Adding and syncing an LED Strip to the Xiegu G90's Display Backlight
Adding a front-panel LED strip to the Xiegu G90 that dims in sync with the display backlight. Plus a debugging detour into why the first MOSFET I tried left it running dim, and what the datasheet actually said about it.
Backlighting the Xiegu G90 Front Panel
The G90's front panel buttons have no backlight at all. Fine in daylight, but genuinely annoying on a night operating session when I couldn't see what I was pressing. I found a video showing someone add an LED strip around the buttons (Xiegu G90 Backlit Buttons) and liked the idea, but wanted to go one step further: have the strip dim together with the display, instead of running on its own separate switch.
A quick disclaimer before you start: this mod means opening the radio, soldering onto an existing circuit, and tapping a signal straight off the microcontroller. Do this at your own risk. It can void your warranty, and if you tap the wrong pin or short something while probing, you can damage the radio, in the worst case bricking the front panel entirely. Take your time, double check every measurement before you solder, and if you're not comfortable with a soldering iron and a multimeter around a live board, this isn't the mod to start with.
Finding the right pin
I pulled the full G90 schematic from yo3hjv.blogspot.com, which shows the display's microcontroller (U1) driving the backlight through pin 62, labeled BL_PWM. That signal feeds a small transistor (Q1) which switches the stock backlight LEDs on and off.
Rather than trust the schematic alone, I confirmed it on the bench: probed each of Q1's legs with a multimeter while cycling the display brightness setting. One pin's voltage tracked the brightness change exactly. That's the base, and that's the tap point.
The circuit
Simple low-side switch: BL_PWM drives the gate of an N-channel logic-level MOSFET (Q2), which switches the LED strip's negative lead to ground. Two resistors round it out:
- R1 (220Ω), between the tap point and the gate, to limit inrush current.
- R2 (10kΩ), gate to ground, so the strip doesn't flicker on before the MCU boots up and takes control of the pin.
Power for the strip comes from the DB9 connector's pin 1, which carries a regulated 9V. My strip is rated for 12V, so it runs a bit dimmer than full spec, but plenty bright for an accent light. I went with a 3000K, 12V, 0.5m narrow LED strip, warm white to match the rest of the shack lighting.
The MOSFET I got wrong first
I started with a part I already had, an IRFR4104. It worked, but way dimmer than expected. Measuring across the MOSFET's drain and source while it was "on" showed 1.5V still sitting there, meaning it wasn't switching cleanly.
Turned out to be the gate threshold. That part needs close to 10V to fully turn on, and BL_PWM only gives 3.3V logic. It was on, just barely, wasting most of the voltage as heat instead of light.
Swapped it for an IRLR8721, a genuine logic-level MOSFET that turns on properly off 3.3V. Same footprint concerns applied (I wanted something between a tiny SMD part and a bulky TO-220), and this one comes in a DPAK package that hits a good middle ground. Problem solved, full brightness, MOSFET stays cool.
Pro tip: slip a small piece of black heat shrink over the three stock front panel pilot lamps (Power/RX/TX status, FUNC, and ΔF) before closing it up, stops the strip's light from bleeding into them.
Mounting everything
Inside the G90 there's no real anchor point for extra parts like these, so I held the MOSFET, LED strip and resistors in place with a dab of UV-cure glue rather than hot glue or tape. It sets hard in seconds under a UV light, doesn't sag from heat the way hot glue can, and holds well enough that nothing shifts around when the case goes back together. If you ever need to redo the work, a bit of gentle heat plus a plastic spudger gets it back off cleanly enough, isopropyl alcohol helps loosen the more stubborn spots.
The schematic
Here's the full circuit as I built it:
What I'd tell someone doing this
Check your MOSFET's gate threshold against whatever's actually driving it before you solder anything in. A part rated for 10V gate drive will technically switch on 3.3V logic, just badly. And measure drain to source voltage under load any time a switched LED or motor circuit seems dimmer or weaker than it should be. That single measurement is what pointed me straight at the actual problem instead of chasing the wrong thing.