Cree 2018 '60W' Led Bulb Teardown
Vložit
- čas přidán 1. 12. 2018
- More details:
electronupdate.blogspot.com/2...
A look at a bulb from 2018. Bought from Home Depot
Model # TA19-08027MDFH25-12DE26-1-12 Store SKU #1003004374
A good example of a modern controller drive for costs far below anything I have seen before. - Věda a technologie
Your cross-sections and microscope views are delicious. I love what you do and I'm glad you still make videos from time to time.
It may be possible with a camera but you'd certainly need one with raw capability. Automatic white balance would make any measurement impossible.
And a very good high magnification macro lens.
Something has gone wrong here...
The comment at the top of the thread here isn't the one I replied to originally by GooglePleaseStahp (ironic) about measuring CRI with a camera.
The "transformer" might be a choke. I would check the inductance, and also check to see if the extra pins are just for mechanical stability.
Had the same thought. Direct off-line CC buck is pretty common topology for LED drivers.
I'd have presumed that he'd have noticed if two of the legs didn't have a wire going to them, but....
EDIT: looked at the video again, there are three distinct copper pads connected to the transformer - two of the leads are connected together. Tapped buck?
I'd be curious to see some impedance measurements on those terminals
I agree. I dropped an LED bulb once and it stopped working. Opening it reviled the choke was held by two leads and one lead was broken off. Four leads would make a better mechanical connection.
Yes, usually on the output of these non-isolated Buck drivers there is some form of choke. At 4:57 you can see that only only 2 of the pins have windings coming down from the former, so this is probably just a regular inductor. Other comments below say they see 3 leads, but nevertheless it's certainly not a "transformer" to alter voltage.
I've found that the manufacturers tend to use inductors wound onto a bobbin with E or C-core rather than using a circular inductor (like used on the input side), which may well be related to getting a low enough ESR or some other technical/cost reason.
Likely a buck converter that has a choke with two extra legs for mechanical stability. This would also make sense with regard to the circuit. Putting a secondary winding of a transformer between V- and the LED cathode, while having the anode connected to V+ would make no sense. The secondary outputs would be across the LEDs. Easy to check: are there any windings actually physically connected to the two pins that are commoned to the negative bus?
it does say T1 next to the part though, what is usually used for chokes?
@@fredlllll Usually L is used for chokes, but that doesn't really say anything. For example, they might have been lazy and used an existing transformer footprint in the board design software because it already had 4 pins, instead of creating a new footprint just for this device. But again, the only way to know for sure is to open it up and look if anything is connected to those pins.
It could also be a sense winding that's placed on the inductor for a buck converter that works in CCM. Technically that's a coupled inductor, but would normally be called a transformer.
@@misterhat5823 It could not be a sense winding in this case based on how it's connected on the board.
Thank you so much for tearing down this new Cree bulb. What a difference in construction after just a few years!
We purchased a bunch of LED bulbs last month, including this exact bulb under teardown.
The new Cree (60W, 90+CRI) has an impressive finish and the heatsink is heavy. We tried them in several rooms and we did not like the light quality of either the 2700K or 5000K; we felt the colors were wrong and rooms felt irritating for whatever reason. We found the similar Philips bulbs (60W, 2700K dimmable, 80CRI) to be much more pleasant...similar to regular incandescent bulbs. Not sure what Philips is doing special. . . but that might be worth some teardown also!
We found the 5000K bulbs from both Cree & Philips to be terribly irritating, even when mixed with several 2700K bulbs. Comparing both Cree & Philips to incandescent bulbs, the 60W equivalent ratings seemed a bit ambitious & both LEDs had light distribution biased heavily upwards.
We also picked up some new Cree candelabra (60W, 90+CRI, 2700K, clear). These Cree bulbs look really neat with LED srips, the light distribution is good, and we find the light very pleasant. The bases are impossibly small so I don't know if flicker would be an issue.
I prefer 3000K. 2700k is too yellow.
@@simontay4851 That makes sense. We tried to mix a 5000k bulb with some 2700k bulbs to reduce the yellow but that failed
Thanks for returning to LED bulb tear-downs.
It's not a transformer, it's a buck converter, so just an "tapped inductor". The last pin are for mechanical stability, and because the transformer bobbin comes with 4 pins as standard. Look back into the video, you can clearly see that there is wires soldered to three of the pins.
Thanks for making these videos! I always learn something interesting.
I did a Cree LED teardown a while ago and disappointingly, all of the 6 LED bulbs I bought failed within a year. I won't trust that 22 year claim.
Yea the newer designs are just terrible. I haven't had all 6 fail out of the box but I have had a high failure rate.
Thank you for continuing to do these teardowns. It's rare to find led bulb reviews that give useful information.
Love your very in-depth teardowns! That transformer also had me quite perplexed.
Electron Dude, thanks for the very informative video.
It was only a matter of time before Cree became 'cost optimized' :(
Lovely video! And perfect cuts!!
That does seem to be a 4pin choke, so, a buck converter we're looking at
Interesting teardown👍
Thanks for sharing👍😀
Hi. What about flicker when is at 50% (with dimmer)
Osram (here in UK/Ireland) has frosted glass LED bulbs, which are basically LED filament bulbs with a frosted glass. Based on my usage, the light is about as close as I've come across to the old incandescent frosted bulbs. I even had someone comment wondering where we got the old bulbs (as they're banned in the EU) and surprised to hear they were LED.
I would be interested in the temperature rise on the casing vs. other LED bulbs. Some of these operate near 100C even without being in an enclosed fixture.
IME, both the Cree's and the Philips' metal heatsinks gets proper hot. After 10 minutes of use on my table lamps, I found the bulbs too hot to remove with the bare hand. Turning off the light for a minute or two, the heatsinks cooled significantly and bulbs were just warm to the touch. I suppose that indicates plenty of heat is generated and that the heatsinks work...to an extent. As the bulbs are an enclosed ecosystem, I would prefer a finned heatsink that had better ability to dissipate heat so run cooler, extending life of the internal components. . .
What are the differences between the old dimmer and the new one. They both use the classic diac+triac dimmer circuit, i assume.
I have two or three early (maybe even 1st gen?) Cree bulbs that have all developed some sort of failure with (I suspect) the power supply. Cree reliability early on was terrible, the bulbs were purchased at different times but they ended up failing a lot. Would love to send them over and understand why, if you're interested. Great channel, been watching for years now and I love the lighting teardowns!
Been waiting for you to take a look at this. How are your original Cree bulbs doing? Most of mine have died or flicker terribly. I wonder what failed. I have one left if you would be interested in taking a look, just let me know
Probably failed capacitors.
How old are they?
All my LED bulbs are soo hot they will burn you if you touch them then if you put them in a enclosed bulb holder they bake even hotter its going to be the capacitors or the LEDs got baked from heat Those bulbs just cant dissipate their heat
Cree's first few generations of bulbs all had terrible long-term reliability. A significant portion of mine failed, and at this point I've replaced them all with non-cree bulbs. I wonder if moving to a generic Chinese design will help or hurt reliability?
It's worth noting that you can still find attempts in ultra low cost LED bulbs to spread light to the sides: I cracked open a Walmart bulb recently to steal its diffuser, and it had a metal reflector inside.
Ha, I just offered to send in two or three Cree bulbs that died in various ways, many not being abused in closed fixtures or anything.
I picked two of these up the other day, and to my surprise the color (while being 5000k for both styles of bulb) is slightly different than the 4Flow style I have right beside it in the same room.
Having the metal external heatsink cree bulb, then 4flow, and now this I must say the changes have been extreme in design style.
It’s quite disappointing that Cree was once the leader with a mechanical design that avoids hand soldering and now Cree uses hand soldering.
I highly doubt that Cree had anything whatsoever to do with the design of this product, other than supplying the artwork for etching onto the PCB and for the packaging.
this style of bulb with the leds near the edge and the dome slightly more than hemispherical seems to distribute light well enough for most applications and is cheap and easy. i have a couple Ecosmart branded bulbs from ~2012 that have the leds in the middle and a smaller dome that's a bit more clear and virtually no light comes from the base end of those
do these change colors cause u bought some but im lost
I wonder how long it will be until laser diode based illumination supersedes LEDs. They should be able to produce the same amount of light as a bulb like this with
What? Laser diodes are less efficient than regular LEDs, and like LEDs they are monochromatic so for white light the phosphor would still dominate the efficiency calculation, assuming you want high quality white light. Lasers are highly directional and produce coherent light, which are disadvantages in lighting applications, since you want to spread the light out, not burn someone's retina, and you don't want weird shimmering on reflective surfaces. For lighting applications, the only reasonable use case I can think of would be a decorative dot pattern using optic fibers, think a starry night sky with individually lit stars.
@@Gameboygenius nah, laser diodes are way more efficient at gain extraction and don't suffer from efficiency droop like leds. Laser diodes have much higher input power densities than LEDs and parasitic nonradiative recombination processes, including those likely responsible for efficiency droop in LEDs, are clamped at low rates at the lasing threshold. You can pump 2,000 X more current into a laser diode than you can in an identically sized led. The other things you mention aren't problems. The phase coherence of the light is destroyed upon fluorescent conversion by the phosphor so speckle isn't an issue and the directionality of the original beam is easily mitigated with microdiffusers or simply by making a large number of individual laser nanoemitters on a surface. Look up laser diodes for general lighting. There's a reason high end auto manufacturers are already doing it for headlights. They're going to supersede all other forms of lighting eventually.
Concerning the inductor/transformer, I think it's probably an inductor for a buck regulator and not a transformer. Because there are many LEDs in series, there usually is something like 60v which is depending on country half or quarter of the mains voltage. Btw if it was a transformer, having no insulkation isn't much a problem because no galvanic isolation is needed. Lastly I would say that I'm a bit concerned about the size of the control chip and especially its power transistor.
That transformer might be a choke part of a buck converter as it's usually the case with led drivers. I wouldn't consider that a safety issue since the circuit is enclosed safely inside a thick enclosure.
Is this IC a custom Ic from Cree.
I am Not able to get the Data sheet of this IC.
Now this is a teardown.
Some of those older dimmers may cause flicker issues with LEDs. The newer dimmers may be better but still can be optimized.
What was the model of the dimmer?
Can you measure CRI with a camera and photoshop analysis or is a spectrophotometer necessary?
You need a spectrophotometer
You also need that the calibration is checked / confirmed. The photosensors are non-linear depending on both wavelength and intensity, so the CRI calculations need to have been made on a correctly adjusted (and normalised) intensity@wavelenth values. Small variances in measurements at certain wavelengths can make huge differences to the calculated CRI. In other words, without being measured in a certified lab, the results are liable to be completely erroneous.
Adding to that, the whole "CRI" and "Ra" measurement method is completely and inherently flawed for LEDs anyway, so a high stated CRI doesn't necessarily translate to colour accuracy. You can have a LED bulb with a "black hole" in the spectrum yet it'll be calculated as having CRI>80. The non-continuous spectrum of LED bulbs mean that even is there is literally zero Red (R9 value) the CRI can be high (especially when calculated with the older Ra systems), yet the bulb will not reproduce skin tones and other red objects correctly.
Nice 👏👏
Pretty nice!
7:07!! hmm if it live wire that not switch off😁
amazing what inside a tiny chip.
All you're lacking is an integrator sphere to measure lumen output.
BTW too bad they can't get Cree bulbs here. I guess they're only marketed in America.
I didn't know it was possible to dim LED bulbs with conventional dimmers. How does the dimming work? Won't the rectifier and switching regulator in the bulb negate the effects of the dimmer?
You might suffer quite a bit of stress and flicker on the bulb with the older dimmers. The newer LED dimmers are more expensive and work better with LEDs in my experience but they are still being refined. . .
Newer LED bulbs have circuitry inside the controller chip that detects the chopped waveform and decreases light output
@@AmericanLocomotive1 Thanks for the reply. So they only work with triac-based dimmers, and not resistive or capacitive dimmers? Would be interesting to know how the detection circuitry works. All this happens in the controller IC?
Considering that the IC is after rectification and step-down, how does it sense the chopped waveform?
Hou yhea! A video like the old times...
Seems like they've further pared down the costs, ripped apart a new Cree bulb and the PCB looks almost identical to this but there's a LOT less metal in the heat sink.
It's a shame the efficiency of these lights are so abysmal. barely 80 lumens/watt. Modern linear LED "bulbs" are around 140 lumens/watt these days.
Can't you notice that the base, "GETS TOO HOT"
.
Curious why the filament lamps haven't taken over these. They seem to be vastly simpler and have an excellent reproduction of the old lamps in terms of light output.
They are essentially reusing the designs and tooling for the directional PAR and BR led lamps for these A-shaped lamps. The same factory can easily pump out both types of bulbs. Filament LEDs are often made with old machinery used to make the glass envelopes for incandescent lamps. Companies that previously manufactured incandescent bulbs happen to own this machinery or able to subcontract it out cheap will have an advantage over a manufacturer like Cree who don't have that tooling.
6:22 flicker rate
I prefer filament type LED lamps over all others. They might not be the most efficient, but they emit light equally in almost all directions.
They are even MORE efficient than this style. An LED filament 800 lumen bulb will consume 7 watts.
@@Muonium1 yup, no diffusor to waste light into
10 year warranty. Send it back under warranty and see if you can get your money back. :D
I really like these videos but I have to say watching the decline of Cree's manufacturing really saddens me.
I've actually switched my house lights from LED to halogen and some incandescent. Much better lighting.
Solar cells don't exactly have the best AC performance...
Fortunately, neither do your eyes
@@CerealKiller951 well yeah, but if you're gonna do a scientific test, you might as well use a good high bandwidth photodiode since some people get sick with PWM'd lights even if they can't perceive any flicker