The screen finally looks the way it should. The photo above is the same display before and after the change — the brightness and contrast we had been missing are there. Getting to it raised two new problems, and we would rather show you both now than wait until they are tidy.
Fixing one thing on a hardware wallet routinely uncovers two more. The board is small, the power rails share space with the signal lines, and a change that solves a display problem is entirely capable of introducing an audio one.
What the display was missing
Update 12 described the two changes. We moved the display connector to the top of the board, which takes a lot of cost out of assembling the PCB, and we added a DC/DC boost that steps +5V up to +12V to act as a charge pump for the screen. What we could not show then was the result. This is it: the panel is bright, the contrast holds, and you can read it at a glance instead of hunting for the text.
The booster whines
The new +12V supply makes a faint, high-pitched whine, and it is coming from the inductor. We think this is a compensation problem: the loop is tuned against a load we assumed rather than the current the display actually draws. That is an ordinary fix rather than a redesign — match the compensation to the real load and it should go quiet.
The USB-C only works one way up
This one is new. The fix from Update 12 holds — the signal is stable and the port works — but the device will not connect to the desktop wallet app, and we have narrowed down why: it depends on which way up the cable is plugged in. One orientation behaves, the other does not. A Type-C plug is reversible by design, so a board has to work identically both ways round, and ours does not yet. That symptom points straight at the USB-C connector and how we have wired it — a part we had not worked with before. It is the first place we are looking, and this one gets a proper investigation rather than another guess.
Why we report problems in the week we find them
It would be easy to sit on this update until the whine was gone and the USB behaved, and publish one clean post about a working device. We would rather not build that habit. A hardware wallet is judged on whether its makers tell you things they would prefer you did not know, and the moment to establish that is while the stakes are still a noisy inductor rather than a security disclosure. Anyone can be transparent about a finished product. The record that means something is the one written before the outcome was known.
That is the honest state of the board on 9 August: the screen is right, and two things behind it are not yet. We will report both as they land — and the prototype is still headed for Bitcoin Thailand Conference 2026.