Working as a teacher of educational robotics for schoolchildren, in three years of practice I dealt with several dozen "cubes" of EV3 and NXT control modules. In the autumn of 2015, when deploying equipment at a new location, a strange thing came to light: 2 of 12 “computers” refused to turn on, without any reaction to the buttons, to replace the battery and connect to a computer. That fall, no solution was found; the modules were put into the box. Periodically, I tried to revive them, but to no avail. And the solution is found. Welcome under kat for faced with a similar problem.

1. Anamnesis
Today, once again, one of the modules was removed from the cabinet. As usual, no reaction to the buttons, battery replacement, computer, reset button under the battery. But the following was noticed: when you try to turn on, if you do not remove your finger from the central button, the module starts, and after 3-5 minutes it is successfully loaded. One has only to release the button, and the module is immediately turned off, as if the power supply to the module depends on pressing the button. It was established experimentally that the module is also successfully turned on when you press any button along with the central one. After loading, the central button can be released, and then it becomes possible to select items in the menu. An attempt was made to reflash, which ended in failure, and the module stopped loading before the menu. When loading with the central button pressed, the inscription "Starting ..." appeared on the screen, and nothing after it. When loading with the central and right buttons, the inscription "Updating ..."
2. The diagnosis
As soon as we release the central button, the module is immediately turned off. It's all about the buttons! Rather in their design. Having examined the module, I carefully looked at the design of the buttons. Quite nontrivial for modern electronics, which I dealt with. Structurally, the "buttons" that we see outside, under the top cover, are a board with concentrically applied contacts under the key locations. The plus circle is edged with an insulator around which the minus contact is located. Similarly, through shorting the contacts, the action of the keys of many keyboards is implemented. With the difference that in keyboards the contacts are short-circuited with a metallized layer inside the silicone pad that is pressed to the contacts with the key. The shorting of the contacts of the EV3 Brick board is implemented differently: a flexible dome made of thin stainless tin is placed above the contact pad. At rest, the top of the dome is raised above the center contact of the button, and the edges are tightly pressed against the edge contact. When we press the plastic key of the module, the dome is pressed inward and the contacts close. Metal domes on all 6 buttons are not soldered to the board, but bonded by gluing on a thin film, which is applied on top of the board and holds each part in its place.
')
3. Treatment
In my case, one of the domes gave a constant contact to the button. Apparently, the module firmware incorrectly handled this click, turning the EV3 Brick into a real "brick". First, with the domes cleaned together with the film, I achieved a stable launch of the module without a constant impact on the buttons. Then the module was reflashed in the standard way, the ports were checked, everything works. I recommend everyone who faced the sudden death of the module, the “white screen”, to check the buttons under the top cover.
Done!
PS: By the way, the design of the buttons of the NXT module does not have such features. It uses a completely standard silicone with metallization.
PPS: Also, from the notes about the design of the EV3 module from the inside, we can note the good quality of the screen plume, which suffered greatly during all the above-described repair attempts. However, it remained intact, the screen works fine. Among the notable electronic components are 1000mkf and 6.3V capacitors, I wonder how many years they will be enough.
