re sparse. Sub-cycle excursions that a monthly meter read is never going to catch, which makes this a monitoring problem, not an inspection problem.
So if this evidence only exists while the fault is acting and the fault acts on its own schedule, then you can’t schedule a truck roll against it. You have to be watching continuously. That’s the patient work our recorders do. The Seeker, the Guardian, the Bolt: plug-in recorders inside the house. You put one at the service or at the receptacle, and it streams RMS data on both legs up to Canvas, where you can look at weeks of service behavior in one strip chart.
Figure Three: A Week of Min and Max Data
Which brings me to figure three over on page two. And I want to be honest with what this figure’s asking of you. This is a week of min and max data from a different service from the one we’re looking at. Somewhere in there, every spike on one leg pairs up with a dip in the other. It’s a handful of narrow marks in a whole field of clean voltage.
Now we’ve documented this manual workflow thoroughly. White paper four oh two is the most recent one. White paper sixty-six goes all the way back. The strip chart view, the RMS capture, and the dedicated loose neutral report. They all screen for exactly this behavior, and they work.
But look at what the workflow assumes. It assumes an engineer already suspects its service, and that he’s got an afternoon to get refit. Because this isn’t really a power quality problem anymore, it’s an arithmetic problem about somebody’s workday. A utility fleet of continuously recording monitors produces far more strip charts than a PQ department can sit down and read. And a loose neutral’s early evidence is the worst possible case for annual triage. A few brief mirrored excursions scattered across weeks of perfectly clean data.
Merlin™ in Context
So Merlin, and I want to put Merlin in context because this isn’t a product launch, it’s the next step in a chain we’ve been building for a long time. ProVision that you all are probably pretty well familiar with, put deep waveform and strip chart analysis on the engineer’s desktop. PQ Canvass moved that analysis to the cloud so the entire fleet of recorders streams into one place, and you can review it from anywhere. Each of those steps widen the view. Merlin is the step that reads every recording every time, eyes on every piece of data.
Figure Four: The Merlin Dashboard
Figure four on page three, it’s a capture of the dashboard of the same recording. Same two days, same service, same house. Nothing here is staged. And let’s see what Merlin handed back. The headline is a diagnosis itself in plain language. Loose neutral with severe flicker, sags.
Loose neutral came back as a top issue at the severity eight out of ten. Described an intermittent mirrored leg and undervoltage. And the numbers underneath it are specific. Across thirty-one captures, it counted eleven sags, thirteen swells, and seven rapid voltage changes. All of them with the same mirrored leg signature. One leg rising to a hundred and forty-five, a hundred and fifty volts, while the other one fell to ninety-four and a hundred volts.
In the RMS record, it found seventy-nine single leg sags inside one sixty-six minute window. And flicker clearly failing IEEE fourteen fifty-three on both legs. The PST plateaus reaching seven-point-seven, but the limit is one point oh.
Compliance: Voltage Regulation vs. Flicker
Look at the compliance panel on the same dashboard of the ANSI C84.1. Voltage regulation reads fully compliant. Ten-minute RMS on both legs stayed inside range A for a hundred percent of the interval, and right underneath that, flicker sits at ten out of ten. Read those two panels together by voltage regulation. This service passes perfectly, but flicker, it’s a ten out of ten. That’s the whole story of this fault. It looks healthy on averages, and then it misbehaves in moments.
Auditability and Deterministic Calculations
One more thing about this screen because I know who I’m talking to. Every number in the narrative comes from the same deterministic standards-based calculations you run by hand. Merlin generates the language not by math, and every finding on the screen is one click from the capture it came from. It’s all about audits.
Figure Five: Individual Capture Detail
Figure five, right next to it on page three, pulls one of those thirty-one captures out on its own, number twenty-two. One leg is sitting near ninety-five volts while the other one holds a hundred and forty-nine. Merlin attributes it to reestablishment of a high impedance neutral path. Classifies it per IEEE eleven fifty-nine, and notes the ITIC and SEMI F forty-nine ride-through implications. The point here is just that the classification work is already done by the time you open the card.
Figure Six: Merlin Chat
Figure six on page four. Because that report is only the front door, Merlin Chat lets you question the findings in plain language, and I put this exchange on paper. I asked it the question everyone else gets to ask. “Is this on our side of the meter or the customer’s?” And it answered like an engineer. It put the defect in the shared neutral path that’s common to the whole service, and it weighted the utility side slightly higher than the customer’s side.
But then it named the limit of its own data. I’m going to read that part word for word. “We can say the problem is on the common service secondary neutral, not just the branch neutral. We cannot say exactly whether that weak point is before or after the metering.” The one recording at the one service can’t localize the weak point. For that, you need a second recorder, and Merlin said so instead of guessing. An analyst that tells you what the data can’t support is worth more than the one that always has the answer.
Field Inspection Priority List
Then I asked it what to inspect first, and it came back with a priority ordered field list. Starts at the transformer’s secondary neutral hardware and works toward the customer panel with reasoning attached to each step.
Merlin’s Document Writer
It’s also got the least engineering in it of anything I’ve shown you, and it might save you the most time, because a finding isn’t finished until somebody who isn’t an engineer can act on it, and that’s usually where the hours go. Merlin’s document writer has a template for every audience that finding has to reach. A plain language customer letter, talking points for customer service rep, an internal escalation email, a regulatory filing, a damage claim response, and engineering deep dive.
I gave it that deep dive template one sentence, and produced a seventeen-page engineering report on this recording with every claim linked back to that numbered data snapshot. So detect, explain, discuss, document. That whole loose neutral workflow now lives inside PQ Canvass, and the engineer’s time goes to confirming dispatching instead of searching.
PQ Canvass and Merlin Live Demo
And we can dive into some PQ Canvass and Merlin examples. So here I have a recording that I uploaded to PQ Canvass. This one I ran a Merlin analysis on it, and it came back listing severe flicker with neutral-driven RMS swings, single leg swells from loose neutrals.
We come over here to click on Flicker. It rates that it’s a ten out of ten, and it will give you the full detailed analysis of what it found throughout this recording on flicker. There’s a chat assistant if you need to ask it any questions.
But I would like to open up the document writer here. This is one that I wrote last week. And then here you can actually choose the template. So after you’ve gone through and seen the Merlin diagnostics of the loose neutral recording, you can come in and choose. You can have an internal escalation email, executive summary. You can write a customer letter in plain language, something that anybody can understand. A letter for any audience, a report for any audience.
So we come back here. Here’s one that I actually had to generate. This is a report that I would write for another engineer to read, so more technical. We’ll start out with the executive summary, go into the ANSI C84.1 and touch on the IEEE standards throughout this fourteen fifty-three. And it goes through everything that it finds in the recording. What I told it was just tell me everything you could find in loose neutrals about this recording. So that’s what it covers in here in great detail.
Closing Summary
To finish it, “A loose neutral is a small connection failure with house-sized consequences. Its mechanism hides it from breakers and multimeter reads. Because it comes and goes, only continuous monitoring will catch it early. ProVision, PQ Canvass give engineers graphs and fleet. Merlin gives them an analysis. A fault that once depended on a lucky truck roll and patient manual review can now be caught and written up from a desk. Utility engineers who book the pipeline work can resolve flicker complaints at their true cause, prevent customer equipment damages, and in the cases that matter most, get the failing neutral repaired before it becomes a fire.”
Q&A: Single 120 Volt Loads with Poor Neutrals
The question is: How might single one hundred and twenty volt loads with poor neutrals show up on the graph? If you want to go back to what we’re showing up here, this is a two hundred and forty volt split-phase service. So you have two hundred and twenty volt lines a hundred and eighty degrees out of phase referenced to neutral. In this situation, there might be rusted connection, loose lug at the service, and that might cause a loose neutral or a broken neutral depending on the severity of it.
In this example, we had a loose neutral showing up intermittently. Instead of a computer and the light being in parallel across that service, they turn into in series because of the neutral going out. So with a hundred and twenty volt load, you’re gonna have much higher voltage across the light than you are across the computer, which you might see the computer drop out at that instance. But for the light, you actually might see the light bulb blow up.
So on the graph coming down here, you can see that when the loose neutral occurs, you’ll see one side of the split-phase service go up. So this is going all the way up from a steady state hundred and twenty on either side of the event, all the way up to about a hundred and twenty-six volts here at the peak. And below you can see the exact opposite happened on the other split phase. So that drops down to a hundred and fourteen from the nice steady hundred and twenty volts.
Thank you for attending. You can reach us at eight hundred to nine six forty-one twenty or emailing support@powermonitors.com if you have any further questions. Thank you for attending today, and have a good day.