GE Refrigerator LED failure

[rcbuck]

I have a GE refrigerator and the LED light strip in the freezer compartment went bad. The part number is WR55X26671. From what I found online, the board has a nominal operating voltage of 4 volts. With the bad board plugged in I measure 7.6 volts at the connector. With the board unplugged I measure 32 volts. There must be a dropping resistor somewhere on the board that feeds the voltage to the LED board.

I purchased a new board online and ran some tests. I connected a regulated 1 amp walwart that was sourcing 5.6 volts to the board. The walwart voltage dropped to 3.8 volts. (I'm sure the Chinese just made a mistake when they rated the supply at 1 amp.) The LED strip was pulling 640 mA and the LEDs very quickly got too hot to touch.

The LED board has a very bad reputation for failure. GE is aware of the problem but ignores the issue. A large number of the boards fail after 2 to 3 years. In my case the refrigerator is 4 years old so I guess I am one of the lucky ones. I suspect the high failure rate is due to the 7.6 volts being applied to the board. There are two 2 ohm resistors in parallel feeding the LEDs on the board. I don't have any idea of what the LEDs are or what their wattage rating is. I do know their forward voltage drop is 1.6 volts as I measured that on the new board. The LEDs on the old board measure open circuit.

I decided to do something to try to extend the life of the new board. I removed the connector from the old board. I soldered a 11 ohm, 3 watt resistor between that connector and the connector on the new board. The refrigerator cable plugs into the old connector ahead of the 11 ohm resistor. With the new board plugged in, I measure 8.2 volts at the source connector and 3.2 volts at the LED board connector. The board is pulling about 250 mA. The light is plenty bright enough for the freezer. Hopefully the LED board will now last past the lifetime of the refrigerator.

This board is used in dozens of different GE refrigerators. Has anyone seen a similar failure with your refrigerator. If so, did you just replace the board and call it good? I've read online where lots of people say the replacement boards also go bad after a couple of years.

[amyk]

It could be a constant-current supply. They're probably relying on the fact that it's in a freezer, and that most of the time the LEDs won't be on for long, to let them get away with driving the LEDs harder and with less heatsinking.

[rcbuck]

Possibly. I removed one of the LEDs from the old board. It only has about a 5 x 5 mm heatsink area underneath the LED. Due to the high failure rate they are obviously driving the LEDs too hard. My resistor in series may help since it reduces the voltage to the LEDs. I said I used a 22 ohm resistor but that was a mistake. It is actually a 11 ohm resistor. I will edit and correct the original post.

[jrobertg]

Hey guys...I'm a total hack, but my buddy is always my personal source of all things electrical.

I've had the refrigerator (GSS25IYNEHFS) since Feb 2020, and am on the 3rd LED board. I was able to read a feed voltage of 3.1 VDC when the board is connected and energized.

My friend's reply regarding his analysis:
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
I did a bit more measuring and then a quick analysis of the LED circuit for your refrigerator (attached).  I'm convinced the weakness is the design.  It is VERY sensitive to supply voltage given that it is fed with a very low voltage (3V) with LEDs that have almost the same forward voltage drop of 2.953V.
This drove the designers to use very low value current limiting resistors for each LED of only 1.5 ohms (two 3 ohm resistors in parallel).  I even measured 0.9 ohms for them.

Given an exact 3.0V supply, the current flow in each LED will be (assuming a 1.5 ohm resistance) 31.5mA.  That implies a forward drop through the LED of 2.953V.  Assuming this is constant over a range of current, which is isn't exactly, but will be close to constant, then if the supply voltage increases only 5% to 3.15V, the current through the LEDs will increase to about 131mA.  That's a 400% increase.  A 4x change will likely blow the LEDs.

This is why they keep blowing out.

A solution is harder:  You could try some sort of voltage clamp on the supply, but these are usually themselves not very precise nor sharp in their response.  Another approach would be to insert a resistor in series with the board to drop voltage and reduce the overall light, which leads to longer life in all LEDs when run more dimly.  Otherwise you could complain to the company and try to get them to improve the design.  They need to use a circuit with higher voltage and higher resistance.  Good luck with that.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[cosmicray]

Given an exact 3.0V supply, the current flow in each LED will be (assuming a 1.5 ohm resistance) 31.5mA.  That implies a forward drop through the LED of 2.953V.  Assuming this is constant over a range of current, which is isn't exactly, but will be close to constant, then if the supply voltage increases only 5% to 3.15V, the current through the LEDs will increase to about 131mA.  That's a 400% increase.  A 4x change will likely blow the LEDs.

The implication is that all the LEDs are in parallel. If one LED fails, that may allow additional current thru the others. Depends on the circuit topology.

[Zero999]

It probably uses a constant current supply, not voltage. As mentioned above, having all of the LEDs in parallel is a problem, because if one fails, the rest will carry more current, making them more likely to fail.

The design is fundamentally flawed. You could buy a new LED board and reduce the current through the LEDs by connecting a resistor in parallel with it, which will reduce the current through the LEDs.

[rcbuck]

Mine is still running fine after 16 months with the added 11 ohm, 3 watt resistor in series with the supply line. I measured 250 mA of current through the new board with the resistor in series, which still seems excessive. But not knowing the LED part number, we don't know what current it is rated for. The resistor in series drops the voltage applied to the LEDs so maybe that will be enough to keep my board from failing.

If it fails in the future, I will try adding a parallel resistor as Zero999 suggested. If the board does fail, I will measure the constant current into a dead short. I will then calculate the resistor needed to drop the current through the LEDs to 50% of the constant current.

[fzabkar]

I would rewire all 3 LEDs in series and then adjust the value of the current setting resistor on the main PCB. Also, I would add a polyswitch in series for safety. An added benefit might be reduced power dissipation in the current source, depending on the topology.