VFD ground fault on a PowerFlex 525: what F013 is actually measuring

The worst version of this call always sounds the same. Third drive in four months, same slot on the same machine, F013 every time. Motor meggers clean. Cable meggers clean. The replacement runs fine for a few weeks and then dies exactly the same way. I have watched crews go through five and six units like that before anyone stopped to ask what the drive is actually measuring when it calls a VFD ground fault.

It is rarely what people assume. Here is what F013 means on a PowerFlex 525, the order to isolate it in, and the causes of a VFD ground fault still standing after the motor and the cable have both tested good.

What a VFD ground fault actually measures

Rockwell keeps the description just as tight: a current path to earth ground has been detected at one or more of the drive output terminals.

That is the entire test. So a VFD ground fault reads better as "these three phases stopped carrying the same current" than as "something is shorted to ground." A short will certainly do it. Plenty of things that are not a short will do it too.

Publication 520-DU001 keeps the description just as tight: a current path to earth ground has been detected at one or more of the drive output terminals. F013 is Type 1, so it resets, and it carries a footnote that catches people out. Auto-restart attempts F013 exactly once and ignores whatever sits in A541 [Auto Rstrt Tries]. Set that to five and you still get one attempt.

Code	Name	Type	What the drive is telling you
F013	Ground Fault	1	A current path to ground at one or more output terminals. Resets. Auto-restart tries once and ignores A541.
F038 / F039 / F040	Phase U / V / W to Gnd	2	A phase to ground fault located between the drive and the motor, on that specific phase. Will not reset.
F041 / F042 / F043	Phase UV / UW / VW Short	2	Excessive current between two output terminals. Phase to phase, not to ground.
F012	HW OverCurrent	2	Output current past the 200% hardware limit. Often shows up alongside F013 from the same root cause.
F021	Output Ph Loss	1	Output phase loss, and only if you turned it on with A557 [Output Phas Loss En].
F059	Safety Open	1	Both safety inputs not enabled. Check the jumper across terminals S1, S2 and S+. Nothing to do with ground.

Codes that get mistaken for a VFD ground fault. A Type 2 fault will not reset, which by itself says the drive believes the problem is hardware.

How to isolate a VFD ground fault in the right order

There are five steps to this. The value in them is not the steps, it is that each one cuts the problem in half.

Split the motor leads at the motor junction box, insulate them, run the drive. Still faulting means the motor and its junction box connections are cleared. A drive that suddenly behaves has just told you where the ground is.

Then the drive end. Take everything off U/T1, V/T2 and W/T3 and run again. A bare output that still throws F013 puts it on the drive. If it runs clean, the cable between them is your answer.

I go in that order every single time, and I still meet people who skip straight to a new drive because the panel is a shorter walk than the motor.

Isolating a VFD ground fault, plus the three settings worth checking before you condemn a drive.

Meggering at the wrong voltage is why the motor tested fine

1000V is the floor Rockwell will accept, not the recommendation. Rockwell is blunt about staying there: if 1500V or 2000V is not used, the megger test may not reveal that the motor is bad. That sentence accounts for an awful lot of clean readings on motors that were not clean.

Reflected wave on a PWM output stresses the motor terminals well past anything a 1000V test reaches, so a winding can pass at 1000V and still fail in service. A megger that tops out at 1000V has not ruled the motor out. It has ruled out the obvious.

What you are testing	Test voltage	Minimum acceptable	Source
Any motor, Rockwell's absolute floor	1000V	10 MΩ per phase to ground	Rockwell
Motor on a 460V system	1500V	10 MΩ per phase to ground	Rockwell
Motor on a 575V system	2000V	10 MΩ per phase to ground	Rockwell
Machine windings rated under 1 kV, general practice	500V	5 MΩ	IEEE 43

Megger settings for a VFD ground fault. Rockwell's 10 MΩ threshold is stricter than the general IEEE 43 floor, and the test voltage matters more than the number you write down.

Five causes of a VFD ground fault when the motor and cable are clean

Now the part that costs people money: a VFD ground fault that survives all of that. Output terminals bare, and the drive still faults.

Wrong motor data is the one I run into most often, and it's usually the first thing worth checking. The 525 builds an internal model of the motor from P031 through P037. Feed it the wrong nameplate values and the currents it commands stop matching what the motor actually draws. Say a 7.5 kW drive goes onto a 5.5 kW motor and the power parameter gets left at the drive rating. Nothing about that looks alarming on a panel schedule, and the drive is now regulating to a motor that does not exist.

Tuning belongs in the same bucket. Improper tuning is a known cause, and V5.xx firmware on the 525 addresses it.

Then Vector mode. A 525 in Vector Control can trip F13 Ground Fault or F12 HW OverCurrent because the speed loop is too aggressive. The fix is cutting A511 [Freq 1 BW] to 3 Hz, with A513 [Freq 2 BW] available too. Check those parameter numbers against your own drive before you change them. The blunter fix, when the load does not truly need vector torque, is P039 [Torque Perf Mode] back to V/Hz and a real autotune with P040.

Lead length is third. A motor run too long for the output with no reactor is a common cause, and the fix is an output reactor. Long cable, high carrier frequency, charging current going somewhere it should not, and the phases stop matching.

Fourth is moisture, which is boring and real. Rockwell names the motor conduit box and Quick Connect cables specifically. Wash-down areas, outdoor gearboxes, any conduit that breathes.

Fifth is the MOV jumper, pure installation. The rule is plain: the MOV to ground jumper must be removed if the drive is installed on an ungrounded (IT mains) or resistive grounded distribution system. Leave it in on a high-resistance ground and you have left a deliberate path to ground across the drive's own power structure.

Cause	How to check it	What fixes it
Motor data entered wrong, or never tuned	Compare P031 through P037 against the motor nameplate, not against the drive rating. Check whether an autotune was ever run	Correct the values, V5.xx firmware on the 525, then a stationary or rotating autotune with P040
Vector mode speed loop too aggressive	Read P039 [Torque Perf Mode]. Vector without a good tune is the pattern	Reduce A511 [Freq 1 BW], or run V/Hz if the load does not need vector
Motor lead length too long	Measure the actual run. Compare against the drives that are not faulting	Output reactor, and lower the carrier frequency
Moisture in the conduit box or Quick Connect cable	Open the box and look. Megger cold, then again after the machine has run warm	Dry and reseal, add a space heater or a drain fitting
MOV jumper wrong for the grounding system	Identify the distribution system first, then look at the jumper	Remove the MOV to ground jumper on ungrounded or resistive grounded systems

Causes of a VFD ground fault that survive a clean megger test.

Pulling the drive is the last step in chasing a VFD ground fault, not the first. By the time it is on the bench, the motor, the cable and the parameters should all be cleared.

The swap test, when a VFD ground fault survives everything

Three identical drives on three identical motors, and only one keeps faulting. Stop testing components and start testing positions.

Move the load. Take the motor leads off the faulting drive onto a healthy neighbor, put that neighbor's leads on the faulting drive, run both.

A VFD ground fault that follows the motor and cable to the other drive means the load side is guilty no matter what the megger said. One that stays with the original drive, now on a different motor and a different cable, narrows it down to the drive, its parameters or its supply. Either way it is a real result, and it costs an hour instead of another unit.

Do the parameters before the wiring, though. Read the faulting drive against the working one line by line off the displays, not from memory. The mismatch usually turns up in the motor nameplate block.

FAQ

What causes a VFD ground fault?

Rockwell's list for F013: a loose wire, insulation rubbed through in the motor junction or conduit box, motor data entered incorrectly, the motor wired incorrectly, a lead length too long for the output with no reactor, moisture in the conduit box or a Quick Connect cable, and a drive never properly tuned to the motor.

How do you test a VFD for a ground fault?

Megger each output phase to ground with the motor leads disconnected at the junction box. Use 1500V on a 460V system and 2000V on a 575V system, and treat anything under 10 MΩ per phase as a failed motor. Then repeat on the cable alone, both ends free.

Can you disable a VFD ground fault on a PowerFlex 525?

No. The 525 fault table names an enable parameter for the faults that have one, A557 for F021 Output Ph Loss and t105 for F059 Safety Open. F013 has none. A drive that faults with bare output terminals is telling you something true about itself anyway, so switching detection off would only move the failure somewhere more expensive.

Does a VFD ground fault mean the drive is dead?

Not on its own. F013 is Type 1 and it resets. If it comes straight back the moment you hit start with nothing landed on U/T1, V/T2 and W/T3, the drive is finished. If it only faults under load, keep looking. Repeated faults while the drive sits idle, after being jostled or vibrated, mean Rockwell wants it replaced.

When the isolation test lands on the drive

If the isolation test lands on the drive and you need one back on the line today, we keep PowerFlex 525 drives and control modules on the shelf at IQElectro, tested and ready to ship. If it lands anywhere else, a new unit will not fix a VFD ground fault that a wrong parameter or a missing reactor caused. It will fault again in a few weeks, same as the last one.