Choosing a PWM Carrier and Reading It in Kilohertz
A switching frequency almost always arrives as a raw hertz figure: 4 000 sitting in a drive parameter, 24 000 buried in an ESC menu, 500 000 printed on a regulator datasheet. The conversation about that setting, though, happens in kilohertz — nobody argues about a sixteen-thousand-hertz carrier. Getting the candidates onto the same scale is the first step before you weigh audible whine against heat in the switches.
What the Carrier Setting Actually Decides
The whine you hear from the machine
Heat paid per transition
Ripple and the size of the magnetics
What a camera sees in a dimmer
From a Parameter Value to a Carrier You Can Discuss
The routine is the same whether the figure came off a drive keypad, a flight-controller configurator or the front page of a regulator datasheet.
Type the raw switching figure
Put the hertz value into the left field — 2 000, 4 000, 24 000, whatever the parameter list holds — and the kilohertz reading follows as you type. Spaces inside a figure like 16 000 are ignored, and a comma is accepted as the decimal point.
Place the result against the hearing band
Anything landing between 2 and 5 kHz will be heard clearly in an occupied room; 8 to 12 kHz thins the tone down to a hiss; 16 kHz and above is silent to most people but hottest for the bridge.
Copy the bare figure back into the parameter
The copy button puts the plain number on the clipboard with no unit and no spacing, which is what a keypad entry, a configurator field or a spreadsheet of drive settings expects. Ctrl + C inside a field does the same.
Turn a kilohertz spec back into hertz
Press the swap button (↔) when a regulator is quoted at 500 kHz and a ripple or loop calculation needs 500 000 Hz. Typing straight into the right-hand field does the same job without swapping sides.
Carrier Frequencies Used in Drives, ESCs and Converters
The settings you meet in the field, shown as the hertz value stored in the equipment and the kilohertz figure people use to describe it — plus the reason someone landed there.
| Stored value | Carrier | Where it turns up | Why it was chosen / what you hear |
|---|---|---|---|
| 1 000 Hz | 1 kHz | Simple LED PWM dimmer | Below the 1.25 kHz low-risk line — visible on camera and in peripheral vision |
| 2 000 Hz | 2 kHz | Drive set for maximum efficiency | Loud tonal whine, lowest switching loss, full current rating available |
| 4 000 Hz | 4 kHz | Common drive factory default | Sits in the ear's most sensitive band — the classic motor-drive whine |
| 8 000 Hz | 8 kHz | Balanced drive setting | Thinner, higher tone; noticeably more heat in the power stack |
| 12 000 Hz | 12 kHz | Pump and fan drives in occupied areas | Faint hiss; this is usually where derating tables start to bite |
| 16 000 Hz | 16 kHz | Upper drive setting | At or past the hearing limit of most adults; 62.5 µs period |
| 24 000 Hz | 24 kHz | Multirotor and hobby ESC | Silent, smoother phase current, more heat per FET |
| 32 000 Hz | 32 kHz | High ESC setting | Inaudible; chosen where throttle response matters more than efficiency |
| 100 000 Hz | 100 kHz | Older buck regulator | Large inductor and output capacitor, very low switching loss |
| 500 000 Hz | 500 kHz | Modern point-of-load buck | A 2 µs period keeps ripple small with a physically tiny inductor |
Two and a half orders of magnitude separate the top and bottom of that list, yet the same trade sits behind every row: quieter and smoother costs heat, and heat costs rated current.
Handy While Comparing Carrier Options
Walk the carrier ladder in one field
Type 2 000, then 4 000, then 8 000 and 16 000 without clearing anything — the kilohertz side tracks each candidate as you work through the drive's allowed range.
Reverse for a regulator quoted in kHz
The swap button flips to kHz → Hz, the direction you want when a spec says 500 kHz and the inductor sizing formula wants the plain count of switching cycles per second.
Figures that go straight into a keypad
Copy returns digits only, so what lands in a parameter table or a commissioning record is exactly what the drive will accept.
Carriers that have run past a megahertz
The searchable dropdowns include MHz, so a wide-bandgap converter switching at 2 000 000 Hz reads back as 2 MHz beside a 4 kHz drive setting.
Switching Frequency Questions
Should the carrier sit above the audible range or below it?
It depends on which problem hurts more. Above about 16 kHz (16 000 Hz) the tone vanishes for most adults, which matters in offices, hospitals and public spaces — but the bridge runs hotter and the drive may have to be derated. Down at 2 to 4 kHz the power stage is happiest and the full current rating stays available, at the price of an obvious whine. Plant rooms and hot cabinets tend to stay low; equipment near people gets pushed up.
How much extra heat does doubling the carrier cost?
Switching loss is roughly proportional to how often you switch, so moving from 4 000 Hz to 8 000 Hz roughly doubles it while conduction loss is unchanged. On a small drive that is a warmer heatsink; on a large one it can be kilowatts of extra dissipation inside the cabinet. The manufacturer's derating table is the honest answer for a given model — it states how much continuous output current the higher setting costs.
What dimming frequency keeps LEDs clean to the eye and on camera?
Cheap dimmers often run somewhere between 200 and 800 Hz, which shows as banding under a rolling shutter and as a stroboscopic effect on moving objects. IEEE 1789-2015 puts the low-risk boundary near 1 250 Hz (1.25 kHz) regardless of modulation depth, and lighting intended for motion-sensitive spaces or video work is commonly taken to 3 000 Hz (3 kHz) or beyond.
How much of the period does dead time eat at a high carrier?
Dead time is fixed by the devices and the gate drive, not by the carrier, so raising the carrier makes it proportionally larger. Take a bridge needing 2 µs of blanking: at 4 kHz the period is 250 µs and that costs 0.8 % of it, but at 16 kHz the period is only 62.5 µs and the same 2 µs is 3.2 %. It shows up as voltage distortion at low output, and it also sets the shortest usable pulse the modulator can produce.
Why does a datasheet start in hertz and finish in kilohertz?
Because writing 500 000 in body text is unreadable, while a parameter field cannot hold anything but the plain count. Mains-related and control-loop numbers stay small enough to leave in hertz; anything above a thousand cycles a second gets the kilohertz label. That mismatch is where mistakes creep in — the manual discusses a 12 kHz carrier while the keypad wants 12000, so it pays to convert deliberately instead of counting zeros.
No comments yet. Be the first to comment!