How to meg a motor: what to disconnect first, and when the reading actually fails it

The megger question I get most is not how to use the thing. It's "the motor megged fine, so why does the drive keep tripping?" Usually the motor never got megged on its own. Somebody lifted the cable off the drive output, put one probe on a phase and the other on ground, saw a big number and wrote "motor OK." What they really tested was the cable, the motor, the joins in the terminal box and whatever moisture lives in that box, all at once.

So here's how to meg a motor and get a number that means something: what comes off first, which voltage, how long to hold it, and which readings fail it.

Decide what you're testing before you clip on

A megger reads everything on the far side of its probes. Every foot of cable and every device still connected adds its own leakage path, and the more you include, the lower the number gets. A wet cable can make a good motor look bad. And with a long cable in the circuit you can’t trend the motor’s own number, because the cable’s leakage is mixed into every reading.

That doesn't mean you strip it down every time. If the whole circuit reads high, each piece of it is at least that good and you're done. Split it only when the combined reading is low, or when you need a number for the motor alone. Same logic as any split-the-circuit troubleshooting, just with 500 volts on it.

Three ways to clip on, and only the bottom one is how to meg a motor if you want the motor's own number.

Which setup is right depends on why you're testing. That choice comes before anything else in how to meg a motor:

What you're after	Where to clip on	What comes off first
Quick ground check before a restart	Load side of the starter or disconnect, one phase to ground (the windings are tied together, so one phase covers all three)	Drive or soft starter, if there is one
Motor tripped a drive, or a drive just failed	Motor terminal box	Cable at both ends, joins in the box
Maintenance trending	Motor terminal box, identical setup every time	Cable, plus the same joins every test
New or rewound motor before install	Motor leads	Everything: test the bare motor

 

After a drive failure I always go straight to the motor box. A drive that let go may have had help from the motor, and I want that answer from the motor alone, not from motor plus 150 feet of cable.

Get the drive and anything electronic out of the circuit

First rule of how to meg a motor with electronics in front of it: drives and soft starters come off before the megger goes on. No exceptions for "it's only 500 volts." Leave a drive landed and you get two problems. The reading now includes the drive's own output section, so it means nothing. And you've put a DC test voltage across electronics that were never designed to take it. I've watched good motors get condemned because the "ground" the megger found was really the drive.

Lift the motor leads off the drive output terminals (U/T1, V/T2, W/T3 on a PowerFlex) and keep the ends apart. The same goes for anything else hanging on the motor leads, like power factor capacitors or surge capacitors and arresters at the motor. Space heater and thermistor leads in the box stay out of the winding test. I tie them to the frame while I test the winding. Then I join the leads of each circuit together and test that circuit to the frame at 500 V, never across the sensor itself. One megohm or better is fine for them.

Then lock it out and prove it dead at the motor leads. Plenty of testers with a built-in voltmeter refuse to start when they see more than roughly 25 to 30 V on the leads. Nice backup. It doesn't replace the lockout.

Take the joins apart in the terminal box

This is the step that separates knowing how to meg a motor from just owning a megger. With the joins made up, all three phases are one piece of copper. Put the megger between two leads and you'll read next to nothing, which tells you nothing about insulation. That's where the whole "can you meg phase to phase" argument comes from. Between phases only means something winding to winding, with the joins apart. On a connected motor the tool for lead to lead is a low-resistance meter, and what you're checking there is balance.

Before anything comes apart, take a phone photo of the box and the nameplate connection diagram. Then tag every lead, including the ones whose markers fell off years ago. Nameplates get painted over, and putting a 9-lead motor back together from memory is how you end up with a motor that runs hot and loud, if it runs at all.

How far you can split the winding depends on the lead count, and that decides how many tests the job really is:

Motor leads	Take apart	Groups you end up with	To frame	Winding to winding	Total
3	Nothing, joins are internal	1 (whole winding)	1	0	1
6 (wye-delta start)	All six leads free	3 windings	3	3	6
9, delta	4-7, 5-8, 6-9	3 (1-4-9, 2-5-7, 3-6-8)	3, from 1, 2, 3	3	6
9, wye	4-7, 5-8, 6-9	4 (1-4, 2-5, 3-6, and the 7-8-9 star)	4, from 1, 2, 3, 7	6	10
12	Every join, including 10-11-12	6 single coils	6	15	21

 

On a 9-lead wye, leads 7, 8 and 9 share the internal star point, so testing from lead 7 covers all three of those coils at once. The delta version works differently: each of leads 1, 2 and 3 carries two coils joined inside the motor. You can't split those pairs, and no test from the box will tell you which coil of the pair is the weak one.

How to meg a motor with nine leads: separate 4-7, 5-8 and 6-9 and it splits into three groups (delta) or four (wye). The coil pairs inside a delta can't be split from the box.

Three-lead motor, common on smaller single-voltage ones: one test to frame and the megger part is over. Check phase-to-phase resistance with a low-ohm meter instead. On a random-wound motor the three readings should land within 2% of their average. A regular two-wire meter can't resolve that on a winding measured in tenths of an ohm, its own leads are in the same range.

Pick the test voltage off the nameplate

Voltage is the easy decision. The DC test voltage follows the winding's rated line-to-line voltage:

Winding rated voltage	DC test voltage
Under 1,000 V	500 V
1,000 to 2,500 V	500 to 1,000 V
2,501 to 5,000 V	1,000 to 2,500 V
5,001 to 12,000 V	2,500 to 5,000 V
Over 12,000 V	5,000 to 10,000 V

 

On a 480 V motor, 500 V is the standard test and 1,000 V is common too. Either one finds moisture and dirt, because those show up at voltages well below what the motor sees in service.

A motor that keeps tripping a drive is the exception. Insulation that's clean and dry but aged, or nicked during a rewind, can pass at low voltage and only start leaking under higher stress. A PWM output puts a lot more stress on the winding than a sine wave ever did (the VFD cable piece covers why). For a motor on a PowerFlex that's throwing F013, the number to work to is 1,000 V minimum, 1,500 V on a 460 V system, 2,000 V on 575 V, with 10 MΩ per phase as the floor. The full isolation routine for that fault is in VFD ground fault on a PowerFlex 525.

Two cautions. Before you put more than 1,000 V on an old or rewound motor, check what its maker allows. A test is supposed to find a weak spot, not make one. And try a step test before you condemn anything: 500 V, discharge, then 1,000 V. A healthy winding reads about the same at both. A noticeable drop at the higher voltage is a weakness the low test couldn't see.

How to meg a motor, step by step

1.    Lock out, tag out, and prove zero at the motor leads.

2.    Disconnect the drive or soft starter, plus anything else on the leads from the list above.

3.    Open the terminal box, photograph it, mark the leads, take the joins apart.

4.    Wipe the terminal board and the lead ends. A damp, dirty board leaks on its own and drags the reading down.

5.    Check the tester: probes shorted reads zero, probes open reads its maximum.

6.    Ground lead on bare frame metal (not paint), line lead on the first winding group.

7.    Hold the test for one minute. Write down the 30-second and the 1-minute readings.

8.    Discharge that group to the frame before you touch it, for at least four to five times as long as the test ran. Most testers do this themselves if you let the test finish instead of yanking the lead.

9.    Repeat for every group to frame, then every pair of groups.

10. Record winding temperature, ambient temperature and humidity.

11. Reconnect per the nameplate, check the hardware is tight, close the box.

Step 8 is the one people skip. Don't. A motor winding holds more charge than you’d think, and it can bite.

What reading actually fails a motor

You can do every step of how to meg a motor right and still misread the result. The minimums below are for the one-minute reading on the whole winding, corrected to 40 °C:

Winding	Minimum at 40 °C
Random-wound stator (almost every low-voltage motor), and form-wound coils rated under 1 kV	5 MΩ
Form-wound AC windings built after about 1970	100 MΩ
Windings built before about 1970, and field windings	kV + 1 MΩ (rated kV, so 5.16 MΩ for 4,160 V)

 

So a 480 V random-wound motor fails below 5 MΩ at 40 °C. That's the floor. A healthy dry motor reads far above it, often into the thousands of megohms. If a new one shows single digits, something's wrong even though it technically passes.

Temperature is where people fool themselves. Insulation resistance drops as the winding warms, and the rule of thumb is that it halves for every 10 °C rise. To compare a reading with the 40 °C minimum, or with last year's number, correct it first:

Winding temperature at test	Multiply the reading by (to get 40 °C)
10 °C	0.125
20 °C	0.25
30 °C	0.5
40 °C	1
50 °C	2
60 °C	4

 

Measure the winding, not the room. A motor that just shut down is a lot warmer than the air around it, and its reading will look worse than it is. Some motor makers publish their own correction curve with a gentler slope; if the manual for your motor has one, use theirs.

One reading by itself says less than people want it to. A motor that went from 2,000 MΩ to 200 MΩ in a year, same temperature, same test voltage, still passes. It's also telling you something, and the trend is what you act on.

Why the reading keeps climbing, and what PI tells you

If you’ve never watched a big one, this part will throw you. Watch the display and the number climbs: 50 MΩ in the first seconds, 250 by twenty seconds, still creeping at a minute. That's normal. The winding charges like a capacitor first, then the insulation itself slowly soaks up charge, and good dry insulation keeps rising for 5 to 10 minutes. Wet or dirty insulation doesn't. Leakage current swamps the effect and the reading sits flat.

That rise is what the two ratios measure. Neither cares much about temperature, so they work on a motor you've never tested before:

Ratio	How to get it	Pass
Polarization index (PI)	10-minute reading divided by 1-minute reading	2.0 or higher, Class B insulation and up
Dielectric absorption ratio (DAR)	60-second reading divided by 30-second reading	1.5 or higher on random-wound motors

 

On a random-wound motor the absorption is over in seconds, so the PI often lands near 1 and means very little. Use the 60/30 ratio there. If the 1-minute reading is already above 5,000 MΩ, skip the PI, the ratio stops meaning anything that high. A PI below 1 is different: the reading fell while you held the test, and I don't put that motor back in service.

Timing matters here: dry insulation keeps soaking up charge for minutes, wet insulation leaks it away. The ratio between two readings is the test.

A low reading: wet, dirty or damaged?

The megger can't tell you which. Moisture, a dirty winding and real damage all look the same on the display. The pattern across your readings is what separates them:

What you see	Most likely	Next move
Low on every group, PI flat, motor sat idle or lives in a washdown area	Moisture	Dry it out, retest
Low with the joins made up, normal once they're apart	The joins themselves: wet tape, a chafed lead, a tracked terminal board	Fix the box, not the motor
One group low, the rest fine	Local damage in that group or its leads	Check the leads where they leave the frame, then send it out
Good at 500 V, drops at 1,000 V or higher	Aged or mechanically damaged insulation	Motor shop. Don't trust it on a drive
Passes cold, trips the drive once warm	A fault that only shows hot	Meg it again within minutes of shutdown and compare
Motor sat in seawater	Salt in the winding	Plan on a rewind

 

That second row catches a lot of people. Low reading with everything joined, hundreds of megohms per group once it's apart, and somebody orders a motor. The windings were fine. The leakage was in the joins.

For drying, a common bake-out is 90 °C for 12 to 16 hours, then 105 °C for 6 to 8 hours, but follow the motor maker if they publish a schedule. Expect the reading to drop at first as the winding heats, then climb as the water leaves. I track it with the PI while it dries, which saves the temperature math.

Write it down the same way every time

Log the date, the test voltage, where you clipped on (starter or motor box, which joins were apart), the 30-second and 1-minute readings (and the 10-minute one if you ran a PI), plus the winding temperature, room temperature and humidity. Skip one of those and next year’s reading is harder to compare with this one, which was the whole point of learning how to meg a motor in the first place.

FAQ

How to meg a motor that runs on a VFD?

Never through the drive. Get its output leads off and apart, then test the motor at its terminal box with the cable free, and the cable by itself. 500 V covers a routine check. If the motor is tripping a PowerFlex on F013, test a 460 V motor at 1,500 V (never below 1,000 V) and fail anything under 10 MΩ per phase.

What is a good megger reading for a 480 V motor?

The minimum for a random-wound motor is 5 MΩ at 40 °C, one-minute reading. A healthy one reads far higher, often thousands of megohms. Compare it with that motor's own history at the same temperature, that tells you more than any single number.

Can you meg a motor phase to phase?

Yes, but only between separated windings, so the terminal box connections come apart first. Otherwise the phases are tied together and the megger reads close to zero however good the insulation is. On a connected motor, compare lead-to-lead resistance on a low-ohm meter instead; on a random-wound motor the three should agree within about 2% (about 1% on form-wound).

How to meg a motor from the starter?

Open the disconnect, take any drive or soft starter out of the circuit, and test one load-side phase to ground. That covers the cable and all three windings at once. A high reading means you're done. A low one means you split it: motor box first, then the cable.

How long should you meg a motor?

One minute for the standard reading. Note the 30-second reading too if you want the DAR, and hold it 10 minutes for a PI. Then discharge to the frame for four to five times the test time.

When the motor passes and the drive is the problem

If the motor reads clean at the box at a proper voltage (and hot, if it only trips hot) and the cable does too, the winding is probably not your problem, and the drive or its setup is next. A clean motor and cable is a useful answer too: it tells you where not to spend money. When it does come down to the drive, we keep PowerFlex 525s on the shelf, like the 25B-D013N104 (7.5 HP) and 25B-D017N104 (10 HP), the PowerFlex 753 20F11ND065AA0NNNNN for 50 HP motors, and soft starters including the SMC-3 150-C37NBD and SMC Flex 150-F251NBR. And if a drive or soft starter got megged while it was still landed, check it before you trust it.