Flicker, EMF, and Clean Light: Arjen Helder Answers the Questions Most Manufacturers Won't
A technical Q&A with Arjen Helder, PCB and optical engineer, AriHelder — Chiang Mai, Thailand, 2026
There is a category of question that most red light device companies route around. Questions about flicker percentages, waveform shape, switching frequency, electromagnetic field sources, and whether USB-C power actually changes anything. The answers require circuit-level knowledge, and honest answers sometimes reveal uncomfortable differences between products.
Arjen Helder designed the PCB inside the Helder 2. Before that, he co-created the FlexBeam — among the first consumer wearable near-infrared devices — which means this is his second time building a consumer photobiomodulation device from the component level up. We filmed this Q&A in multiple sessions at our bench in Chiang Mai, with an oscilloscope and a solar panel sensor as the opening demonstration. The questions came from our community.
This is the full transcript, edited for readability. Where Arjen pushed back on a question's premise, that is preserved. Where he said he did not know something, that is preserved too.
Before the Questions: Live on the Oscilloscope
Before the structured Q&A, Arjen ran a live bench demonstration using a small solar panel from a previous project as a photodiode sensor, pointed at the Helder 2 electronics and connected to an oscilloscope.
His explanation of why a solar panel works for this: it converts light variation directly to a voltage signal, which the oscilloscope displays in real time. No special equipment needed — the physics of a solar cell makes it a natural flicker detector.
At full Body Mode power: essentially a flat line on the oscilloscope. No measurable flicker. A small amount of noise was present, but Arjen switched the device off and the noise remained — confirming it was ambient RF from the environment, not from the device.
He then pressed the Head Mode button. A signal appeared immediately. He auto-scaled the oscilloscope and the trace resolved into a clean sine wave at approximately 50 kHz.
His comment: "That's what you want. Sine waves are natural. Square waves are man-made. Nothing is square in nature — you walk into any forest and as soon as you see a square, you know one thing for sure: a human made that. So that's what I try to do. I try to make things as close to nature as possible. And if I do have some noise in my output, I want that to be a natural shape."
He then probed directly on the PCB to show what the driver itself produces at full power — a small residual pulse visible at very high oscilloscope magnification (200 millivolts range). This demonstrates the difference between what the driver chip generates and what the capacitors deliver to the LEDs. The capacitors absorb and smooth that driver-level signal before it becomes light.
Key finding: Optical output in Body Mode — no measurable flicker. Head Mode — a clean sine wave at approximately 50 kHz, above any perceptible or biologically relevant threshold.
Why Do LEDs Flicker in the First Place?
The LED itself does not flicker by design. The driver does.
LEDs are current-controlled devices, not voltage-controlled like a traditional incandescent bulb. An incandescent runs on whatever voltage you supply; an LED will draw as much current as it can and destroy itself if nothing limits the flow. Every LED product therefore uses a driver chip — a controller that limits current to a defined value.
The flicker you see in cheap LED bulbs comes from the driver, not the emitter. Specifically, it comes from a driver that is not adequately buffering the mains supply. Mains power in a 50 Hz country produces a pulsing DC signal at 100 Hz after rectification. Higher-quality LED bulbs historically used a 400V capacitor to buffer that signal; the cheap ones do not. If the driver passes that 100 Hz pulsation through to the LEDs without smoothing it, you get 100 Hz flicker — exactly the range where human vision is most sensitive.
This is not an LED problem. It is a driver design and component quality problem.
What Causes Flicker Specifically in Red Light Therapy Panels?
Red light therapy panels use the same LED fundamentals, but add a second flicker source: PWM dimming.
PWM — Pulse Width Modulation — is the standard method for controlling average power through an LED without wasting energy as heat. Rather than running the LED at partial voltage, the driver switches the LED fully on and fully off at high speed. At 50% duty cycle — equal on and off time — you get half the average power. The analogy Arjen uses: controlling water flow not by partially closing a valve, but by opening and closing it rapidly at a fixed interval. Inefficient for plumbing; highly efficient for electronics.
The problem is low-frequency PWM without filtering. A driver switching at 200 Hz with no smoothing produces a 200 Hz square wave at the LEDs — a square wave of light intensity that the nervous system can detect even if you are not consciously aware of seeing it pulse. This is measurable, and it shows up on slow-motion phone footage.
The solution is not to abandon PWM. The solution is to switch faster and filter properly.
What Is the Difference Between Flicker Frequency and Flicker Percentage?
Both matter, but they measure different things and people frequently conflate them.
Flicker frequency is how fast the light switches — measured in hertz. Flicker percentage is how deep the modulation is: how far does the light output drop between the peak and trough of each switching cycle?
The most familiar form is mains-induced flicker. At 220V AC, rectification produces a bouncy DC waveform at 100 Hz in 50 Hz countries. Cheap drivers pass that straight to the LEDs. The other category is PWM-induced flicker — switching at a low frequency without adequate filtering.
Key distinction: 100 Hz at 80% flicker percentage is worse than 1,000 Hz at 30%. Frequency alone does not tell the full story. You need both numbers to evaluate a device.
What Is the Flicker Index and How Is It Different from Flicker Percentage?
Flicker percentage is a single number that captures modulation depth. It says nothing about waveform shape.
The Flicker Index, defined in IEEE standard PAR1789, captures shape. It runs from 0 to 1. A device that flickers with a smooth sine pattern scores far lower than one producing sharp brief spikes — even at the same overall percentage — because brief intense spikes are more physiologically disruptive than gradual undulations.
Arjen notes this from personal experience designing LED bulbs: high-end products used transformer-based supplies with near-zero flicker; cheap modern bulbs optimise for cost and skip the filtering. A 10-watt LED bulb today sells for under a dollar wholesale, which is why you almost cannot find well-filtered products at that price point.
A Flicker Index below 0.02 is considered excellent. The Helder 2's sine wave output in Head Mode scores well here because capacitor filtering converts the PWM square signal into a gradual waveform before it reaches the LEDs. Most manufacturers do not publish Flicker Index values. The percentage is easier to cherry-pick, and both are easier to omit from a spec sheet than to explain.
What Is PWM — and Is It Actually the Worst Way to Dim an LED?
Arjen pushed back on this framing directly: "I don't really think that's the right way to ask the question. PWM is not the worst way to switch an LED. It all depends on frequency and filtering."
PWM is how LED driver chips are designed to operate. The problem is not PWM itself — it is low-frequency PWM without adequate filtering. Switch at 200 Hz with no smoothing and you get a square wave at 200 Hz at the LEDs. Switch at 30–50 kHz with appropriately sized capacitors and you get something close to a sine wave, at a frequency far beyond any perceptible or biologically relevant threshold.
Other dimming approaches — DC voltage control, digital brightness control — ultimately rely on an internal switching chip operating at 500 kHz to 1 MHz with an output inductor, making filtering trivial. The practical outcome for light quality is similar if the PWM design is properly engineered. PWM is a problem in the hands of someone optimising for cost, not quality.
What Is Constant Current Regulation and Why Does It Matter?
LEDs have an avalanche effect. As they heat up, their forward voltage drops and their effective resistance decreases — so they want to draw more current. The hotter they get, the more they pull. Without a current limiter, an LED will thermal-runaway into failure.
Every competent LED design uses a current-controlling driver chip. You set exactly how many milliamps the LED is allowed to draw, and the driver adjusts voltage to enforce that regardless of temperature.
The Helder 2 extends this with a temperature sensor on the board feeding back to the microcontroller. Above 45°C the fan speeds up. Above 55°C it speeds up again. If temperature still climbs beyond the safe threshold, the device shuts itself off automatically. This is adaptive thermal management — not a fixed fan-always-on condition, which would be noisy and accelerate fan wear.
For red and NIR LEDs specifically — no phosphor layer — if the die temperature is kept around 60–70°C, rated lifespan can reach 30,000 hours. Controlling fan speed also reduces noise in most ambient conditions; the fan only runs hard when cooling is actually needed. Proper thermal management is essentially free longevity.
How Do Capacitors Physically Smooth the Current?
A capacitor is best understood as a very fast, very small reservoir — a bucket with an inlet and a hole in the bottom.
Imagine filling that bucket with a cup rather than a continuous stream — scooping water in, dumping it, scooping, dumping. That is like PWM at 50% duty cycle. But because the bucket holds water between scoops, the flow out of the hole at the bottom is smooth and continuous — the buffer absorbs the pulsing input and delivers steady output.
Capacitors do exactly this for electrical current. When the driver pulses on, the capacitor charges. When the driver pulses off, the capacitor discharges — filling the electrical gap. At 50 kHz, the pulses are short enough that even small capacitors in the microfarad range adequately fill each trough.
The Helder 2 uses capacitors on the main driver board and again directly on the LED board per LED string. This two-stage approach smooths the signal at the driver level and once more immediately before the LEDs receive it — converting the PWM square wave into the sine wave visible on the oscilloscope. Not a theoretical claim; a measured one.
What Is Ripple Current and How Does It Relate to Flicker?
Ripple current is exactly what the name describes: instead of a flat DC line, the current fluctuates — a wave visible on an oscilloscope as an undulation rather than a straight horizontal line.
For LEDs, ripple current creates two distinct problems. First, it damages the LED. Each current peak is a mechanical and thermal stress event on the die. Cheap products frequently exceed the LED's rated peak current, accelerating wear. Second, it affects the user. Ripple in current means ripple in light output — optical flicker. Below approximately 200 Hz, this sits in a range where human physiology responds, even if the flicker is not consciously visible.
Minimising ripple current is simultaneously good engineering and good product design. It extends LED lifespan and reduces unnecessary biological load on the user — two benefits from one design decision.
Does Flicker Get Worse When Running the Device at Lower Power?
Yes. This is predictable from first principles.
At full power the driver is effectively fully on — minimal switching modulation, minimal ripple. As you reduce power by increasing the off-time proportion, modulation depth increases. At exactly 50% power — equal on and off intervals — you have the highest possible flicker percentage for any given switching frequency. This is the theoretical worst case.
Head Mode on the Helder 2 runs at approximately 50% power. However: switching at 50 kHz with capacitors sized for that frequency, the result remains a sine wave at a frequency far beyond any perceivable or measurable biological effect.
If you want to evaluate the actual worst-case flicker output of any LED device, always set it to 50% brightness before measuring. Full power can look clean on any device. The 50% point is where insufficient engineering reveals itself.
Is There a Flicker Frequency at Which Biology Simply Does Not Respond?
The current research consensus is approximately 3,000 Hz. Above that threshold, most people show no measurable biological or perceptual response to flicker. The Helder 2 switches at 50 kHz — more than sixteen times above that boundary. At those frequencies, the cells of the eye and skin simply cannot respond to individual pulses.
Below 120 Hz, flicker is clearly problematic for most people. Between 120 Hz and 3,000 Hz is a grey zone where individual sensitivity varies significantly. Migraine sufferers can respond to frequencies well above what the general population notices. Arjen notes that 50–60 Hz flicker — still found in some older neon lighting — is for him personally "pretty much unbearable."
One important clarification: the visible stepping that occurs when a device transitions between discrete brightness levels is not flicker. That is a firmware step-size issue — how large each power increment is — not a switching frequency problem. The two should not be conflated.
Can Flicker Interact with the Photobiomodulation Effect Itself?
This is the most genuinely unsettled question in the flicker section.
The body is full of oscillatory systems — brain wave frequencies (EEG), heart rhythm variability (HRV), digestive motility, circadian oscillators. We also know this from music: certain frequencies and sounds affect how we feel. The research question is whether those same principles extend to light frequencies delivered to tissue.
Some research suggests yes. PEMF at 16 Hz has shown bone healing effects. The mechanism is unknown, but the effect has been replicated. The 40 Hz frequency appears in Alzheimer's-adjacent research. Pulsed delivery has shown advantages over continuous wave in some dental applications. 10 Hz, 40 Hz, 16 Hz — these are all frequencies that have shown measurable effects in humans, in the brain and for healing purposes.
The distinction Arjen draws clearly: intentional pulsing at a researched frequency with a specific therapeutic target is a design choice. Uncontrolled flicker is noise. One is designed to produce an effect; the other might occasionally do so by accident. The engineering position: remove uncontrolled flicker first. If the evidence on targeted pulsing matures to the point of practical protocols, it can be added deliberately to a clean platform. You cannot add precision to a device that already has undefined noise in its output.
How Should Flicker Be Measured — and Why Do Consumer Electrosmog Meters Get It Wrong?
Consumer electrosmog meters are designed to detect electromagnetic field strength. When pointed at a red light device, they produce a reading — but that reading is an EMF measurement being misapplied to a different phenomenon. It does not measure optical flicker at all. The number is not meaningless; it is simply answering a different question than the one being asked.
To measure optical flicker properly, you need a photodiode or solar panel connected to an oscilloscope. The sensor converts light variation to voltage; the oscilloscope displays that signal over time. From the trace you read frequency, amplitude, and waveform shape. That is a real measurement.
For a rough functional check without equipment: connect a small solar panel to a speaker through a capacitor. If the device is flickering at an audible frequency, you will hear it. Not calibrated, but immediately informative.
Smartphone cameras in high-frame-rate mode can detect low-frequency flicker visually — useful for a quick check. For anything quantitative, an oscilloscope is the correct tool. Consumer meters with colour-coded danger indicators and no stated units, no frequency range, and no reference baseline are not producing data. They are producing anxiety.
Measurement hierarchy: solar panel connected to an oscilloscope is a valid engineering test. A calibrated photodiode with data logging is research grade. A consumer electrosmog meter pointed at a light source is not a flicker measurement.
Why Don't More Companies Publish Flicker Specifications?
Because there is no regulation requiring them to, and silence is the default when a number is unflattering.
There are no enforced standards specific to red light therapy devices. Lighting products have some industry benchmarks, but therapeutic devices operate in a regulatory gap. If a manufacturer skips filtering components to save cost, most users will not notice, and there is nothing requiring disclosure.
Arjen's position: publishing specifications is a basic design practice, not a marketing strategy. If a device is designed well, the numbers are worth publishing. If a device is not designed well, the numbers explain why they are not published.
What Types of EMF Does a Red Light Device Actually Emit?
The dominant EMF source in any LED therapy device is the driver chip — the switching component that controls current. These chips switch hard and fast by design, because high-frequency switching is the most thermally efficient method. That switching generates EMF.
PCB layout is the primary variable in containing that field. How components are positioned relative to each other, how close inductors are to driver chips, what the ground plane looks like around switching components — all of these determine how much of the switching field escapes the board rather than being absorbed internally.
Two frequency categories are relevant. Low-frequency EMF follows the PWM switching rate — at 50 kHz switching, capacitors suppress this effectively and it does not propagate significantly. High-frequency EMF from switch-mode components can extend into the hundreds of megahertz. The general engineering benchmark is to keep radiated emissions below 30 MHz, where standard EMC compliance frameworks begin to apply. Arjen checks this with a spectrum analyser; the signal from the Helder 2 is low, attributed to the board layout choices.
What Is the Difference Between Electric Field and Magnetic Field EMF — and Which Matters at Treatment Distance?
Electric fields are produced by the presence of voltage. Magnetic fields are produced by current flowing. Both drop with distance, but they behave differently.
Magnetic fields are more penetrating and more relevant at treatment distances. The standard unit for magnetic field strength is microtesla; for electric field, volts per metre. Both should be measured at the actual treatment distance — 20 cm in the case of the Helder 2 — not at the device surface, which will always read substantially higher.
On USB-C power specifically: Arjen's answer pushes back on the received wisdom. Even with AC mains power, the device still rectifies that input and runs it through an internal switch-mode supply to produce the DC the LED driver needs. The EMF profile at the LED driver — the dominant source — is determined by the driver design, not whether the upstream power came from a wall socket or a USB-C charger. The driver produces its own switching field regardless of what feeds it.
The real advantages of USB-C are practical rather than electromagnetic: universal compatibility with off-the-shelf chargers, power banks, and laptop adapters; cable replaceability; no proprietary connector dependency. These are genuine user benefits. They are not EMF benefits in the way sometimes implied.
What Is nnEMF — Non-Native EMF — and Why Do Some Wellness Practitioners Treat It Differently?
Arjen's answer here was direct and worth reproducing: he said immediately that he did not know, and committed to researching it before Part III of this series.
For context: nnEMF is a term used in integrative wellness and some biohacking communities to describe artificially generated electromagnetic frequencies, as distinct from naturally occurring EMFs such as the Schumann resonance or solar radiation. The premise is that the human body evolved in the presence of natural EMFs and may respond differently to artificial pulsed frequencies, particularly from wireless devices. The theory is not without controversy — mainstream EMC frameworks treat all EMF at equivalent frequencies as equivalent regardless of source. The nnEMF distinction reflects a precautionary biologically-informed perspective rather than a regulatory or engineering category.
Arjen's candid admission that he needed to research it further before answering is characteristic of this series. He does not claim expertise he does not have. That honesty is reproduced here intentionally.
Does the Helder 2 Have Bluetooth or Wireless Connectivity?
No. No Bluetooth, no Wi-Fi, no app. This was a deliberate design decision for three converging reasons.
Utility: the device turns on, runs its session, and turns off. There is no function that wireless connectivity would meaningfully improve. Adding a wireless chip to solve a problem that does not exist is feature bloat, not engineering.
Cost: wireless chips, the certification requirements they trigger, and app development all add cost that the user pays for. If the feature adds nothing, the cost justifies nothing.
EMF: Wi-Fi and Bluetooth broadcast continuously at 2.4 GHz or 5 GHz. The Helder 2 is a device intended for regular close-proximity use on the body. Adding a continuous GHz-range broadcaster to a device used against skin is a choice that is difficult to justify, particularly given growing (if not yet definitive) evidence that prolonged close-proximity wireless exposure is not beneficial.
There is a secondary concern that goes beyond EMF. Any device whose core functions are gated through an app is a device whose features can be retroactively restricted, moved to subscription, or disabled entirely if the company changes direction. No app means no dependency on continued company cooperation. Simple is not a limitation — it is a specification.
How Does EMF Behave with Distance from the Device?
The same physics that governs light output governs EMF propagation: the inverse square law. As distance doubles, field intensity drops by a factor of four — not two. The relationship is exponential, not linear.
Practically: at 5 cm from the device, field intensity is sixteen times higher than at 20 cm. This is one of the reasons the Helder 2's designed operating distance of 20 cm is a repeated emphasis in our materials. Distance matters twice: once for receiving the correct optical dose — the 30° lens geometry means the beam has spread appropriately by 20 cm — and once for field exposure. Every centimetre closer increases both.
Arjen used his own phone behaviour as an illustration: his phone is almost never in his pocket against his body, because a device broadcasting at close range absorbs roughly 50% of its radiation into the adjacent tissue. A metre away, that fraction drops dramatically. The same principle applies to the Helder 2 at its driver frequencies. 20 cm is the designed intent. It is not a preference; it is an engineering parameter.
What Questions Remain Open?
Four EMF questions remain pending filming: whether high EMF can interfere with the photobiomodulation effect itself; electromagnetic hypersensitivity (EHS) and its relevance for wellness-oriented users; how EMF should be properly measured on a device of this type versus what consumer meters actually report; and whether the Helder 2 has undergone or is planned for formal EMC chamber testing.
One question — nnEMF — Arjen explicitly flagged as requiring more research before he will answer it on camera. Part III will be added here on completion.
Formal EMC chamber testing is a longer-term task. When completed, the results will be published in full.
Key Specifications Referenced in This Interview
Switching frequency: approximately 50 kHz Optical flicker in Body Mode (full power): no measurable flicker on oscilloscope Optical flicker in Head Mode (50% power): clean sine wave at approximately 50 kHz Target flicker percentage: below 5% Flicker Index target: below 0.02 (IEEE PAR1789 excellent threshold) Operating distance: 20 cm — designed for both optical dose and field exposure Wavelengths: 631 nm red, 856 nm NIR (independently verified, integrating sphere) Total optical output: 10.47 W Wireless connectivity: none — no Bluetooth, no Wi-Fi, no app Thermal management: adaptive fan control via microcontroller with temperature sensor; automatic shutoff above thermal threshold LED lifespan at rated die temperature: approximately 30,000 hours
Arjen Helder is the co-founder and PCB/optical engineer at AriHelder, Chiang Mai, Thailand. Wavelengths and optical output independently verified by integrating sphere test. arihelder.com