[IC] FSSK Metal Plate | Completed

[Icarium]

Maybe this question is kind of off topic but if you're redoing the plate and PCB why not add more keys to the bottom row?

[Muirium]

I don’t think editing the PCB is a good idea. Changes need tested in prototype rounds, et cetera et cetera.

Those who want to use this as a Unicomp FSSK kit need quite a lot of changes after all, and someone doing the design who actually has such a keyboard! Better to make it a separate project.

/says impatient SSK modder.

[DMA]

Well, some of us are just sadomasochists then.

I’ve flashed my Xwhatsit’s dozens of times with many different firmwares, let alone keymaps.

Yes, but have you _built_ those? xwhatsit doesn't require firmware build for layout changes And flashing is done by the same utility that's used to change the layout.

My AT has run Xwhatsit’s original, Pandrew’s QMK, Arkku’s QMK, and just days ago back to Pandrew’s again. It’s not alone.

Same question.

Let's face it - nobody builds firmware themselves. And if you do - you're likely doing PCB design and to do that you're basically _required_ to run Windows. Yes, since Cypress was devoured by Infineon there's a certain risk that toolchain will go the way of the dodo. But c'mon, atmega 2560 is EOL for like 3 years and the company was eaten by (of all things) microchip long time ago and that doesn't seem to bother anyone

QMK Toolbox is a great little GUI tool even a GUI using tool like me can breeze.

Oh my god. Not even 10 years have passed!

But in any case - not sure why CS was even dragged into this.. I mean.. I even have a reflow oven now, so small-batch manufacturing isn't a problem, but I can bet a whole penny everybody will just vote for xwhatsit because QMK.

[DMA]

Maybe this question is kind of off topic but if you're redoing the plate and PCB why not add more keys to the bottom row?

LOL. [traumatized mr. Incredible.png]
You see, due to a string of unfortunate hardware design decisions xwhatsit requires _really_ uniform matrix readouts across the column. Which means that the PCB must be incredibly uniform electrically. That's why you see those nonsensical tracks that don't lead anywhere - it was found that they are "improving" the output (the quotes are there because nobody at the time was sure how it really worked, and adding those tracks kinda improved the situation and it was a time when every prototype iteration costed $5 per square inch so 3 boards were way north of $100).
So now there's a religious fear of changing the PCB - even though the prototype iteration is now less than $40 for 5 boards. Things change, people don't

[taylorswiftttttt]

I will be ordering 10 SMD Model F Controllers. Those who want them will have them included with their kit, and those who want to use alternative controllers will not. I can find homes for the extras.

Unfortunately the 74AHC595D,118 shift register used in the SMD Model F Controller is out of stock on JLCPCB's in-house parts inventory. It is in stock from a distributor from their Global Parts Sourcing service, but their website appears to be... broken? and I literally cannot view my cart. It just takes me to a blank page. I'll have to talk to their customer service to see what the issue is.

So, I don't have an accurate quote yet. I expect it will be under $20 per controller, plus the cost of a USB-C Pro Micro clone (~$5?)

If you can provide the dimensions I can measure my Mini M and confirm if they’ll work.

Can you measure the distance from the right side of the furthest right barrel to the right side of the barrel plate, and the left side of the furthest left barrel to the left side of the barrel plate?

Even if the PCBs don't support Mini M this time, the plates might as well.

Since we're editing PCBs - please add vias between floating pads and the pads under them - one small via, somewhere in top 3mm of the pad.
I guarantee you this will result in way better signal when key is pressed, without affecting the resting state - which, in turn, will virtually guarantee lack of xwhatsit-specific problems with sensing.

I have been told by multiple people who are far smarter than me to not use VIAs on thin, curved PCBs, so I hope you understand my reluctance. I could try this out in the future, as I will definitely be designing more capacitive PCBs, but for now I want to try to limit the scope of this project: the more innovations I try to include, the longer it will get pushed out.

The FSSK PCB is a known working PCB, and although I'm sure it can be improved, I don't think now is the time, at least not for me.

I understand that you know way more about electrical engineering than I do and I really appreciate your feedback; please don't take this as an insult.

Maybe this question is kind of off topic but if you're redoing the plate and PCB why not add more keys to the bottom row?

Adding keys requires means redoing the matrix and routing, which I am not comfortable doing on diptrace. The FSSK PCB I am using is actually the FEXT v1.00B, which is a fullsize PCB that has been designed so that the numpad does not share any columns with the rest of the board and can literally just be deleted.

[DMA]

Unfortunately the 74AHC595D,118 shift register used in the SMD Model F Controller is out of stock on JLCPCB's in-house parts inventory.

I can serve as a PCBA shop if you send me the parts, PCBs and a stencil (although with no QFN parts stencil isn't really required, I can just hand-solder it all). Will even do initial programming. May even be cheaper - depending on parts costs.

I have been told by multiple people who are far smarter than me to not use VIAs on thin, curved PCBs, so I hope you understand my reluctance. I could try this out in the future, as I will definitely be designing more capacitive PCBs, but for now I want to try to limit the scope of this project: the more innovations I try to include, the longer it will get pushed out.

The curvature is pretty small, so risk is virtually zero - but ok, I semi-understand. You are driving that part, after all. Double-check, however, that you don't touch any traces with your edits - xwhatsit (at least the exemplar I had my hands on) _severely_ overdrives the columns, and column traces locations and lengths _are_ pretty critical. Same goes for extra keys - unlike the vias, it's a _real_ minefield for xwhatsit-based designs.

I understand that you know way more about electrical engineering than I do

I don't. But I have 3 F122 PCBs in my garage with pretty interesting modifications. So the vias advice comes from practice - it boosts the readout about 6x, which boosts the SNR about 6x.

please don't take this as an insult.

None taken - you're driving this, it's your authority.

[Muirium]

Same question.

Let’s take this argument to a new thread if we mean to continue it; this thread is already for a whole hardware project in its own right. They get plenty busy.

[Muirium]

Searching my storage, looks like I already used the Xwhatsit Model F controller I thought I had spare. So +1 for a controller, please.

[thefarside]

If you can provide the dimensions I can measure my Mini M and confirm if they’ll work.

Can you measure the distance from the right side of the furthest right barrel to the right side of the barrel plate, and the left side of the furthest left barrel to the left side of the barrel plate?

The right side appears to be a little over 7.5mm from the rightmost barrel to the edge of the barrel plate.

DB91C4F5-2BFD-459D-B118-82EDE14894A3.jpeg (1.35 MiB) Viewed 20305 times

The left side appears to be 6.5mm from the leftmost barrel to the edge of the barrel plate.

3940D46C-E13D-42C5-9AFE-3BAAFF213D07.jpeg (2.13 MiB) Viewed 20305 times

Hope this helps!

[taylorswiftttttt]

I have ordered 10 PCBs and 10 SMD Model F Controllers.
The final cost of each PCB is $6.63 and the final cost of each controller is $6.63 plus $5.01 for a USB-C Pro Micro clone.
My current thinking is that the controller and pro micro will each be optional.
I cleaned up the PCB slightly before ordering. This was the design I submitted:


I also ordered 5' of some 0.156" pitch wire from digikey (https://www.digikey.com/en/products/det ... 00/7809902)
I have not tested the wire yet, so I don't know how much each person realistically needs.

The wire should arrive this week and the PCBs should arrive in about 2 weeks (they are shipping from China).

Next steps: order the plate, foam, and hardware.

Also, I don't think this is a GB anymore. The total cost of everything isn't even that much so I'll probably just front the cost.

I can serve as a PCBA shop if you send me the parts, PCBs and a stencil (although with no QFN parts stencil isn't really required, I can just hand-solder it all). Will even do initial programming. May even be cheaper - depending on parts costs.

I appreciate the offer. I managed to get the shift registers, so I went ahead with JLCPCB.

The curvature is pretty small, so risk is virtually zero - but ok, I semi-understand. You are driving that part, after all. Double-check, however, that you don't touch any traces with your edits - xwhatsit (at least the exemplar I had my hands on) _severely_ overdrives the columns, and column traces locations and lengths _are_ pretty critical. Same goes for extra keys - unlike the vias, it's a _real_ minefield for xwhatsit-based designs.

I have 3 F122 PCBs in my garage with pretty interesting modifications. So the vias advice comes from practice - it boosts the readout about 6x, which boosts the SNR about 6x.

I will have to try this out the next time I order capacitive PCBs.

The right side appears to be a little over 7.5mm from the rightmost barrel to the edge of the barrel plate.

The left side appears to be 6.5mm from the leftmost barrel to the edge of the barrel plate.

Thanks! The distance on either side of my barrel plate is just under 9 mm. Judging by your pictures, this is a little less than the width of the back plate, so, the plate will probably fit, but again, I cannot confirm it.

[inozenz]

I have ordered 10 PCBs and 10 SMD Model F Controllers.
The final cost of each PCB is $6.63 and the final cost of each controller is $6.63 plus $5.01 for a USB-C Pro Micro clone.
My current thinking is that the controller and pro micro will each be optional.
I cleaned up the PCB slightly before ordering. This was the design I submitted:


I also ordered 5' of some 0.156" pitch wire from digikey (https://www.digikey.com/en/products/det ... 00/7809902)
I have not tested the wire yet, so I don't know how much each person realistically needs.

The wire should arrive this week and the PCBs should arrive in about 2 weeks (they are shipping from China).

Next steps: order the plate, foam, and hardware.

Also, I don't think this is a GB anymore. The total cost of everything isn't even that much so I'll probably just front the cost.

I can serve as a PCBA shop if you send me the parts, PCBs and a stencil (although with no QFN parts stencil isn't really required, I can just hand-solder it all). Will even do initial programming. May even be cheaper - depending on parts costs.

I appreciate the offer. I managed to get the shift registers, so I went ahead with JLCPCB.

The curvature is pretty small, so risk is virtually zero - but ok, I semi-understand. You are driving that part, after all. Double-check, however, that you don't touch any traces with your edits - xwhatsit (at least the exemplar I had my hands on) _severely_ overdrives the columns, and column traces locations and lengths _are_ pretty critical. Same goes for extra keys - unlike the vias, it's a _real_ minefield for xwhatsit-based designs.

I have 3 F122 PCBs in my garage with pretty interesting modifications. So the vias advice comes from practice - it boosts the readout about 6x, which boosts the SNR about 6x.

I will have to try this out the next time I order capacitive PCBs.

The right side appears to be a little over 7.5mm from the rightmost barrel to the edge of the barrel plate.

The left side appears to be 6.5mm from the leftmost barrel to the edge of the barrel plate.

Thanks! The distance on either side of my barrel plate is just under 9 mm. Judging by your pictures, this is a little less than the width of the barrel plate, so, the plate will probably fit, but again, I cannot confirm it.

Here is saylor twift talking, plz accept donations even if you are going to sell at cost or my heart wont be at ease for all the work youve put into

[Muirium]

Great stuff!

I’m kinda mesmerised by these RP2040 capsense controllers–check those curves–but tried and tested is good, too.

[taylorswiftttttt]

Here is saylor twift talking, plz accept donations even if you are going to sell at cost or my heart wont be at ease for all the work youve put into

I love you, Inozenz

The wire I ordered from digikey came in a couple days ago. It was a failure. It has too many strands per wire and they won't fit in the holes. Ribbon cable with fewer strands per wire may fit in the holes. However, ribbon cable like that may not work anyway, because the PCB sits flat against the back plate and the wire has to be soldered in a way that doesn't lift the PCB up too much to avoid over-stressing the PCB. Then, because of the tight clearances in the top of the SSK case, it has to make a very sharp bend downwards.

I used a 40-pin IDE cable for my SSK. It is also not ideal but it works.
At this point, I think I'd like to forget about including wire and let you guys connect the controllers however you see fit.

[DMA]

to avoid over-stressing the PCB.

Stop stressing about that. PCB isn't going anywhere.

Then, because of the tight clearances in the top of the SSK case, it has to make a very sharp bend downwards.

Why not just solder it so it goes down from the start?

I used a 40-pin IDE cable for my SSK. It is also not ideal but it works.

Why use anything else? It's a time-tested solution and dirt-cheap as you can get those cables at your nearest PC recycler by weight.

At this point, I think I'd like to forget about including wire and let you guys connect the controllers however you see fit.

Right call.

[Muirium]

Aw! What will I use? Coat hangars are good right? :authentically_clueless:

[taylorswiftttttt]

Plates and foam ordered.



Here is the final plate design. The orange lines (which are not part of the cad drawing) indicate where I cut with a vertical bandsaw after rolling the plate. Slip rollers inherently leave a distance unbent on the leading and trailing edge, approximately equal to the distance between adjacent rollers.



Here is the final foam design.



PCBs, controllers, and pro micros have arrived. I will epoxy the USB ports on all the pro micros and then start soldering them to the controllers.

[inozenz]

Oh boy can't wait to have my PCB and plate also damn awesome of you being so transparent

[Muirium]

Agreed. Quality work and a quality fellow. Christmas has come for my SSK! Really looking forward to it.

[Muirium]

January 6? Anticipation ramping up…

[thefarside]

I’m 99% sure I completed the interest form when it was open but if not, I’d be happy to participate.

[snoopy]

looks so good... can't wait...

[adamcobabe]

Seems like I'm way to late to get in on this. Is that true? Took a break from DT at the wrong time, I guess.

[Muirium]

^ Know how that feels. I missed the original round despite following the buildup! Bad timing.

Anyway, all the more ready and willing for this one. The final countdown.

[taylorswiftttttt]

Two items of bad news:
1. Ponoko hasn't even shipped it yet. I guess they're busy over the holidays... all my previous orders were made in exactly the estimated time
2. I ordered 10 PCBs from JLCPCB when I need about 13. I ordered 5 more. Although they cost more each I will match the price of the first 10 because its my fault

I’m 99% sure I completed the interest form when it was open but if not, I’d be happy to participate.

You completed it on time

looks so good... can't wait...

Thanks - I hope it lives up to everyone's expectations.

Seems like I'm way to late to get in on this. Is that true? Took a break from DT at the wrong time, I guess.

There are no extras, unless someone backs out

[taylorswiftttttt]

Plates and foam arrived.



I completed the first plate and did a quick test fit to verify dimensions.



Quick video of the plate:

Obviously, the top and bottom edges, which are cut with a bandsaw, are rougher than the sides, which are laser cut. I will be filing the bandsaw cuts to remove any sharp edges, but I'm not going to spend a whole bunch of time trying to make them perfect.

I am working on the rest of them. It will take a while.

Once they are all complete, I will start sending invoices in DMs.

[Muirium]

Yes! Bring it MFing on!

[inozenz]

Quick video of the plate:

Can i have "ST" instead of "TS", your dearest Saylor Twift

Looks amazing!

[thefarside]

Looks awesome!

[dcopellino]

Great update for a great project. This MF things and their derivatives are becoming more and more popular here on DT even though it's already been several years since it was introduced by i$.
My 2 cents advice, regarding my previous experience with jlcpcb, is this: pay attention to the circuit path highlighted here in the attached pic.

Screenshot_20230121_101353_Chrome.jpg (737.66 KiB) Viewed 18078 times

Mine was interrupted by the hole, and I had to patch it up

[Muirium]

Good point. Looks like an accident waiting to happen.

I see the affected key is the “extra” pad, between Control and Alt on an OG SSK. Does a flaw in that trace disable that one spot or does it take out other (actually populated) keys on the matrix?