Smart lightbulb LED runs at 37 volts?

[rasmasyean]

Hi,

I was puzzled by this finding.
I opened the cap of a smart LED light bulb and probed the individual SMD LEDs. It seems to go up to 37 volts at max brightness. This is on the color LEDs as well.
I was thinking of replacing one of the colors because it was flashing. It looks like LEDs are 3V.

Can someone explain this?

Thanks.

 

[JoakimFlorence]

I was thinking of replacing one of the colors because it was flashing.

What exactly was flashing? One of the LEDs or all of them?

The interesting thing about constant current drivers is that, when a long string of many LEDs is being powered, you can often remove one and all of them will continue to work just fine.
Constant current power supplies will adjust the power supply to maintain a specific current (mA) level, by modulating the voltage.
(But they often have a limited voltage range, so this ability to adapt is not unlimited)

Red LEDs typically have a voltage of 1.7 to 2v. Which likely will not matter to this situation.

I don't really know what I'm talking about and could be wrong, but it could be possible you are not able to accurately measure how many volts it has at full brightness. Explaining exactly how you came to the conclusion it goes up to 37 volts could be important. Remember, the power supply might be reducing voltage to maintain a constant current, and if your voltameter has less resistance than the total LEDs, we might expect it to measure less voltage.

 

[Dave_H]

Hi,

I was puzzled by this finding.
I opened the cap of a smart LED light bulb and probed the individual SMD LEDs. It seems to go up to 37 volts at max brightness. This is on the color LEDs as well.
I was thinking of replacing one of the colors because it was flashing. It looks like LEDs are 3V.

Can someone explain this?

Thanks.

By "smart" bulb I presume you mean WiFi controlled?

SMD LEDs used in ac light bulbs typically consist of multiple LEDs connected in series. I have seen 3 LEDS (vf = 8.5-9v) up to 9 LEDs (around 27v). This can vary according to vf of individual LEDs, current level, and temperature. So 37v looks like 12 LEDs per package.

RGB bulb I am looking at now has three SMD per colour, each totals 33v. The driver will adjust total voltage for the set current. The $1 Ikea 40W eq. bulb I opened supplies about 80v dc to nine SMDs, which means three LEDs per package.

Dimming bulb with PWM will reduce the average voltage measured across each part. Accurate vf needs full brightness.

If you replace SMD LED with one with a different total Vf it might work but you need to be careful.

Dave

 

[rasmasyean]

OK, so I actually blew that WiFi RGB bulb away while testing it. lol

It's OK, I just so happen to have another with really bad Blue flickering and opened it. I think it's a manufacturing batch problem.

And I probed the different color SMD's with a multimeter and found these voltage ranges at brightness extremes.

White: 10v - 16v
Red: 18v - 37v
Green: 17v - 32v
Blue: Jumps around (seems to briefly go up to 28v)

The Blue SMD's seem to flicker in unison. Would this mean it's likely the electronics, and not the individual SMD's that cause the problem?

Also, if these are driven by PWM, would you need an oscilloscope to see what the actual peak voltage it's using?

 

[rasmasyean]

I decided to open up and probe one which doesn't flicker and I get:
Blue: 18v - 34v

The low range is actually fairly large from 16v-18v, depending on the specific SMD. Is this an indication of current control?

 

[Dave_H]

OK, so I actually blew that WiFi RGB bulb away while testing it. lol

It's OK, I just so happen to have another with really bad Blue flickering and opened it. I think it's a manufacturing batch problem.

And I probed the different color SMD's with a multimeter and found these voltage ranges at brightness extremes.

White: 10v - 16v
Red: 18v - 37v
Green: 17v - 32v
Blue: Jumps around (seems to briefly go up to 28v)

The Blue SMD's seem to flicker in unison. Would this mean it's likely the electronics, and not the individual SMD's that cause the problem?

Also, if these are driven by PWM, would you need an oscilloscope to see what the actual peak voltage it's using?

If the blues are all flickering, it could be the driver, or one or more of the LEDs themselves. If your bulb has linear driver it is likely an 8-pin SOIC with markings. Mine is KP18058.

Your red, green and blue look similar to mine, multichip with total vf around 30v or so. White looks like 6 LEDs per chip (you might see two die with light off or highly dimmed). If you can mange an image of the PCB that would be great.

Measuring at less than full brightness is mildly interesting but limited useful data. I would also do an ac reading (200v range) . With no dimming , dc should show peak unless there is a lot of ripple.

If you plan to use a scope, great care is necessary. You would need to use a properly rated isolated probe, an isolation transformer, or other suitable setup. If in doubt, hold off doing it.

Even with DMM, lethal voltages are present; my bulb shows 154vdc for raw supply before regulator, typical of linear regulation.


Dave

 

[rasmasyean]

If the blues are all flickering, it could be the driver, or one or more of the LEDs themselves. If your bulb has linear driver it is likely an 8-pin SOIC with markings. Mine is KP18058.

Your red, green and blue look similar to mine, multichip with total vf around 30v or so. White looks like 6 LEDs per chip (you might see two die with light off or highly dimmed). If you can mange an image of the PCB that would be great.

Measuring at less than full brightness is mildly interesting but limited useful data. I would also do an ac reading (200v range) . With no dimming , dc should show peak unless there is a lot of ripple.

If you plan to use a scope, great care is necessary. You would need to use a properly rated isolated probe, an isolation transformer, or other suitable setup. If in doubt, hold off doing it.

Even with DMM, lethal voltages are present; my bulb shows 154vdc for raw supply before regulator, typical of linear regulation.


Dave

What to you mean by those dangers? Is it if you touch those 154vdc leads by hand? Or can it transfer through the DMM somehow and do something? Can I wear like rubber gloves or something as a precaution?

I've had limited experience with low voltage electronics, but I've never worked with anything related to 120v AC type of stuff. I just thought I'd try to figure out these light bulb things for fun. It's not a "budget thing" so perhaps I should stop?

 

[rasmasyean]

 

[Dave_H]

Great pic, interesting, bulb is somewhat similar, but different from the WiFi bulb I opened: Sylvania Smart+ RGBW which are available OTC here at $5 for two.

BP5778 8-pin IC is 2-channel linear constrant-current driver, from datasheet found which is mostly in chinese. Exactly how the LEDs are connected is not obvious, but I imagine R, G, and B are on separate series chains. Four each at 32v each makes sense.

For white flexibility, looks like 7 warm white, 7 cool white, and 7 yellow LEDS around perimeter. Likely they are driven by circuitry "below deck" through P1 pins. You would think interspersing/spreading these would create good mixing, but they are grouped together.

Not sure what is on P1 pins, could be low-voltage PWM to control R/G/B drivers, plus drive voltage (possibly hazardous) for white and yellow LEDS. Without figuring out pin assignments,
measuring would be of limited use, and could change depending on the light setting.

On that note insulating gloves are a good idea, plus some eye protection. Consequence of shorting with DMM probes could range from no effect (current limited) to damage, smoke/sparks (with some risk of fire), burn from hot metal, and electric shock. Even after bulb is powered off, a capacitor could hold high voltage for a short period.

As for diagnosing the blue LED issue, not sure how far you want to go. If it's a driver problem, finding replacement chip and removing/reinstalling could be challenging. It would likely involve a hot air gun, as chip has a bottom pad soldered to the PCB. As for checking individual LEDs without removing them, I have some ideas, let me know if you want further detail.



Dave

 

[rasmasyean]

I'm thinking it's gotta be the Blue LEDs they used. It seems to happen eventually to all of my bulbs after a couple of months.

I'd reckon it's something like this, where they have to order a batch for a run.

https://www.bridgelux.com/sites/default/files/resource_media/DS252%20SMD%202835%201W%2036V%20Data%20sheet%20Rev%20A%2020190725.PDF

I doubt I'd be able to find a replacement LED for sale in small quantities anyway.

I did try to use a heat gun to melt the solder and it didn't do much. In one, I melted it to the point where the plastic package started to warp and it didn't do anything to the Blue LEDs, but seemed to destroy the Green LEDs. rofl.

How do you check individual LEDs? I was thinking you'd need a variable power supply and just give it voltages and current. I don't actually have one, but was thinking of finding a cheap one just to satisfy the experiment itch since I've already done all this.

 

[Dave_H]

How do you check individual LEDs? I was thinking you'd need a variable power supply and just give it voltages and current. I don't actually have one, but was thinking of finding a cheap one just to satisfy the experiment itch since I've already done all this.

Most common DMMs can source enough voltage/current to light single LEDs directly, when set on ohms or diode-check; but modern bulbs typically use compound LEDs which have much higher Vf.

To check LEDs on board, make sure bulb is not powered and disconnected. Set variable PSU to around 40v (for these LEDs) and use 1k (nominal) series resistor. If PSU has current limit adjust, set it low, 100mA or less. Determine polarity of LEDs first. Probe each LED, look for a dim glow. Fine DMM probes should be OK.

Alternatively, you could probably use multiple cheap 9v zinc-carbons clipped together in series; four fresh ones in this case. For lower Vf LEDs adjust number of batteries and series resistor to limit current to a few mA.

Dave

 

[Dave_H]

I did try to use a heat gun to melt the solder and it didn't do much. In one, I melted it to the point where the plastic package started to warp and it didn't do anything to the Blue LEDs, but seemed to destroy the Green LEDs. rofl.

I've done PCB chip removal in lab without damage (up to 54-pin TSOP) using hot air gun with temperature control and fine nozzle, plus technique (a bit of practice on old PCB). This type of LED may not be amenable to easy removal without damage, without specialized equipment. If the LEDs can be sacrificed, as long as you don't damage the PCB, replacement is possible.

Dave

 

[LEDphile]

I've done PCB chip removal in lab without damage (up to 54-pin TSOP) using hot air gun with temperature control and fine nozzle, plus technique (a bit of practice on old PCB). This type of LED may not be amenable to easy removal without damage, without specialized equipment. If the LEDs can be sacrificed, as long as you don't damage the PCB, replacement is possible.

Dave

SMT LEDs (all power levels) can be replaced with a hot air rework station and proper technique, although the task is certainly easier if the PCB can be separated from the heatsink first.

 

[Dave_H]

Here is inside "AuraLED" A19/E26 RGBW bulb sold at Home Depot, about $5 each in 4-pack. SMT colour LEDs are similar to OP's bulb (but not WiFi, these use IR remote).

I probed each LED using four series 9v batteries (37v total) with 10k ohm series resistor, worked fine. Vf's are all around 30v or so at low current. White LEDs have lower voltage, lighted them OK although at slightly higher current.

The KP18058 LED driver (8-pin SOIC) has five channels. Unlike most other similar chips, I don't see any current set resistors. Currents are controlled by digital commands from the little 6-pin IC (connected to the IR receiver) over a 2-wire I2C bus...neat!

Dave