A solar garden light has a brightness ceiling that was set in a factory, by one component, before anyone put it in a box. You can restore brightness the fixture has lost — and most dim lights have lost a lot — but you cannot add brightness it never had, unless you open it and change a part. Knowing which of those two problems you have saves an afternoon.
One component decides how bright the light is
Open a $15 stake light and there is almost nothing inside: a solar panel, one AA-size Ni-MH cell, an LED, a small coil, and a four-pin part marked YX8018, QX5252 or 5252F. That four-pin part is a boost driver — a Joule Thief on a chip. It has to lift 1.2 volts from the battery up to the 3 volts a white LED needs, and it uses the coil, an inductor, to do it.
Analog Devices’ teardown of one of these describes the YX8018 as “more or less just a gated oscillator which runs at approximately 200 KHz driving an open drain NMOS switch.” The chip switches the inductor on and off. How much current reaches the LED depends on how much current the inductor builds up before the switch opens — which is set by the inductance value. Nothing else in the circuit regulates it. There is no resistor, no feedback loop, no dimmer.
So the inductor is the brightness setting. Here is what people who have measured these circuits report:
| Inductor | LED current | Source |
|---|---|---|
| 560 µH (factory part) | well under 1 mA in the unit tested | dzrmo teardown |
| 390 µH (marked 330 µH) | ~5.0 mA at 1.25 V; ~4 mA at 0.9 V | Mamezou bench test |
| 330 µH | 6.3 mA measured at 1.5 V | EEVblog thread |
| ~70 µH | ~20 mA | Mamezou, calculated |
| 33 µH | 22–26 mA | dzrmo, after modification |
Read the second row again, because it is the most useful number in this article. Across the entire usable voltage range of a Ni-MH cell — 1.25 V fully charged down to 0.9 V nearly flat — LED current moved from about 5 mA to about 4 mA. That is roughly a 20% change in light output across the whole battery. A brand new cell does not make a working light dramatically brighter. It makes it stay lit longer.
Restore first, in this order
If the light was brighter last year than it is now, something is recoverable. Work down this list; it is ordered by payoff, not by effort.
1. Clean the panel, and treat shading as a separate problem
Warm water, a drop of dish soap, a microfiber cloth, rinse with clean water. That is the whole procedure. Our guide to cleaning garden light panels covers the details.
What matters more is the difference between dirty and partly covered. A uniform film of dust costs you roughly what it blocks. A leaf, a bird dropping, or a shadow across one corner costs far more than its area, because the four cells in a garden light panel are wired in series with no bypass diodes: the weakest cell throttles the whole string. In a study of larger PV modules, soiling covering just 0.5% of the module surface in one corner produced a 9% power loss. On a 2-inch panel with four cells, a single sycamore leaf can take out a quarter of the array. Move the light before you scrub it.
2. Work out whether the panel is dirty, crazed or yellowed
These are three different faults with three different answers, and most advice on the internet runs them together. Rub a wet finger across the panel and watch what happens.
- The cloudiness disappears while wet and comes back as it dries. The surface is crazed: years of UV have opened microscopic cracks in the clear resin over the cells, and each crack scatters light back out before it reaches silicon. The water is temporarily filling those cracks, which is why the haze vanishes — and that is also the fix.
- The panel is amber or brown all the way through and water changes nothing. The encapsulant has yellowed in bulk. No coating reaches that, and adding one only puts more absorbing material in front of the cell. Replace the panel or the light.
- It wipes off. It was dirt. You are done.
For a crazed surface, a thin coat of water-clear UV-stable acrylic lacquer or clear spray varnish works by refractive index matching: it fills the cracks with something optically close to the original resin, so light passes straight through instead of scattering. Hobbyists have been doing this since at least 2012 and report cloudy panels going clear again. Expect one to three seasons before it needs redoing.
Not clear nail polish, which is the version of this the forums recommend. The principle is right and the bottle is wrong, twice over: nail polish is nitrocellulose, which yellows on its own, and it is carried in ethyl and butyl acetate — esters, which sit on Covestro’s list of chemical classes that damage polycarbonate. You would be filling the cracks with something that clouds and attacking the housing while you do it.
Two honest caveats. Any coating you add absorbs some light of its own, mostly at the blue and violet end — so the gain is the scatter you remove minus the absorption you add, and on a panel that is only lightly hazed that trade can come out flat. And tiny garden panels are sometimes amorphous silicon, which does most of its work in exactly that blue-green band, so it loses more to a coating than a crystalline cell would. Do one light, leave the rest alone, and compare them for a week before you commit to a set of twelve.
Wet-sanding with 800–1000 grit and polishing is a different job. It works on a lens — the clear dome over the LED, usually polycarbonate or acrylic — where there is material to spare. On the panel it is risky, because the hazed layer is often only a few thousandths of an inch thick and the cells are directly underneath. Our article on yellowed plastic on solar lights goes further into which plastics respond to what, and the cleaning guide covers what should never touch either part.
Do not use ammonia-based glass cleaner on polycarbonate. Ammonia attacks polycarbonate by aminolysis — it chemically breaks the polymer chains whether or not the part is under stress. This is often reported as “stress cracking,” which is the right warning for the wrong reason. Skip WD-40 too: it leaves an oily film that collects pollen within a week.
3. Replace the battery — for runtime, not brightness
A dead or badly degraded Ni-MH cell is a real fault and worth fixing. Most of these lights use one AA at 1.2 V, 300–700 mAh. Replace like for like with a decent low-self-discharge Ni-MH. What you will get back is a full night instead of two hours; what you will not get back is more light, for the reason in the table above. Higher capacity does not raise LED current either, which is the question we answer in full in can I use a higher mAh battery in solar lights.
Never put alkaline cells in a solar light. The panel will try to charge them.
4. Check the contacts
Battery springs and terminals in outdoor fixtures corrode, and corrosion adds resistance exactly where the circuit can least afford it — a boost driver starting from 1.2 V has no headroom to spare. Scrape the terminals bright with a fiberglass pen or fine abrasive, and check the two panel wires where they enter the housing. Intermittent or no operation rather than dimness usually points somewhere else; see solar lights not working and how to fix a solar light sensor.
5. Move the light
These panels are horizontal and cannot be aimed, which is the worst orientation for both harvest and self-cleaning. Four hours of genuine direct sun beats eight hours of dappled shade by more than most people expect, because a panel in shade is not producing a fraction of its rating — it is producing almost nothing. Do solar lights need direct sunlight to charge has the detail.
The modification route, with the real numbers
You can change the inductor. It is the only change that actually raises brightness, and it costs you runtime at close to a one-for-one rate.
Take a typical stake light: a 2 V, 80 mA panel — about 0.16 watts — and one 600 mAh AA. Here is what that budget buys, assuming a white LED at 3.0 V and around 85% converter efficiency:
| Inductor | LED current | Draw from battery | Drawn over a 10-hour night | Sustainable on a 0.16 W panel? |
|---|---|---|---|---|
| 330–390 µH (stock) | 5–6 mA | ~15–19 mA | 150–190 mAh | Yes — this is what it was designed for |
| ~70 µH | ~20 mA | ~59 mA | 590 mAh | No — one full cell, one night, no margin |
| 33 µH | 22–26 mA | ~65–76 mA | 650–760 mAh | No — exceeds the cell’s entire capacity |
A 2 V, 80 mA panel lying flat in a yard returns somewhere around 150–280 mAh on a good summer day, and much less in November. The factory inductor is not a cost-cutting choice. It is the value that makes the night’s draw match a realistic day’s harvest. Quadruple the current and the light runs bright for about two hours and then the low-voltage cutoff shuts it down — which is exactly the pattern people describe when they buy “bright” no-name lights.
If you still want to do it, three warnings that matter:
- The inductor’s current rating. A low-value coil pulls a much higher peak current. The tiny drum inductor in a stake light is not rated for it and will saturate, which turns your efficiency gain into heat.
- The LED. Cheap 5 mm LEDs in these fixtures are nominally 20 mA parts with no heatsink and no potting. Running one at 25 mA in a sealed plastic dome in an Arizona summer shortens its life considerably. If you are changing the inductor, change the LED to a modern high-efficacy part at the same time — that is where the actual brightness gain is.
- Adding LEDs does nothing. The driver is a current source. Two LEDs in parallel on the same driver each get half the current. You get the same total light from two dimmer points.
If the circuit interests you more than the yard does, building a solar light from scratch is a better use of the evening than fighting a $15 enclosure.
Why the lumen figure on the box is usually fiction
Listings for single-LED stake lights routinely claim 7 to 10 lumens with an 8-hour runtime. Run the energy: a small LED of this type manages roughly 40–80 lumens per watt. Ten lumens therefore needs something like 0.13–0.25 W at the LED, for 8 hours — call it 1.0 to 2.0 watt-hours. A 600 mAh, 1.2 V Ni-MH cell holds 0.72 watt-hours in total, before boost converter losses.
The claim is arithmetically impossible. Either the light is not running all night at that level, or the number was measured for an instant at full charge, or it was never measured at all. Treat single-cell stake lights as one-to-three lumen devices and you will be right far more often than the spec sheet is. This is also why “brighter battery” advice keeps disappointing people — see how long do solar lights last for how the same budget governs lifespan.
Perceived brightness is a different lever, and it’s cheaper
Lumens measure output. What you actually notice at night is luminance in your field of view, and at these levels your eyes are not working the way a light meter does.
Garden lighting sits in the mesopic range — average scene luminance roughly 0.005 to 5 cd/m² — where both rods and cones contribute. Rods peak at 507 nm, cones at 555 nm, so as light falls your spectral sensitivity shifts toward blue. Lighting engineers handle this with the S/P ratio: warm white LEDs sit around 1.2, cool white around 2.0, and under CIE 191:2010 a higher S/P ratio earns a more favourable correction at low luminance. Per lumen, a cool white light genuinely does read brighter in a dark yard.
That is not the whole story, and it is not usually the deciding factor:
- Aim beats output, by a lot. A fixture that throws light sideways into your eyes is a glare source; the same fixture shielded and aimed at the path puts every lumen on a surface you can see. A well-aimed 8-lumen light is more useful than a badly aimed 15-lumen one, and lower glare means your eyes stay dark-adapted.
- Beam pattern. A clear dome that scatters in all directions and a faceted lens that concentrates downward can contain the same LED. One lights the path; one lights the sky.
- Contrast. Warm light on brick, cedar, flagstone and mulch renders those surfaces with more contrast than cool white does, which is why warm often looks better even though cool measures brighter.
Practical translation: shield, aim down, raise the fixture, and put light on a surface. Clustering three fixtures at a corner and a doorway usually reads better than spacing nine of them evenly down a path.
When the honest answer is “this one can’t be fixed”
Stop working on a light when all of the following are true: the panel is clean and clear, the battery is new, the contacts are bright, it gets four hours of direct sun, and it is still dim. At that point the fixture is doing what it was built to do, and the ceiling is the panel.
The number to shop by is panel area, not advertised lumens. A fixture with a separate panel on a 6–10 foot lead, mounted where it can be aimed at the sun while the light points at the ground, will beat any integrated stake light of the same price, because it collects several times the energy. That is the whole difference between decorative and functional. Our solar spot lights and solar wall lights are built that way; pathway lights and step lights are the right tool when the job is marking an edge rather than illuminating it. If you want a comparison across types first, the best solar spot lights and solar uplights for trees cover what each format can realistically do.
And if the light works but stays on when it shouldn’t, or won’t come on at all, that is a sensor or switch problem rather than a brightness problem — the on/off switch on solar lights explains what that switch is actually for.
Sources
- Analog Devices StudentZone - Hacking an LED solar garden light ez.analog.com The YX8018 as a gated oscillator running at about 200 kHz driving an open-drain NMOS switch, and that the datasheet carries an inductor-value versus output-current table
- Mamezou Developer Site - Trying out the solar LED driver YX8018 developer.mamezou-tech.com Measured inductor of about 390 microhenries; LED current of 5.0 mA at 1.25 V falling to 4 mA at 0.9 V; the calculation that 70 microhenries would yield 20 mA
- dzrmo - Garden Light (updated) dzrmo.wordpress.com A 560 microhenry factory inductor, 200 kHz operation, and 22-26 mA after dropping to 33 microhenries; also that the driver senses darkness through the panel
- EEVblog - How to calculate the current in a PWM LED driver circuit eevblog.com 6.3 mA measured at 330 microhenries and 1.5 V input, and that the topology is open-loop with no current feedback
- antalife - Research: behavior of QX5252F antalife.com Measured daily converter efficiency of 86.9%, about 133 kHz during LED operation, and that the IC has no maximum-power-point tracking
- Bessa, Micheli, Almonacid & Fernandez - Monitoring photovoltaic soiling (Heliyon, 2021) pmc.ncbi.nlm.nih.gov That soiling covering 0.5% of a module's surface in one corner caused a 9% power loss - non-uniform soiling costs disproportionately more than uniform dust
- All Things Lighting - Understanding mesopic photometry allthingslighting.org The 0.005-5.0 cd/m2 mesopic range, the 555 nm photopic and 507 nm scotopic peaks, and S/P ratios of about 1.2 for warm white and 2.0 for cool white LEDs
- POLAB - Mesopic photometry and night vision polab.se The Purkinje shift, and that under CIE 191:2010 a higher S/P source receives a more favourable mesopic correction at the same photopic level
Questions people ask
Why are my solar lights not as bright as usual?
Almost always the panel, not the LED. A clear coating that has hazed with UV, a film of dust, or a shadow across one corner all cut charging current sharply - and because the cells in a tiny panel are wired in series with no bypass diodes, partial shading costs far more than its area suggests. Check the panel first, then the battery, then the contacts. If it was never bright, cleaning will not help: LED current is fixed by a component inside.
Why are my solar lights dull?
Either they are discharged early or they were always dull. A light that starts bright and fades within two hours has a worn battery or a panel that is not harvesting enough during the day. A light that is evenly dull all night is working as designed - the boost driver holds LED current near-constant at roughly 5-8 mA, and the whole usable voltage range of the battery only shifts that by about 20%. Dull all night is a design limit, not a fault.
How do you rejuvenate solar garden lights?
Four steps, in order. Wash the panel with warm water, a drop of dish soap and a microfiber cloth. If haze returns as the panel dries, the surface is crazed - a thin coat of clear lacquer fills the micro-cracks. Replace the Ni-MH cell with a like-for-like low-self-discharge one. Scrape corrosion off the battery terminals. Then move the light somewhere it gets four hours of genuine direct sun.
Does WD-40 clean solar lights?
No, skip it. WD-40 is a petroleum-based penetrant, not a cleaner, and the film it leaves behind attracts dust and pollen within days, so the panel ends up dirtier than it started. It also does nothing for UV hazing, which is the usual cause of a cloudy panel. Warm water and a drop of dish soap removes everything WD-40 would, and leaves nothing behind for dirt to stick to.
What is the best cleaner for solar lights?
Warm water with a drop of dish soap, applied with a microfiber cloth, then rinsed with clean water and dried. That is it. Avoid abrasive powders and scouring pads, which scratch the clear layer over the cells. Avoid ammonia-based glass cleaners on polycarbonate parts - ammonia degrades polycarbonate by aminolysis, breaking the polymer chains whether or not the part is stressed. Avoid solvents such as acetone entirely.
Is Dawn dish soap safe to clean solar panels?
Yes. A drop of Dawn in warm water is a mild, non-abrasive, non-ammoniated detergent and is safe on the plastics and coatings used in garden lights. Use a microfiber cloth rather than anything abrasive, rinse with clean water so no surfactant film is left to attract dust, and dry the panel. The same approach is fine for rooftop panels, though there the greater risk is walking on the array.
Will a higher-mAh battery make my solar lights brighter?
No. The driver chip sets LED current from the inductor value, not from battery capacity, so a 2,000 mAh cell produces exactly the same brightness as a 600 mAh one. What it changes is runtime - and only if the panel can actually refill it, which on a 0.16-watt panel harvesting 150-280 mAh a day, it usually cannot. An oversized cell in a small-panel light often sits permanently half-charged.
