Samsung UN48J5000: Backlight Strip Burn

[Sleddriver]

New to me free. First time to work on one. PSB is a BN44-00852A. No joy on both SM or schematic despite extensive searching. First power up resulted in one Hz/sec buzzing/arcing and smell of electricity.

Removed front pieces to expose backlight LED strips (2015 SVS48 FCOM FHD DOE A RIGHT REV1.2 150605 LM41-00120Q_LM4100149A) and noticed bad burn mark on one LED strip! Right between two LEDs.

I tested all LEDs using a bench supply with current limiting. All 32 were fine. Von ~ 2.75Vdc. I made a wire bridge over the burned area by soldering on both sides, but it didn't work and burned out the end LED to the right. So I cut the strip to remove the burned area and shorted it to complete the series ckt. Then reinstalled it in the backlight panel and powered on the set using a VARIAC to limit the voltage to about 85 V. Slowly raised it to 100V, then off, then back on to restart. Only upper two rows of LEDs light. They still pulse on off on off. Heard & smelt more arcing and have now lost another LED due to burning. Quickly turned off the power.

Began PCB tracing on backside of PSB. Though four pairs of wires connect the PSB to the backlight, it's all one series ckt, with a single driver, a D9N40 MOSFET. Tested in ckt. it's not shorted. Something is wrong though. It appears the LED driver ckt voltage is way too high to cause such a burned out area. The PSB indicates "130V @750mA". I initially meas. over 200Vdc on a single pin and zero on the others.

Anyone here seen such a fault before? We're a long way from the boot chime & welcome screen.

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

" I initially meas. over 200Vdc on a single pin and zero on the others."
Is that reading with working LED strips connected in place?
So I do nt quite understand the condition of the LED strip, are all the bad LEDs replaced or they are bypassed with wire? If the bad LEDs are replaced, do they have them same rating as the ones that burnt?

[Sleddriver]

Re: Samsung UN48J5000: Backlight Strip Burn

No, they were not connected. I disconnected the backlight plug to prevent further damage.
No LEDs have been replaced. They're very, very tiny on these strips. Barely over 1mm square. Far smaller than I've ever seen on any repair video.

BTW, I'll be posting additional photos of both sides of the PSB.

[CapLeaker]

Re: Samsung UN48J5000: Backlight Strip Burn

So you have a bunch of blown LED's on the strip and you didn't replace any LED's? Since LED's are current driven, the PSU will try to always send an X value amount of it unless it goes out of x amount of spec and turns off. So if a diode fails, more power is getting sent to other LED's in that string. They don't like that and are going to fail. By the sounds of things, first see if you can find replacement back light strips.

[JohnCT]

Re: Samsung UN48J5000: Backlight Strip Burn

So if a diode fails, more power is getting sent to other LED's in that string. They don't like that and are going to fail. By the sounds of things, first see if you can find replacement back light strips.

I've only seen one LED drive circuit that used a fix voltage for driving the LED string (small, low end no name TV). In that one example, a single shorted LED would indeed increase the current to the remaining as the voltage was regulated.

However, in these Samsungs it's the current that's tightly regulated. As individual LEDs short, the current is maintained but the voltage *across the string* will change to maintain string current while the voltage drop across each individual LED remains the same.

John

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

No, they were not connected. I disconnected the backlight plug to prevent further damage.
No LEDs have been replaced. They're very, very tiny on these strips. Barely over 1mm square. Far smaller than I've ever seen on any repair video.

BTW, I'll be posting additional photos of both sides of the PSB.

That is why it went up to 200V due to no load. LED is driven with constant current.

[Sleddriver]

Re: Samsung UN48J5000: Backlight Strip Burn

So you have a bunch of blown LED's on the strip and you didn't replace any LED's?

No.

[dick_barton]

Re: Samsung UN48J5000: Backlight Strip Burn

JohnCT

However, in these Samsungs it's the current that's tightly regulated. As individual LEDs short, the current is maintained but the voltage *across the string* will change to maintain string current while the voltage drop across each individual LED remains the same.

Isn't this conflicting. If the voltage across the string changes then the voltage across each led must also change?.

[arenat]

Re: Samsung UN48J5000: Backlight Strip Burn

32 led for a 48" at 130V .75A = 97,5W 3W each led . J is code for junk.I saw this before , put a regular led and explode instantly

[Sleddriver]

Re: Samsung UN48J5000: Backlight Strip Burn

As promised, here are PSB photos.

Q on R9842 (9 Ohms), will increasing it or decreasing it lower backlight current? The LED driver MOSFET (D940) appears OK. At least it's not shorted D to S. However I may have to pull it for a more thorough test.

I'm not sure just how common a LED backlight driver ckt is between manufacturers or within the same brand. I've looked at several Samsung PS schematics with much lower AN44-00xxx numbers and noted more than a single driver. Perhaps Samsung is departing this 5000 series to save $$$?

A few more items from my notes:

1. With BL unplugged, 227Vdc was meas on pin 1+. All others were 0V. So this must be the max V available with an open ckt.

2. Using a Kill-A-Watt, I meas. set current at 90mA (6W) with BL unplugged and 1A (43W) with BL plugged in. Line Vac = 121V.

3. Top row of 8 LEDs draws 97mA with 22Vdc from a bench supply. All 8 very bright.

4. Second from top row: 103mA with 22V. All very bright.

5. Third from top: 1.9mA at 22V. All very dim. Obviously something wrong with this one. High resistance impeding current flow. This meas. was taken with the burned area shown in the photo still in place. It's since been cut out and the ends shorted (removing one LED) to complete the ckt. New meas: 440mA @ 22V! Holy Cow! That's Nuts! Small wonder the PSB board doesn't like that load! Losing a single 3V LED couldn't do that. There are these tiny (3.2 T x 2.8mm W) blocks between each LED with two side terminals marked + and - that I used to apply voltage to each LED to test it. The two silver pads on top are not connected to the sides. One of these was also removed when cutting out the bad burned section. Perhaps they also serve another function?

6. Bottom row: 127mA @ 22V. All very bright.

7. Obviously something's still FUBAR about strip #3. With the burned section present, the end LED and the 'tiny block', it was only drawing 1.9mA @ 22V. Now with these removed and the traces shorted together where I cut out the bad section, it's drawing 440mA at the same V? What's up with that? The 'block's are not resistors. Clearly I don't understand how these LED strips are wired.

That needs figuring out before I power up the BL though. Ideas?

[CapLeaker]

Re: Samsung UN48J5000: Backlight Strip Burn

No.

You said "No LEDs have been replaced."

Not sure anymore what goes on, maybe you wanna clear things up.

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

https://www.badcaps.net/forum/attach...5&d=1525221835
What do you get when using meter in Ohm mode to read the resistance of the component with the (+) and (-) in both directions?
What do you get when using meter in Diode mode to check the component with the (+) and (-) in both directions?
It looks to be protection Diode.

[Sleddriver]

Re: Samsung UN48J5000: Backlight Strip Burn

https://www.badcaps.net/forum/attach...5&d=1525221835
What do you get when using meter in Ohm mode to read the resistance of the component with the (+) and (-) in both directions?
What do you get when using meter in Diode mode to check the component with the (+) and (-) in both directions?
It looks to be protection Diode.

In Ohm mode, - to -, + to +, I read OL and LED barely lights, with polarity swapped, I read OL and LED doesn't light.

In diode mode, matching polarity as in above, I read OL and LED barely lights, opposite polarity, I read a diode drop of .78V and no beep.

When repeated on the short section with the burn mark, the open LED, the meter reads OL in all orientations and both Ohms & diode modes.

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

OK, so it is protection diode (5V ZENER is typical for 3V LED) connected in anti parallel with the LED.
So did you test each one of those diode on all the strips?
So are you going to replace the BAD LEDs strip?
BTW, I am not so sure about 750mA of LED current, if it is then it is over 2W for the LED and I never seen 2 ~ 3W LED in that small form factor.

[JohnCT]

Re: Samsung UN48J5000: Backlight Strip Burn

However, in these Samsungs it's the current that's tightly regulated. As individual LEDs short, the current is maintained but the voltage *across the string* will change to maintain string current while the voltage drop across each individual LED remains the same.

JohnCT
Isn't this conflicting. If the voltage across the string changes then the voltage across each led must also change?.

No, not in the context of shorting LEDs in a string powered by a current regulated source.

For argument's sake, let's assume the back light is set to a predetermined level and fixed (that is, not active to scene content) so as not to throw any variables into the discussion.

The TV monitors the current and adjusts accordingly. As LEDs short out in the string over time, the voltage *across* the string will necessarily drop to compensate for the shorted LED in order to keep the current constant, but the voltage across each functioning LED will remain the same if the current is the same.

As an extreme example, let's say we have an LED strip on our bench that has 10 3V LEDs on it that we will test with a typical LED tester. Before we apply the test leads, the open circuit voltage is almost 300V. When connected, the voltage reading drops to approx 25V (2.5v X10 at 20ma). But suppose we place shorting jumpers across half the LEDs in the string and remeasure with our tester.

In order to maintain current, the voltage on our test strip will drop to roughly half (say 12.5v), the current will remain the same (the tester is current regulated to 20ma), and the voltage across the remaining functioning LEDs will stay at 2.5V.

John

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

When the LED in one of the string shorted out, the current is still be maintaining but the Vdrops on the active control element, PWM Controlled MOSFET) wihich is driven by the LED controller IC to maintain constant current on the low side, will increase so there will be more power dissipation on the MOSFET.
The power supply Voltage does not vary (The boost converter is not tightly regulated and no needed to since it is dealing with the constant current which is being done with the active control element to maintain the constant current), the Vdrops on the MOSFET for each LED string is varied to maintain constant current for that strip.

[Sleddriver]

Re: Samsung UN48J5000: Backlight Strip Burn

OK, so it is protection diode (5V ZENER is typical for 3V LED) connected in anti parallel with the LED.
So did you test each one of those diode on all the strips?
So are you going to replace the BAD LEDs strip?
BTW, I am not so sure about 750mA of LED current, if it is then it is over 2W for the LED and I never seen 2 ~ 3W LED in that small form factor.

I'm not so sure about a 5V zener. 5V would seem too high for any protection. Whatever it is, it didn't do much of a job of protecting that strip from over-voltage serious enough to cause arcing. I've spent a few hrs reading other posts in the PS Forum (didn't know it was there) and found TV PS's discussed as well. Also discussed were burning LED backlight strips w/ photos! Far more serious than my case. LG was mentioned often. Very alarming no protections were in place to prevent such damage long before it got to that point. Electric arcs burn very hot! Very illuminating and educational to say the least given my very recent introduction to what's going on 'behind the screen' in these modern sets.

I used those "test points" to verify the individual LEDs worked by applying 2.75V. That and the recent meas. you requested were all of the testing I did on them as I wasn't sure just why they were there. All of the repair videos I've watched are for much older sets that didn't include these.

Re: Strip replacement. Yes, if I can find a good deal and not have to purchase an entire set of 8. One outfit wants $50 + shipping. A bit rich for a free-to-me-TV. Given my measurements, I need only a left & right section of a single strip. Not to replace all of them. So I'll continue hunting as time allows. If you know of other places to shop, I'd appreciate it.

Thanks for your time & to the others who chimed in here! I appreciate your assistance.

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

You are reading 0.78V vF in forward bias of the Zener, any Vf of the Zener will be the same as Vf of the didoe, to read Zener Voltage you will have to connect the Zener in reverse bias to variable power supply with current limiter to read the Zener Voltage.

"LG was mentioned often. Very alarming no protections were in place to prevent such damage long before it got to that point. Electric arcs burn very hot! Very illuminating and educational to say the least given my very recent introduction to what's going on 'behind the screen' in these modern sets." This situation is command in LG and Samsung that I know of due to bad design, I am surprised that no house burns down yet.

https://www.badcaps.net/forum/showth...ht=LG+47LN5790

[budm]

Re: Samsung UN48J5000: Backlight Strip Burn

Example of LED with built-in protection Zener diode.

[arenat]

Re: Samsung UN48J5000: Backlight Strip Burn

When the LED in one of the string shorted out, the current is still be maintaining but the Vdrops on the active control element, PWM Controlled MOSFET) wihich is driven by the LED controller IC to maintain constant current on the low side, will increase so there will be more power dissipation on the MOSFET.
The power supply Voltage does not vary (The boost converter is not tightly regulated and no needed to since it is dealing with the constant current which is being done with the active control element to maintain the constant current), the Vdrops on the MOSFET for each LED string is varied to maintain constant current for that strip.

There´s not Vdrop in the low side of the string . This is not how PWM work , when the mosfet is ON the string get the full current , when is OFF get zero , nada .The current is an average given those times on and off .