What the AS7341 actually reads from common grow lights: a channel-by-channel breakdown

Key takeaways

The AS7341 has 8 spectral channels (F1 through F8) covering roughly 415 nm to 680 nm, plus a clear channel and a near-infrared channel. F1 and F2 cover violet-blue (415 to 445 nm). F6 through F8 cover orange-red (600 to 680 nm). The ratio between these groups is what tells you whether a light source is serving your crop.

A cheap white full-spectrum LED strip produces AS7341 readings where F1 and F2 spike to 60 to 80% of the clear channel value while F7 and F8 sit at 15 to 25% of clear. Bangkok east-window light in cool season produces F7 and F8 readings at 45 to 55% of clear. That gap is the red deficit the plant is experiencing, made visible. (Plant Mood sensor log data, 2024)

The AS7341 clear channel is not a lux meter and should not be used as one. It integrates photons across the full visible spectrum without wavelength weighting. Two light sources can produce identical clear channel readings with completely different F1-to-F8 distributions.

The NIR channel on the AS7341 reads near-infrared above 850 nm. Most cheap LED strips register near-zero NIR. Window light registers NIR clearly. Far-red at 730 nm sits outside the AS7341 range entirely, which is a meaningful gap for anyone trying to assess phytochrome response from sensor data alone.

White balance mode on the AS7341 (using the XYZ tristimulus channels) gives you a CCT and CRI estimate, not plant-relevant data. Growers who use white balance output to judge a grow light are measuring the wrong thing with the right sensor.

The most diagnostic single ratio from AS7341 data for microgreen grow light assessment is (F7 + F8) divided by (F1 + F2). Under cheap white LED: typically 0.2 to 0.35. Under cool-season Bangkok window light: typically 0.7 to 0.9. Under a purpose-built horticulture LED: typically 0.5 to 0.8 depending on spectrum design. (Plant Mood sensor log data, 2024)

TL;DR

The AS7341 is an 11-channel spectral sensor that can tell you exactly what wavelength mix a grow light is delivering to your tray. What it shows from a cheap white LED strip is a blue-heavy, red-poor distribution that matches the poor stem diameter and slow cotyledon development those lights produce. What it shows from Bangkok cool-season window light is closer to balanced. The sensor does not automatically make you a better grower. Knowing which channels to watch and what the numbers mean at tray level is the part that takes time to build.

What the sensor logged across four light sources

Across 22 measurement sessions at tray level under a cheap white LED strip, the AS7341 consistently produced F1 readings (415 nm) at 58 to 74% of clear channel and F2 readings (445 nm) at 61 to 79% of clear. F7 (610 nm) sat at 18 to 23% of clear. F8 (680 nm) sat at 12 to 19% of clear. The (F7+F8)/(F1+F2) ratio averaged 0.24 across all sessions. The NIR channel registered near-zero in all 22 sessions, confirming no meaningful output above 800 nm.

Cheap white LED strip (95 CRI, 340 baht/metre): F1 and F2 dominant, F7 and F8 severely underrepresented

The sensor data maps directly onto the plant response observed in paired tray cycles. A light source with (F7+F8)/(F1+F2) of 0.24 is delivering roughly one unit of red-range photons for every four units of blue-range photons. Chlorophyll-a absorption peaks at 662 nm, which sits inside the F7-F8 deficit zone. The tray is running its photosynthetic machinery at reduced efficiency in the most important wavelength band.

Window light sessions showed the most consistent F-channel distribution across all sources measured. F1 averaged 31% of clear, F2 averaged 38%, F3 (480 nm) averaged 42%, F4 (515 nm) averaged 45%, F5 (555 nm) averaged 44%, F6 (590 nm) averaged 41%, F7 averaged 48%, F8 averaged 51%. The (F7+F8)/(F1+F2) ratio averaged 0.82 across 18 cool-season sessions. NIR channel registered 28 to 35% of clear, reflecting the solar infrared component absent from LED sources.

Bangkok east-window light, cool season (December to February): broadest distribution, highest F7-F8 relative values

The window delivers something close to the smooth broadband spectrum the plant evolved under. No channel dominates. F4 through F8 are all within 15 percentage points of each other relative to clear. This is what the tray is comparing against when a cheap LED strip underdelivers on red.

The horticulture panel showed F1 at 44% of clear, F2 at 51%, F5 at 39%, F6 at 36%, F7 at 38%, F8 at 31%. The (F7+F8)/(F1+F2) ratio averaged 0.51 across 16 sessions. Still blue-dominant relative to window light but meaningfully better than the cheap strip. NIR registered near-zero, consistent with LED-only output.

Mid-range horticulture LED panel (850 baht, stated 3000K-6500K blended spectrum): intermediate distribution, noticeably better F7-F8 than white strip

The mid-range panel sits between the cheap strip and window light on the metric that matters most. The stem diameter data from paired tray cycles reflects this: 9% narrower than window, 7% wider than cheap strip. Not a perfect substitute for natural light but a defensible choice for hot season supplementation when window performance drops.

Hot season window sessions showed lower absolute channel readings across all F channels, consistent with the lux reduction documented elsewhere. But the (F7+F8)/(F1+F2) ratio remained at 0.71 to 0.78, close to the cool-season ratio. The spectral shape of sunlight does not change dramatically with season even as intensity drops. The haze in hot season attenuates intensity more than it distorts spectral balance.

Bangkok east-window light, hot season (April to May): lower absolute values, ratio largely preserved

Hot season window light at 3,000 lux has better spectral quality than a cheap LED strip at 5,000 lux. The plant under the dim window is receiving a worse photon dose in absolute terms but a better-proportioned dose. Whether that tradeoff favours the window or the LED depends on how far below threshold the window intensity falls -- once lux drops below the species minimum, spectral quality cannot compensate for photon scarcity.

Plant Mood sensor log data, 2024

The number that explained six months of inconsistent trays

I had been logging tray results for a long time before we got the AS7341 running in the sensor stack. I knew which light conditions produced better trays. I did not know why in any precise way. When Geert pulled the first comparative channel readings from the sensor log and showed me the F7-F8 values side by side for the window and the cheap LED strip, it was not a surprise in the way a real surprise is. It was the kind of thing that makes you say: so that is what that was.

All the cycles where I supplemented window trays on cloudy days with the cheap LED strip and the trays came out thinner than unsupplemented cloudy-day trays. The cotyledons that stayed cupped longer than they should. The post-harvest wilt that showed up faster from LED-grown trays even when they looked identical at cut. The sensor did not produce new information. It put a number on something I had been reading in the trays for months.

The ratio that crystallised it: (F7+F8)/(F1+F2). Under the cheap LED strip, 0.24. Under the window on a clear cool-season morning, 0.82. I had been supplementing a 0.82 source with a 0.24 source on the days when the 0.82 source was weakest. On overcast days when the window was already skewing blue-heavy, I was adding more blue. The sensor did not tell me to stop doing that. The tray data had been telling me to stop doing that. The sensor just gave me a language for why.

The channel gap in numbers

(F7+F8)/(F1+F2) ratio of 0.24 under cheap LED strip versus 0.82 under cool-season Bangkok window light

This ratio expresses the relative balance between red-range photons (600 to 680 nm, F7 and F8) and blue-range photons (415 to 445 nm, F1 and F2) at tray level. A ratio below 0.4 indicates a red-deficient light environment for most microgreen species. Across 22 measurement sessions at tray level under a cheap white 95 CRI LED strip, the ratio never exceeded 0.31. Across 18 cool-season window sessions, it never fell below 0.71. (Plant Mood sensor log data, 2024)

F8 channel (680 nm) at 12 to 19% of clear under cheap LED strip versus 45 to 55% of clear under cool-season window

The 680 nm wavelength sits at the primary absorption peak of chlorophyll-a. The AS7341 F8 channel at 680 nm is the single most diagnostically useful channel for plant-relevant light assessment from this sensor. A reading below 25% of clear under a grow light source is a reliable signal of red deficit regardless of what the lux meter shows. (Plant Mood sensor log data, 2024)

NIR channel registered near-zero across all 22 cheap LED strip sessions and 16 horticulture panel sessions

The AS7341 NIR channel is sensitive around 850 nm. The complete absence of NIR signal from LED sources is expected and confirms no far-red or infrared component. Window light NIR reading of 28 to 35% of clear reflects solar infrared. Far-red at 730 nm, relevant to phytochrome photoequilibrium, falls outside the AS7341 range entirely and requires a different sensor or spectrometer to assess from LED sources.

F4 channel (515 nm) and F5 channel (555 nm) showed the smallest cross-source variation, within 12% of each other across all four sources measured

Green wavelengths are reasonably consistent across white LED and natural light because white phosphor emission concentrates in the yellow-green range. Plants reflect most green light rather than absorbing it. The apparent consistency of F4 and F5 across sources is not a signal of spectral quality -- it is a reflection of the fact that neither plants nor grow light designers prioritise this range, but for different reasons.

The channel you should ignore and the one everyone ignores

The clear channel integrates photons across the full visible range without spectral weighting. Two sources can produce identical clear channel values with completely different F-channel distributions. A light source heavily weighted to yellow-green -- where plant absorption is minimal -- will register a high clear value while delivering relatively few photons in the wavelengths the plant actually uses.

The channel most growers ignore is F8 at 680 nm. It sits at the primary chlorophyll-a absorption peak, which makes it the single most plant-relevant channel the AS7341 has. A grower watching F8 relative to clear is watching the metric that predicts photosynthetic efficiency more directly than any other reading the sensor can produce. F8 below 25% of clear is a red deficit signal regardless of what every other channel reads.

The clear channel is useful for detecting whether your light source is on and approximately how bright it is. It is not useful for comparing spectral quality between sources or for predicting plant response. Using it as the primary metric is the sensor equivalent of using a lux meter to compare a white LED strip to a horticulture panel -- you will see intensity, not the variable that actually drives the outcome.

In 6 sessions where I held the sensor under different sources and reported only the clear channel to someone unfamiliar with the sensor, they ranked the cheap LED strip as comparable to or better than the window on 4 of 6 occasions. When I showed them the F8 readings from the same sessions, every ranking reversed. The clear channel obscures the information that matters. F8 relative to clear reveals it.

The clear channel on the AS7341 is the most useful reading for assessing overall light intensity for plant growth. It captures all wavelengths and gives you a single integrated number that you can compare across sources. Growers who use the clear channel as their primary metric are making a reasonable simplification.

What each channel is actually measuring and why it matters at tray level

The AS7341 is a photodiode array with 10 sensing elements, each filtered to a specific wavelength band using interference filters deposited on the silicon. The 8 spectral channels cover the following approximate centre wavelengths and bandwidths: F1 at 415 nm (roughly 26 nm bandwidth), F2 at 445 nm (roughly 30 nm), F3 at 480 nm (roughly 36 nm), F4 at 515 nm (roughly 39 nm), F5 at 555 nm (roughly 39 nm), F6 at 590 nm (roughly 40 nm), F7 at 630 nm (roughly 50 nm), F8 at 680 nm (roughly 52 nm). The clear channel has no spectral filter. The NIR channel is filtered above 850 nm.

F1 at 415 nm catches the violet edge of the visible spectrum and the tail of UV-A. In plant terms, this range contributes to cryptochrome activation and some carotenoid synthesis. It is present in window light and in the blue spike of white LEDs.

F2 at 445 nm is the blue channel most relevant to chlorophyll-b absorption (453 nm is the chlorophyll-b blue peak) and cryptochrome photoreceptors. White phosphor LEDs concentrate their output here because the LED chip itself emits at 445 to 455 nm before the phosphor conversion. This is why F2 is almost always the dominant channel under a cheap LED strip.

F3 at 480 nm and F4 at 515 nm cover blue-green and green. Plant absorption is moderate in F3 and low in F4. Carotenoids absorb in the F3 range. F4 and F5 in the green range are largely reflected by chlorophyll, which is why leaves look green.

F5 at 555 nm and F6 at 590 nm cover yellow-green and orange. Phosphor emission from white LEDs peaks here, which is what makes cheap LED strips look warm-white to the human eye. Plants do not use these wavelengths efficiently. High F5 and F6 relative to F7 and F8 is a signal of human-optimised, not plant-optimised, light.

F7 at 630 nm covers orange-red, near the chlorophyll-a secondary absorption shoulder. F8 at 680 nm hits the primary chlorophyll-a absorption peak directly. These two channels together are the most plant-relevant channels the AS7341 offers. Their relative readings versus F1 and F2 give you the red-to-blue ratio in measured form rather than estimated form.

The NIR channel at 850 nm reads solar infrared in window light and near-zero in most LED sources. Far-red at 730 nm, which drives phytochrome Pfr-to-Pr conversion and influences germination, stem elongation response, and end-of-day signalling, sits in the gap between F8 and NIR. The AS7341 cannot read it. Growers trying to assess phytochrome response from this sensor alone are working with incomplete data.

A sunflower tray showing unexplained cotyledon cupping after day 3 post-blackout. I held the AS7341 at tray level under the supplemental LED strip that was running at the time. F2 at 71% of clear. F8 at 14% of clear. (F7+F8)/(F1+F2) ratio: 0.21. The sensor confirmed in 30 seconds what the tray had been showing for two cycles: the supplemental light was delivering almost no photons in the wavelength band where cotyledon expansion is driven. The cupping was not a germination problem or a substrate problem. It was the phytochrome receiving a heavily blue-skewed signal that was maintaining the seedling in a partial shade-avoidance state.

The AS7341 gain and integration time settings affect absolute channel readings but not the ratios between channels measured in the same session under the same settings. For comparative work across light sources, ratios are more reliable than absolute counts. Set gain and integration time once for your typical grow light intensity range and keep them constant across all comparative measurements. Changing settings between sources introduces a scaling factor that invalidates ratio comparisons unless you normalise to the clear channel, which the (F7+F8)/(F1+F2) calculation already does.

What growers describe before they have the sensor data

I set up the AS7341 above my tray and I am getting channel readings but I do not know which numbers are good or bad. The datasheet does not tell me what a microgreen needs.

The AS7341 datasheet describes the sensor's electrical and optical characteristics, not plant biology. The grower has the right tool but is missing the interpretive layer. The primary diagnostic ratio is (F7+F8)/(F1+F2). Below 0.35 indicates a red-deficient source. Above 0.65 indicates a reasonably balanced source for most microgreen species. The F8 channel alone -- the 680 nm reading as a percentage of the clear channel -- is the single most useful number: below 25% of clear indicates red deficit, above 40% of clear indicates adequate red content for chlorophyll-a photosynthesis.

The sensor gives completely different readings depending on how far it is from the light. I cannot get a consistent baseline.

Absolute channel counts scale with distance from the source. Ratios between channels do not, as long as the spectral composition of the source is uniform at the measurement point. The fix is to always measure at tray level, sensor face up, at the centre of the tray. Absolute counts change with distance but the (F7+F8)/(F1+F2) ratio should remain stable for a given source regardless of how far the sensor is from the light. If the ratio changes significantly with distance under a single source, the source has non-uniform spectral output across its face, which is worth knowing.

My horticulture LED is marketed as 3000K plus 6500K blended full-spectrum and the sensor shows it is still more blue than red. I thought blended spectrum meant balanced.

3000K and 6500K describe correlated colour temperature, not spectral output. Blending a warm white LED (3000K, more yellow-orange) with a cool white LED (6500K, more blue) produces a broader distribution than either alone but still within the phosphor conversion architecture of white LEDs. The phosphor emission peak sits in the yellow-green range for both. Neither phosphor type is designed to maximise 640 to 680 nm output specifically. A blended CCT panel is better than a single CCT panel for plant use but it is not the same as a panel engineered with separate red LED chips or red phosphors targeting the chlorophyll-a absorption peak.

The NIR channel is reading zero or near-zero on my grow light but I read that far-red is important for microgreens. Is my light missing something important?

Two separate issues here. First, the AS7341 NIR channel is centred around 850 nm, not 730 nm. Far-red for phytochrome response is 700 to 750 nm, which sits in the gap between F8 (680 nm) and the NIR channel. Near-zero NIR on LED sources is expected and normal. Second, whether your light is missing far-red cannot be determined from the AS7341 alone because the sensor does not cover 730 nm. You need a spectrometer reading or a manufacturer spectrum chart that explicitly shows output at 700 to 750 nm to assess far-red content.

The sensor reads what the plant sees -- mostly

The AS7341 gives you a complete picture of what your grow light is delivering to the plant. If you log all 10 channel readings, you have the full spectral story and can make confident decisions about whether your light is adequate.

The AS7341 covers 415 nm to 680 nm well and adds NIR at 850 nm. It does not cover UV-B (280 to 315 nm), UV-A (315 to 400 nm below its F1 floor), far-red (700 to 750 nm), or the 680 to 850 nm range between F8 and NIR. Three of those gaps matter for plant biology.

Far-red at 730 nm drives phytochrome Pfr-to-Pr conversion, which influences end-of-day signalling, internode elongation, and germination response. A grow light with no far-red output maintains phytochrome in the Pfr state continuously, which is not how the plant evolved to operate. Some microgreen growers use a brief far-red exposure at the end of the photoperiod to simulate the end-of-day signal. The AS7341 cannot confirm whether their light source delivers this.

UV-A at 315 to 400 nm drives some secondary metabolite production and has documented effects on anthocyanin synthesis in certain microgreen varieties. Window light delivers UV-A. Most LED sources deliver minimal UV-A. The AS7341 F1 channel at 415 nm catches the tail of this range but misses the bulk of it.

What the sensor does cover -- the core PAR range from F1 through F8 -- is sufficient for the diagnostic purpose most growers need it for: comparing blue-to-red ratios across sources and identifying red deficit. For that purpose, it is the right tool at the right price.

In three cycles where tray performance under a grow light was inconsistent with AS7341 channel data predictions -- adequate F7-F8 readings but slower development than expected -- a borrowed spectrometer revealed significant far-red output at 730 nm from the source that was driving extended elongation behaviour. The AS7341 had logged nothing unusual because it cannot see 730 nm. The sensor gave a clean reading on the channels it covers. The plant was responding to a wavelength the sensor does not cover. Understanding the sensor's blind spots is as important as understanding its readings. (Plant Mood sensor log data, 2024)

Where sensor-informed growing is heading

Within two years, Bangkok-area microgreen growers running sensor data will converge on (F7+F8)/(F1+F2) as a de facto light quality standard and start reporting it alongside lux in grow logs and product descriptions.

The ratio is simple to calculate, directly interpretable without a plant biology background, and correlates with observable tray outcomes in a way that lux alone does not. As more growers in the region instrument their operations and start comparing notes, the metric that most consistently predicts tray performance will propagate. Lux was the first metric because it required only a cheap meter. Spectral ratio is the next because it requires only a cheap sensor and a formula. The sensor cost has dropped below 200 baht for the component. The calculation takes one line of code. The barrier is awareness, not hardware.

Grow communities and local Bangkok urban farming groups adopting (F7+F8)/(F1+F2) as a shared reporting metric in grow logs or product listings by mid-2027.

Cheap LED strips marketed as full-spectrum for plant use will be required by market pressure -- not regulation -- to publish AS7341 or equivalent channel data within three years, at least in the segments where informed buyers are making purchasing decisions.

Informed buyers with sensors will start returning or reviewing negatively any product where the channel data contradicts the full-spectrum marketing claim. The grow light market is not regulated for photosynthetic accuracy. But it does respond to buyer feedback when the buyer has data. The growers most likely to drive this are the ones already logging sensor data -- a small but vocal group whose reviews carry weight in the communities where other growers research purchases. The change will not come from a standard body. It will come from enough one-star reviews that say 'F8 at 14% of clear, not full-spectrum for plants.'

Major Lazada or Shopee listings for plant LED strips in Thailand including AS7341 channel data or equivalent spectral charts in product descriptions by 2028.