Yuji BC 5mm LED Stress Test - Graph and Images

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
If you don't care about the method or detailed results, just skip to the pictures at the end.

Intro

With literally several thousand of these emitters now in circulation among CPF members, I thought it was worth taking some data on their performance when overdriven.

Before continuing, I want to thank those who have organized group buys or even outright shared Yuji 5mm emitters with others, including scout24, ma tumba, lampeDépêche, and K2-bk-bl-rd. This is part of what makes CPF a great community!

Disclaimers

1.) Treat my data as approximate, primarily for comparison to itself. I used a cheap, salvaged photocell I came up with my own rough calibration method for. My budget solderless breadboard also clearly had intermittent contact issues from handling that caused small fluctuations in some of the data. I'd be surprised if voltage and current measurements were off by more than 5%, but I'd be happy if I were able to confirm my Candela data are within +/- 50%.

2.) Anything above 30mA exceeds manufacturer specs. Worsening life span (potentially both hours and on/off cycles) should be expected the higher the currents go, but I have no data on how much worse. My testing lasted maybe 4-5 minutes at each current level as I moved my multimeter between measurement points, waited for stable readings, and recorded data.

3.) I tested a single sample. There may be variation in performance between samples.

Disclaimer to the disclaimers - at the group buy pricing, there's no tears to shed if you burn out one or a handful experimenting with something interesting. These are great to learn with.

Method

Power was controlled by a linear power supply. An inline shunt resistor was used to derive current from voltage drop over the resistor. Voltage was increased on the power supply until the expected voltage for each current level was achieved. The multimeter was then moved to measure forward voltage of the LED, and resistance of the photocell. Separate data were taken approximately correlating photocell resistance with luminous intensity. Checks made on that data showed fair agreement with expected values.

Results

In a dark room, a faint glow was perceptible starting around 2.35V, at microamps of current. Measurements started at 2.53V and roughly 0.2mA. This batch of LED's is rated for a forward voltage of 2.8-2.9V at 20mA. This sample tested at 2.81V at 20mA.

Light output peaked at 90mA and 3.23V. Output appeared stable, and tint minimally less warm than at lower currents. A pink tint becomes noticeable around 110mA, increasing to dominate the tint as current rises. Current ceased to be stable with respect to voltage at 3.49V and initially at 160mA. No more measurements were taken, and voltage was increased over the space of a few seconds until the LED went completely dark at about 200mA.

After a short rest, the LED was powered on again at 20mA. The LED was clearly seriously damaged. Brightness was roughly equivalent to the previously measured 5mA level, and the forward voltage necessary to achieve 20mA seemed to have risen to 4.4V.

Graph and images follow. 4500K camera white balance was chosen to approximate naked eye appearance.

NJUQxfimLwtobX-VEPNf6zefnP3cEVN_4mykj1rkJLiOYDSmtW2uk9j8yW4NhSJBEMpTiv2BqHrGqTqnc_8GjPZuaEzuyt3tfpYPM4RmTOG4jMkMoTDlzlCZRBrjQsxWYVj30a9HZJIm2TrBse-j6yzeRJ6rLO0Z3ev8QbNxBA1bBM1b-A-jDrNdpIu5P3jmWgGXu1Pqdz1TmOeYWqJBWknMxjgYk3s5PPdw0KXl0UzdnhqymWvsgjzCQJDMS352CJtE4h198L7edP5vNeEGety1DWcqxUnA2G1J0rBEAXVhW6oH16Zf94tIBW8C2ocrksC2r2zmZgu4sO1jlzYAJGC6Fx1KIRcug4_EhwnG0pRjq9s_Zg9ltD58AqdHhOmtZh7tZwxPkjYBe0wzmOUoJG0w6BN5zUOdI39hcOSE9-xj9oSnbsXoEQZQLEcg6zeP1erM2MTZCV2RmVhHJVtShs1vcpXL_kQsZ8n5EBmaAaDUnNsVO-SflukUw_I0TOE073ihca7PRxBa0nnvUpTqrLfbbVonQWREM2laYFDrDFgtEFar6gEk0AMf4dVtaAp1j54oC9o15fjn1ARz-re6Ty_V4oPkjs5FRHvA5AU9zasUObRjLJ0OFQkhUFSN_-kvs_2kGTHwuO_PWY2TS4xu7qo9lcSxp7_a=w720-h465-no


Standard rated current of 20mA, before and after overload test:
7jM1qFIO-T1TtWiEyJx9XW_E-WCgarm01PntpUY-iTo5pVtpGb4XiX_JcGdIT-9bfUijuBU_VZDyT-hbh6nOBQaggnjQHSBIn88igFHB9ikRClL2ca7yy-QAjalqXmW3ne5tUihsdaXDxFXQaLjSR2HYAcJvtjE8W4J7-sykJz_LTp8etqn95oc6e3F6GIAYme-0uhHe0JhCMBc2U2uSg0K3X5oCa3XSsATLrjXo_0ZgzeFfUwWAOATiYJJTNo1bIiPPWy1Z9XymW6jCNTjxLXxa0OmwL3hUFzRPbEuApNhpzrhXM_QaAJgrZdZhyqtn7NtrEcCiTJQucudHD9NLbRo3aJJoA6w-eXbp3_o5hmHBPxNWONOlAFnMIKQ3j9TLgvBcNzphqKyfoUL7C6mErfot9fWk2oxtYGnBvRmpElpWSTIms2Lr-_E0iSh1RrgpLL1hUom6PqtV28CxPVMSP1dH7Dv3HVyf5ueeTq_t307zjGBAthhuMlzPLajvGyVX8anNAGsO9pBOlnr3bU4dVb9ViiurzGmsIQ5Hy5cpcSnSThKnKS_nHuiTeFaCluKxcNPay4MTnfHfYLsR2MM-lCsxsNNE3a843Wt8_TbiQVGYeRrHkVoUIkVFs0zHe3n9PqDZ4k_zICGsZ_pynmN8moUy_h-DcuY0=w900-h600-no


The full test was photographed at fixed camera settings (1/20s, F/10, ISO200, 4500K WB). The following image should display as an animated GIF:

F4YtSz8wTwlwT_dBko4TjoMQa7NI6NSBbpFFOK-nBQDOOahkAdlOrccz4Vk9nd99-PIU-f8CkrGxPkpwC90lZ2X6eAIZ-4f_HcxsfA7owkaZTmZ9iZBuzeNz2-7Rz3UeQRaGfCYB0ULygUA1lhA9U0-ERU9PikUd6OP_V2YVwQ5CXBhGZitpTR--zopjdyxKimg1NcSK_vu8Eaa01tm9Y-51V2n1V2en7Sj3q3e1-dpCAreS4jSfi35LHWzZcaLOiTa93C603XEiqJd9uVWOgZcnLI9roWBiTs2oECHeeP339-rViI2uTR2GUa8sojFhNwaAK70QMyGaIbd2W0be6KVPJ6m4kGd7XALlfLkIeuhhI-cQFZvIZ6_OaMTPYjgAuGl6lPwVBpb7I0oBQj8uP7KnR2l8PrAT0wa7KLXLf4-GZSqAy1urqGzD11Qsb48EXX_RAvhIQYzTgW_d1GXOBL4bgAfzCylYnhHvFc_Lqggw-4llrUsqoEHXONx6KfN240psgjKq7r_EKJ5hCA9ZdDNcCDh7hxCokU41_AfZLeqR3UbK8slFUDZNEcG3HirOPh1AHOTjfFn4UrmZgu-1ak1xtGX1mGgQRudL7e3Ssmq2BZLEuSOHvH3FIzKDl9HpzNRgnv_KJwHb7p661_s4VFhPoEMsWgbh=w400-h500-no
 
Last edited:

Hondo

Flashlight Enthusiast
Joined
Oct 26, 2005
Messages
1,544
Location
SE Michigan
Awesome work iamlucky13!

What I would like to try is a 50 mA overdrive long term lumen maintenance test. But the only equipment I have is my free Harbor Freight multimeter. I know that is about the level a pair of broken-in Energizer lithium AA's drives one of these, so I suppose I could just turn it on and do periodic comparisons to a resting light. If I could go through a pair of batteries without noticing a loss of brightness, I would call it a win.
 

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
The thought of doing the same had occurred to me. I don't know if there's any concerns about leaving my power supply on unattended for a week straight, but if not, that would be a relatively straightforward test.
 

snakebite

Flashlight Enthusiast
Joined
Mar 17, 2001
Messages
2,721
Location
dayton oh
light one at just enough current to get a tiny glow and see what shape the die is.
i need to get some 3200k and a few of the 5600k to compare with the ones i got off ebay which i suspect are the same led yuji sells.
mine have a big rectangular die.
 

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
Perhaps whoever Yuji contracts with to apply their phosphor to an existing LED sells their standard LED on Ebay, but I can't imagine Yuji bothering to do so, given that they seem to consider the overhead of even their own web store only worth the hassle for the sake of getting their products in the hands of potentially high volume OEMs. I assume if Yuji takes on the cost of getting their phosphor applied to an LED, they would only sell it with their name on it.

As far as die shape, it is impossible to see under the coating of diffuser placed over cavity.

Here's a closeup where you can see this. Even the wires that power the LED are impossible to see after they enter that coating. A Yuji 3200K is on the left. The LED on the right is an unknown cool white, narrow beam LED taken from an Inova Microlight.

Hskka5wU_mjJxvXCuZmerN3STWpOAQG-7w9WehN7ihq4-1XYvXN3ojl_mTRpahgOOH5Pt_MLCA0gd7L1va88m0a7t3B8E1ZoignEo1Ay2MO8Oc6Ce8MvpqhZryivucAxjkOPt90xyDaOE4bOjEmhjbfWa8U1P-TCSWpbPM5_d9uLlyspOFiyYBlUvFD2qr38c7B5iU4oMOFlw5irZnU4IXPyWPlf08xzrqxs6IybmcAw9BuJWflS0IcdPHGc78ZHx6xrLRDXKCIc8wmI64jhJSRwfzk-IhY8lVxoiac1ER3W9h3vcCiDH0tb95PcoRzqirmk-dAyx8Q7bnwbjxMU-Rp-4YiLX8BtpV4HgWl6lEhPaq2y7I4cG17v1VrUCMebtnHLfRRqnvBon7VXNGS2UrryjbhUZjFI-2rVGowVCQJvUkkSk5OdbQiB4LKhNzqnD3aDumMXm0G9dfAkg0qi9__VOq7430N3dlp3RUSyV5VjDoig6WAA0rhvGEE1BbmzbwpJcim0V1hbri1CNW1EMSnsKc2PTv2TW84yXwH_TqcyAw4Us8T1ZOYvi5a9wpFUYKusbsInuyTL6AbT4l21x7jfTcUlAT-pPaJcN0cotOeGA8q3PR8l=w900-h600-no
 

snakebite

Flashlight Enthusiast
Joined
Mar 17, 2001
Messages
2,721
Location
dayton oh
that pic shows me that the ones i have are not the exact same led.
mine have a crosshatch pattern stamped into the metal.
to see the die shape you must power the led at a very low level.
then you can see its shape.
i need to get my hands on a few for testing/comparison purposes.
i had some in hand about an hour that a customer had me swap into a 100 led photo light just before a really important trip to do work for a client.
these first gen photo lights had a nasty green/yellowish green peak and were very low on red.ridiculous they were sold for photography with such poor leds.
and worse yet in the few hours they got used for mainly work areas they faded down a lot.
i could not tell any real difference between the ebay leds and the yuji in the short time i played with the unit to test it.
i did up several of his panels with the ebay leds and he stated that they seemed equivalent to the yuji.
i never bothered to get any yuji due to the high cost+expensive shipping.
he is getting fresh macbeth charts due to his being 2 years old and will send me the old ones to play with.
too bad the ebay source does not have 3200k!
 

bykfixer

Flashaholic
Joined
Aug 9, 2015
Messages
20,351
Location
Dust in the Wind
Thanks for doing this 13.

In a completely unscientific platform using 54 year old analog eyeballs, I set out to do a dimming test with a Yuji'd minimag fed by the supplied Duracells.
IMG_20180311_100138.jpg

f3.7, 1/14s, ISO 800
Started this morning around 9:00am (that's 8:00am for those of you without the forced government daylight savings time).

By 10:00 I would've noticed dimming with a light bulb.

So far no dimming has been noticed in the output of the 3200k Yuji from the Scout24 group buy.
Note: my unscientific method is to look at the original photo about once every hour. Taking notice of the outter edges of lit area.
Photo was taken using a Hasselblad 4116 motomod point n shoot camera attached to my celphone.

Edit: 12 noon no dimming noted. Will do a full on runtime I suppose. End edit.

3Pm edit: some dimming, but not a lot.
IMG_20180311_154732.jpg

f4, 1/14s, ISO 400

6:45pm edit
IMG_20180311_190354.jpg

f1.7, 1/14s, ISO 1600 stock phone cam this time.
Still no discernable dimming.
Note, the series of photos have been taken near a window. The 6:45 pic shows the sun coming in the window was nearly none.

9:45 edit:
IMG_20180311_214458.jpg

Still going strong...
f1.7, 1/14s, ISO 1600

IMG_20180311_224141.jpg

f1.7, 1/14s, ISO 1600
One more before calling it a day. No significant dimming noted yet.

IMG_20180311_224357.jpg

f1.7, 1/14s, ISO 2500
Put it in a tube tailstand style and called it a day. Left it on overnight knowing if it turns off or burns out it'll make a mighty fine hurricane lamp using about 50¢ worth of fuel per 12+ hours.

24hrs later edit:
IMG_20180312_094330.jpg

f1.7, 1/14s, ISO 12500
This'll be the last photo. But suffice to say, these Yuji's are certainly fuel sippers. And the output is plentiful in darkness.
At 4:30am I checked on it and thought "holy smokes, imagine this in a Mag C or D incan that uses their bi-pin. The runtime could last for weeks perhaps."
 
Last edited:

ma tumba

Flashlight Enthusiast
Joined
Oct 7, 2013
Messages
1,344
Location
Russia
I put 12x 3200k yujis to a cheap uv host. With an imr battery the light, with all internal resistance the current at tail was 600ma or 50ma per led. In a few days of infrequent use three leds died and the fourth was damages so it would severely flicker unless a 3v battery was used.
 

snakebite

Flashlight Enthusiast
Joined
Mar 17, 2001
Messages
2,721
Location
dayton oh
bummer.
the ebay ones can be beat on like that with impunity.
you might rebuild the board with fresh leds and somehow add maybe a 1 ohm 1/2w resistor to tame it.
maybe eliminate the center led and put the resistor there.slice traces and reroute to the + from the spring pad to the + of the rest of the array.it has the worst heat path to the board traces of all so no big loss.
i did that to some of my 9 led mods to stretch the runtime.
on a few i used a 2.2 ohm.
brightness drops a bit but not that much.
the ebay leds i linked https://www.ebay.com/itm/Hot-500Pcs-F5-5MM-White-Color-White-Light-Round-Superbrigh-t-LED-Lamps/131899349854?hash=item1eb5d05b5e:g:QFgAAOSw5ZBWHcBR
are still twice as bright as the cheapies that come in the free hf lights.

I put 12x 3200k yujis to a cheap uv host. With an imr battery the light, with all internal resistance the current at tail was 600ma or 50ma per led. In a few days of infrequent use three leds died and the fourth was damages so it would severely flicker unless a 3v battery was used.
 
Last edited:

Hondo

Flashlight Enthusiast
Joined
Oct 26, 2005
Messages
1,544
Location
SE Michigan
Last edited:

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
^ Yep. I didn't mention it before because I just assumed the group buy would have been noticed. While still more expensive than I'd guess you paid for the Ebay emitters, it's a very reasonable price to get your hands on some of these Yuji's to tinker with.

I put 12x 3200k yujis to a cheap uv host. With an imr battery the light, with all internal resistance the current at tail was 600ma or 50ma per led. In a few days of infrequent use three leds died and the fourth was damages so it would severely flicker unless a 3v battery was used.

That's useful to know. It seems likely the LED's shared the current unequally due to normal variation in forward voltage, so presumably the first one to die was getting more than 50mA.

I still haven't had a chance to set up an overdriven endurance test. I was thinking 50mA might be a good starting point, but based on that, maybe I'll try 40mA, and wait to increase it if I don't see any significant dimming over a couple days.

Hondo, thanks for your test, too. I figure these should run a long time in a mini-Maglite, just very gradually dimming as the voltage drops.
 

snakebite

Flashlight Enthusiast
Joined
Mar 17, 2001
Messages
2,721
Location
dayton oh
current hogging is always a possibility with a bunch in parallel.
it would be interesting to see vf variations in the yujis.
i would assume for a high end product they would use high quality dice,ect.
not that these are bad they just may have vf variations and may not be very abuse tolerant.
i saw the gb for these and was waiting to be able to get 3200 and 5600 at the same time.
comes down to shipping even at usps 1st class being a significant part of the cost of a small qty.
i dont really need 5600k as the ebay ones seem really good its more for a-b comparison.
i have a use for some 3200k so i may just forget the 5600.
or if someone like minded got a bunch of the 5600 a swap could serve both parties.
it would be especially nice if someone had a spectrometer.
Craig at the led museum had one but not heard anything for a long while.
That's useful to know. It seems likely the LED's shared the current unequally due to normal variation in forward voltage, so presumably the first one to die was getting more than 50mA.

I still haven't had a chance to set up an overdriven endurance test. I was thinking 50mA might be a good starting point, but based on that, maybe I'll try 40mA, and wait to increase it if I don't see any significant dimming over a couple days.

Hondo, thanks for your test, too. I figure these should run a long time in a mini-Maglite, just very gradually dimming as the voltage drops.
 

ma tumba

Flashlight Enthusiast
Joined
Oct 7, 2013
Messages
1,344
Location
Russia
Also, when one led in parallel configuration fails it puts more stress on the rest increasing the odds of the next weakest to fail and so on
 

bykfixer

Flashaholic
Joined
Aug 9, 2015
Messages
20,351
Location
Dust in the Wind
IMG_20180309_173402.jpg

3200 vs 5600.
No scientific data to prove it but in side by side minimag and P60 testing the 3200 seem brighter in low light such as a back yard in the city with all that light pollution). Better throw too.
In total darkness things seem equal.
 
Last edited:

staticx57

Flashlight Enthusiast
Joined
Dec 2, 2011
Messages
1,749
Location
NJ
current hogging is always a possibility with a bunch in parallel.
it would be interesting to see vf variations in the yujis.
i would assume for a high end product they would use high quality dice,ect.
not that these are bad they just may have vf variations and may not be very abuse tolerant.
i saw the gb for these and was waiting to be able to get 3200 and 5600 at the same time.
comes down to shipping even at usps 1st class being a significant part of the cost of a small qty.
i dont really need 5600k as the ebay ones seem really good its more for a-b comparison.
i have a use for some 3200k so i may just forget the 5600.
or if someone like minded got a bunch of the 5600 a swap could serve both parties.
it would be especially nice if someone had a spectrometer.
Craig at the led museum had one but not heard anything for a long while.
I have 5600k which ill try to get to tonight but here is a 3200k I tested. Note I did test this in a modded battery junction 2 coin cell deal. Ill hook the 5600k and retest a 3200k in a more controlled manner.
uKzdInY.png
 
Last edited:

staticx57

Flashlight Enthusiast
Joined
Dec 2, 2011
Messages
1,749
Location
NJ
Test Equipment and yes I did not test like this, I did turn off all the lights and blocked any incident light ;) :
BbkMJg3.jpg




Results: Do note I did try a few LEDs from the batch seeing as the 2700k seemed low but they all provided the same results. 5600k is clearly superior. Also note, the 2700k I posted previously is from ma-tumba's order from last year so it's different.
XaBj2iC.jpg
 
Last edited:

dealgrabber2002

Flashlight Enthusiast
Joined
Nov 9, 2007
Messages
1,908
Location
California, Los Angeles
IMG_20180309_173402.jpg

3200 vs 5600.
No scientific data to prove it but in side by side minimag and P60 testing the 3200 seem brighter in low light such as a back yard in the city with all that light pollution). Better throw too.
In total darkness things seem equal.

Bykfixer,

You just insert the Yuji into a Minimag 2AA? Looks way brighter than the 5mm of eBay I used in my 2AA maglite.
 
Top