Variable frequency drive: what happens inside, and why your meter says 300 V on a 600 V motor

"Motor is only getting half voltage. Drive is bad." I get some version of that call more than any other drive call. Half the time I walk up and the drive is running the motor at 30 Hz, and 300 V on a 600 V motor is right about what it should be putting out.

A variable frequency drive does several things that look like faults if you read it with a plain meter and a line-starter mindset. Below is what happens inside, what the numbers should be, and how to tell a real problem from a normal reading. My examples are mostly on a PowerFlex 525, since that is the drive I see most, but the physics holds for any PWM drive.

What a variable frequency drive does between the line and the motor

Three stages, always in this order:

1.    Rectifier. A diode bridge turns the incoming AC into DC.

2.    DC bus. A bank of electrolytic capacitors holds that DC. It is charged higher than the line voltage, not lower.

3.    Inverter. IGBTs switch the bus on and off thousands of times a second. By changing the width of those pulses (PWM) the drive builds an output that behaves like AC at the commanded frequency.

So the motor never sees a clean sine wave of voltage. It gets pulses of full bus voltage, and its winding inductance smooths the current into something close to a sine. The pulse rate is the carrier frequency: a 525 is rated at 4 kHz and adjusts from 2 to 16 kHz. Turning it up quiets the motor whine, but above 4 kHz the drive gives up output current, and at low speed it may drop the carrier by itself unless A540 [Var PWM Disable] says otherwise.

Bus voltage runs about 1.35 to 1.41 times the line voltage:

Line voltage (3-phase)	DC bus, approximately
230 V	325 VDC
400 V / 480 V	680 VDC
600 V	848 VDC

 

Two things follow. The bus is a capacitor bank at hundreds of volts, and it does not care whether the disconnect is open. And the drive cannot make voltage it does not have: the bus sets the ceiling and the line sets the bus. Feed a drive 380 V and a 440 V motor never sees 440 V at full speed. It runs, just with less flux near the top than the nameplate expects.

Inside a variable frequency drive on a 480 V line. About 680 VDC on the bus, pulses of full bus voltage on the motor leads, and two different ways of reading the output.

Why the output of a variable frequency drive reads "wrong"

Usually one of three things, and the first two stack on top of each other.

Voltage follows frequency

In V/Hz mode the drive holds the volts-per-hertz ratio of the motor: nameplate volts at nameplate hertz, half the volts at half the hertz, a straight line in between. A 460 V, 60 Hz motor at 30 Hz gets roughly half, around 230 to 240 V depending on boost. That keeps the flux where it belongs, which is why torque capability holds up as speed comes down.

At the bottom the line gets a lift called boost, or there would not be enough voltage at a few hertz to push current through the winding resistance. In V/Hz a 525 ships with 5% boost under 5 HP and 2.5% from 5 HP up, in A530 [Boost Select]. In SVC, the factory default, that table is not used at all; the drive sets low-speed voltage from what autotune measured.

Where the line tops out is set by P031 [Motor NP Volts] and P032 [Motor NP Hertz]. Say somebody enters 120 in P032 for a 60 Hz motor. At 60 Hz output it now gets half its rated voltage and feels weak under load. Nothing is broken. The curve just moved.

V/Hz curve for a 460 V, 60 Hz motor on a variable frequency drive. At 30 Hz the motor gets about half its nameplate voltage, and that is correct.

Your meter and the drive are measuring different things

A decent true RMS meter on the motor leads can read as much as 20 to 30 percent higher than the drive display. Neither one is wrong. The meter gives you the heating value of the whole pulse train, carrier harmonics included, while the display only shows the fundamental, the part at output frequency that makes torque.

Meters with a low-pass filter mode for drive outputs land on the same number as the display. Without one, I read b001 [Output Freq] and b004 [Output Voltage] off the keypad and leave the leads alone.

Measure the input with the motor loaded

Plenty of "bad drive" calls start upstream. A half-open fuse or a backed-off lug reads full line voltage while the drive sits stopped, then sags the moment the motor pulls current. On a 525 that shows up as F004 UnderVoltage (bus below its minimum) or F003 Power Loss, the drive seeing single-phase operation with too much load. Check all three input phases with the motor running. Idle readings prove very little.

What you measured	What it usually means	What to check next
About half the nameplate voltage at half speed	Normal V/Hz behavior	b001 and b004 on the keypad: the ratio should track the nameplate
Low output voltage even at full speed	P031 or P032 does not match the nameplate, or the line is lower than the motor rating	Nameplate entries first, then line voltage at the drive input
True RMS meter reads as much as 20-30% above the display	The meter includes carrier harmonics	Low-pass filter mode on the meter, or go by the display
Input fine at idle, F004 or F003 under load	Weak fuse, loose connection or a lost phase upstream	All three input phases, measured with the motor loaded
Motor hums, will not start or stalls under load, no phase-loss fault	One output phase open at a terminal or in the motor box	Current on each phase with a clamp meter; A557 [Out Phas Loss En] ships disabled

 

That last row catches people. Output phase loss detection on a 525 is off from the factory, so a motor on two legs can pull a believable current while the drive stays quiet until overload. Where the application allows it, I switch A557 on.

The DC bus is still live after the disconnect opens

This is the part that hurts people. On a 480 V variable frequency drive the bus sits near 680 VDC, and the capacitors hold that charge after the line is gone. A dark display proves nothing about the bus.

On a 525 my routine is:

1.    Kill power at the disconnect and lock it out.

2.    Wait at least three minutes.

3.    Check L1, L2 and L3 line to line and line to ground, to prove the mains is really off.

4.    Measure DC+ to DC-. It has to read zero.

5.    Measure every power terminal to ground (L1-L3, T1-T3, DC+ and DC-) and hold the probes there until the reading goes to zero. That can take several minutes.

Other drives print their own wait time on the cover, and some ask for longer. Read the label instead of carrying a number over from a different drive.

On 480 V and 600 V drives use a meter rated at least CAT III 600 V; CAT IV 600 V / CAT III 1000 V is better. The 100-120 V single-phase 525s do not bring DC+ and DC- out at all, so there the terminal-to-ground checks are all you get.

DC+ and DC- on a PowerFlex 525 power terminal block. Zero volts between them, and from every terminal to ground, before anyone puts a screwdriver in there.

A variable frequency drive that sat on a shelf

Electrolytic bus capacitors do not like sitting unpowered for years. After long storage they have to be brought up gently before full voltage, or the first power-up becomes the stress test. People buying surplus ask me about this more than anything, and the answer is usually less work than they expect.

Time unpowered	Allen-Bradley PowerFlex	ABB ACS580, ACH580, ACQ580, ACS880 (except -104)	ABB ACS800, ACS850, ACQ810, ACSM1, ACS880-104
Under 2 years	No reforming needed	Power on 30 minutes, no load	Power on 30 minutes, no load
2 to 3 years	Rated voltage for 30 minutes, no load	Power on 30 minutes, no load	External DC supply or diode-bridge method
3 years and over	External DC supply on the DC terminals, stepped up to 50%, 75% and 100% of rated voltage, 30 minutes at each step	External DC supply or diode-bridge method	External DC supply or diode-bridge method

 

On the Allen-Bradley side, the DC supply comes up to 1.35 to 1.45 times the rated AC voltage with current limited to 500 mA. The ABB DC and diode-bridge methods run for hours, not minutes.

So the first question about a drive off a shelf is not whether it powers up. It is when it last ran. A drive that stays in service normally runs past ten years, some past twenty.

V/Hz, SVC or vector: picking the control mode

On a 525 this is P039 [Torque Perf Mode], set to SVC out of the box. Right default for most single-motor jobs, but only if the variable frequency drive knows the motor: SVC reads voltage and current against a motor model, and the model comes from autotune.

P039 setting	What it does	Speed regulation, no encoder	Good fit
0 "V/Hz"	Fixed volts-per-hertz curve plus boost	±1% of base speed over 60:1	Fans, pumps, several motors on one drive
1 "SVC" (default)	Estimates motor behavior from voltage and current	±0.5% of base speed over 100:1	Most single-motor loads, after autotune
3 "Vector" (525 only)	Velocity vector control: regulates speed, cannot control torque	±0.5% of base speed over 60:1	Tighter speed with an encoder (±0.1% over 100:1)
4 "PM Control" (525, firmware 5.xxx and later)	Permanent magnet motors, open or closed loop	±0.5% of base speed up to 20:1	PM motors only
5 "SynRM" (525, firmware 7.xxx and later)	Synchronous reluctance motors	Not listed	SynRM motors only

 

Autotune is P040. Option 1 is a static tune for a motor that cannot be uncoupled. Option 2 rotates it, so take the load off first; the shaft can turn the wrong way. Enter every nameplate parameter before you start, and give the start command within 30 seconds or it faults out. Swap the motor and you tune again, the model belongs to the old one.

Setting 2, "Economize", is an SVC variant with the same regulation spec. For fans and pumps I tune in SVC too. If one already runs on V/Hz, A530 has variable torque boost curves (the VT settings) for exactly that.

"Vector" sounds like the top of the range, and on a 525 it is not what people picture. It regulates speed, cannot control torque, and wants an encoder and a rotating tune. Holding a load at zero speed, a hoist or anything hanging weight on the shaft, needs a drive built for torque control at zero speed with encoder feedback.

Several motors on one drive

One drive on a row of conveyor motors or fans is common and works fine, with rules:

·       Run V/Hz. SVC and vector need a single motor to model, and parallel motors break that.

·       Size the drive for the sum of all motor full-load amps plus 20%.

·       Give every motor its own overload. The drive only sees total current, so one motor can be badly overloaded while the total looks fine.

·       Skip very small drives, roughly under 3 HP, for multi-motor setups. Charging current from several motor cables eats too much of their rating.

·       Standard motor protection circuit breakers age early on a drive output, worn down by the reflected voltage pulses. Use breakers designed for drive outputs, and keep the carrier at 4 kHz or below.

Slow speed, load type and heat

Whether slow speed is a problem depends first on the load.

Load type	Examples	Torque vs speed	Power vs speed
Constant torque	Conveyors, feeders, screw compressors	Stays constant	Proportional to speed
Variable (quadratic) torque	Centrifugal pumps, fans	Proportional to speed squared	Proportional to speed cubed
Constant power	Winders and rolling, where the diameter keeps changing	Inversely proportional to speed	Stays constant

 

Then the motor. A standard motor cools itself with a fan on its own shaft, so slowing the shaft slows the cooling and the continuous torque drops with it. Typical numbers for a standard cage induction motor on a drive (the motor maker's data wins if you have it):

Speed, % of base	Continuous torque, self-cooled	Continuous torque, separate blower
0%	About 70%	100%
20%	About 80%	100%
40%	About 88%	100%
60%	About 95%	100%
80%	About 98%	100%
100%	100%	100%
200% (above base speed)	About 50%	About 50%

 

Say a conveyor runs at 30% speed fully loaded. Constant torque, so it still needs 100%, and a self-cooled motor at that speed is good for roughly 85% continuous. It runs fine for a while, then cooks. A fan at 50% speed needs only about a quarter of rated torque and never gets close. Fixes for the conveyor: a separately powered blower, a bigger motor, or gearing so the motor runs nearer base speed.

Short peaks are different. Motor thermal time runs from about 15 minutes on small motors to hours on big ones. A drive heats up in minutes. A 525 at heavy duty allows 150% for 60 seconds and 180% for 3 seconds, so on a short overload the drive usually runs out first.

What goes between the drive and the motor

Ideally nothing but cable. A variable frequency drive starts and stops the motor from its control inputs. A contactor or disconnect on the output that drops and reapplies the motor to start and stop it should not be there.

When you do need a disconnect at the motor for lockout, give it an auxiliary contact that disables the drive at the same moment, into a digital input set to Aux Fault (a 525 trips on F002) or a coast-to-stop input. I use early-break auxiliaries, the kind that open before the main poles, so the drive is already off when the power contacts part. Otherwise you are breaking motor current under power, or reclosing the leads onto a drive that is still running.

Keep the drive as close to the motor as the layout allows. At 240 V and below reflected wave is not a concern. At 480 V and up, long runs can need output protection, and allowed length depends on the drive, the motor insulation and the carrier.

Left: a motor disconnect whose aux contact shuts the variable frequency drive down before the power contacts open. Right: one drive in V/Hz feeding three motors, each with its own overload wired back to the drive.

FAQ

Is a variable frequency drive the same as an inverter?

In everyday use, yes. VFD, AC drive, adjustable frequency drive, variable speed drive, inverter: same box. Strictly, the inverter is only the output stage that turns the DC bus back into AC.

Variable frequency drive or soft starter?

A soft starter ramps voltage with back-to-back SCRs at line frequency. Torque goes roughly with the square of voltage, so on a typical motor held to 350% current it gives about 34% of full-voltage starting torque. Some have a couple of fixed slow speeds, good for minutes before things get hot. For a gentler start on a light-starting load, the soft starter is cheaper. For speed control, or high starting torque without the current spike, it has to be a drive.

Can you put a variable frequency drive on any motor?

Any standard three-phase induction motor, within limits. A 525 also runs PM and synchronous reluctance motors with the right P039 setting and firmware. DC motors need a DC drive. On an ordinary AC motor the limits are low-speed cooling (table above) and, on long 480 V or 600 V runs, insulation stress from reflected wave.

Can a variable frequency drive damage a motor?

It can, three usual ways: a self-cooled motor run slow at full torque until it overheats, reflected-wave spikes on long 480 V or 600 V leads, and current escaping through the bearings. The first is a setup problem. For the other two, cable length and type matter, and shielded cable helps keep shaft voltage and bearing currents down on some installs.

Can a three-phase variable frequency drive run on single-phase power?

Only derated. On single-phase 60 Hz the bus ripple drops from 360 Hz to 120 Hz, current through the input diodes roughly doubles and input distortion goes past 90% THD instead of about 40%. The derate is model-specific, so take it from the table for that exact drive. Cleaner still: a drive built for single-phase input, like the 240 V single-phase PowerFlex 525s.

When the drive really is done

Check the nameplate entries, the input under load and the bus before you order anything. If the variable frequency drive still turns out to be dead, we keep Allen-Bradley drives across the range: 480 V PowerFlex 525s like the 25B-D4P0N104 (2 HP), 25B-D013N104 (7.5 HP) and 25B-D017N104 (10 HP), single-phase 240 V PowerFlex 525s like the 25B-A2P5N104 and 25B-A4P8N104, the PowerFlex 4 22A-A8P0N104 and PowerFlex 4M 22F-D2P5N103 for small machines, and the PowerFlex 753 20F11ND065AA0NNNNN, PowerFlex 755 20G11ND5P0AA0NNNNN and PowerFlex 70 20AD052A0AYNANC0 for bigger ones. Whatever you pick, ask when it last ran and use the storage table above before you put full voltage on it.