Short answer, in plain words. A chip's name on the box tells you very little. Samsung's LM301B and LM301H were the same chip with different paperwork: in 2019 a Samsung rep told us so, and our own lab test of LM301H vs LM301B builds measured the same, while the H parts cost about 20% more. What really decides the light is the brightness bin, how hard the chip is driven, how hot it runs, the driver, and whether the chip is genuine at all. Samsung stopped making every LM301 part at the end of 2025. Ask for the reel code, the lab test and the DLC listing, and measure the watts yourself.
The deeper version. The datasheets show the LM301B and LM301H share one package, one set of voltage and flux bins and one thermal resistance, and differ by a single character in the part number [1][2]. A genuine low-bin chip pushed hard on a budget driver ends up near 2.3 µmol/J at the fixture, about 26% better than double-ended HPS; a top-bin chip run gently on a good driver reaches about 3.1 µmol/J, about 45% better [1][2][3][4]. The tricks below are the founder's first-hand account from years of sourcing in Shenzhen [5], with the public evidence beside each.
Part of the series: LED grow lights, properly. 1. LED vs HPS: the heat and power-bill myth · 2. Samsung LM301B vs LM301H, bins and fake chips · 3. LED drivers explained · 4. How much grow light your space needs
What are the tricks in the LED grow-light business?
This section is the founder's first-hand account from years working with Shenzhen fixture, diode and driver suppliers, labelled as such, with the public evidence we could find alongside each claim [5]. GenetixBay criticises practices, not named businesses.
Trick 1: "Samsung LM301B/H" on the box, something else on the board
Founder's field experience: fixtures advertised as "Samsung LM301B" or "LM301H" are often built with other, cheaper diodes, or with a mix [5].
Public evidence: we found no Samsung public statement on counterfeit LM301 parts. What we did find makes the claim harder to trust, not easier. Samsung's end-of-life notice of 31 December 2024 covers every LM301B, LM301H, LM301B EVO and LM301H EVO part: last orders 30 June 2025, last shipments 31 December 2025 [6]. Any fixture sold in late 2026 as newly built with "Samsung LM301H EVO" must be using stock bought before that date, or the claim deserves a hard look.
How to check: ask for the full 18-character product code from the reel label (see the checklist below), and compare it with Samsung's datasheet code structure [1][2][7].
Trick 2: bottom-bin diodes sold as "top bin"
Founder's field experience: reels of lower-bin diodes are routinely sold as "top bin" at top-bin pricing [5].
Public evidence: Samsung's own datasheets show why the claim is easy to make and hard to check. For the LM301B and LM301H at 5000K, 65 mA and 25 °C, the flux bins are SK 36–38 lm, SL 38–40 lm and SM 40–42 lm [1][2]. The cheapest ordering code, S0 ("whole bin"), is a reel that can contain any of SK, SL or SM [1][2]. Only codes ending SL (SL only) or, on the LM301H, L2 (SL plus SM) narrow it down [2]. So "top bin" on a listing means nothing without the code. Supported (datasheet).
Colour temperature matters too. Samsung's LM301H datasheet lists calculated efficacy from 2.83 µmol/J at 2700K to 3.08 µmol/J at 5000K, and marks those figures "Calculated, Reference only" [2]. A warm-white board can be honestly lower than a cool-white one.
Trick 3: "anti-sulfur" and "H" sold as premium upgrades
GenetixBay first-hand account, 2019 (report no longer on file) [8]: when we were developing our own fixtures in 2019, we lab-tested LM301H builds against LM301B builds. They measured the same. The H parts cost about 20% more. We no longer have the lab report, so we give no numbers from it.
Samsung's own rep then confirmed why. In 2019, after the LM301H launch, a Samsung LED sales representative wrote to GenetixBay [9]:
"The real story with the 301H is that they are identical to the 301B. They are just marketed to different markets and their data is presented accordingly. They both have the same photon or luminous efficacy (depends how you want to look at it). There is no tweaking or special phosphor used, the anti sulphurization coating and stronger bind [sic: bond] wires (which is key in many harsher environments, but especially horticulture) is the exact same on both parts. But since they were gaining so much traction in the horticulture market with the 561c and 301b years back, they decided to cater specifically to that market and give them a unique data sheet that works for the specific industry; micromoles rather than lumens for the case of the 301H."
The public datasheets agree. The LM301B (2018) and LM301H (2022) data sheets show the same 3.0 × 3.0 × 0.7 mm package, the same absolute maximum ratings (200 mA forward current, 110 °C junction), the same forward-voltage bins (AY, AZ, A1), the same 7.5 °C/W thermal resistance and the same SH–SM flux bins [1][2]. Their product codes differ by a single character: SPMWHD32AMD5XAR0S0 (B; "D" = ANSI standard chromaticity) versus SPMWHD32AMH5XAR0S0 (H; "H" = Horticulture) [1][2]. The H datasheet adds µmol figures; the B datasheet gives lumens [1][2]. A German fixture maker that measured both reached the same conclusion: the gap they saw came from binning, not from a better chip [10].
The honest nuance. The anti-sulfurisation coating and the stronger bond wires do matter. Grow rooms are humid and full of fertiliser and pest-control chemistry, and the rep calls these features key for horticulture [9]. The point is that every genuine chip has them. Samsung's own product pages listed "anti-sulfurization" as a feature of the H and EVO parts [11], but a seller charging extra for "anti-sulfur" or for "H" over "B" is selling you nothing extra. The founder adds that teardown analysis shows the coating on every chip anyway (founder's field experience) [5].
- 3.0 × 3.0 × 0.7 mmSame package
- 7.5 °C/WSame thermal resistance
- SK–SMSame flux bins (36–42 lm at 65 mA)
- 1 characterPart-number difference: D vs H
- ~20%Extra cost of the H parts in our 2019 builds
Source: Samsung datasheets [1][2]; GenetixBay first-hand account, 2019 [8].
What has changed since 2019?
The 2019 statement covers the original LM301B and LM301H. Three things came later.
- LM301H EVO (introduced 2022) is a genuinely revised part. Samsung introduced it with a "plant-centric" spectrum and a 437 nm blue peak [12]. Its datasheet shows a taller 0.8 mm package (versus 0.7 mm), a new PPF-based bin table, and Samsung listed it at 3.14 µmol/J versus 3.10 for the LM301H (65 mA, 5000K) [7][13][11]. So the EVO is not simply a relabelled LM301B.
- "H" did not become a top-bin-only part. The 2022 LM301H data sheet still sells the whole-bin S0 reel (SK, SL or SM) at 5000K, plus SL-only and a new L2 (SL plus SM) code [2]. The LM301H EVO data sheet has a single, wide PPF bin at 5000K: PR, 0.520–0.587 µmol/s at 65 mA, with ±5% measurement tolerance [7]. There is no separate "top bin" EVO at 5000K to pay extra for. Supported (datasheets).
- The whole LM301 line is end-of-life. Last shipment was 31 December 2025 [6]. Crescience also reports that Samsung has left the LED market [10].
Real differences between genuine chips come from bin, drive current, temperature and authenticity, not from the letter after "301".
Trick 4: diodes and drivers overdriven to hit a wattage claim
Founder's field experience: to hit a headline wattage with fewer diodes, builders run each diode at high current and push drivers near or past their limits. Diodes and drivers then fail early, and fixtures run hotter and hotter as they degrade [5].
Public evidence, piece by piece:
- Diodes lose efficiency at high current. Radiative efficiency falls as current rises ("current droop") [14]. Samsung rates the LM301 family at 65 mA typical with an absolute maximum of 200 mA [1][2][7].
- Diodes lose efficiency when hot. Efficacy drops about 10% as junction temperature rises from 25 to 85 °C (77 to 185 °F); better thermal management raises both efficacy and life [14].
- Light output fades faster at high current and temperature. LM-80 tests measure a diode's light maintenance at stated case temperatures and currents, and TM-21 projects it forward only for those conditions, and no more than six times the test lengthbeing reviewed [15]. A fixture run hotter or harder than the tested conditions has no valid projection.
- Drivers age faster when hot. Mean Well rates its HLG drivers at a life expectancy above 62,000 hours only when the case temperature point is about 75 °C (167 °F) or less [3][16]. The electrolytic capacitors inside drivers roughly halve their life for every 10 °C (18 °F) hotter they run [17].
- Why a degrading fixture heats more at the heatsink: at the same input power, every photon that no longer leaves the fixture becomes heat inside it. Hotter diodes are less efficient and fade faster, which pushes the temperature up again. Total room heat per wall watt stays the same [18]; the fixture itself runs hotter. We found no published measurement of how much hotter; the mechanism is standard physics, the size of the effect is the founder's field experience [5].
Does every "Samsung" LED beat HPS? The bin and drive-current math
Not by the same margin, and not always by much. Even with genuine Samsung chips, a low-bin chip pushed hard on a budget driver ends up near 2.3 µmol/J at the fixture, about 25% less power than double-ended HPS for the same light. A top-bin chip run gently on a good driver reaches about 3.1 µmol/J, about 45% less [1][2][3][4]. An unknown diode below about 2.2 µmol/J at its datasheet rating, built the same cheap way, does no better than HPS. Supported (arithmetic on datasheet curves; assumptions labelled).
Plain words: the same name on the chip can mean a very good light or a so-so one. Brightness bin, how hard the chip is pushed, how hot it runs and how good the driver is all take a bite.
- Chip bin2.94–3.26 µmol/J
- Drive current+3% gentle, −9% at 150 mA
- Heat−1% to −3%
- OpticsBare boards ≈ 0.99
- Driver0.89–0.96
- Fixture µmol/J2.28–3.16
Source: Samsung datasheet curves [1][2]; driver datasheets [3][4]; budget driver is founder's field experience [5].
Step 1: from lumens to µmol/J at the chip
Samsung sells the LM301B and LM301H in the same flux bins at 65 mA and 25 °C: SK 36–38 lm, SL 38–40 lm, SM 40–42 lm [1][2]. To turn lumens into plant photons we use Samsung's own figures for 5000K at 65 mA: the LM301B page lists 38.8 lm at 220 lm/W, and the LM301H datasheet lists 0.544 µmol/s at 3.08 µmol/J [2][11]. Both imply the same forward voltage, about 2.72 V (38.8 lm ÷ 220 lm/W ÷ 0.065 A ≈ 2.71 V; 0.544 ÷ 3.08 ÷ 0.065 ≈ 2.72 V), which is one more sign they're the same chip.
- Photons per lumen (5000K, CRI 80): 0.544 ÷ 38.8 ≈ 0.0140 µmol/s per lm (assumption: applies across these bins).
- Chip power at 65 mA: 0.065 A × 2.717 V ≈ 0.1766 W (assumption: mid voltage bin; an A1 chip at 2.85 V is about 4% worse).
| Bin (5000K, 65 mA, 25 °C) | Bin midpoint | PPF per chip | Chip efficacy |
|---|---|---|---|
| SK (lower) | 37 lm | 0.519 µmol/s | 2.94 µmol/J |
| SL (mid) | 39 lm | 0.547 µmol/s | 3.10 µmol/J |
| SM (top) | 41 lm | 0.575 µmol/s | 3.26 µmol/J |
The SL result matches Samsung's 3.10 µmol/J headline for the LM301H [13]. Bin alone moves chip efficacy by about 10% between SK and SM.
Step 2: drive current (the overdriving penalty)
The LM301H datasheet graphs relative light output and forward voltage against current at 25 °C [2]. Reading the curves (approximate):
| Drive current | Relative light | Forward voltage | Efficacy vs 65 mA |
|---|---|---|---|
| 40 mA (gentle) | ~64% | ~2.70 V | about +3% (curve suggests 3–6%; we use 3%) |
| 65 mA (datasheet) | 100% | ~2.75 V | 100% |
| 150 mA (hard) | ~219% | ~2.86 V | about −9% |
| 200 mA (absolute max) | ~285% | ~2.94 V | about −13% |
Kusuma and colleagues give the same picture: low drive current can reach about 104% of the rated efficacy, and efficiency falls as current rises ("droop") [14].
Step 3: heat, optics and driver
- Heat: the datasheet's temperature curve shows light falling to about 94% and voltage to about 97% as the solder point goes from 25 to 85 °C (185 °F), about −3% efficacy [2]. We use 0.99 at 45 °C (gentle drive, big heatsink), 0.985 at 55 °C and 0.969 at 85 °C (overdriven, small heatsink). Kusuma's broader estimate is about −10% from 25 to 85 °C at the junction [14]; we show that as a worse case.
- Optics: bare boards about 0.99; a protective cover costs about another 8% [14].
- Driver: HLG-600H about 0.96 at 240 V and 0.947 at 120 V; XLG-320 about 0.95 and 0.92 [3][4]. Unbranded driver: 0.89 at 120 V and 0.90 at 240 V, an assumption from the founder's field experience [5]; no datasheet exists to check.
Step 4: the fixture, side by side with HPS
Fixture µmol/J = chip efficacy × drive factor × heat factor × optics × driver. Saving for the same light = 1 − (1.7 ÷ fixture µmol/J), against double-ended HPS at 1.7 µmol/J [19][14].
| Build (all genuine LM301B/H, 5000K) | Chip | Drive | Heat | Optics | Driver | Fixture µmol/J | Watts for 1,000 µmol/s | Power vs DE HPS |
|---|---|---|---|---|---|---|---|---|
| Double-ended HPS (independent tests) | 1.70 [19][14] | 588 W | baseline | |||||
| Top bin (SM), gentle 40 mA, big heatsink, HLG at 240 V | 3.26 | ×1.03 | ×0.99 | ×0.99 | ×0.96 | 3.16 | 317 W | 46% less |
| Same, at 120 V | 3.26 | ×1.03 | ×0.99 | ×0.99 | ×0.947 | 3.11 | 321 W | 45% less |
| Mid bin (SL), 65 mA, normal heatsink, XLG at 240 V | 3.10 | ×1.00 | ×0.985 | ×0.99 | ×0.95 | 2.87 | 348 W | 41% less |
| Same, at 120 V | 3.10 | ×1.00 | ×0.985 | ×0.99 | ×0.92 | 2.78 | 360 W | 39% less |
| Low bin (SK), overdriven 150 mA, small heatsink, budget driver at 120 V | 2.94 | ×0.91 | ×0.969 | ×0.99 | ×0.89 | 2.28 | 438 W | 26% less |
| Same, driven at 200 mA | 2.94 | ×0.866 | ×0.969 | ×0.99 | ×0.89 | 2.17 | 461 W | 22% less |
| Same as 150 mA, with Kusuma's −10% heat loss | 2.94 | ×0.91 | ×0.90 | ×0.99 | ×0.89 | 2.12 | 471 W | 20% less |
| Break-even: unknown diode, built like the budget row | ~2.19 | ×0.91 | ×0.969 | ×0.99 | ×0.89 | 1.70 | 588 W | no saving |
| Unknown diode at 2.0 µmol/J, built like the budget row | 2.0 | ×0.91 | ×0.969 | ×0.99 | ×0.89 | 1.55 | 644 W | 9% more power than HPS |
Every factor except the chip bin ranges and DE HPS figure is an assumption or a reading from a published curve, labelled above. Board wiring losses are ignored, which flatters every LED row slightly.
Old and low-end fixtures: often worse than HPS. In the 2014 independent tests, ten LED fixtures measured 0.89–1.70 µmol/J, and eight were below the best double-ended HPS at 1.70 [19]. Older-generation and "blurple" panels, and no-name boards of low-bin diodes overdriven on cheap drivers, still sit around 1.3–1.7 µmol/J in the founder's experience [5]. At 1.5 µmol/J they need more watts than HPS for the same light.
What "replaces a 1,000 W HPS" really means
A 1,000 W double-ended HPS fixture draws about 1,045 W [20]; at 1.7 µmol/J that's roughly 1,780 µmol/s of light [19].
- The top-bin build drawing about 581 W at 120 V (the "better build" in the driver guide's worked example) makes 581 × 3.11 ≈ 1,810 µmol/s: it really does match the HPS.
- The low-bin budget build drawing about 584 W makes 584 × 2.28 ≈ 1,330 µmol/s: about 25% less light than the HPS it claims to replace.
Same wall watts, same heat [18], very different light. The "more light than HPS" promise holds for the top tier and the middle; for an overdriven low-bin panel it doesn't, and for unknown diodes it may not even save power. Supported (arithmetic); the budget-driver efficiency is founder's field experience [5].
How do you spot a fake or overstated LED grow light?
Ask for proof, measure what you can, and treat silence as an answer. Grower practice, built on the public standards and datasheets cited.
What to ask the seller
- The diode product code and bin. Ask for a photo of the reel label or the full 18-character code (for Samsung, it starts SPMWHD32AM). Check the last two characters against the datasheet: S0 means "whole bin, any of SK/SL/SM"; SL or L2 means a narrower, brighter selection [1][2]. On the EVO, check the PPF bin code (for 5000K, PR) [7].
- B vs H markings. "LM301H" on a genuine part means the horticulture datasheet of the same chip as LM301B, per Samsung's rep (2019) [9]. Don't pay a premium for the letter. LM301H EVO is a different, later part [7][12].
- The LM-80 report and TM-21 projection for that exact diode, at a case temperature and current close to how the fixture runs it [15].
- A third-party PPF and µmol/J test of the whole fixture, naming the lab, the test method and the date. A supplier PDF with no lab name is a brochure, not a test.
- The DLC listing. Look up the exact model (not just the brand) at qpl.designlights.org [21]. Since April 2025, new listings must reach about 2.5 µmol/J [22].
- The driver brand and model, with a photo of the driver label. Download that datasheet yourself from the maker and check rated power, efficiency at your voltage and the derating curves [3][4][16].
- Measured wall watts at 120 V and at 240 V, at full power after warm-up.
- Safety listing. A UL 8800 certification (or equivalent) by a recognised lab, such as an OSHA NRTL in the US or an SCC-accredited body in Canada [23]. See grow-room electrical safety.
- Heatsink mass and design. Ask the fixture weight and the heatsink material. A large, thick aluminium heatsink keeps diodes cooler and brighter for longer [14]. We found no published rule for heatsink mass per watt; in the founder's field experience, a fixture that feels light for its wattage is a warning sign [5].
- Warranty terms in writing: length, what's covered (driver and diodes separately), who pays shipping, and whether a running fixture with a dead driver counts.
What to measure yourself
- Wall watts. Use a plug-in power meter on 120 V, or a clamp meter on 240 V circuits. Run 30 minutes at 100%, then read. Compare with the advertised watts and the driver rating.
- Driver case temperature. An IR thermometer on the driver's marked tc point after an hour. Compare with the datasheet's life-expectancy temperature [3][4].
- Diode markings. A loupe photo of a few diodes. You can't read a bin off a chip, but obvious mismatches (different package sizes, mixed phosphor colours) are a warning.
- Light. A quantum PAR meter gives PPFD at points across the canopy; divide average PPFD × area by wall watts for a rough check on µmol/J (see measuring PPFD).
Red flags
- "1000 W" in the model name, with 100–200 W on the driver label.
- "HPS equivalent" or "replaces 1000 W HPS" instead of measured µmol/s.
- "Top bin" with no product code.
- "Samsung LM301H EVO, new production" in 2026 [6].
- "Anti-sulfur" or "H" offered as a paid upgrade [9].
- A PPF map measured at a short hanging height over a smaller area than advertised.
- No driver brand, or a driver label scraped off or covered.
- No DLC listing, no lab name, no LM-80, and a seller who changes the subject.
- A heatsink that is thin, light or hot to the touch beyond the driver's stated limits.
Greenleaf's aside. A good LED is a fine tool. I run them. But a light is a heater that happens to glow, and a sales sheet is a story that happens to have numbers in it. Trust the meter, the lab and the datasheet, in that order.
Common mistakes
- Buying on diode name instead of measured fixture µmol/J and a DLC listing.
- Paying more for "LM301H" or "anti-sulfur" over a genuine LM301B of the same bin [9].
- Accepting "top bin" without the 18-character product code [1][2].
- Believing "new production Samsung LM301H EVO" in 2026 [6].
- Trusting a supplier PDF with no lab name as a "test".
