Red Light Bed Irradiance Explained: What the Numbers Really Mean

Two red light beds sit side by side on a comparison page. One lists 100 mW/cm2. The other lists 60 mW/cm2 and costs more. The first one looks like the better buy, and it may not be. Those two numbers were almost certainly taken at different distances, with different meters, at different points on the panel. We have measured light output on salon equipment since 1990, and the single most useful thing we can tell a buyer is this: an irradiance figure without its measurement conditions is a decoration, not a specification. Here is how the numbers are produced, why they drift so far apart, and how to make two beds comparable.

The three numbers that matter

Every red light bed can be described with three quantities. Everything else on a spec sheet is derived from these or is marketing.

• Wavelength, in nanometers (nm). This is the color of the light, or for near infrared, the part of the spectrum just past what the eye sees. An LED is sold by its peak wavelength, but it emits a band around that peak, typically 20 to 30 nm wide at half its maximum output. A bed labeled 660 nm is putting out light from the mid 640s to the mid 670s.
• Irradiance, in milliwatts per square centimeter (mW/cm2). This is the rate at which optical power lands on a surface. It is the figure that changes most with how and where it was measured, and it is the figure sales pages lead with.
• Dose, in joules per square centimeter (J/cm2). This is irradiance multiplied by time. It is the number that describes a session, and it is the one the controller is really setting when it counts down minutes.

The arithmetic, once. A milliwatt is a thousandth of a watt, and a joule is one watt for one second. Take a bed that delivers 50 mW/cm2 at the body. That is 0.05 W/cm2. A 10 minute session is 600 seconds. Multiply: 0.05 times 600 equals 30 J/cm2. Change the irradiance to 25 mW/cm2 and the same session delivers 15 J/cm2. Change the session to 20 minutes and it is back to 30. Dose is the product, which is why a spec sheet that lists irradiance without a distance has not told you the dose at all.

Why irradiance claims vary so much

Five things move the number, and a seller controls all five. None of them is dishonest on its own. The problem is that listings rarely say which choices were made.

Measured at the LED surface versus at the body

The highest reading any bed will ever produce is with the meter pressed against the acrylic or the diode lens. Nobody lies there. In a bed, the body rests on the base acrylic and the canopy is some inches above it, so the relevant number is the irradiance where the person actually is. Independent testers who have put a calibrated meter at a stated working distance have found advertised figures running well above what the device delivers there, in some published comparisons by roughly double. The seller's figure was not invented. It was taken at the glass.

Distance falloff

Light spreads as it travels. For a small source, irradiance falls with the square of distance, so doubling the distance cuts it to a quarter. A large flat panel does not behave like a point source at close range, which softens the falloff, but the direction is the same and the drop over the first several inches is steep. A bed that reads 100 mW/cm2 at the surface may read 40 to 60 at a working distance a few inches away, and lower again at the far side of a body. The only fix is a number taken at a stated distance.

Peak versus average across the panel

LED arrays are not uniform. Output is strongest where diodes are densest, usually center panel, and weaker toward the edges, between arrays and around hinges and vents. A single reading at the brightest spot is a true number for one square centimeter. A buyer wants the average over the area a body occupies. Testing labs handle this by reading several points in a pattern, often center plus four or more around it, and reporting both the center value and the average with a uniformity ratio. A spec sheet that gives one figure has almost always given you the peak.

Meter type and calibration

This is the quietest source of error and the largest. There are three common instruments.

• Solar power meters are inexpensive handheld units built to read sunlight. Their sensors respond across a wide band and sum everything that lands on them, including room light, reflections and infrared heat from the panel. On narrow band LED light they tend to read high. Several independent reviewers have reported solar meter figures 30 to 70 percent above what a spectroradiometer records on the same panel, and most manufacturer figures are produced with this class of meter.
• Photodiode meters are fast and precise at low power, but a photodiode's sensitivity changes with wavelength. The meter is only accurate if it has been calibrated at the wavelength being measured and the user sets it to that wavelength. Set to the wrong band, it reads confidently and wrong.
• Thermopile meters absorb the light as heat and respond nearly evenly across the spectrum. They are slower and less sensitive at low power, but they do not care what color the light is, which makes them a fair instrument for mixed red and near infrared output.

The reference instrument is a spectroradiometer, which separates the light by wavelength and reports irradiance for each band. That is how an accredited lab produces a report, and it is why "lab tested" figures from two beds are comparable in a way two solar meter figures never will be. Whatever the instrument, it has a calibration date, and a meter that has not been calibrated against a traceable standard is a guess with a display.

Pulsed versus continuous

Some beds drive the LEDs continuously. Others pulse them, switching on and off many times a second. A pulsed array has a peak irradiance while the diodes are on and a lower average over the whole cycle. If the duty cycle is 50 percent, the average is half the peak. A listing that quotes the peak of a pulsed bed next to the continuous figure of another bed is comparing two different quantities. Dose over a session follows the average, not the peak, so the average is the number to ask for.

Wavelengths you will see on spec sheets

Manufacturers commonly list red in the range of roughly 630 to 670 nm, with 630, 633, 660 and 670 nm the peaks that appear most often, and near infrared in the range of roughly 800 to 880 nm, where 810, 830 and 850 nm are the usual peaks. Many beds combine two to five of these in one array, and the spec sheet should say how the diodes are divided among them. A bed with 20,000 LEDs split five ways has 4,000 at each wavelength, and the irradiance at any one wavelength is a fraction of the total. Near infrared is invisible to the eye, so a canopy that looks dimmer than another is not necessarily putting out less light. Only a wavelength-resolved measurement tells you how much is red and how much is near infrared. We describe these as the ranges in common use and nothing more; the choice of wavelength is between the buyer and the manufacturer.

Watts in is not watts out

The wattage figure is the easiest number to find and the easiest to misread. Full-body red light beds are commonly listed at 1,500 to 3,000 watts, and the first question is which watts. Some listings quote the measured draw at the wall. Others quote the sum of the maximum ratings of every LED on the board, which is a larger figure than the array ever draws, because LEDs in a bed are driven below their rated current for life and heat. The data plate on the bed gives the electrical rating, and our electrician spec sheet tool turns that plate into a circuit spec. Treat the marketing wattage as a separate number until you know which kind it is.

Whatever enters the bed as electricity leaves as light or heat. Red LEDs are among the more efficient diodes made; manufacturer datasheets for high-output 660 nm parts list wall-plug efficiencies in the range of 60 to 70 percent at their rated test conditions, and the figure falls as the diode warms and as it is driven harder. Near infrared parts have their own curves. Then the power supplies and drivers take a cut before the current reaches the board. In round numbers, a bed drawing 2,000 watts is producing well under half that as light and the rest as heat that has to go somewhere. That is what the fans are for.

Cooling is an output question, not a comfort question. Published datasheets put the output loss from a diode warming from 25 to 85 degrees C at roughly 10 percent, and red diodes lose more to heat than blue ones. A bed with weak airflow, a blocked intake or a failed fan runs its diodes hotter, so the irradiance measured in the first minute of a session is higher than the irradiance in the fifteenth. Two beds with identical cold readings can deliver different doses over a session because one holds its temperature and the other does not. Over months, heat also shortens LED life, and the stated tens of thousands of hours on a spec sheet assume the diode was kept at its design temperature.

How to compare two beds honestly

You do not need a lab. You need the same five facts from both sellers, in writing.

1. Irradiance at a stated distance, with the meter named. Ask for mW/cm2 at a specific distance from the acrylic, the make and model of the instrument, and whether the figure is a center reading or an average. If the answer is a single number with no distance, ask again.
2. Panel coverage area. The illuminated length and width of base and canopy, in inches. A high irradiance over a small area and a modest irradiance over a full body are different products.
3. LED count and type. How many diodes, their rated power, the wavelength split, and whether they are single-chip or multi-chip packages. Two listings that both say 20,000 LEDs can differ by a factor of several in actual output if the diode packages differ.
4. Continuous or pulsed, and the average if pulsed. Along with the frequency and duty cycle the controller uses in its standard programs.
5. Cooling specification and uptime. Number of fans, airflow rating, clearance the manufacturer requires around the bed, and whether the unit has a duty cycle limit between sessions. A commercial bed that must rest between sessions is a scheduling fact a salon needs before it builds a price list.

Once you have both sets, convert each to irradiance at the same distance and average over the same area. If a seller will not or cannot provide the figures, that is itself information. Our red light equipment page covers the beds and hybrids we sell and service, and the installation cost guide covers what the circuit, the route in and the startup involve.

What a flat meter misses

There is a gap underneath all of this that sales pages never mention. Every meter reads a flat plane. A body is not flat. It is a curved shape lying on one surface with another surface some inches above, and the light reaching it arrives at angles that change from the chest to the sides to the far edge of a shoulder. A reading taken by holding a flat sensor square to the panel describes the plane the sensor was in. It does not describe what the whole surface of a person receives.

This is a problem we have worked on in print, in a different light source. In 2010 Stan Pope and Dianne Godar published "Solar UV geometric conversion factors: horizontal plane to cylinder model" in Photochemistry and Photobiology (PubMed 20059727). The paper took the standard way of measuring sunlight, a meter facing straight up on a flat horizontal plane, and worked out the geometric factors needed to convert those readings into the average irradiance over a cylinder, the simplest shape that stands in for a human body, and a half cylinder for the face, shoulders and tops of hands and feet. The flat plane and the body did not agree. The study's conclusion was that people receive less than half the dose on their bodies that the flat meter reports. An earlier paper, Pope, Holick, Mackin and Godar in the same journal in 2008 (PubMed 18513232), dealt with a related measurement issue: the same sunlight produces different dose numbers depending on which weighting function the meter or the analysis applies, and body geometry has to be accounted for before the numbers are useful.

We want to be exact about what those papers are and are not. They are about sunlight and ultraviolet, measured outdoors, and they say nothing about red or near infrared LEDs or about what any light does to a person. We cite them for the measurement logic alone, because it transfers directly. A flat sensor reading is a starting point that still has to be converted to the geometry of the thing receiving the light, and the conversion is not small. A red light bed spec sheet gives you the flat-plane number at best. What a body in that bed receives is lower and uneven, and no published bed figure we have seen accounts for it. That is not a reason to distrust the beds. It is a reason to insist on measurement conditions, and to treat irradiance as the beginning of the comparison rather than the end.

What we check at installation

A bed that has been crated, trucked and reassembled is not the bed that left the factory until someone confirms it. Our startup on a red light bed installation covers the output side as well as the electrical side.

• Panel output consistency. Every array lights at full output with no dark segments, dim strings or color shift between panels. A string that reads visibly different from its neighbors on a mixed array is checked before the bed goes into service.
• Fan function. Every fan spins, pulls air at the intake and pushes it at the exhaust, with the clearances the manufacturer requires. A fan that runs but is blocked is as bad as one that does not run.
• Thermal behavior over a full session. We run the bed for a full program, not a two minute test, and watch canopy and base temperatures settle. Output that is stable from the first minute to the last is the goal. A bed that climbs through the session has a cooling problem that will show up as a repair later.
• Controller settings. Programs, session lengths, any pulse modes and lockouts are set to the manufacturer's specification and the owner's instructions, and we record what we set so the salon has a baseline.

On service calls we repeat the same checks, because the usual complaint, that a bed "does not feel as strong as it used to," almost always traces to a cooling fault, a failed driver or a dimmed array, all of which show up on inspection. Stan's background is on the about page, and the measurement habits there are the ones we bring to every bed.

Related guides

• Red light bed installation for homes and salons
• Red light bed installation cost: circuit, delivery, assembly and startup
• Red light beds and hybrid equipment we sell and service
• Electrician spec sheet: from the data plate to a circuit spec