F104+SSK+122+62+77+50+Ergo orders now open! New Kishsaver+Industrial Model F Keyboards

[beepbloop]

Received my F77 last week and am thoroughly enjoying it! I ordered the split right shift option, industrial gray case, and unprinted keys. Feels great to type on, smoother and more effortless than my old XT I have mine set up with a mashup of keycaps including APL keys and some from a Model M and aforementioned XT. I love how insanely heavy this thing is, zinc rules. Will probably be tweaking the layout as I continue using it, thankfully changing the keymap is super easy and quick with the xwhatsit software.

Here's some pictures and a typing test video ( too lazy to learn how to type properly sorry :] )

[dcopellino]

Looks amazing. The finish on the case seems to leave a bit to be desired though no?

emh, what the earth are you saying Willie? That's a refined finishing effect once upon a time you could find solely on some very rare keyboards reserved to selected choosy banks.

[consolation]

Well... now I feel old, I remember when manufacturers would slap crinkle finish on anything that they wanted to "say" quality.
It used to be a way to offer a more durable protection, when anodising was black magic, and was put on early technical and electric devices. Since they were reserved for the rich and industry, it became associated with quality.

[beepbloop]

The sun shining on the case may make it look strange but the finish is excellent in person. I really have no complaints. Excited to order some new keys once Ellipse finalizes the dye-sublimation setup. He has seriously done an amazing job.

... I also have a craving for the F62 now too

[dcopellino]

The sun shining on the case may make it look strange but the finish is excellent in person. I really have no complaints. Excited to order some new keys once Ellipse finalizes the dye-sublimation setup. He has seriously done an amazing job.

... I also have a craving for the F62 now too

I'am waiting that all this F buy sale finishes, so Ellipse will put sales on....

[Hypersphere]

I've been away from DT for quite some time due to work demands.

Looking forward to receiving my new Model F keyboards.

[Ellipse]

Thanks beepbloop for the review with photos and video! Nice APL key set too!

Another new variation:

2020-03-13_18-04-33.jpg (536.13 KiB) Viewed 11689 times

2020-03-13_18-04-20.jpg (343.92 KiB) Viewed 11689 times

[Scarpia]

Thanks beepbloop! Out of all the early boards reviewed so far, yours is the one closest to my own order, and wow it looks fantastic like that! At this rate I’ll probably not be receiving mine until 2021, but what a glorious keyboard it is!

[Ellipse]

Two more variations going out:

The first variation below is the first assembled keyboard photographed with the custom Silver Gray case color. I paid for some extras of this color beyond what was ordered by everyone so I have about 10 extra cases in stock. This custom color will not be made in the final round so this is your only chance to order it. The color is difficult to photograph but the Pantone Silver Gray color is what I would describe as a gray color with a small amount of blue.

2020-03-16_20-06-53.jpg (326.87 KiB) Viewed 11207 times

2020-03-16_20-07-06.jpg (389.79 KiB) Viewed 11207 times

2020-03-16_20-12-24.jpg (313.43 KiB) Viewed 11207 times

2020-03-16_20-12-34.jpg (320.85 KiB) Viewed 11207 times

[Ellipse]

Maledicted posted a nice review with high res photos over on GH and I wanted to share some repair notes over here too:

Regarding re-seating springs: just a note for those who come across the review that in nearly all cases you do not need to take apart the keyboard to fix keys that do not click or spend a few minutes pressing a troublesome key - I have posted some videos on this thread as well as on the web site blog detailing a quick spring adjustment and key re-seating guide that requires less than one minute per key to do. Very important to reattach the key as shown in these videos, with the keyboard positioned as shown in the videos (vertically, with the space bar row up). The goal is to have the spring touch the 12 o'clock position of the barrel when the keyboard is positioned that way. If the spring end is not positioned at 12 o'clock (per the video) and the spring does not touch the barrel, buckling error is more likely to occur.

Regarding squeaky space bars - that is nearly always the case of adjusting/slightly stretching or replacing the spring rather than due to the space bar stabilizer wire. This is something I test on all keyboards as part of QC but I hope to improve on this. On a similar note, especially with rattling space bar wires as well as to prevent the space bar from getting stuck on the end of the metal tab, I recommend pressing down on the metal tabs if you re-seat the space bar (after confirming no squeak and proper buckling when pressing the space bar). I do this as needed during QC. With the final production round, the metal tabs will be adjusted a bit to minimize the need to get them out of the way manually.

[Ellipse]

Dye sub testing update:

I think we are finally on par with a 50th to 90th percentile IBM XT for dye sublimation quality! This was the original goal of the new dye sublimation.

Color and bleed/sharpness are close to my reference IBM XT keyboard (original key from that keyboard is on the left; the right side key is the latest test). Looks great under the stereo microscope too. (As a reminder please disregard the legend positioning. The full sublimation transfer sheets have pin registration with the jig for proper placement)

2020-03-19_15-20-35.jpg (297.53 KiB) Viewed 10735 times

[webwit]

Any news on the problem I have? It's a bit of a bummer on an otherwise fantastic keyboard. Hope it's software and solvable but I don't know.

[Twst]

Now that is just showing off on the dye subs!
Congratulations - it looks really perfect!

Looking forward seeing a whole freshly dye subbed set for the babies

[Invisius]

Well done on the dye subs, they truly look up to spec with original IBM quality. I can see these becoming more popular than Unicomp replacements if you decide to produce them after the new-F runs are done!

[tentator]

Any news on the problem I have? It's a bit of a bummer on an otherwise fantastic keyboard. Hope it's software and solvable but I don't know.

I can confirm I see the same on mine.
I was thinking to replace the xwhatsit with a commonsense from @dma if it would be able to do space-fn but so far, despite promises, it has never worked unfortunately.. but it might pretty much solve at least the problem you're mentioning a couple of posts agoo.

Kr,
Tent:wq

[Shifty]

webwit, I think that's just how the xwhatsit controller functions, I complained about this here a while ago when I installed one in my AT - viewtopic.php?f=2&t=20494&p=433522#p433522

It might be possible to make a custom firmware for the xwhatsit controller to fix this, but I haven't had any success on that front. I ended up just using autohotkey to do what I wanted to do. Sadly it's not a perfect solution, some applications don't really play nicely with it.

[swampangel]

It might be possible to make a custom firmware for the xwhatsit controller to fix this, but I haven't had any success on that front. I ended up just using autohotkey to do what I wanted to do. Sadly it's not a perfect solution, some applications don't really play nicely with it.

You don't happen to have any source code kicking around, or notes on what you tried, do you?

I don't think I'll see my F77 for a while yet but if nobody's solved it by then, I'll have a go.

[0100010]

Isn't the source code on Toms old site?

ETA: [url]https://static.wongcornall.com/ibm-capsense-usb/[/utl]

[Shifty]

That's just the compiled firmware, but the source code is available here - https://github.com/BASLQC/xwhatits-capsense-controller

I didn't really spend any time on it swampangel, but if I get some free time I might take another look. Having an F62 on the way might give me more motivation, it would be nice to have macros stored on the keyboard rather than having to swap autohotkey scripts all the time.

[tentator]

dual action keys would be just awesome (if you press a key you get for instance arrow up, but if you keep it pressed you get another layer like shift key)...

[webwit]

webwit, I think that's just how the xwhatsit controller functions, I complained about this here a while ago when I installed one in my AT - viewtopic.php?f=2&t=20494&p=433522#p433522

It might be possible to make a custom firmware for the xwhatsit controller to fix this, but I haven't had any success on that front. I ended up just using autohotkey to do what I wanted to do. Sadly it's not a perfect solution, some applications don't really play nicely with it.

Thanks for this, I hadn't seen this before. This confirms it is also a problem on original keyboards, not just Ellipse ones. And that I'm not crazy and the only one with this problem. For reference this was also discussed here:
viewtopic.php?p=460637#p460637
Seems independently exactly the same problem and exactly the same autohotkey fix.

It does unfortunately mean that especially the HHKB version (where the layer is used a lot) is currently essentially flawed out of the box. If it outputs extra characters you can't trust it.

As seen in the link above, I'm offering a bounty for a fix! One could argue if Ellipse should do this, but that's up to him. For me, since it's a generic issue with all keyboards using xwhatsit and layers, and it's open source, I don't mind offering this for the advancement of the software.

I'm putting a bounty of one IBM 4704 50-key* (not refurbished) on the best working fix for above problem in the xwhatsit IBM Capsense released under GPLv3:



* If you fancy something else that's negotiable too of course.

[pandrew]

[EDIT : removed the attachment to avoid causing trouble for anyone else]
[EDIT : please see viewtopic.php?p=461737#p461737 for a working pure xwhatsit solution,
also, please see viewtopic.php?p=461764#p461764 for a solution that includes joc's improved debouncing ]

I'm putting a bounty of one IBM 4704 50-key* (not refurbished) on the best working fix for above problem in the xwhatsit IBM Capsense released under GPLv3:



* If you fancy something else that's negotiable too of course.

Hi webwit,
I gave it a try, please see the attached .hex file, compiled for BOARD_TYPE=MODEL_F
The changes are over here:
https://github.com/purdeaandrei/ibm_cap ... 9ab0bce438

I am not really able to test it, because my model F has not yet arrived, and I don't have another xwhatsit-compatible keyboard.
Please test it and let me know if it fixes your issue.

Cheers,
Andrei

[webwit]

That makes Windows go berzerk like a million keys are pressed at once, and bricked the keyboard, as I can't even reflash it or reset it (Windows can't handle all the presses before loading capsense). I have no idea how I'm gonna fix this...

[pandrew]

Dammit! That's what you get when you're too confident.
Sorry, man, let's fix your board first, and then I'll try to figure out what went wrong.
Xwhatsits should have a programming switch on them, see this screenshot from the schematic.
Can you find this on your board?

Andrei

[webwit]

Found it, bottom right.

f62-2.jpg (523.15 KiB) Viewed 10159 times

P.S. On the smaller case the soft tinier screws are even a bigger problem... I got one now that's completely stuck, can't even move it with pliers. Since it's only one I could still rotate the bottom to get to the inners.

f62-1.jpg (722.16 KiB) Viewed 10159 times

[pandrew]

Could you try shorting the PROG pads while you plug in the keyboard?
Not sure it will have any effect at all, I can't find the code that would put it into bootloader mode, maybe the bootloader looks at that pin?
Anyway, it's worth a try.

Also, how does windows react to the keypresses exactly? Does it throw out the USB device itself?
If not, then one option would be to lock the keyboard input, for example teamvier used to have (and may still have) a way to diable all input on the machine you're connecting to.

[webwit]

How do I practically short it? With a piece of wire?

There's something about shorting on Ellipses blog:
https://www.modelfkeyboards.com/blog/
" if can’t install, can also try shorting PROG on the xwhatsit controller and keeping it shorted, then connecting it to USB, wait 1-2 seconds MAXIMUM, then release prog. will cause error device descriptor failed if you hold it too long"

Windows is starting a lot of programs, installed a different theme, went into admin mode, starting running out of memory etc. I.e. hell on earth. When I unplugged, other keyboards did not work and I had to reboot.

I already tried some progs to lock keyboard input like keyfreeze but they are too soft and it does not have effect.

[pandrew]

Yes, exactly, try to short it with a wire, or metallic tweezers, or a fork, or anything, and keep it shorted while the USB cable is being attached.

By the way, I figured out my mistake, it was really stupid: https://github.com/purdeaandrei/ibm_cap ... a330d2c400

I will have a new hex file, but let's fix your board first.
Andrei

[webwit]

Thanks, that did the trick. Somehow It still pressed keys but this time together with a key locker I could launch capsense and do an emergency halt and reflash. Then still lots of key pressed but a reboot did it.

[pandrew]

Thanks, that did the trick. Somehow It still pressed keys but this time together with a key locker I could launch capsense and do an emergency halt and reflash. Then still lots of key pressed but a reboot did it.

Alright, I'm relieved

So I managed to reproduce the many keypresses problem using another atmegau4 board. (without a keyboard, just a development board). The problem was that I chose an unsigned datatype, and because of that there was an underflow, and that caused a false keypress on every key, nonstop. No wonder windows went crazy from that. Anyway, I fixed that datatype error, and the false keypresses are now gone.

I've attached a new .hex with the fix.

I still can't test on a keyboard, if this works as intended, but the keypress hell is gone,
I hope this hex solves the issue with the layers. Please let me know if it helps (if you are brave enough to try it after my previous failure)

Andrei

[EDIT: NOTE: the attached file is based on pure xwhatsit 0.9.0]
[EDIT: If you are looking for a solution that includes joc's improved debouncing, and this layer fix, then please see viewtopic.php?p=461764#p461764 ]