Project Logs

Collapse

• Beginning Diagnostics

  Colin Pate • 04/19/2025 at 03:14 • 0 comments

  With the broken speaker torn apart, first order of business was to check both PCBs for popped capacitors or any other evidence of component destruction. There were no obvious signs of component failure, but I was pleased to find a number of very handy features on both PCBs - power LEDs for each supply rail of the PSU on the analog board, and two 0.1" headers on the digital board, one with JTAG pins labelled! 

  Everything looked normal checking the rail voltages, so my suspicion turned from the PSU section to either the digital board or the control board (with the buttons and LEDs) up front. The most obvious step seemed to be to swap the digital board from the working speaker into the broken speaker. If this brought the broken speaker to life, the failure was almost certainly in the digital board. It felt a little risky, as I couldn't be sure that there wasn't some fault with the analog board that had fried the original digital board and would fry the working one when I swapped it in. After checking the voltage at the V+ input on all the chips that I could identify, I couldn't find an obvious fault so I decided to pull the trigger.
  
  Thankfully, the broken speaker did not fry the working digital board. And with the swap, it worked! So the fault (presumably) had to be somewhere in the original digital board. This was both good and bad. Good, because digital stuff is fun, and the architecture was pretty old school - 8-bit MCU, discrete flash, discrete SRAM - so I was fairly confident I could reverse engineer quite a bit by following the traces and probing signals with my basic USB logic analyzer. It was also bad, because if the issue was corrupted flash memoery or a dead chip, I wasn't sure I'd be able to recover the original programming or program a new replacement part. However, the JTAG header, relatively simple architecture, and having a known working board gave me hope.

• Teardown

  Colin Pate • 04/18/2025 at 23:50 • 0 comments

  The first thing to do was to power up the problem speaker and try to find all the symptoms of the problem. The seller had told me that it wouldn't turn on when plugged in, so I figured I probably could plug it in safely without risking any further damage. 

  Upon plugging the speaker in to 240V (the rated voltage), I was thankfully not greeted by any magic smoke. However, I was a little disappointed to see that the JBL logo on the front dimly lit up. This meant that the problem couldn't be as simple as a blown fuse or loose PSU connection. This was also where having the working speaker started helping to narrow down the issue. When plugged in up, the functional speaker starts with the JBL logo dimly lit for a split second, then brightly lit while the VU meter and button LEDs show an animated test pattern, then goes back to dimly lit until the speaker is woken up with the power button or IR remote. The broken speaker's JBL logo never progressed from the dimly lit state, and was completely unresponsive to button presses or the factory reset procedure with the DIP switches on the rear panel.

  I popped open the busted speaker and was quite impressed by the internals, having only disassembled consumer speakers in the past. The whole back panel is a hefty cast aluminum heatsink with the port integrated. The interior of the speaker was lined with acoustic padding and all cables were wrapped with anti-vibration foam. I was surprised to find that despite the fancy DSP features and digital audio input, the power amplifier stage is old-school class AB fed by a toroidal transformer. That explains how each speaker weighs 12kg!
  
  

  The signal chain is split into two helpfully labelled boards: The smaller digital board, with ribbon cables to the front panel (disconnected in the photos) and to the analog board, and the analog board with the PSU and power amp section and all of the rear-panel IO (analog audio, digital audio, USB, control). 

View all 2 project logs