Olympia People (Fujitsu Leafspring Gen1) USB Conversion
Posted: 06 Jul 2018, 01:33
Got this Olympia People Computer Keyboard on ebay.de recently. I have wanted to try Fujtsu leafspring for a while to see whether they'd live up to the hype.
Seebart has made a showcase of the keyboard here, so I'll mainly focus on the conversion and my opinions about the switches here.
When I first got it the keyboard was very dusty, hence the switches were extremely scratchy, so I desoldered all of them for ultrasonic cleaning. However they were still scratchy after that, so I had to painstakingly clean all switch housing with a cotton swab and ethanol, which finally made them pretty smooth. Because I don't have a logic analyzer nor the knowledge to reverse engineer the protocol, I converted it the brute force way; via a controller replacement. The switches themselves all have integrated jumpers in a mirrored L shape, but tracing the matrix was fairly straightforward, although tedious. Unfortunately the matrix wasn't completely connected to the main controller directly, instead most columns went through some TTL stuff first. This made the wiring of the controller considerably more annoying, which was further compounded by the very tight case insides; the only space to put the controller was below the bottom row. In the end I desoldered the main controller, 3 minor ICs and 2 resistor packs (those probably could have stayed). I had to use a Teensy++ 2.0 because of the large amount of rows/columns (29 in total). I then created a KLE file for the layout, and used it as input for https://kbfirmware.com/ (my config here), which is much more efficient to use for me with these kinds of projects than writing the matrix layout directly in code. I ended up having to resolder a single switch after this, because I put it in the wrong way around, causing the internal jumper to connect a row to a column in an undesired way. In the end I think this was a fun project, but for me the switches didn't live up to the expectations: They have short travel of about 3.3mm, which I surprisingly didn't see any of the opinions I read about them mention, and the difference between preload and bottom out force is high, which results in not that comfortable weighting for me. These switches also have an immsely loud upstroke. Other linears such as vintage MX black or alps SKCC green feel better in my opinion.
Here is a short sound demo of me doing a WPM test on this keyboard, and here is a video (with horrible sound quality).
Seebart has made a showcase of the keyboard here, so I'll mainly focus on the conversion and my opinions about the switches here.
When I first got it the keyboard was very dusty, hence the switches were extremely scratchy, so I desoldered all of them for ultrasonic cleaning. However they were still scratchy after that, so I had to painstakingly clean all switch housing with a cotton swab and ethanol, which finally made them pretty smooth. Because I don't have a logic analyzer nor the knowledge to reverse engineer the protocol, I converted it the brute force way; via a controller replacement. The switches themselves all have integrated jumpers in a mirrored L shape, but tracing the matrix was fairly straightforward, although tedious. Unfortunately the matrix wasn't completely connected to the main controller directly, instead most columns went through some TTL stuff first. This made the wiring of the controller considerably more annoying, which was further compounded by the very tight case insides; the only space to put the controller was below the bottom row. In the end I desoldered the main controller, 3 minor ICs and 2 resistor packs (those probably could have stayed). I had to use a Teensy++ 2.0 because of the large amount of rows/columns (29 in total). I then created a KLE file for the layout, and used it as input for https://kbfirmware.com/ (my config here), which is much more efficient to use for me with these kinds of projects than writing the matrix layout directly in code. I ended up having to resolder a single switch after this, because I put it in the wrong way around, causing the internal jumper to connect a row to a column in an undesired way. In the end I think this was a fun project, but for me the switches didn't live up to the expectations: They have short travel of about 3.3mm, which I surprisingly didn't see any of the opinions I read about them mention, and the difference between preload and bottom out force is high, which results in not that comfortable weighting for me. These switches also have an immsely loud upstroke. Other linears such as vintage MX black or alps SKCC green feel better in my opinion.
Here is a short sound demo of me doing a WPM test on this keyboard, and here is a video (with horrible sound quality).