11
\$\begingroup\$

I know this is not a huge deal, but still I want to do it because for Three reasons: 1, it only costs 1 usd. Yep, this is a very cheap led bulb, and I need to take it apart to see if it's trust worthy. 2, It's a great opportunity to learn something new and exciting. 3, I need some parts for my Led project.

(Update: I actually found out that this $1 bulb is made by uninex. A comparable 8 watts costs $7 usd in their website, so I think I got a good deal here. It's probably being subsidized like one commenter has said.)

With that being said, I took apart the thing, and my first impression is impressively surprising. This bulb use better parts than all those other more expensive bulbs that I have took apart over the years. I am super skeptical whether this is really a 11w led bulb or not because counting all those SMT leds doesn't really add up to 11w anyway, and that the outer case is actually plastic, the only heat sink is that Aluminium disc. See photo below, Q1: Can this really dissipate 11 watts?

enter image description here

Other than that, the manual soldering connectors are crude. Not really complaining here, it should works fine. The PCB looks very nice and the parts look New and high quality to me.

enter image description here

I also notice there are few parallel resistors pairs, like a lot of them, which is a very cool design that I probably will be copying soon in my future project.

enter image description here

The One part that I don't really know, but I am sure that I have seen them in other PSU before, is that green round thing. Q2: what is that?

enter image description here

Finally, There is my trace circuit for this LED bulb. Other than that green thing (I assume it's some kind of fuse for the moment), the top part is easily understood, but the bottom part is more than confusing. I know the bottom part is a high frequency switch, but Q3: how does it really work? (Note: I am not too sure if B is C or C is B, but E for the NPN is correct.) There is also a 63v 330uf capacitor between the 473 resistor. The circuit seems very simple, but I don't really get it. It's not an isolated design, So how does it step down a 120v to 65v and less?

enter image description here

Update with corrections and voltage read-out: One of the diode is actually a 51 voltage zener if I am not wrong. The transformer measures 0.4:3 while in the PCB. So for Q3: How does it drop down from 104vdc (Point B) to 50vdc (Point LED-)? I don't really see a current path here. (The leds are 6 in series and 3 pairs in parallel, total 3x6=18. 3.6vX6=21.6v, but voltage read out is 50v, very strange.)

enter image description here

To aid discussion: enter image description here enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ I suspect Q2 is asking about a resettable fuse: en.wikipedia.org/wiki/Resettable_fuse \$\endgroup\$ – calcium3000 Nov 14 '17 at 22:34
  • 3
    \$\begingroup\$ The "green thing" is likely a surge suppressor. \$\endgroup\$ – Misunderstood Nov 14 '17 at 22:42
  • 1
    \$\begingroup\$ It,s possible this is being subsidized by the power company. New power plants cost about $5 a watt give or take. It is far cheaper to help consumers save a watt than to build a new power plant. Also, there's no reason to buy Cheese cheapies, I often see quality namebrand (i.e. General Electric) 25,000 hour units for under $3. Once I got a box of 8 BrightStix for $6.80 at Home Depot. GE stuff is exactly what it says on the tin. Target sells GE and their own brand. \$\endgroup\$ – Harper Nov 14 '17 at 23:41
  • 1
    \$\begingroup\$ @Atmega328 that sounds like a rationalization. Why can't you fix quality products? Do you really fix cheapies or do they pile up in your someday pile? How fixable are they really? Cheap PCBs that crack, cheap traces that lift. None of which addresses my core observation, which is that I can get quality cheap too. \$\endgroup\$ – Harper Nov 15 '17 at 4:19
  • 1
    \$\begingroup\$ @Atmega328 No, expensive, Western made products tend to be easier to tear apart, because of all their labor saving designs. Check out US-made CREE bulbs, those things have clips and spring contacts instead of labor-intensive hand-soldered wires and screws you show. \$\endgroup\$ – user71659 Nov 17 '17 at 19:12
5
+50
\$\begingroup\$

I give it a shot, and I try to reverse engineering your LED driver circuit.

First what I did was to draw components on the PCB photo.

enter image description here

And now "without any problems" I could draw the circuit diagram.

schematic

simulate this circuit – Schematic created using CircuitLab

As you can see we have modified version of a singe BJT blocking oscillator (joule thief).

http://mmcircuit.com/understand-rcc-smps/

https://skootsone.yolasite.com/led-en.php

http://www.powerelectronictips.com/teardown-60-w-equivalent-led-bulbs/

Also can you post the diodes marking? So we will be able to tell exactly what type of a diode was used.

\$\endgroup\$
  • \$\begingroup\$ Part number for 3 diodes: D4=M5; D5=E51 (This should be a 51v zenzer); D3=ES1JL \$\endgroup\$ – Atmega 328 Nov 19 '17 at 22:14
  • \$\begingroup\$ Part number for 3 diodes: (D3)D4=M5; (D2)D5=E51 (This should be a 51v zenzer); (D1)D3=ES1JL; I have double checked your Schematic. It should be correct. \$\endgroup\$ – Atmega 328 Nov 19 '17 at 22:30
3
\$\begingroup\$

I just wanted to add a few opinions about the light in general. I think the above analysis of the driver is very good.

A lot can be learned from the UL markings on the bulb, not the least of which is the actual manufacturer. The wattage rating on the outside of a product is the maximum power used, usually during startup and may not reflect the running wattage of the the product. This is for safety so you calculate worst case scenario for circuit load in your house.

If the bulb is marked 11W, that's the power consumed, not the power used for the LEDs. There is minimum efficiency required of about 83% on small AC/DC converters. 83% of 11W is 9W with 18 LEDs at 0.5W each. Things seem to add up pretty well on that front.

Since it's UL approved the heat of the product must meet certain requirements. Max 85C for ABS etc. The materials for the driver PCB (see note below) and LED PCB are tightly controlled. The driver itself needs to dissipate about 2 watts (83% efficiency) and the LEDs are producing light (let's assume 15% efficiency) meaning the LED PCB/heat sink must dissipate about 7.7W, which is a lot lower than 11W.

A comment about "Hot" to the touch. Nerves aren't good thermometers but they are good heat sensors. We know something is hot when we touch it, but we don't know how hot. An LED PCB at 60C, 75C or 85C is all very hot to the touch and can't be distinguished by touch alone. 85C would be a bit hot for the LED board, but it would still be considered safe.

The Driver PCB wouldn't have been hand soldered but wave soldered. You can tell it was wave soldered by the red glue under the SMD components. They are glued down, then put through the wave soldering machine. It looks well soldered to me.

All the major components of the driver also have their own UL approval and of reasonable quality. The PCBs must also have UL approval (you can see the RU mark on the aluminum LED PCB). There is a case where the FR4 PCB wouldn't need it's own certification, and that's if the temperature during testing is low, the UL engineer can choose specify the material only if there is no safety concern about the temperature.

I can speculate the light bulb could be produced for around $1, but sales channels need profit, there is shipping to consider also. I would think this bulb in an online setting should retail for about $2. This is my opinion only, not fact.

I wanted to add a few more comments.

It's unfortunate this is being labelled a bad bulb. The construction quality is pretty good. Hot to the touch should not be the basis to decide the quality of a bulb. Use a thermocouple and measure the actual temperature. A UL approved product is safe and won't cause any damage through heat. This bulb was built to a price, but it is far from low quality. The majority of bulbs die from the driver failing, not the LEDs failing. LEDs typically "fail" by slowly emitting less light. The failure point is called the L75 rating. This is when a bulb emits 75% of the light it did when it was sold. This is the rating in thousands of hours we see on all LED products we buy. Generally the driver will fail long before this is ever a concern.

\$\endgroup\$
  • \$\begingroup\$ The design uses an electrolytic capacitor. Best aluminum caps have durability rating of about 10000 hours at 105 deg.C. The bulb is destined to die in about one year. How the "UL approval" can change this? \$\endgroup\$ – Ale..chenski Mar 22 '18 at 8:58
  • \$\begingroup\$ Why would it see 105C at the capacitor? at 80C it would last many times longer than that rating. I would doubt they use caps rated for more than 5000hrs. \$\endgroup\$ – B. Barrett Mar 23 '18 at 10:06
  • \$\begingroup\$ Sensing the outer shell temps will show the minimum inner temp of the cap . Arrhenius effects will be worst on the caps and 50k LED specs are usually fake \$\endgroup\$ – Sunnyskyguy EE75 Mar 23 '18 at 15:37
2
\$\begingroup\$

My question is about whether that tiny disc can dissipate 11 watts of power from those LEDs

I trust your measurements validate the power consumed is 11W.
6V white LEDs use 2 chips in series on the substrate.

My Rule of thumb for convection cooling is 1 sq.in/W in free air. I see the 1" Edison and estimate the heatsink approaches this metric. The only problem is the lack of free air inside the bulb with unknown surface area on the external heatsink, if anything.

Conclusion:

Thumbs-down

  • sell cheap for rookie buyers to deplete stock and obsolete the warranty and dump on eBay.

.if you agree or understand then up vote.

\$\endgroup\$
  • \$\begingroup\$ Kinda suks, but most LED bulbs that I bought have this over-heat problem anyway. Even those with a large heatsink get very hot to touch after 30 minutes. So I guess, it will take a huge heatsink to make these led bulbs last, and that means much expensive bulbs, which transfers to harder to sell. \$\endgroup\$ – Atmega 328 Nov 20 '17 at 2:45
  • \$\begingroup\$ Worse yet, many are put into enclosed ceiling fixtures with no convection flow to ceiling. The best have large surface area fins. \$\endgroup\$ – Sunnyskyguy EE75 Nov 20 '17 at 23:00
  • \$\begingroup\$ Well is this worth 1 dollar? According to the package, 22.8 years life based on engineering test, but they only give you a 3 years warranty with normal use, and I believe normal use is 3 hours per day and 7 days a week. As a normal bulb, I am not sure if it's worth it, but for me, it's pretty nice parts and education there. \$\endgroup\$ – Atmega 328 Nov 21 '17 at 9:47
  • \$\begingroup\$ $1 is certainly less than competitive bulbs using ($0.09/W) its a bargain and very low margin above material cost. Hence a selloff at cost. \$\endgroup\$ – Sunnyskyguy EE75 Nov 22 '17 at 1:59
  • \$\begingroup\$ The disc only needs to dissipate 9W, the driver itself is dissipating the other 2W. \$\endgroup\$ – B. Barrett Mar 23 '18 at 10:10

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.