I'm trying to repair this battery-powered LED bike light but do not really understand the circuit.
LED circuit

Positive terminal middle left, negative bottom left. Momentary push switch at the top, transistor on the right, capacitor in the middle, 5 LED array at the bottom. The lights are not supposed to blink: just on/off. The switch works, as do the lights (when transistor is shorted). Capacitor seems to work.

  1. What is likely under the epoxy? Just water resistance for resistors or is it actually a chip? How does this design work? Is there a need for a capacitor in a battery-powered LED circuit?
  2. I believe the transistor is broken. My diode tester measures 800 ohms from both positive top to middle and positive bottom to middle but also 680 ohms negative top to middle and 1260 ohms negative top to bottom (others are infinite). From this I guess that it is PNP with middle being base. Does this sound correct?
  3. The part number is S8550 D 331. Only results are Alibaba without data sheets. Close match by name is SS8550 from Fairchild (https://www.fairchildsemi.com/products/discretes/bipolar-transistors/small-signal-bjts/SS8550.html). Dimensions are slightly off. Does this seem a reasonable replacement? Any clue on the differences between models, other than leg spread and ammo vs bulk packaging?
  • \$\begingroup\$ Under the epoxy is likely some controller IC for the light. If the LED's are lighting, then it's unlikely the problem is in the transistor pass element... you could measure some voltages when the led's are on... Hey maybe you could hack in a blinking circuit? \$\endgroup\$ – George Herold Jan 15 '16 at 15:10
  • \$\begingroup\$ To be frank, I consider it not blinking any more a feature. It is annoying to no end, because it keeps drawing your attention, and where your attention is focussed, that's what you drive towards. Much better off not having any blinking for all parties involved. \$\endgroup\$ – Asmyldof Jan 15 '16 at 15:18

The transistor is jellybean Chinese PNP transistor. The specs (and even the pinout) vary somewhat from maker to maker- no JEDEC standardization here, but it's broadly similar to the Fairchild SS8550. They seldom fail if not abused.

Anyway, the thing under the blob is an IC chip (Chip-On-Board), and chances are very good that's where the problem lies (quite possibly in the wire bonds to the chip). In which case, the unit is not economically repairable. That construction technique is not particularly good (reliability-wise) for this kind of application but it is very cheap in high volume.

You could always hack in a mechanical switch (toggle or whatever) across the transistor and ignore the electronics. It's mostly there to allow an almost free momentary switch and allow blink modes. I'd try to disconnect the chip itself - I find that the one I have slowly drains the 4 AA cells whilst allegedly off. Or if you're ambitious beyond all reason, hack in a small microcontroller to drive the transistor.

  • \$\begingroup\$ The non-infinite transistor resistance in both directions is not suspicious? \$\endgroup\$ – Sam Brightman Jan 15 '16 at 15:47
  • 1
    \$\begingroup\$ Not when it's still soldered into the circuit. \$\endgroup\$ – Spehro Pefhany Jan 15 '16 at 16:02
  • 1
    \$\begingroup\$ Indeed, having removed it the reverse test shows infinite resistance. I guess it must be the chip. Given that there are no blink modes, why is it there - to toggle state with the momentary switch instead of using a real switch? \$\endgroup\$ – Sam Brightman Jan 16 '16 at 13:39
  • \$\begingroup\$ Yes and maybe to give different brightness by PWM or possibly control the LED current in a linear fashion. RIP bike light. \$\endgroup\$ – Spehro Pefhany Jan 16 '16 at 16:59

It's probably a chip under the epoxy blob. It's a technique used a lot in high volume products, called "chip-on-board". The chip will control the flashing of the LED and any other low-power modes you may have. There's not much chance of being able to replace or repair that.

  • \$\begingroup\$ There is no low power or flashing mode. My theory is that it is ok and the transistor is at fault. \$\endgroup\$ – Sam Brightman Jan 15 '16 at 15:26

Quick Schematic based on that circuit. Super simple. The Capacitor is just for smoothing. I'm assuming the epoxied Chip-On-Blob IC does some sort of PWM, even if it doesn't blink or flash, or dim. Even if it doesn't, the cap doesn't hurt.


simulate this circuit – Schematic created using CircuitLab

You can test the transistor easily. First find the emitter, using a multimeter's continuity tester. The pin that's connected to B+ will undoubtedly be the emitter. That leaves you the base. Desolder the base pin, and then, with a say 1k resistor, touch it to ground. If the leds come on, the transistor is fine. If not, then replace it with any common PNP transistor, it really doesn't matter. Those LEDs, assuming common white 5mm, should only be drawing 100 to 200 mA max. Even something like the super cheap 2n3906 should work.

  • \$\begingroup\$ Emitter and base both disappear into the blob. There may be a current sense resistor on the chip. \$\endgroup\$ – Spehro Pefhany Jan 15 '16 at 16:25
  • \$\begingroup\$ @SpehroPefhany A multimeter can tell which is which. The one that's directly tied to B+ is the emitter. I highly doubt there is current sensing involved. \$\endgroup\$ – Passerby Jan 15 '16 at 16:27
  • \$\begingroup\$ (aside) The capacitor is unusual. Unless I'm missing something the two conductors just go into the chip blob. The top one can't go in then out directly to the battery + because the switch conductor is in the way. So I think it goes through the chip. Or maybe I'm missing something. \$\endgroup\$ – Spehro Pefhany Jan 15 '16 at 16:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.