Author Topic: advice for building a very feature-heavy keyboard (Read 8676 times)

Volny · « **Reply #50 on:** Fri, 14 May 2021, 12:23:13 »

When you say to short/solder "to the diode" do you mean to the black part, to one of the wires coming in/out of the black part, or to the rectangular pad?

suicidal_orange · « **Reply #51 on:** Fri, 14 May 2021, 12:59:47 »

You need to connect to the left pad of the diode where it (should be) connected to the switch pin. You could test on the other end but if you solder there it will bypass the diode and you'll get ghost keypresses...

This is weird to talk about clearly so here's a pic:

If we say F11 is where the pink dot is and the other dots are on a dead switch you would test by shorting pink to orange. If that works short pink to green. If that works the diode is connected, if not you need to solder between green and orange to reconnect it and re-test the switch. If the switch still doesn't work (after soldering or because pink to orange already worked) you need to solder pink to red to substitute for the trace on the other side of the board.

You might get away with one jumper from F11 to fix two switches so test all of them after soldering the columns

TK0 · « **Reply #52 on:** Fri, 14 May 2021, 21:02:02 »

Maybe I arrived a bit to late to this thread. But, if you ever need/want to replace your corsair, maybe consider a 96% keyboard that is QMK compatible, or any layout you prefer, but with QMK compatibility.

For the Macros that we so much love and need, I suggest you configure layers in QMK. Than, you would have one layer with all your normal layout, and, with a designated keypress, you can change your whole layout, in which each key is a different macro. This way, you can configure a lot of layers (iirc 32 in total) giving you more macros in a single kb than you can fit keyboards on your table.

Exactly because of layers (imo) this hobby is filled with smaller layouts. When you get used to your layers, you will basically never need to use your macropads/stream decks.

And, as already has been said, with QMK you can configure RGB as well, if your PCB supports it. I'm out of the loop for RGB QMK compatible fullsize/96% keyboards, but I'm sure a Google (or duckduckgo

) search would solve it.

Anyways, good look with your attempt! Hope it all goes well.

Volny · « **Reply #53 on:** Sat, 15 May 2021, 06:18:58 »

I like that idea in theory, but experience has taught me that 'blind' macros like what you're talking about have a very finite limit before they become unmanageable, or at least decreasingly useful. By 'blind', I mean that you must remember what they are without any labels or cues.

I'm all about trying to optimise my 'flow state' while I'm working. A big part of that is removing as much cognitive load as possible from my poor, pitifully limited working memory (mine is possibly more limited than average, but no one has particularly much of it at all - that's why most phone numbers and melodies are 7 digits/notes or fewer; our paltry brains struggle to hold as little as 8 bits of data in working memory at once). The moment I need to stop and think about whether some hotkey is layer 2 & F5 or layer 3 & F6, I'm using up cognitive load that I'd prefer to use for actual work. Plus it slows me down and shakes me out of my flow.

I do actually do something similar to what you suggest, but using software (autohotkey) rather than firmware. I have my capslock key work as a 'layer' key that turns some 15 other keys into macros. It's excellent for a few simple things that remain intuitive and easy to remember (eg. Capslock + up = capitalise selection) , but using a blind system like that would never cut it for the literally hundreds of macros I currently have set up across my streamdecks and macropads.

gh_pp · « **Reply #54 on:** Sat, 15 May 2021, 17:35:30 »

Easiest solution is to get a QMK enabled per key RGB TKL or full-size , and a QMK enabled macro/numpad+encoder to put on the left/right.

You will have to program the layers and per key rgb highlight for your requirement #1.

Look at QMK massdrop/Ctrl/keymaps/endgame as an example to implement #1

You will want a controller with large firmware size like ARM(e.g drop ctrl/shift) if you need lots of custom code

Volny · « **Reply #55 on:** Sun, 16 May 2021, 04:15:15 »

ok, so the pads and diodes all seem fine. So I guess the problem(s) are broken traces on the rear side, as you said suicidal_orange. By soldering a jumper between F11 and -_ it fixed -_ [{ '" /? and Fn. By soldering a jumper between PrntScrn and INS, it fixed INS, DEL, and Left.

This leaves just one bad key: num3. By trying different shorts while plugged in and a multimeter while unplugged, the absolute closest matching pin I could find to num3 was X, which seems a little crazy. Does it seem feasible that I might indeed need to run a jumper over such a long distance?

suicidal_orange · « **Reply #56 on:** Sun, 16 May 2021, 04:39:53 »

There is the sensible way of designing PCBs where rows run straight across the board and columns are nearly vertical, then there's saving pins by using two rows per physical row and U shaped columns whicn I can understand the idea behind, then there's professional PCB design which either involves lots of drugs or really weird automatic connection of the switches. This board looks to be the last option and if there is no key in the bottom row connected to X there is a gap in the matrix so why not connect num3 there? You can run the trace around the edge of the PCB without crossing anything which would need vias so that makes it 'easier' to design...

There is no problem having a really long jumper just make sure it doesn't get stuck between a switch pin and the case or caught in any of the mounting holes. Tape works for a while before falling off which could be annoying if you move the board (it will rattle around) and you'll need to redo it if you take it apart again in the future, hot glue is perfect for the job if you have access to a glue gun.

Volny · « **Reply #57 on:** Mon, 17 May 2021, 00:13:34 »

Every time I fix something then plug it in and test it, more keys are broken that were working before. This is getting disconcerting. It's not like I'm playing football with it - I'm trying to be gentle while I work on it. I'm soldering de-soldering at 370 degrees, and everything appears to be going smoothly. I place the pcb on some non slip shelf liner while I'm soldering. I try and ground myself regularly (though not religiously, I admit) and I've got some neoprene and rubber inside the case that the pcb rests on while it's plugged in. I don't have all the ribbon cables plugged in, but I'm pretty sure they're just for the side and back leds.

All of the broken keys mentioned in previous posts are fixed, but since then a new column stopped working (... L-O-9-F9) which I've also fixed. Now yet another column is broken on the west side of the alphas. As well as the main enter key and num1, which I may need to bridge all the way to f1, which will be a pain.

It's starting to feel like the sorcerer's apprentice here. If it continues much longer, there'll be more jumper wires than keyboard.

What could I be doing wrong?

suicidal_orange · « **Reply #58 on:** Mon, 17 May 2021, 01:09:42 »

Is the neoprene too thick so you have to use force to get it screwed in? That could be causing the PCB to bend and cracking the traces, other than that your technique sounds good.

Volny · « **Reply #59 on:** Mon, 17 May 2021, 03:00:42 »

I'm not even screwing it in. Just gently sandwiching it between both sides of the case (with neoprene and rubber as a soft protective layer) and resting the layers like that on the desk.

The only thing I can think of is that I've been too rough with sliding the stabilisers in and out as I've been swapping them. I didn't think I was, but it's possible, and those wires are hard metal after all.

I take it my experience is fairly uncommon? Like, if I eventually brick this one and have to replace it with some other brand of keyboard, can I at least reasonably hope that putting these switches into that keyboard won't be as painful?

suicidal_orange · « **Reply #60 on:** Mon, 17 May 2021, 07:41:26 »

It is usual to break a couple of pads, if you're lucky they only take out one switch each, sometimes they take out part of a row or column (like your original breaks) and if you're unlucky you break the exact pad where the whole row or column links back to the controller chip on a badly designed board which is not fun. The extra random breakages you had are not common.

It is possible to scratch the traces using the sharp ends of stabiliser wires, the smooth middle bits probably not though maybe if you keep rubbing the same place it could wear through.

May as well keep firefighting the problems - you wont see al those wires once the case is back on.

Volny · « **Reply #61 on:** Mon, 17 May 2021, 08:42:17 »

Thanks for the moral and technical support. I've definitely needed it.

I've fixed everything so far, with just one dead switch left at the moment. It's one I didn't notice before, but has probably been broken this whole time (it's a macro key which isn't recognised by most software, including the software i was using to test the board thus far.)

It had a bad switch plugged into it (confirmed broken with a smultimeter) so I happily replaced it, but turn out it *also* has a lifted pad - my first one from what I can tell. It kind of feels like whatever can go wrong did go wrong

Anyway, it seems lifted only on one side, and after a bit of cleaning and scraping it seems like I might still be able to get a connection. I soldered really carefully and testing the solder joints with a multimeter shows promise. I'll test it in the PC when I wake up in the morning (the pcb is drying overnight).

If it turns out the pad is too broken to work, what's the procedure? This one seems in a fortuitous spot. It's in a column of 6 macro keys. The multimeter reveals a successful connection between all of their right pins, including the keys above and below the bad key. (see pic)

suicidal_orange · « **Reply #62 on:** Mon, 17 May 2021, 08:49:24 »

I hate to suggest it but the diode for that switch appears to be missing part of it's top revealing some shiny insides - not sure if that would stop it working

Shorting to the other side of the diode would confirm if that is the problem.

Volny · « **Reply #63 on:** Mon, 17 May 2021, 09:08:18 »

I think that was just a combination of a bad photo and shiny goop. From what I can tell with the multimeter it behaves the same as the healthy diodes, and up close it looks more or less the same too (see new pic).

If the key doesn't work tomorrow due to the lifted pad (not visible in the pics because it's currently soldered) what's the procedure? Do I try and fix the pad (if so, how), or can I just run a jumper from somewhere to bypass the pad (if so, where?) thanks!

suicidal_orange · « **Reply #64 on:** Mon, 17 May 2021, 09:15:20 »

That's good because I'm not seeing any spares on this board.

There's no trace to reconnect the pad to so just jump it - if it's the one in the big brown puddle the switch above or below looks very likely, if it's the other pad to the end of the diode in the puddle.

Volny · « **Reply #65 on:** Mon, 17 May 2021, 09:25:57 »

Yes, the lifted pad is in the ~~scene of my finest soldering handiwork~~ big ugly brown puddle. So one end of the jumper will go to the matching pin of the above or below switch. And the other end will go to the lifted pad, is that right? So this way the switch pin will be connected to the correct column even though its own pad is dead - am I understanding correctly?

Volny · « **Reply #66 on:** Mon, 17 May 2021, 23:19:05 »

Finally, it is done. It took a ridiculous amount of time and money, and the PCB looks like a murder scene, but I've rid my K100 of the speed linear switches I hated so much. The keyboard is all sealed up again and - so far, at least - is working perfectly. My measured WPM has increased by almost 20%(!) practically overnight, thanks largely to a great reduction in typos (switching to DSA has also played a part - turns out I also hate the pointy corners of OEM).

Because I'm ~~anally retentive~~ thorough, I have a pretty complex layout of switches that I planned using numerous blind tests and several spreadsheets. In total, I now use some 8 different switches - tactile, linear and even clicky - depending on the keys' function, use pattern, typo risk factor, and stem tightness requirements. And I've selectively used opaque housings (or blackening with a permanent marker) to get rid of RGB underglow except in those few areas where I want it.

I've got a mix of Glorious Pandas, Koalas, SP Purple Stars, Gateron Blacks, Hako Clears, Holy Bobas, and Kaihl Box Thick Clicks, including a few spring variations. So far it feels great (and if it doesn't, too bad - I don't think I'll ever be game to reopen this damn thing again). I also upgraded the stabs, filled the case with dense rubber and neoprene for sound dampening, (partially) fixed one of the rubberdome media keys that I had broken early on, and raised the typing angle. If this thing continues to work (I won't be surprised if it doesn't - it seems to want to break at any opportunity) then this will be close to my ideal keyboard, and certainly an upgrade from the stock version.

No matter what, a big thanks to everyone who helped me with advice on this thread (especially suicidal_orange who got me across the finish line)!

Leslieann · « **Reply #67 on:** Tue, 18 May 2021, 12:16:47 »

Quote from: Volny on Mon, 17 May 2021, 23:19:05

Because I'm ~~anally retentive~~ thorough, I have a pretty complex layout of switches that I planned using numerous blind tests and several spreadsheets. In total, I now use some 8 different switches - tactile, linear and even clicky.

No matter what, a big thanks to everyone who helped me with advice on this thread (especially suicidal_orange who got me across the finish line)!

S.O. is great at figuring this stuff out, I can sometimes do it but they have a knack for it.

As for your crazy switches, I've used 4 in a 65%, various clicky and tactiles and at least 4 springs, mostly it was I need a stiffer switch here or there to avoid accidental clicks and that switch just happened to be handy but still, whatever works for you and gets the job done.

Congrats, and don't sweat the underside, no one ever looks there but you and the next time you do this it will be easier.

suicidal_orange · « **Reply #68 on:** Tue, 18 May 2021, 12:21:05 »

That is a very pretty looking collection of switches!

Glad to hear it's finally working, enjoy

Maledicted · « **Reply #69 on:** Wed, 19 May 2021, 12:11:10 »

That looks great.

You've probably got the most custom Corsair around now. I don't care about dampening and doubt I'll ever try multiple switch types in a single board, so I don't think that will ever change.

The best part is that it is your keyboard now. There are no others like it.

yui · « **Reply #70 on:** Thu, 20 May 2021, 01:46:19 »

went from being quite shy about the modification idea to 300% fast

nice one, when i tried a similar setup i regretted using browns as my main alpha block, looks like you will not have that problem

Volny · « **Reply #71 on:** Thu, 20 May 2021, 09:43:18 »

Tell me about it! In a month I went from having a Corsair K100 I was kinda dissatisfied with to having a Corsair K100 I'm satisfied with..... and a soldering station, and a new macro pad, and a lube station, and enough odd keycaps and switches to kit out 10 more keyboards, and several switch testers, and a whole new shelf in my cupboard dedicated to storing all this new keyboard paraphernalia

I definitely could have gone about this with a bit more self discipline, but I am happy with the outcome


120/100g linear Zealio R1 GMK Hyperfuse 'Split everything' perfection	MX Clear SA Hack'd by Geeks EasyAVR mod

News:

Author Topic: advice for building a very feature-heavy keyboard (Read 8676 times)