- Why bother fixing it?
- The symptoms
- Tear down
- A closer look
- Diagnosing and fixing the problem
- Final thoughts
I’ve upgraded my bike headlight to a rechargeable NiteRider Lumina 1200 Boost at the end of 2019 to make my long commute on unlit streets rides safer and to stop using batteries. Unfortunately, only after six months of active use, the light has failed on me. It’s rather disappointing to have such an expensive device fail so soon, but instead of buying a new one I’ve decided to take it apart and try to fix it in order to avoid creating more e-waste and to prolong it’s useful life.
Why bother fixing it?
While I think it could have been repaired under warranty, it would still have cost me additional money (i.e. postage) as well as I would have been left without a light for a long period of time.
Since I already have a small home electronics lab, I felt I’d put it and my EE skills to good use rather than shipping the device back and forth across half the country. A nice bonus is that if I’m successful I’ll know how to maintain it indefinitely.
The light initially started to flicker or just would shine very dimly irrespective of the level set. Turning it on and off sometimes helped, but eventually the light just stopped working, with the exception of the indicator light under the button. Charging would also work when plugged into an USB power source so it was not completely dead yet.
The light is build with an aluminium core that has the front and back plastic shells attached to it with three Allen bolts (2mm sockets), two at the top and one at the bottom.
Unscrewing them will allow for the removal of both plastic shells. The rear shell houses the main PCB assembly and the rechargeable Li-Ion battery.
The front shell holds in place the lens assembly in front of the high powered LED.
Waterproofing the shell and the USB port is solved by a rubber seal. The front part can be further disassembled by removing the two Phillips screws next to the LED. Removing these screws will release the PCB assembly from the light’s aluminium casing.
When removing the PCB assembly from the aluminium housing some double-sided tape has to be peeled off carefully. First the one that sticks the battery and the main PCB together. Then another one on the back of the LED’s PCB which serves as the thermal coupling to the housing. This will allow sliding the PCB out to the front.
A closer look
After taking the light apart I’ve spent some time analyzing the details of the construction. The 3.7V 3000mAh Li-Ion rechargeable battery does not seem to be anything special. Though I’ve not been able to find the part by it’s number (KP-070-24349) on the web, it’s likely possible to find a compatible replacement if needed.
The circuit is build by soldering together two PCBs, one containing the control logic and one which has the LED. The two PCBs are soldered together in a 90° angle.
The main board is powered by a PIC16F1503 microcontroller. There is no dedicated charger IC, but it’s not really needed either, since the uC has all the peripherals to be singlehandedly responsible for handling user I/O via a GPIO pins, driving the power transistor with a PWM signal for the LED, and controlling the charging current to the battery with the help of an ADC. The board doesn’t even have a crystal on it, as it’s clocked using the internal source. Truly minimalistic and functional.
Since the LED module is clearly the same as it was in the previous version of this model, the Lumina 1100 Boost, it’s very likely that most devices in this line have an identical hardware, with the only difference being the battery capacity and a look up tables in the firmware to achieve similar single charge lifetime by using different brightnesses. Again this is not surprising, this is a common practice in the industry.
The indicator lights are individual LEDs on the two sides of the tactile switch: red (D3) for charging and blue (D4) for showing on/off state. It’s probably cheaper than using a single RGB LED.
The other IC on the board is a 050NE2LS N channel MOSFET, used to control and deliver the high current to the LED directly from the battery, as none of the microcontroller I/O pins would be capable to supply the needed amount (see later pictures).
It would be interesting to see if the firmware can be modified to reduce the brightness levels a bit (e.g. to those of the previous generation) of this light to achieve longer lifetime on a single charge, since the programming pads are easily accessible. I would also consider replacing the blue coloured LED with something more friendly, like green. Maybe another time.
Diagnosing and fixing the problem
To find the problem I felt it would be safer if I disconnected the battery and used my bench power supply instead, with leads attached to the PCB in the right points (I simply used alligator clips for this, not pretty but it has worked).
Using my multimeter I could not find obvious issues on the main PCB. However when measuring the power endpoints near the LED I could not see any voltage there. Upon closer inspection and some mechanical fiddling around, sure enough the problem became obvious. There was a solder joint where the two PCBs meet, which has broken most likely due to fatigue caused by vibrations. This type of breakage is also perfectly in line with the experienced failure symptoms mentioned previously.
Both middle connections seemed to be in a bad shape, so I’ve re-soldered them to form a solid connection again. The challenge here was to avoid accidentally creating a short circuit by soldering front pad and the thick copper on the back together, since the latter is on ground potential. The extra thick panel is most likely used to improve heat transfer from the LED to the aluminium case, which acts as a heatsink (it also explains the ribbed design exterior).
Before reconnecting the battery I did a few short circuit measurements with my multimeter, then I’ve quickly validated that both the light and charging are working.
Once all seemed in order, I’ve reassembled the headlight, and now I have a fully functional device again after spending just a few hours on it. Hopefully this time it will last longer than 6 months.
Since it can be easily taken apart and put back together, as well as has a clean PCB design, where all parts have a clear silkscreen marking, nor are the chip identifiers scratched off (unlike what most Chinese companies do with their products), this device is quite repairable. The materials used are also high grade. These should make it possible for it to last long.
The only part that could be improved is the way the two PCBs are connected. I think using some angled through hole pins would’ve provided for a more robust connection or perhaps tighter holes in the PCB for the teeth.
Overall I feel the design and quality are quite good, despite it’s premature breakage. Maybe I just got a lemon or the roads are just too bad around here. A bit expensive for what it is, but since I was fortunate enough to get it at a heavily discounted price and before the pandemic induced inflation, for me it was still worth it.