My sister-in-law asked me to look at a set of Christmas lights - the type with LEDs wired with alternate polarity allowing a variety of flashing patterns. One of the kids had run over the wire between the controller and the first LED with an electric scooter causing it to get snagged in the wheel and break. But on examination, it wasn't just a question of repairing a broken wire - it looked like it'd shorted out, damaging the controller.

I had to cut open the controller with a dremel, but inside, it seemed fairly simple: an H-bridge driving the LED string with either polarity, and controlled by a microcontroller with attached serial NVRAM. The H-bridge uses 6 SOT transistors, 4 of which were dead (testing the B-E and B-C junctions with a multimeter on the diode range). The circuit wasn't hard to trace (or so it seemed), but the circuit I came up with couldn't possibly work. Clearly something wrong either with my tracing or with the component identification. Here's a photo of the board and my attempt to trace the circuit:

I've marked the inputs from the microcontroller IN1 and IN2 (presumably 5V logic levels), and the outputs to the LED string OUT1 ad OUT2. The supply is 31V.

I could understand the circuit if Q3 and Q4 were PNP and the emiter and collector reversed, but with the red testmeter lead on the pin 1 (base) they give around 0.7V on diode test on the other two pins. (These are the only 2 out of the 6 trasistors to have survived.) The abbreviated marking J6 appears most plausibly to be NPN trasistor S9014.

The flaw in my logic is escaping me, perhaps partly due to a brain befuddled by an excess of turkey and Christmas pudding. Can someone please point out my silly mistake?

 

The circuit looks correct to me, the inputs are pwm from the microcontroller and the leds go across the two outputs, so one set of leds light up one way, and the other set light up the other way. This circuit is used a lot on these multi function led lights in 400 leds in anti-parallel in rows of upto 8, the output voltage swings between +/- 15V ( 30 V total) .

 

The circuit looks correct to me, the inputs are pwm from the microcontroller and the leds go across the two outputs, so one set of leds light up one way, and the other set light up the other way. This circuit is used a lot on these multi function led lights in 400 leds in anti-parallel in rows of upto 6, the output voltage swings between +/- 15V ( 30 V total) .

But if IN1 is low, Q5, Q4 and hence Q1 will all be off, and if it's high (5V) they will all be on, and Q1 and Q5 will instantly die!
Not sure where you get your +/- 15V from. If OUT1 is high and OUT2 is low, one set of antiparallel LEDs will come one, and the other set if the reverse. If both are either high or low none will be on. Hence the circuit can implement PWM.

 

IN1=5V.
Q5 turns on. Q4 turns on.
If Q4 is on, Q1 turns on.
Q1 and Q5 on together is a dead short across the supply.

Your mistake is that Q1 and Q6 form one side of the bridge and Q2 and Q5 form the other.
(It does have a "self-destruct" mode if IN1 and IN2 are simultaneously high.)
It must be very price sensitive if they decided that they needed a cheaper transistor for Q3/Q4 than for Q5/Q6.

 

Clearly something wrong either with my tracing or with the component identification.

I agree with Ian0. Looks like Q1 collector goes via 0 Ohm link to Q6 collector, not Q5.
Btw, D2 looks cooked.

 

I agree with Ian0. Looks like Q1 collector goes via 0 Ohm link to Q6 collector, not Q5.
Btw, D2 looks cooked.

Yes. Picture attached.

 

I agree with Ian0. Looks like Q1 collector goes via 0 Ohm link to Q6 collector, not Q5.
Btw, D2 looks cooked.

Thank you @Ian0 and clarified by @Alec_t - that's it! Each input turns on the top of one side and the bottom of the other side of the H.
D2 tests ok but if Schottky diodes could speak it certainly would have said Ouch! It looks like short-circuit protection in the PSU saved its bacon.

 

Thank you @Ian0 and clarified by @Alec_t - that's it! Each input turns on the top of one side and the bottom of the other side of the H.
D2 tests ok but if Schottky diodes could speak it certainly would have said Ouch! It looks like short-circuit protection in the PSU saved its bacon.

D1 looks like a Zener diode with R1 to drop the supply for the microcontrollers.

 

D1 looks like a Zener diode with R1 to drop the supply for the microcontrollers.

Yes, I'd come to the same conclusion.

I'm surprised they don't use an integrated H-bridge instead of 6 discrete transistors. Such a thing must exist and would surely be more economical on production coss and probbly on BoM as well. There are plenty around for brushed motor control, though I couldn't find a low power one at either Mouser or RS. I have a USB-powered set of lights which appears to contain one - there are just 2 8-pin SOT devices, the other of which must be a microcontroller. But I guess this H-bridge controller is only rated for 5V. It's in a frosted transparent housing so I can't see any markings. I might try cracking it open when I take those lights down.