ICL Keyboard with Buckling Sleeves and conductive PCB
Posted: 10 Jan 2022, 01:13
Long time lurker, first time poster. I recently came across this beauty on eBay. What caught my eye was the nice color scheme, and the absolute level of jank. F-buttons ranging from 1-11, a seemingly lost “) (“ key, stepped keys everywhere and odd clustered mods, not too dissimilar to the ones found on some “newer” Televideo keyboards. I expected to find RMD 973 switches in this keyboard, based on similarity with other boards using this switch, but I was curiously surprised to find them to be some kind of buckling sleeve switches. I’ve not been able to find this switch documented anywhere, which is quite interesting.
The cleaned keyboard
It arrived in dirty condition, but due to the nature of the buckling sleeves “enclosing” the slider, only a little dirt was found on the conductive pads/sliders.
Overview
A 102-button ICL keyboard, that came with the ICL Quattro computer in 85. This was before Fujitsu took over the manufactoring of ICL hardware. I’ve been unable to locate any OEM brand names inside of the keyboard, so I will assume this was produced by ICL in-house.
The keycaps are very thick double shot ABS keycaps in an interesting sculpted profile. It has 11 relegendable function keys, with F1-F11 printed on them - pad printing, I think. The keycap mount is a type of cross-mount, not fully MX compatible. I was able to mount a cap on an MX switch, but this slightly cracked the stem, and the keycap is too thick to go nicely over the housing without hitting slightly, making most keypresses jagged.
The switch: ICL Buckling Sleeves
The switches are conductive buckling sleeves. The buckling sleeve is held in place between the plate and the keycap. The keycap is attached to a slider, which at the bottom of it, has a conductive pad, suspended with rubber in order to give cushioning and overtravel. This results in a very soft tactile switch feel, that actuates around the bottom of the tactile event, and is finally heavily cushioned. I recently tried Topre at a local meetup, and if I remember correctly, the ICL feels quite similar, with a slightly more cushioned bottom.
Off-center keypresses are not handled greatly, and causes increased perceived switch weight. This is especially pronounced on the bottom row as these switches are very curved, and this makes the force applied to the switch angled. This in turn makes me type more heavy-handed than I would type naturally.
The domes themselves have strongly varying force, making some switches more tactile than others. I suspect this is due to wear. When I first received the keyboard, the spacebar was extremely light and sloppy, which was fixed by changing the sleeve.
A single switch is a dummy switch, having an extra plastic element placed beneath it rather than the conductive rubber pad. The “barrel-plate” has a fixed layout, and has inserts on the non-used pads to prevent dust from getting in.
The stem, showcasing the conductive rubber-pad suspended by a small elastic band.
The pcb:
The PCB has a diode for every pad, allowing for full NKRO. There are more pads than there are keys on the keyboard. The pcb states “TSS-Mark III”, that I first assumed to be an OEM manufactorer, but given the lack of search results, I dropped that assumption.
Converting to USB
I located Vcc/Gnd on the pcb and saw that they did not match up with a standard AT DIN plug. The PCB had a leaking capacitor, which explains why my initial attempt at using soarer’s analyzer to decode the signal did not work out. Fortunately the columns/rows are exposed, allowing for a manual conversion. I used a stm32f401-based “Black Pill” as it’s cheap and has enough ports. I desoldered all existing components, added a few extra wires to the PCB, allowing the development board to be mounted directly on the PCB. The PCB consists of 8 columns / 15 rows (or the other way round, if you’d like), which does not give an intuitive matrix, so I didn’t keep count of which columns/rows I wired where, only whether they were columns/rows. I modified the layout slightly, by moving the “dummy-slider” from the bottom-left key, to the “char del” key (two spaces right of F11).
For firmware , I decided to go with ZMK. I’m currently using ZMK in my esoteric 30/40% split wireless keyboards and I like it. I could also have gone with QMK, which is the more traditional choice for wired keyboards at the time of writing. I was lucky enough that Deemen17 had already made a template for starting out with ZMK on a stm32f401-based Black Pill (https://github.com/Deemen17/zmk-config-blackpill_f401), which I shamelessly copied and modified.
For mapping the keymap to the actual layout, I used the ZMK feature of matrix transform. Finding out how to transform the matrix was quite tedious. I loaded up Karabiner-Eventviewer (xed would be a linux alternative), and remapped the keys one-by-one, making many errors along the way.
Conclusion
Very cool keyboard, interesting keyfeel, beautiful keycaps. It has been my first time converting an existing PCB to use a modern development board, and was a fun experience. The varying weight based on sleeves and off-center keypresses makes the keyboard tedious to use, but given all the hours I’ve spent on this, I’ll have to use it. My thirst for rubberbucklers has awoken.
More pictures:
Similarities with other keyboards
Future work
The cleaned keyboard
It arrived in dirty condition, but due to the nature of the buckling sleeves “enclosing” the slider, only a little dirt was found on the conductive pads/sliders.
Overview
A 102-button ICL keyboard, that came with the ICL Quattro computer in 85. This was before Fujitsu took over the manufactoring of ICL hardware. I’ve been unable to locate any OEM brand names inside of the keyboard, so I will assume this was produced by ICL in-house.
The keycaps are very thick double shot ABS keycaps in an interesting sculpted profile. It has 11 relegendable function keys, with F1-F11 printed on them - pad printing, I think. The keycap mount is a type of cross-mount, not fully MX compatible. I was able to mount a cap on an MX switch, but this slightly cracked the stem, and the keycap is too thick to go nicely over the housing without hitting slightly, making most keypresses jagged.
The switch: ICL Buckling Sleeves
The switches are conductive buckling sleeves. The buckling sleeve is held in place between the plate and the keycap. The keycap is attached to a slider, which at the bottom of it, has a conductive pad, suspended with rubber in order to give cushioning and overtravel. This results in a very soft tactile switch feel, that actuates around the bottom of the tactile event, and is finally heavily cushioned. I recently tried Topre at a local meetup, and if I remember correctly, the ICL feels quite similar, with a slightly more cushioned bottom.
Off-center keypresses are not handled greatly, and causes increased perceived switch weight. This is especially pronounced on the bottom row as these switches are very curved, and this makes the force applied to the switch angled. This in turn makes me type more heavy-handed than I would type naturally.
The domes themselves have strongly varying force, making some switches more tactile than others. I suspect this is due to wear. When I first received the keyboard, the spacebar was extremely light and sloppy, which was fixed by changing the sleeve.
A single switch is a dummy switch, having an extra plastic element placed beneath it rather than the conductive rubber pad. The “barrel-plate” has a fixed layout, and has inserts on the non-used pads to prevent dust from getting in.
The stem, showcasing the conductive rubber-pad suspended by a small elastic band.
The pcb:
The PCB has a diode for every pad, allowing for full NKRO. There are more pads than there are keys on the keyboard. The pcb states “TSS-Mark III”, that I first assumed to be an OEM manufactorer, but given the lack of search results, I dropped that assumption.
Converting to USB
I located Vcc/Gnd on the pcb and saw that they did not match up with a standard AT DIN plug. The PCB had a leaking capacitor, which explains why my initial attempt at using soarer’s analyzer to decode the signal did not work out. Fortunately the columns/rows are exposed, allowing for a manual conversion. I used a stm32f401-based “Black Pill” as it’s cheap and has enough ports. I desoldered all existing components, added a few extra wires to the PCB, allowing the development board to be mounted directly on the PCB. The PCB consists of 8 columns / 15 rows (or the other way round, if you’d like), which does not give an intuitive matrix, so I didn’t keep count of which columns/rows I wired where, only whether they were columns/rows. I modified the layout slightly, by moving the “dummy-slider” from the bottom-left key, to the “char del” key (two spaces right of F11).
For firmware , I decided to go with ZMK. I’m currently using ZMK in my esoteric 30/40% split wireless keyboards and I like it. I could also have gone with QMK, which is the more traditional choice for wired keyboards at the time of writing. I was lucky enough that Deemen17 had already made a template for starting out with ZMK on a stm32f401-based Black Pill (https://github.com/Deemen17/zmk-config-blackpill_f401), which I shamelessly copied and modified.
For mapping the keymap to the actual layout, I used the ZMK feature of matrix transform. Finding out how to transform the matrix was quite tedious. I loaded up Karabiner-Eventviewer (xed would be a linux alternative), and remapped the keys one-by-one, making many errors along the way.
Conclusion
Very cool keyboard, interesting keyfeel, beautiful keycaps. It has been my first time converting an existing PCB to use a modern development board, and was a fun experience. The varying weight based on sleeves and off-center keypresses makes the keyboard tedious to use, but given all the hours I’ve spent on this, I’ll have to use it. My thirst for rubberbucklers has awoken.
More pictures:
Spoiler:
Spoiler:
Spoiler:
