dV/dt Filter vs Sine Wave Filter: When Each Earns Its Cost

A dV/dt filter slows the voltage rise at the motor terminals but leaves the drive output as a PWM pulse train. A sine wave filter goes further and rebuilds a near-sinusoidal waveform. The first is small, cheap and nearly lossless. The second costs four to eight times as much and is the only one of the two that makes a standard motor safe on a long cable.

The trouble starts when you look for the cable length that tells you which to buy. Published thresholds for that same decision disagree by a factor of ten. One manufacturer puts a sine wave filter at 50 m. Another does not reach one until 450 m.

If you have two tables open and they contradict each other, that is not your mistake, and it is not a typo in either document.

This article explains why the tables differ, gives you the four parameters that actually decide the answer for your drive, and covers the cases where neither filter earns its cost. Input-side device selection is a separate decision, covered in line reactor vs harmonic filter, with the strategy behind it in harmonic mitigation for VFDs.

Key Takeaways

• A dV/dt filter and a sine wave filter are not two strengths of one product. One slows the voltage edge, the other rebuilds the waveform, and the difference in cutoff frequency explains the entire gap in size, loss and price.
• Published thresholds for the same cable-length decision range from about 120 m to over 450 m for a sine wave filter. Four parameters explain the spread: switching frequency, cable type, motor insulation class and voltage class.
• A load reactor and a dV/dt filter are not interchangeable. A reactor alone holds voltage rise to roughly 1,000 V/µs. A dV/dt filter reaches roughly 500 V/µs.
• Sine wave filters cost 4 to 8 times a load reactor and add 1 to 2 percent continuous loss. At 100 kW that is 1 to 2 kW of heat inside your enclosure.
• On a short run with an inverter-duty motor and shielded cable, the honest answer is often neither filter.

dV/dt Filter vs Sine Wave Filter: The Short Answer

A dV/dt filter limits how fast the output voltage rises by adding inductance and an RC network, which reduces reflected-wave stress and insulation heating. A sine wave filter is an LC low-pass filter that removes the switching frequency and delivers a near-sinusoidal voltage. Choose the dV/dt filter for moderate cable runs and inverter-rated motors. Choose the sine wave filter for long runs, non-inverter-rated motors or multiple motors on one drive.

	dV/dt filter	Sine wave filter
What it is	Inductance plus an RC network	LC low-pass filter with damping
Cutoff frequency	Above the switching frequency, typically tens of kHz	A few hundred Hz, below the switching frequency
Waveform at the motor	PWM pulse train with slower edges	Near-sinusoidal
Voltage rise achieved	Roughly 500 V/µs; some sources cite 200 V/µs or better	50 V/µs in some sources, 200 V/µs in others
Effect on output harmonics	Minimal, because the cutoff sits above switching	Substantial, the fundamental is reconstructed
Physical size	Small, often mounts beside the drive	Large, frequently a separate enclosure
Losses	Roughly 0.5 to 1 percent	Roughly 1 to 2 percent
Cost index, load reactor = 1.0	1.3 to 2.0 times	4 to 8 times
Switching frequency limit	None significant	Filter must be tuned below the drive’s switching frequency

The row that explains all the others is the cutoff frequency. A dV/dt filter’s cutoff sits above the switching frequency, so it barely touches the output spectrum and can stay small. A sine wave filter’s cutoff sits below the switching frequency, so it carries the full motor current at a much lower frequency, and that is why it needs large inductors, large capacitors and damping resistors.

A note on the numbers. Rise-time figures vary between sources by a factor of four or more, and the variation is not sloppiness. A dv/dt measurement at the drive terminals is not the same measurement as one at the motor terminals 100 m away, and the second is always worse. Treat any single figure as indicative.

What Each Device Actually Does to the Waveform

A drive output is not a sine wave. It is a train of rectangular pulses whose width varies to synthesise an average voltage. The edges of those pulses are fast, with rise times between about 0.05 and 1 microsecond, and it is the edge rather than the pulse that damages motors.

A dV/dt filter stretches that edge. The pulses stay rectangular and still switch at the carrier frequency. What changes is that each transition takes longer, so voltage at the motor terminals climbs more gently and the reflected-wave overshoot is smaller. Because the filter is not removing the switching frequency, its components are modest: usually a reactor plus a small capacitor and resistor network.

Fitting a sine wave filter to a VFD output is a different category of project. Its cutoff sits at a few hundred Hz, well below the carrier, so the carrier and its sidebands are attenuated rather than shaped, and what reaches the motor is close to a sine wave at the commanded output frequency.

That is a much stronger intervention, and it changes what you can ask of the drive. A standard motor never designed for inverter duty becomes usable, insulation stress from fast edges largely disappears, and audible motor noise drops. The trade is that a low cutoff requires large components that dissipate energy. EASA’s work on filtering electric machines is the best treatment of the retrofit cases where this matters most.

One consequence is the most common specification error with these devices. A sine wave filter must be tuned to sit below the drive’s carrier frequency, which caps the carrier frequency you can select. A drive running at 12 kHz for quiet operation may be incompatible with a filter rated for 4 kHz maximum.

The Reflected Wave: Why Cable Length Is the Wrong Question

Cable length matters because of a transmission-line effect, and understanding it removes most of the confusion in this topic.

A motor cable has distributed capacitance and inductance, which gives it a characteristic impedance. A fast voltage edge travelling down it behaves like a wave. When that wave reaches the motor, whose impedance is much higher than the cable’s, most of the energy reflects back toward the drive.

If the cable is short, the reflection returns and settles before the next edge arrives. If it is long enough, the reflection arrives back at the drive just as a new edge is being launched, and the two superimpose. That reflected wave voltage spike can push the motor-terminal peak toward twice the DC bus voltage.

Published measurements on 480 V drives commonly land between 1,100 and 1,600 V. A motor with a 1,000 V insulation system does not survive that for long.

The length at which this starts is the critical length, and it depends on how fast the edge travels and how long it takes to rise. Faster edges mean shorter critical lengths. Published values span roughly 8 m to about 30 m, and the same logic that makes the filter tables disagree makes these disagree too.

Dermot inherited a conveyor drive when his plant consolidated two production halls. The drive and motor were fine in the old layout, with about 12 m of cable between them. The new layout put the motor at the far end of the hall, 90 m away, and the electrician reused the same drive and motor because both were within their service life. Fourteen months later the motor failed with a shorted winding.
Nothing about the load had changed and the duty cycle was identical. The only variable was the cable. A standard motor on a run long enough for reflected-wave peaks to erode the winding edge by edge will fail eventually, and it takes over a year to show up.

Why the Published dV/dt Filter vs Sine Wave Filter Tables Disagree

Here is the same question answered by seven sources.

Source	Output or load reactor	dV/dt filter	Sine wave filter
SMARTqube	50-100 m	100-200 m	above 200 m
Franklin Water	not stated	4-120 m	above 120 m
INVT, non-shielded cable	50-150 m	150-450 m	450-1,000 m
INVT, shielded cable	30-100 m	100-230 m	230-500 m
Phase Technologies	not stated	100-300 ft	above 300 ft
Hammond Power Solutions	not stated	100-300 ft	1,000-15,000 ft
DPA Magazine	not stated	not stated	30-300 m, depending on drive size

Look at the sine wave filter column. The most conservative entry is 120 m. The most permissive is 15,000 ft, roughly 4,500 m. INVT publishes two different ladders in the same document set, one for shielded cable and one for non-shielded, and the shielded ladder is consistently more conservative.

The DPA Magazine guidance is the only entry that scales its answer with drive size rather than stating a fixed limit, which is the more honest approach.

None of these sources is wrong. Each states the threshold for the hardware it sells, evaluated against a different set of assumptions. The four parameters below are what those assumptions are, and they are the only thing you need to establish to get the right answer for your own installation.

Parameter	What it changes	Where to find it
Switching or carrier frequency	Higher carrier frequency means faster edges and a shorter critical length	Drive parameter list, typically adjustable from about 2 to 16 kHz
Cable type	Shielded cable has higher capacitance per metre, so it reaches the critical length at a shorter run	Cable datasheet, or simply whether the run is shielded
Motor insulation class	An inverter-duty motor to NEMA MG1 Part 31 or IEC 60034-25 withstands far more than a legacy standard motor	Motor nameplate and datasheet
Voltage class	DC bus voltage scales with supply voltage, and so does the reflected-wave peak	Drive nameplate

The pattern is worth internalising: anything that makes the voltage edge faster, or the motor less tolerant of it, moves the threshold toward you. Raising the carrier frequency shortens the permissible run. Moving from shielded to non-shielded cable lengthens it.

This is why a single number cannot answer the question. If your drive manual gives a table, it was computed for that drive’s output stage at a stated carrier frequency and cable type. Find those two conditions, confirm they match your installation, and only then use the number. The motor side is covered in motor compatibility with VFD.

Output Reactor vs dV/dt Filter: The Mistake That Costs Money

These two are constantly treated as synonyms, including in manufacturer documentation, and the error runs in both directions. Sometimes a specifier buys a cheap reactor expecting dV/dt filter performance. Sometimes a contractor fits a dV/dt filter where a reactor would have done.

The measured difference is real, and it is the distinction that decides most of these jobs.

Device	Rise time achieved	What it buys you
Nothing	3,000 to 8,000 V/µs at the motor terminals, depending on drive and measurement point	The reflected-wave problem, unmanaged
Load reactor	Roughly 1,000 V/µs	Cable-length headroom, and a smaller reflected-wave peak
dV/dt filter	Roughly 500 V/µs	A genuinely slower edge, plus reduced bearing current
Sine wave filter	50 to 200 V/µs	The reflected-wave problem removed rather than reduced

The practical reading: a load reactor buys cable-length headroom, while a dV/dt filter buys a slower edge. Those are related but not the same benefit. If your run is slightly over the limit and the motor is inverter-duty, an output reactor is often the cheapest correct answer and will carry you to roughly twice the unassisted length. If the motor is older, or the run is well past the limit, the reactor’s 1,000 V/µs is not enough and the dV/dt filter’s 500 V/µs is what you actually need.

Specifying a long-cable run? Send our application engineers the drive model, cable length and motor nameplate and we will work the reflected-wave numbers against your actual installation rather than a generic table.

What a Sine Wave Filter Costs Beyond the Purchase Price

Purchase price is the number everyone compares and the least useful one for this decision. Every figure below is US list or retail pricing observed in 2026. Copper moves this category, so treat them as dated benchmarks rather than quotations.

Rating	dV/dt filter	Sine wave filter
1 HP, 240 V	$300to$600	$800to$1,500
5 HP, 460 V, 1,000 ft leads	$527to$702 list, likely under $500 net	not quoted
15 HP, 21 A	$1,006list,$653.90 online	not quoted
100 A, low voltage	not quoted	$1,500to$2,000
300 HP, 480 V	not quoted	about $6,382list,$5,425 online
400 HP, 480 V	not quoted	about $17,098
500 HP, 480 V	not quoted	about $22,049
700 HP, 480 V	not quoted	about $28,098

Indexed against a load reactor at 1.0 for representative 400 to 480 V industrial ratings, a dV/dt filter runs about 1.3 to 2.0 times, and a sine wave filter about 4 to 8 times. Engineers working on very long runs sometimes describe the sine wave filter cost as roughly what the drive costs. That comes from practitioner discussion rather than a published price list, so treat it as a rule of thumb.

The costs that do not appear on the quote are the ones that cause trouble later.

Continuous loss. A sine wave filter dissipates roughly 1 to 2 percent of throughput permanently. At 100 kW that is 1 to 2 kW of heat generated continuously inside or beside your enclosure. If the panel was sized around the drive’s own losses, that extra heat changes the cooling calculation.

A switching frequency ceiling. The filter must be tuned below the carrier frequency, which limits how quietly the drive can run.

Capacitors as a wear item. The capacitors age, so they belong on a maintenance schedule and eventually a replacement budget.

Sanne specified filters for a pump station and compared two quotations on purchase price alone. The dV/dt option was cheaper by a clear margin, so that was the recommendation she took to the client. What the comparison missed was the continuous loss. On the larger drive the sine wave filter would have dissipated about 1.5 kW more, continuously, inside a sealed enclosure rated for a fixed heat load.
The revised cost was not the filter. It was the larger enclosure, the upgraded cooling and the switchgear room ventilation, and the total was several times the price difference that had decided the original choice.

When the Answer Is Neither

The most useful thing this article can tell you is that a meaningful share of installations need no output filter at all.

If your motor is inverter-duty, your cable is shielded VFD cable, your carrier frequency is at the manufacturer’s default, and the run is under roughly 15 to 25 m, then reflected-wave stress is small and the case for any output filter is weak. The pressure to buy something anyway is real. Filters are easy to specify, they feel like insurance, and a supplier who sells both options has no particular reason to talk you out of one.

Yusuf was quoted a sine wave filter for a 9 m run on a 15 kW pump. The reasoning in the quotation was that sine wave filters provide the best motor protection, which is true in isolation. What the quotation did not establish was that the run needed any protection at all: the motor was inverter-duty to NEMA MG1 Part 31, the cable was shielded VFD cable, and the drive ran at its default 4 kHz carrier.
A second opinion found the installation already well inside every threshold, and the money went into shaft grounding and a spare motor instead. Both address failure modes the filter would not have touched.

Two application factors override that verdict, and both are worth checking early.

Multiple motors on one drive. The effective cable length is the sum of all the runs, and the motors interact through the shared output. Per-motor filters are often cleaner than one large filter at the drive. The AC drive system overview covers how the output stage behaves in that configuration.

A step-up transformer between the drive and motor. This is the hardest case. The transformer adds inductance and its own reflection behaviour, and low-voltage drives feeding medium-voltage motors routinely need a sine wave filter regardless of apparent cable length.

One qualification on bearing currents, because dV/dt filters get sold on this benefit. A dV/dt filter does reduce the common-mode voltage that drives bearing current, and that is genuine, but it is not a fix. If bearing failure is your actual problem, shaft grounding is the direct solution and costs far less. The filter reduces the excitation; a grounding ring gives the current somewhere to go.

For motor stress more generally see VFD motor life, and for the switching behaviour that creates these edges, VFD working principle.

Frequently Asked Questions

What is the difference between a dV/dt filter and a sine wave filter?

A dV/dt filter slows the rate at which output voltage rises while leaving the switching waveform intact. A sine wave filter removes the switching frequency and delivers a near-sinusoidal voltage. The sine wave filter is larger, costs four to eight times more and adds measurable loss.

How long can a VFD motor cable be before I need a filter?

There is no single number, and any source that gives you one is answering for its own hardware. The dV/dt filter cable length threshold depends on output rise time, carrier frequency setting, cable shielding, motor insulation class and system voltage. Check your drive manual for the figure it computes under your conditions.

Do I need a sine wave filter for a long cable run?

Not automatically. You need one when the run is long enough that a reduced voltage rise is insufficient, when the motor is not inverter-rated, when several motors share one drive, or when a step-up transformer sits between drive and motor. Published thresholds range from about 120 m to well over 450 m.

Is a dV/dt filter the same as an output reactor?

No, and the gap is measurable. A load reactor alone holds voltage rise to roughly 1,000 V/µs. A dV/dt filter adds capacitance and damping to reach roughly 500 V/µs. A reactor is the cheaper way to buy cable-length headroom.

Can I use a sine wave filter with any VFD?

No. The filter must be tuned below the drive’s switching frequency, which imposes a ceiling on the carrier frequency you can select. A drive set to 12 kHz for quiet operation may be incompatible with a filter rated for 4 kHz maximum. Confirm the pairing against both datasheets before ordering.

Does a dV/dt filter reduce motor bearing currents?

It reduces them, because it lowers the common-mode voltage that drives current through the bearings. It does not eliminate them. If bearing failure is the problem, shaft grounding is the direct fix and costs much less.

How do I know which filter my drive manual is asking for?

Drive manuals usually give a cable-length table with columns for no filter, a reactor, a dV/dt filter and a sine wave filter. Find the row for your cable type and the stated carrier frequency, and confirm the voltage class matches. If the manual gives a single number with no conditions, ask what conditions it was computed under.

Is a sine wave filter worth the cost?

It is worth it when the alternative is a rewound or replaced motor. It is not worth it on a short run with an inverter-duty motor and shielded cable, where reflected-wave stress is already low. Compare against the cost of motor failure, not the price of a dV/dt filter.

Conclusion

The dV/dt filter vs sine wave filter decision looks like a cable-length lookup, and it is not. It is a calculation with four inputs, and the published tables you will find are each the output of that calculation for one manufacturer’s hardware under one set of assumptions.

Start with the four parameters: carrier frequency setting, whether the cable is shielded, your motor’s insulation class and your system voltage. Those tell you which part of the published range applies to you, and they take about ten minutes to gather.

Then decide what you are buying. A load reactor buys length. A dV/dt filter buys a slower edge.

A sine wave filter removes the reflected-wave problem rather than reducing it, and charges four to eight times the reactor price plus 1 to 2 percent loss for the privilege. On a short shielded run with an inverter-duty motor, the correct answer may be nothing at all.

Whatever you specify, write the assumption down next to it. A threshold without its conditions is how the tables you have been reading ended up disagreeing.

Talk to our application engineers with your drive model, cable length, cable type and motor nameplate, and we will run the reflected-wave numbers against your installation. We manufacture drives and we do not sell output filters, so the recommendation will be about your motor rather than about a part number.