Desk Lighting for Screen Work: Glare, CRI, and Lux Targets

Desk lighting for screen work has to do two contradictory things at once: put enough light on your desk, keyboard and paperwork to read comfortably, while putting as little as possible on the front of your display. Most bad setups treat lighting as a single brightness dial. It is really three layers with three different jobs, and the fixture that solves one makes another worse if you aim it wrong.

How bright should a desk actually be?

The clearest published guidance for computer work comes from OSHA’s Computer Workstations eTool, which puts office lighting at 20 to 50 foot-candles, with a minimum of 30 for construction-industry employers, and notes that where LCD monitors are in use the same viewing tasks usually need more, up to 73 foot-candles.

US facility documents still use foot-candles; meters and product listings use lux. One foot-candle is about 10.76 lux:

Guidance	Foot-candles	Approx. lux
General office work	20 to 50	215 to 540
Rooms where LCD monitors are in use	up to 73	up to ~785

Measure illuminance at the work surface, sensor flat where your hands and documents are, not up near the fixture. Phone light-meter apps are uncalibrated, so treat their readings as indicative rather than accurate; they still help compare one spot on the desk against another. Land around 300 to 500 lux for mixed screen and paper work and you are inside the documented range with headroom either way. The desk lux calculator turns a task into a target number and a proportional fixture length if you would rather start from the task than from the standard.

For sensor placement, a repeatable sampling pattern and a before-and-after check, follow how to measure lux on your own desk.

Glare is a geometry problem, not a brightness problem

You can hit a perfect lux number and still hate your desk. Direct glare is a bright source inside your field of view: a bare bulb, a window, an undiffused lamp head. Reflected glare, or veiling reflection, is that source bouncing off your screen into your eyes. The second is what ruins screen work, because it drops the effective contrast of everything on the panel.

The governing rule is the mirror test: sit in your normal working position and treat the screen as a mirror. Anything you can see in it is a glare source. Then fix the geometry, which is most of what OSHA’s guidance addresses:

• Put the display face at right angles to windows and other bright sources, never facing one and never with one behind you.
• Run rows of overhead lights parallel to your line of sight, so fixtures sit beside your view rather than in it.
• Tilt the monitor slightly downward so ceiling fixtures reflect toward the floor instead of your eyes.
• Prefer diffuse, well-distributed light, and many low-output sources over a few bright ones.
• Use matte, medium-toned finishes on walls and desk. A white gloss desk under a bright lamp becomes a second source aimed at your face.
• Keep the screen clean. Fingerprint film turns a sharp reflection into a haze.

Panel finish changes the symptom, not the cause: a glossy display gives a sharp reflection you can often kill by moving one lamp, while a matte coating spreads it into a low-contrast veil repositioning cannot fully clear.

Why a monitor light bar solves what a desk lamp cannot

A monitor light bar clamps to the top bezel and throws light forward and down onto the desk in front of the display. Two properties make it a different tool, not a small lamp. Asymmetric optics: the reflector is cut so almost no light travels back onto the panel, whereas a conventional lamp mounted high enough to cover the desk spills onto the screen. Zero desk footprint: the bar uses space above the display, while a clamp lamp eats desk depth or needs an arm.

The trade-offs are real:

• Bezel fit. Counterweight designs assume a flat top bezel in a certain thickness range. Very thin bezels and heavily curved ultrawides leave the bar rocking or angled wrong.
• Top-of-screen wash. Angle the head too far back and you defeat the asymmetric optics, painting a bright band across the top of your display.
• Monitor arm interaction. On an arm, the extra mass sits at the worst lever point. Check rated load and retension the tilt joint.
• Power. Most bars run on USB. Fed from the monitor’s downstream port, the bar usually sleeps when the monitor does, which is either convenient or infuriating.
• Coverage width. A bar lights a pool in front of the display, so it will not cover a wide desk on its own.

Which of those trade-offs you can live with is a spec-sheet question, worked through in the monitor light bar buying guide. If a bar is already fitted and misbehaving, the same failures are diagnosed symptom by symptom in monitor light bar problems.

Bias lighting is a third layer, not a brighter one

Bias lighting sits behind the display and lights the wall, not the desk. It raises surround luminance so your eyes are not adapting to a bright rectangle in a dark room, a mismatch that contributes to evening eye strain and makes dark scenes look worse than they are.

Broadcast formalised this long before it became a desk accessory. Recommendation ITU-R BT.2035 sets a reference viewing environment for assessing programme material: the background behind the monitor is D65 in chromaticity, and its luminance sits at roughly 10 percent of the display’s reference white. That is an evaluation spec, not a desk standard, but both numbers translate:

1. Neutral white point. Aim near D65, sold as 6500 K. A 2700 K strip behind a 6500 K panel makes the screen read cold and blue, the exact shift you were trying to remove.
2. Dim, not bright. The wall should read as a soft glow beside the screen, not a second light source. If it pulls your eye, turn it down.
3. Colour fidelity still matters, because the light bounces off a coloured wall before reaching you. A source with weak red output tints that bounce.

Aim the emitters at the wall and inset the strip an inch or two from the panel edges, for a soft halo instead of visible hot spots. Because these two fixtures get shelved together and recommended interchangeably, bias lighting vs monitor light bar sets out which one actually solves which complaint.

CRI, R9, and what the numbers are worth

Colour rendering index scores how faithfully a source reproduces object colours on a 0 to 100 scale. Per the Department of Energy’s LED Basics, a minimum CRI of 80 is recommended for interior lighting and 90 or higher indicates excellent colour fidelity, with efficiency trade-offs as fidelity climbs.

The catch: general CRI, written Ra, averages eight moderate-saturation samples and says almost nothing about deep saturated red, which is sample R9. A lamp can post a respectable Ra in the 80s while rendering red badly. The ENERGY STAR Lamps specification closed that loophole by pairing the two: an LED lamp had to hit Ra 80 and an R9 above zero. Past tense matters, because EPA sunset the lamps specifications at the end of 2024, so bulbs no longer carry the mark. The pairing survives as a buying rule, not a certification.

For general desk work, Ra 80 or better is fine. For colour-sensitive work, look for Ra 90-plus with a published R9 figure; a product advertising a big CRI number that will not publish R9 is telling you something. The IES TM-30 method, which reports a fidelity index Rf alongside a gamut index Rg, describes colour behaviour more completely and turns up on higher-end spec sheets.

Colour temperature and flicker

Correlated colour temperature describes the light’s own appearance, from warm at 2700 K to 3000 K, through neutral around 4000 K, to cool daylight at 5000 K to 6500 K. Do not mix widely different CCTs in one field of view, or white paper looks yellow beside a white screen. If the fixture is tunable, let time of day drive it: cooler by day, warmer at night. For one fixed value on a mixed-use desk, 4000 K is the safest compromise.

Dimming method matters too. Constant-current dimming lowers drive current and gives steady output. PWM switches the LED fully on and off, varying duty cycle; cheap low-frequency PWM causes flicker many people never consciously see but register as fatigue. IEEE Std 1789-2015 sets out recommended modulation frequencies and depths for LEDs, so this is a documented comfort issue rather than a niche complaint. Quick test: set the fixture low and wave a pencil in front of it, looking for a stroboscopic multiple-image effect, or film it in phone slow motion and look for rolling bands. A good fixture shows neither at its lowest setting.

Putting the three layers together

• Ambient: diffuse ceiling or indirect wall light, positioned so nothing appears in the screen’s mirror image, contributing part of that 300 to 500 lux.
• Task: a light bar or shielded asymmetric lamp filling the desk in front of the display, dimmable so the evening setting sits far below the daytime one.
• Bias: a dim, roughly 6500 K, high-fidelity source washing the wall behind the panel.

Add them in that order and repeat the mirror test after each. The most common mistake is adding a brighter task light to fix glare: it adds a new reflection source and leaves the original geometry untouched.

For the rest of the workstation: display positioning at monitorarmguide.com, workspace setup at homedeskguide.com, gear comparisons at ergoranker.com.