This LED (and 8 others like it) form the lighting component of a ceiling fan assembly. Unfortunately, they put out a very dim and cold light, and I'm interested in replacing them, but I figure I need to determine what I'm starting from before looking for replacements (although any suggestions on what to look/search for are welcome).


  • \$\begingroup\$ See addition to my answer under luminous efficiency. | Please measure vf and Istring as per my answer so we can know where to go next. \$\endgroup\$ – Russell McMahon Nov 28 '11 at 0:33

What Oli says, plus:




See "ANSWERING THE ACTUAL QUESTION" below - but do read the rest as well.

If there are markings on the LEDs it would help greatly to know them.


  • Determine LED forward voltage by measurement.

  • Determine LED string current by measurement

  • Choose appropriate LED.

Some such LEDs have 3 dies in the pkg which are brought out separately on eg pins 1-6, 2-5, 3-4 as viewed here and the user can join them in series. This does not seem to be the case here. The pins MAY be connected underneath but it seems unlikely.

Note that the top left and bottom pins (1 & 4) connect to the adjacent contacts by wider copper tracks while the rest have apparently unconnected tracks.


Be aware that the LEDS may be being fed with DC with a strong 2 x mains frequency component if an attempt is being made to power factor control the load (unlikely but ...).

Also, as part if a ceiling fan, they may have a low voltage winding acting as a transformer winding from the motor. This may lead to unusual waveforms.

If you can determine how steady the DC feed to the LEDs are it would be good. As they MAY be floating wrt ground this may involve a floating differential measurement involving mains. Not hard but you do want to know what you are doing and there are safety aspects to consider.

The following is written on the assumption that the LED supply is reasonably steady conventional DC. Worst case you would get voltage and current readings that are wrong. If LED Vf varies by much from 3.3V say then you may suspect interesting happenings. Lowest Vf you are liable to see with white LEDs is about 2.7V and highest is about 3.7V. More like 3.3 +/- someis usual. (The Raijins I mention below run at about 2.95V at rated current as a part of their overall awesomeness. This low is rare at full current. )


  • Here is how to get the basic LED operating conditions. But, do read the rest as well :-)

Measure voltage across contacts (same as pins 1 & 4).
This will be 3.xx Volt OR a multiple of this.
Record and report.

Multiple occurs if there are 2 or more dies in series in package. Happens.

Name your 9 LEDS LA LB LC LD LE LF LG LH LI say.
Have name increasing in order of connection if in series.

Probably in series but not certain.

Measure voltage from LA in to LI out - ie voltage across string. If this is 10 x the voltage across one LED all are in series. eg if Vf La or Lb or Lf ~= 3.3V then you'd expect 10 in series to be ABOUT 33 volt. There will be some variations. If Vmax across string = ABOUT 5 x Vf = say about 17 Volts then you have series / parallel arrangement. You'd expect them to be all in series.

Hopefully they are not all in parallel - unlikely but happens.

IF all in series and Vsring ~= 10 x Vf of one LED then

  • Set a meter to measure milliamps.

  • Connect meter probes across a single ON LED with all working. Do this briefly, measure current note effect. Remove. It is extraordinarily unlikely that this will do any damage. If it did then your system is prone to such damage at any moment spontaneously as LEDs can and do go short-circuit in use.

  • That LED should go out and meter will measure string current.

  • Note if remaining 9 LEDs get slightly brighter.

  • Remove meter

If LEDs are all in series you now know LED forward voltage and string current. If LEDs are not all in series you can do a version of this as long as not all in parallel.

You now know the voltage and current of LEDs that can be used as replacements.

Tell us what you measure and we can advise best superior replacement.
UNTIL you do the above you could find that a replacement was "just wrong".
It may not light at all or may die instantly.
Both are genuine real-world possibilities.

Look for pin compatible parts, same number of die per package (1 ~= 3V, 2 ~+6.xV, 3 ~= 10V) and. most important, maximum possible number of lumens at available current. If you want a yellower light you'll want a loer "cplour temperature" - say more like 3000 degree K than the 6000+ that bluish LEDs have.

Be prepared to be disappointed IF current is low. IF these are 20 mA LEDs as they might be then may power may be 20 ma x 3.3V say x 10 LEDs = 0.66 Watt. The best LEDs will have outputs combined of the equivalent of 5 to 10 W of incandescent lighting. Higher currents = higher power = more light.

The VERY best LEDs you can buy will give you about 100 lumen at 0.66W.

The best you can get at low currents:

If Vf for one LED is 2.5V to 4V say (either extreme unlikely) and if string current is under 50 mA (say 15 mA - 30 mA most likely) then you will get more light per available energy than from almost anything else on earth in the size range by using Nichia NSPWR70CSS-K1 LEDs - Nichia codename = Raijin.

These are NOT surface mount but may be able to be adapted - the pin positiin match and can be surface soldered - the height may be excessive in your application.

enter image description here

Photos of illumination need great care to be objective and this is rushed, but this is what you can achieve with 1 x Raijin at about 130 milliWatts. More comment on this LED if of interest.

enter image description here

If you have more than 50 mA available then the Raijin is not the best choice.

Luminous efficiency = l/W = lumenr pr Watt.

A Cree CLP6S-WKW or CLP6S-MKW has been suggested as an example of an LED that youd fit in this locaion. It would, but see my comments else where re Vf and current.


Cree make some utterly superb LEDs. This is not one of them, alas.

Vf per LED is strangely high at 4V typical and 5V max !!! - they may even include an internal resistance!. This is too low to be 3 x die in series so each ie must be 4V+!.

To improve brightness you want more lumen so need more lumenr per Watt in a given situation.

Reasonably hidden in the data sheet is the lumen output = 9000 mlm = 9 lumen at 3 x 50 mA on page 2.
This translates into a very low efficiency, alas.
Power = say 4V x 50 mA x 3 LEDs = 600 mW.
Light = 9 lumen so l/W = 9/.6 ~= 15 l/W (lumen per Watt).
This is "not good".
Triple this on the assumption that data sheet is misread and that 9000 mlm is the spec for one LED (which I think is unlikely) and you still only get 45 l/W = not good.

The Raijin I mentioned above gives 160 l/W at 20 mA and 120 l/W at 50 mA.
so at 50 mA a single Raijin outputs just less light than all 3 LEDs in the CREE and at 25% the power (as Vf = 2.95V as opposed to 4V or 5V.)


Please measure vf and Istring as above so we can know where to go next.

A measurement "trick"

If you do not have an oscilloscope or cannot safely measure a floating point then the following MAY help to test whether LEDs are fed with steady DC. They probably are, but ... .

Cheaper multimeters, when measuring DC volts on an AC range, will often display double the actual voltage. eg on a 10 VAC range 3 VDC may display as 6 VAC. We can use this "feature" to advantage.

See if you have a meter that does his. Measure eg a battery on a DCV and ACV range. If ACV = 2 x DCV we have a "magic meter" that can be used as below.

Using a magic meter as above, Measure voltage across ONE LED using DCV and ACV ranges. If ACV = 2 x DCV then voltage is essentially pure DC. If ACV strays from 2 x DCV then AC is probably present. AC readings higher or lower than 2 x DCV may occur depending on AC + DC components. Not a certain test, but a potentially useful one.

NSPWR70CS-K1 / Raijin sources:

LEDRISE - prives in Euros but c... dig ... dig ... aha, Hong Kong based.

Available in 1's here
Nichia Raijin Superflux Warm White LED 25.5lm 70° 50mA NSPLR70CS-K1

Brief review and comments here - the real lifetimes are better than their tests - they are abusing them somewhat thermally.

Anyone know a US or UK source?

  • \$\begingroup\$ Russell, thanks for the incredibly detailed and useful answer. I unfortunately don't have a multimeter available at this location, and won't have access to the lighting fixture again for another month (at which point I'll take a multimeter with me). I'll try to remember to update this question with more information at that time :). \$\endgroup\$ – Travis Northcutt Nov 28 '11 at 4:06
  • \$\begingroup\$ @tnorthcutt - Treat all circuit as potentially at mains. Consider getting a Raijin or similar high efficiency LED and running off say 9V transistor battery. You can then compare relative output per LED to decide what is needed as a replacement. Most LEDs off a newish 9V battery will need a resistor of (Vat-Vfled)/Ima. eg 9V 20 MA 3.3V Vf R = (9-3.3)/0.02A = 285 ohms say 270 ohms std value. Or a Raijin at 50 mA = (9-3)/.05 = 120 ohms series resistamce. If you can find a Raijin do so. It's the benchmark LED that you can compare anything under 50 mA single die against. \$\endgroup\$ – Russell McMahon Nov 28 '11 at 5:27

It looks like a pretty typical high power LED like this.
Here are many more options from the search it came from.

I think that any option from those results will likely be very similar Vf specs wise (3.3-3.6V Vf) but the pinouts/current rating may vary. I think the notch on the bottom right may be for the cathode (e.g. pin 6/bottom right) and the top left pin (pin 1) the anode. What I can make out of the traces looks like it might confirms this, but you need to investigate more.
You can test your LEDs Vf quite easily with a multimeter (also current if necessary) The current in these packages seems to range from 30mA up to ~700mA.
Remove the 6 pin search filter and update the results to look for other options.


It is a 3 die cool or neutral white LED in a PLCC6 package. Is is 5x5 or 3.5x3.5mm?

It looks like it has a CREE style emitter arrangement, where the emitters are placed in a triangle. It could also be a copy/knockoff from any number of second tier LED manufacturers.

The form factor is pretty common. You can probably replace it with any warm white LED in a similar package. It is not common to have common anode or common cathode in a PLCC6 package, but but check to make sure.

You are unlikely to get a significantly brighter LED at the same current. Especially if you go to a warmer white as their lumens output is slightly lower. If you increase the current, the LEDS will run hotter, so you need to make sure you have adequate heatsinking.

Most of these packages are specced at 3 x 25-50 mA. These packages does not have a thermal pad and are cheap to manufacture (and easy to replace), but there is a limit to how much power they can dissipate. Here is one that can take 3 x 50mA, and has a decent CRI for it's type: http://www.cree.com/products/pdf/LEDlamps/CLP6S-WKW&MKW.pdf

  • \$\begingroup\$ That spec sheet is an excellent example of why you MUST do the identifying steps that I listed. Morten's LED is an example of the 3 separate dies brought out to separate ions that I mentioned. It would be normal to connect these in series if fully islated but, as they will (probably) be well matched some may choose to connect them in parallel. My Vf test (just measure one pkg terminals voltage) will establish this. This pkg is good for 150 mA if all in parallel whereas others only allow 20 mA. Making my current test "useful" THEN start looking at LED performance. \$\endgroup\$ – Russell McMahon Nov 28 '11 at 0:16
  • \$\begingroup\$ See my answer add on re lumen/Watt. \$\endgroup\$ – Russell McMahon Nov 28 '11 at 0:22

There is no doubt that it is a PLCC-6 5050 SMD (3-Chip, i.e. 3 LEDs per device). The notched corner denotes the cathodes (-) side of the device.

*For anyone in the future having a similar question, please provide information about the light fixture itself, such as the fixture's total maximum brightness (lumens).

Since different 5050 SMD LEDs produce different amounts of light, it is difficult to pin down exactly which 5050 LED you have from the information given. For example, it would be super helpful (critical actually) to know the maximum light output (lumens) of the fixture. From the output of the fixture, the typical "brightness" of the LED device shown could be determined (divide the total max lumens of fixture by the total number of LED devices in the fixture). From there it would me fairly easy to determine which LED you have, and very importantly its forward voltage and forward current requirements.

The fixture is designed/built to produce the amount voltage and current that this particular LED requires. So you have to replace the existing LEDs with significantly more efficient LEDs (2835 SMD) that operate best with the voltage and current provided by the fixture.

Anyway, below is almost certainly the parameters of the 5050 LED we are dealing with:

Color Temp (at 60mA): Cool White, 4000°K (per the "cold white" description) Size: 5.0mm x 5.0mm x 1.6mm (certainly) Viewing Angle: 120° (certainly) Lens: Diffuse (certainly) Forward Voltage (typical): 3.2V - 3.4V Reverse Voltage: 5V Continuous Forward Current (max): 60mA (20mA x 3 chips/SMD) Pulsed Forward Current (Max): ~400mA to ~450mA Max Solder Temperature: 260°C for 10 sec Brightness (in lumens or mcd): ???

For a brighter and more efficient LED, replace 5050 SMD with 2835 SMD needing VERY SIMILAR Vf (Forward Voltage) and Fi (Forward Current).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.