12V MR16 LED repair

[jmaja]

Just out of interest I'm trying to find out what is wrong with the LED that just failed in my table lamp. It's a 12 V AC 3 W LED with 3 1 W LEDs in series and separate driver PCB. The schematic is attached. +12 V is after a bridge rectifier (which works OK). There is an unidentified SOT123-6 (or similar) package with marking "22 AD". The pins of that are marked in the schematic. PIN3 is NC, but there is a unpopulated place from it to GND. The values of the two ceramic caps are just guessed based on size.

When I power it with 12.2 V DC I get 11.9 V for both +LED and -LED (LED is not connected now). The LEDs work fine, if I input 9 V DC directly into them (~0.3 A ~3 W).

If I measure the eloctrolytic caps (without removing) with DE-5000 LCR I get 260 uF 0.14 ohm ESR for the 220 uF and 305 uF 0.32 ohm for the 100 uF. Both look OK. L1 shows 39 uF 0.46 ohm and L2 46 uF 0.4 ohm.

What has failed? Is this a switching mode buck current driver? Just with switching and thus the coil in the negative side.

[bktemp]

The ic is something like AL8805. The pinout matches, but the marking code is slightly different (xxA6). There are many similar ic, operating the same way, so the exact type doesn't care.
It sounds like the ic is not operating. You can check the voltage at pin 3 if it is not shorted to GND and measure the voltage between 5+4 (current sense). If both are ok, the ic is dead.

[jmaja]

Thanks! The Control, Vin and Set/+LED are all at the same voltage (11.953-11.954 V with 12.20 V DC to the bridge). The -LED and SW are at slightly lower voltage (11.936 V). GND is at 0.252 V. All voltage referenced to negative at the connector (before the rectifier).

Measuring currents from those pins to the negative at the connector I get full current (limited by supply) from +LED/SET, 5 uA from -LED/SW and 2 uA from Control. So the latter two are floating. Just like the chip thought the current is too high already.

I still measured the diode. It has 0.198 V Vf and 0.992 Vr. I guess the Vr comes from the ESD diodes of the converter.

So it seems like the chip is dead? Is that typical? I would have thought it's a cap or a solder failure. But it seems caps aren't very critical in this design and you should get almost full voltage even without caps in a no load situation.

[bktemp]

If Control is high and there is no voltage across pin 4+5 the ic should pull pin 1 to GND until it senses a current above its theshold at pin 4+5, then turn it off again until it falls below the threshold, repeating the operation. Capacitors at the output aren't necessary but help to reduce the ripple seen by the LED. So even without any capacitors the circuit should work if the input voltage is reasonably stable.
If the ic does not turn on its internal mosfet, it is most likely dead.
Based on 1W per LED and 3.3V per LED the LED current should be around 303mA. The current sense resistor has 0.33ohms and the current threshold for AL8805 is 100mV. That gives 303mA. That matches perfectly. So if you want to repair the lamp, AL8805 or similar should work fine.