SK6812
1-wire RGB · OPSCO Optoelectronics · sk68xx
Datasheets disagree
The sources we hold state different values for the facts below. Rather than pick one quietly, both sides are listed with where each was read. Check the datasheet for the exact part number and revision you are buying. How this is shown.
Reset period
The 2015 sheet asks for 80 µs, the 2025 sheet for more than 200 µs.
| Source | Says | Page |
|---|---|---|
| sk6812 | Trst 80 µs | 5 |
| sk6812-012-c5378720 | Reset > 200 µs | 7 |
What we show. Both datasheet sets are listed, the older first. Whether the silicon changed or only the documentation is not known; 200 µs or more satisfies both.
Bit timing
The one-bit split differs between sheets and FastLED.
| Source | Says | Page |
|---|---|---|
| sk6812 | T1H 600 ns, T1L 600 ns, each ±150 ns | 5 |
| sk6812-012-c5378720 | T1H 650 ns–1 µs, T1L ≥ 200 ns | 7 |
| FastLED led_timing.h (driver-tuned) | T1H 900 ns, T1L 300 ns | — |
What we show. All are listed in the timing table. Only a scope on a known reel settles which is true of the silicon.
PWM frequency
OPSCO gives 1.2, 1.0, 4 and 4.5 kHz for parts sold as SK6812.
| Source | Says | Page |
|---|---|---|
| sk6812 | 1.2 kHz (rev 02, 2015) | 4 |
| sk6812-012-c5378720 | 1.0 kHz (rev B/1, 2025) | 6 |
| sk6812rv-f-c5378723 | 4 kHz (SK6812RV-F) | 6 |
| sk6812mini-ka-c7423120 | 4.5 kHz (SK6812MINI-KA) | 5 |
What we show. 1.2 kHz is shown, from the sheet this protocol’s first timing set is read from. The variant sheets are different silicon or revisions.
Supply voltage
The 2015 sheet gives a lower minimum supply than the 2025 sheet.
| Source | Says | Page |
|---|---|---|
| sk6812 | VDD +3.5 to +5.5 V | 4 |
| sk6812-012-c5378720 | VDD 3.7 to 5.5 V | 6 |
What we show. 3.7 V (the newer sheet) is shown because it is the safer minimum.
OPSCO’s counterpart to the WS2812B, and the family that also has the RGBW version. Physically and electrically a drop-in for the WS2812B in 5050, but not strictly better on paper: its reset (80 us in the 2015 sheet, above 200 us in the current one) is longer than the old WS2812B’s 50 us, its supply range is about the same (3.7-5.5 V against 3.5-5.3 V), and its 1.0-1.2 kHz PWM is higher than pre-V5 WS2812B silicon but lower than the V5’s 2 kHz. The two datasheet generations disagree with each other - see the timing sets.
Refresh & flicker
1.2 kHz in the 2015 sheet (rev. 02) and 1.0 kHz in the current OPSCO sheet (rev. B/1); the field records the 2015 figure. Three times the 400 Hz of pre-V5 WS2812B silicon, but BELOW the 2 kHz that WorldSemi’s V5 and later sheets state for the WS2812B-V5, so the PWM advantage over a current WS2812B reel is gone.
Timing
Sources disagree. This is normal and worth understanding: FastLED's numbers are tuned to drive real parts reliably, including clones, while the datasheet gives the manufacturer's nominal figures. Both are recorded here rather than picking one and hiding the conflict.
| Source | T0H | T0L | T1H | T1L | Period | Rate | Reset | ± |
|---|---|---|---|---|---|---|---|---|
| fastled ↗ | 300 ns | 900 ns | 900 ns | 300 ns | 1200 ns | 833 kHz | 80 µs | — |
| Note the reset: 80 us, against 280 us for the WS2812 family in FastLED. Symmetric timings (T0H equals T1L and T0L equals T1H). The current OPSCO sheet asks for more than 200 us (see the third timing set), so FastLED’s 80 us follows the 2015 sheet, not the 2025 one. | ||||||||
| datasheet ↗ | 300 ns | 900 ns | 600 ns | 600 ns | 1200 ns | 833 kHz | 80 µs | 150 ns |
| Note the disagreement with FastLED on the ONE bit: the datasheet splits it 600/600, FastLED drives it 900/300. Both total 1200 ns, so the bit period matches and both work, but the duty cycle differs substantially. The datasheet also states TH+TL = 1.25 us +/- 600 ns while its own four figures sum to 1.20 us - an inconsistency within its own tolerance. | ||||||||
| datasheet ↗ | 320 ns | 800 ns | 700 ns | 200 ns | 1120 ns | 893 kHz | 200 µs | — |
| The CURRENT OPSCO sheet is a different table from the 2015 rev 02 above, and gives a min/typ/max triple instead of “nominal +/- 150 ns”: T0H 250 / 320 / 400 ns, T1H 650 / 700 / 1000 ns, and for T0L (800 ns) and T1L (200 ns) only a MINIMUM with no typical or maximum - so those two values are minimums, not nominals. The reset is above 200 us, not 80 us. The drive is 12 mA typical per channel (10-14.5 mA) at 1.0 kHz PWM. Other OPSCO sheets of the same generation print T0H 200-400 ns and T1H 600-1000 ns, so the windows differ slightly between sibling parts. | ||||||||
Clones & compatibility
“Compatible” is never stated bare here — each relationship carries how strong the claim is, and what bites if you assume too much.
-
drop in
WS2812B
Timings overlap enough that WS2812B code drives it without changes, and the 5050 footprint and pad positions match (in the top view VDD is top-left, DIN top-right, DOUT bottom-left, GND bottom-right on both; only the pin NUMBERS differ between datasheets). Differences that matter: the reset is longer, not shorter - 80 us in the 2015 sheet and above 200 us in the current one, against the WS2812B datasheet's 50 us (the WS2812B-V5 sheets say above 280 us). The supply range is not wider either: 3.7-5.5 V (3.5-5.5 V in the 2015 sheet) against 3.5-5.3 V for the old WS2812B sheet and 3.7-5.3 V for V5. The PWM frequency is 1.0-1.2 kHz, higher than the pre-V5 WS2812B's 400 Hz but lower than the V5's 2 kHz. source ↗
-
protocol compatible of this
Everlight one-wire RGB+IC (5V01)
Timing windows overlap the SK6812's zero bit, but the colour order is RGB rather than GRB and the reset is 50 us. Not tested.
-
electrically similar of this
Guiguang GL01H (unnamed IC)
Inferred from the printed numbers only; no Guiguang sheet names SK6812 or any IC. The 2015 OPSCO SK6812 table (T0H 300, T0L 900, T1H 600, T1L 600 ns +/-150) sits inside all four GL01H windows, as does the 5050 pin order. But the GL sheets send RGB (SK6812: GRB), reset after 180 us (SK6812: 80 us in 2015, above 200 us in 2025) and a 4.5-5.5 V supply. A GRB controller will swap red and green.
-
protocol compatible of this
Inolux PI (single-trace IC)
The four typical values of the older sheets (T0H 300, T0L 900, T1H 600, T1L 600 ns, reset 80 us, GRB, 24 bit, MSB first) are the same numbers the sk6812 protocol records from the OPSCO SK6812 rev. 02 sheet, so the framing and timing are interchangeable on those sheets. None of the Inolux sheets mentions SK6812 or WS2812 by name. The newer windowed sheets move the typical T1H to 640-900 ns, and the PI55TA and PI3210 sheets ask for a reset of 200 us, so this is protocol-compatible, not a clone claim.
-
protocol compatible of this
Kingbright AAAF5050 RGB+IC
Same 24-bit GRB frame and the SK6812's 300/900 ns zero bit; the sheet's one bit is 900/300 (FastLED's SK6812 figure, not the SK6812 datasheet's 600/600). Reset is 200 us against the SK6812 sheet's 80 us. Not tested.
-
protocol compatible of this
SK6112
Accepts SK6812-style 24-bit frames in its default mode and sits in the same 5050 and 3.5 x 3.7 mm packages with the same pad positions, but it is a different IC with extra commands (feedback, current gain, sleep, IDs) and a 250 us reset. Whether a plain SK6812 driver is fully compatible has not been tested.
-
near clone of this
SK6805
The 5 mA sibling of the SK6812: the same single-wire framing and 4-pin GRB order, but 5 mA per channel instead of 12 mA, so a given brightness value is much dimmer, and a typical T1H of 900 ns against the SK6812's 700 ns. The part-number digits encode the drive current (6805 = 5 mA, 6812 = 12 mA). Not interchangeable for brightness.
-
protocol compatible of this
SK6812 RGBW
Identical timing, but four channels rather than three. Driving it with an RGB library shifts colours down the strip, because every pixel consumes one extra byte.
-
near clone of this
SK6812HV
The same 4-pin 5050 outline and pin positions as the 5 V SK6812, but a 9-14 V supply: plugging one into a 5 V SK6812 layout, or the reverse, destroys it or leaves it dark. Data timing windows overlap the SK6812's.
-
electrically similar of this
SK6822
Same family and single-wire scheme, but a 1750 ns bit period against the SK6812's 1200 ns. Not interchangeable at the same timings.
-
electrically similar of this
TONYU 6812 (TR2)
Inferred from the part code and printed numbers only; the TONYU sheet never names OPSCO or SK6812. Windows (T0H 200-400, T0L at least 800, T1H 600-1000, T1L at least 200 ns, reset above 200 us, 12 mA typical, 1.0 kHz PWM, 24 bits GRB) line up closely with OPSCO's current SK6812 sheet, and the package is the same 3.2 x 2.8 mm gull-wing outline and pin order as the SK6812MINI-E. Not tested on hardware.
-
electrically similar of this
TONYU 6812F
Inferred from the printed numbers and the part code only; the TONYU sheet never names SK6812 or any other vendor. Its windows (T0H 200-400, T0L at least 800, T1H 580-1000, T1L at least 200 ns, 24 bits GRB, 1.2 us symbol) overlap OPSCO's current SK6812 sheet (T0H 250-400, T0L at least 800, T1H 650-1000, T1L at least 200 ns). Differences: reset above 80 us against above 200 us in OPSCO's 2025 sheet, and a 1.2 kHz PWM.
-
electrically similar of this
TZ1815
Inferred from the printed numbers only; no TUOZHAN sheet names SK6812. The OPSCO SK6812 2025 sheet's typical timing (T0H 320, T0L 800, T1H 700, T1L 200 ns) falls inside all four TZ1815 windows, and both send 24 bits GRB. TZ1815 resets after 80 us where the 2025 SK6812 sheet asks for more than 200 us.
-
electrically similar of this
TZ2812
Inferred from the printed numbers only; no TUOZHAN sheet names SK6812 or any other vendor. The OPSCO SK6812 2025 sheet's typical timing (T0H 320, T0L 800, T1H 700, T1L 200 ns) falls inside all four TZ2812 windows (T0H 200-350, T0L 550-1200, T1H 550-1200, T1L 200-350 ns) and both send 24 bits GRB. Differences: TZ2812 resets after 80 us where the 2025 SK6812 sheet asks for more than 200 us, supply to 7.5 V against 5.5 V, and a WS2812B-style 400/850 ns stream is NOT inside the T0H window.
-
protocol compatible of this
WL-ICLED (24-bit single wire)
On paper only: nominal 300/900/900/300 ns against the SK6812 datasheet's 300/900/600/600 and FastLED's 300/900/900/300, same GRB order, reset 200 us against the SK6812's 80 us (the current OPSCO sheet also says 200 us). No Wurth part has been driven from a SK6812 driver by this catalogue.
-
protocol compatible of this
Xinglight SK6812B 12 V 5050 (9-14 V, 8.5 mA, BIN)
Only the title says SK6812B. A 12 V (9-14 V) part with VDD1/VDD2 pins, 8.5 mA per channel, a BIN auxiliary input and "intelligent reverse connection protection"; the OPSCO SK6812 is a 5 V part. Nothing in the sheet states compatibility.
-
protocol compatible of this
Xinglight SK6812B with BIN input (3.5-7.5 V, 12 mA)
Only the title (XL-3527RGBC-SK6812B, "XL-3527 (6812)" in the description) says SK6812; the sheet adds an auxiliary input BIN that the sheet describes as receiving the DIN data of the previous chip (i.e. the DOUT of the chip two positions upstream) so that one dead pixel does not stop the chain. The OPSCO SK6812 has no such pin.
-
protocol compatible of this
Xinglight SK6812B with FDIN input (5 mA, 0.25/0.47 us windows)
Only the title (XL-2020RGBC-SK6812B) says SK6812. The sheet adds a second data input (FDIN, pins DI1/DI2) and 5 mA drive; its timing table is the same 0.25/0.47 us set as the Xinglight 1515/2020 "2812B" sheets, which has nothing in common with the OPSCO SK6812 sheet.
Library support
- FastLED
SK6812↗- Adafruit_NeoPixel
NEO_GRB + NEO_KHZ800- WLED
SK6812
Parts using this protocol
Dating evidence
Evidence, not a release date. A library commit proves the part existed by that date and says nothing about how long before.