domingo, 12 de maio de 2019
Update do Spectrum Next
O update completo aqui, nas palavras da própria equipa:
If you have been closely following the saga of the Next keyboard (recap with updates #43, #44 & #45), here’s the latest chapter on the protracted process that has been getting the membrane to work properly.
We have been having problems with the composite keys (when more than one key is needed for a result, such as Caps Shift + 0 for Delete, or Caps Shift + 5 for Left Arrow): in order to keep the keypresses light and in line with modern typing requirements, we adopted a two-layers keyboard matrix approach replacing the original three-layers from the Spectrum+ and 128. We experimented with the three layers and found the keys too ‘heavy’ to press.
To achieve a two-keys press with a single keypress, we made three contacts in the matrix: a split Caps-Shift, the Key itself (0, 6 etc.) on the bottom layer and GND on the top layer. When the key is pressed, GND touches the two halves beneath and the two keys are detected at once.
The test membranes we had until recently had a high error ratio: if the is key pressed slightly at an angle, one of the two contacts would register before the other. When this first contact happens to be Caps Shift, all is fine; but when it’s not, we get a single key press instead, thus the composite key results fail (we get 0 rather than Delete, for instance).
The keyboard manufacturer was certain this effect would disappear once we built the keyboards in the automated assembly line rather than by hand, where the alignment of the silicone dome under each key would be perfect and ensure it hits both contacts at the same time. But unfortunately, once the first units were made last month and tested, the results were not much better than the hand-made prototypes.
The manufacturer then designed a new membrane with the two half contacts closer together, at the limit of their manufacturing resolution, as they deemed this the final solution. Again, the result wasn’t good enough, we still get a lot of errors on real testing.
Last week we commissioned the manufacture of three new membrane designs in parallel, as the ‘let’s try this now’ approach has cost us two months already. These new membranes have composite contacts that we believe will eliminate the effect of one of them getting pressed before the other, and if it happens, the Caps Shift is the one that gets the first touch a few milliseconds before the other contact.
We hope these membranes will be ready this week, and that should be the last of the problems with the keyboard, as all other issues are solved and closed. The manufacturer has a Next board and the testing software to try the new membranes themselves, rather than waste time shipping them to us to do our own testing. If all goes well, this is the last of it -- so fingers crossed.