deskthority

So my ordered Cherry G86-61400 arrived today. I'm typing this message on it currently, with my OS set to US ANSI layout (though I'm typing Dvorak).

First, while it does allow reprogramming of all keys, actually moving them is a problem because the keys are contoured by row. Not only that, but the stems have two anti-rotation locks -- which are DIFFERENT ON EVERY ROW. One always points straight up (except on the modifier/space row), but the other points in some direction almost but not quite straight down. This means if you want to move alpha keys between rows, you'll have to get out wire snips and/or an X-acto knife and remove the lower lock along all but the top 1 mm or so of the stem. Fortunately, I had suitable wire snips but still had to do touch-up work with the pen knife.

They also did this to the arrow keys. Apparently they don't want you moving those somewhere else either, even though the software does allow for it.

There is a moderate difference between Row 1 and Row 2, a small difference between Row 2 and Row 3, and a large difference between Row 3 and Row 4. The arrow keys are all Row 4 profile. The Row 1 keys and the relegendables have the same stems, which is good news for those who want to raise the numerics by one row to get in another row of alphas.

The result is an alpha area that looks like Austin Powers' teeth – visibly out of alignment. In particular, the ' , . C keys (which jump from row 4 to row 2) look particularly bad. They don't FEEL particularly bad though. The plastic is not sufficiently thick to allow them to be carved down, or I'd consider it. Keys moved from Row 2 to Row 4 don't stand out nearly as much. Since the ' , . keys are all consecutive, they sort of look like a set, meaning it is only the C key that really stands out as snaggle-toothed.

The bottom row mods (including `~, which they kicked out of the upper left) have stem locks that don't match ANYTHING ELSE. You can freely move the mods around amongst themselves, but they are not otherwise interchangeable. You would have to remove BOTH locks to make it work, and then you'd have nothing preventing the key from spinning.

There are some bits of good news. The 2U-high keys DO have two stems under them, and thus can be re-purposed as 1U keys. The relegendable keys work quite nicely and really don't feel that much different from the ordinary keys despite looking quite different. If you WANT keys that feel different, just leave the clear plastic top off. This is what I'm doing with the non-arrow cursor keys to make them easier to find by touch. (They're pre-printed AND relegendable.)

The only key on the entire keyboard with a stabilizer is the spacebar. None of the others are really big enough to justify it, especially since the two 2U keys have two stems (the lower of which just doesn't go all the way to the bottom).

The magstripe reader transmits over the keyboard interface. I've already confirmed it with some expired debit cards. This could be useful if you can write to such cards, to give individual users complicated login passwords they don't need to remember, instead swiping their company ID. (Of course if they lose the card and it has their name on it... the security implications are obvious.)

The key action itself is a fairly snappy rubber dome. It's still distinctly non-mechanical, but it's better than most rubber dome keyboards I've used, and there is absolutely no play at the top of the keystroke. I think it is fair for Cherry to refer to these as "premium" rubber domes.

The keyboard is also heavy. If you want to use it on your lap, it will stay put. Modders should be aware that it is assembled with Torx screws. It does not have a choice of angle adjustments, it is always effectively "feet down" with permanent stems about 1 cm long in the back. They could be removed without opening or altering the keyboard, however. [EDIT: It has no such feet out of the box. The previous owner glued them onto this one.]

Other notes: The framed squares that hold the inserts for the relegendable keys are ever so slightly more than 1/2 inch (12.7 mm) on a side. If you make them exactly a half inch, they won't rattle around significantly but you also won't have to be insanely precise with your cutting. (This is almost certainly by design.)

Verdict: If you want a keyboard with programmable keys, but don't mind leaving the alpha area in standard ANSI QWERTY, this will work quite well. If you want any other layout for the alpha area, it's going to be a bit of a mess and require permanent alteration of the keys (though this alteration has no ill effects should you choose to put the keys back). It is not a suitable substitute for a mechanical keyboard, although it's actually pretty decent as rubber domes go.

Attached are pictures of the internals, and of the keyboard as it currently stands. Sorry if they don't load, Binhoster is known for going tits-up for days at a time.


Note the way the different stems are keyed differently. You can even see the stems for the rightmost modifier keys, for which Cherry rotated the entire stem about 8 degrees clockwise. (Why, Cherry, WHY?)


Note the way the double-height keys are stemmed. Only the upper one actually activates a rubber dome. The white stuff is just grease, and it doesn't smell like food so I'm assuming it's original.




Upon closer inspection, the legs seem to be glued on. They probably aren't original, but I like them and will be leaving them as they are. The four holes at the bottom appear to be drainage channels.

I found a tool capable of handling the Torx screws, although I still can't find the one with the interchangeable bits that I know is around here somewhere. Instead, I found a triangular file that had the tip broken off, and the end of it just happened to fit perfectly. None of the screws were very tight, before or after the procedure. They probably don't need to be, because there are a LOT of them (16 for the case, 6 for the PCB).


The steel backing plate. This is only held in place by the plastic back, and is unquestionably the reason the whole unit is heavy.



It's a typical membrane keyboard controller, as far as the way it attaches. The small but rather loud speaker is used to confirm that magstripe swipes have been read.


I was a little surprised to see that the membranes use insulated, crossing layers, but this really does clean up the overall layout compared to most membranes that run gigantic loops around the outside to avoid crossing each other. I'd venture a guess that the resistance is a fair bit lower than the average membrane keyboard because of it.


The extra cable to the controller is for the magstripe reader. I did not bother cleaning the grease off of the rubber dome sheet, as it doesn't seem to be doing any harm.


Most of that grease stays on the key side anyhow. It would be nice if it would stay in the stems more rather than down at the base, but such is life. A few of the keys in the top three rows (not the main typing section) are a bit scratchy.

Permission is granted to use any pictures or other details you wish for the Wiki, since you do have an entry for the G86 series.

Do I like the 18U wide x 8U deep format? I'd like it better if it was 15.5U wide (put the ~` key back where it belongs and eliminate the three columns on the right), but it would be rather self-defeating to market a POS keyboard with no number pad. I'm outside the expected market, so there is little reason for them to cater to my desires. Otherwise, yes, I think putting extra keys up top is a good compromise.

At $40 it was a good deal. Having gotten my hands on it, I wouldn't pay the $120+ asked for one of these new, but that's not because it's a bad keyboard. It's just that at that price point, there are a fair number of mechanical keyboards available.

I suggest you throw this thing away and get a real keyboard!


A Tipro might be right for you.


Example:

7bit wrote: ↑I suggest you throw this thing away and get a real keyboard!


A Tipro might be right for you.

I looked. Tipro keyboards are not popular in the United States (meaning they're not showing up on eBay), and the only ones I did find were all PS/2 interface, which I would be unable to program. They're also relatively bulky, and I wanted something I could easily take with me on the occasional service call — I likely will grab the seven-row version of this for $20 and leave off the special character keys I would never use on someone else's machine. I'm also getting increasingly picky about the layout and don't want the ½U stagger between the Q-row and the A-row (limiting my options to the model with the ¼U stagger there), nor do I want the extra key between Shift and Z. Contoured key caps are also a problem, especially as the contour becomes more severe than that of the G86. The snaggletoothed result is just tolerable when I move keys around now. It would not be on a more severely contoured set of keys. Thus, I'd have to replace the keycaps from the start, adding even more to the expense.

Maybe for the next one, I'm certainly not ruling it out. But not for now.

More updates.

I decided to take a chance and snip the anti-rotation locks off of the 2U high keys entirely, in the hopes that I could then use them as 2U wide instead. It works. They lean slightly downward on the side with the fake stem, but they're functional. Wrapping the fake stem in one layer of electrical tape has reduced the wobble to levels similar to when they were in vertical orientation with their locks still present.

Here's the revised layout that I was only able to do because of this (unless I wanted to source four single keys, which would still be preferable if I can find them):



If I can get two single keys, I'll split Esc (the red DON'T PANIC key) and use the right half for Insert. Splitting the + key (if I can find four single keys) might let me sneak in a Tab, since Excel is where I use the numpad the most.

I was looking around for things that might be the same size as the stem of a Cherry G86 key, since I have a bunch of keys left over from when I chopped a cheap keyboard down to just its cursor and numberpad sections. I found a fair number of things that might suit, but couldn't figure out how I was going to stop them from rotating or pulling out of the keyboard. Then I got the bright idea of putting a crimp in a piece cut from a straw to make a small "key", and went looking for straws. That's how I figured out that some common straws (not McDonalds straws, but ones from other fast food places) fit perfectly and are adequately stiff all by themselves.

To make the "key", I first cut off a piece of the straw, perhaps 7 mm long, and split it open. I folded it in half and pinched it in pliers, then creased it sharply where it exited said pliers. To keep it from unfolding, I wiped the inside with a bit of hot glue while still pinching it in the pliers. This had the side benefit of giving the piece the ability to hug the inside of the tube it was going to fill.

I cut a vertical notch in the end of the straw, and inserted the "key" section, and secured them with a bit more hot glue. Then I carefully rolled the end of the tube back with direct flame. This provides the latching mechanism that prevents the keys from falling out. I checked that this is neither too little nor too much before cutting the tube.

Next, I had to remove the stem from the donor key. This part was no fun, but I was able to leave some of it in place to slip down the straw. This helped a little, though on the keys where it broke off, it doesn't seem to affect their stability.

I cut the tube a little bit over-long compared to an original key, and slipped it into the guide. Then I test-fitted the sacrificial keytop. With a few flicks of the lighter, I was able to select both the level of the keytop, and the angle at which it sits. This went so quickly that it wasn't a hazard to the adjacent keys. Once I was satisfied, I trickled a small pile of hot glue into the underside of the keytop, and plopped it onto the stem. Then I just had to hold it until it became self-stable, and ignore it for a minute or two after that to let the glue truly set up.

The result looks a little something like this:



It's not the prettiest, but it's all completely invisible once the keyboard is assembled. I deliberately left the stem on the Escape key longer than the others to make it easier to find, and inadvertently caused the top of it to crater a little bit in the very center because the glue was excessively hot, but I'm covering it over with a label anyhow. The end result goes something like this:



The donor keys are pad printed and clear coated and the legends are tiny but bright. They're a bit of a mismatch, but it's tolerable given the relegendable nature of so many of the other keys.

I have also figured out that these original keys that feel so nice despite being rubber dome are probably PBT. They certainly seem quite resistant to the flame from a butane lighter, and although I received this keyboard used, there was absolutely no shine on any of the keys. At first I took this to be a sign that the board was lightly used, but given the amount of crud I found inside when I disassembled it, I no longer think this is true. They just don't wear the way ABS does. They also don't stick very strongly to nail polish, and are completely unaffected by cleaning with acetone. Neither is true of ABS -- nail polish might as well be glue for how well it adheres to ABS, and acetone cleaning will leave it quite noticeably marred.

The donor keys are ABS, and will end up wearing down, but it's not like I press any of them frequently (except Escape, which has the "DON'T PANIC" sticker over it anyhow). Even Print Screen, the one I use most frequently of the bunch, has little likelihood of taking on a shine or having its legend wear off. If it does, I'll just slap a label on that too and paint it over in clear nail polish.

The keys also just feel different, but this is undoubtedly due to a combination of factors I can't quite separate. First, the Cherry keys have spherical contours on top, while the donor keys are cylindrical. The ABS keys have a much coarser pebbling, while it's fairly subtle on the Cherry keys. The presumed PBT keys feel almost as if they are slightly oiled. It's slightly strange, but not at all unpleasant. The ABS keys just feel like plastic.

So add that to the plus column for this keyboard. The original keys are quite high quality, which is good because the only second source is SP, and they don't do the relegendable keys (I already asked). I may at some point commission a proper set of contoured Dvorak keys, but not if they'd be a step down in feel. It's also quite nice that the board can easily be reconfigured for Dvorak (or any other physically compatible layout) at the hardware level, making it a good option for people who need to plug-and-play on systems not their own, or boot from CD or other write-protected media they can't reconfigure.

The keycaps most probably are POM. Cherrys black pad-printed MX compatible keycaps are POM and are also quite wear-resistant and feel quite slippery and oily.

Nuum wrote: ↑The keycaps most probably are POM. Cherrys black pad-printed MX compatible keycaps are POM and are also quite wear-resistant and feel quite slippery and oily.

You are probably right. It would make sense for them to be made similarly to the other lines, and they almost certainly are pad printed, though it's a much higher quality of print than that on the donor keys taken from a Logitech Deluxe 250. The print is not raised at all [EDIT: It actually is raised ever so slightly, but not clear-coated], and I thought they were double shots when I first looked at them. Flipping one over quickly revealed that they are not. They're black, so they can't be dye sub. The legends, while not glaringly bright white, are still pretty high contrast, so they wouldn't be lasered either. If they're not pad printed, etch and fill is about the only other possibility I can think of.

Now if only I could understand why they offset the locks on the bottom row. I really don't like having `~ down there, but it won't fit anywhere else (unless I remove the anti-rotation locks entirely), and I don't like labeling over keys unless it's absolutely necessary.

Thanks for the heads-up on the material. Now I know what to look for if I want to order up a set, though I think PBT would probably feel just as good (different, but not inferior).

I know most people here aren't that interested because it's a rubber dome, but there is some overlap between mechanical keyboard fans and alternate layout users. Now I'm in the uncomfortable position of having raised my standards when it comes to extra keys and programmability, such that I think my only mechanical keyboard options are Tipro (which never caught on in the U.S.) and the Cherry LPOS boards (which don't have enough extra keys and are too wide for me). I like the G86 feel, and although its programming features are only moderate (I'm doing the vast majority of the magic in the keyboard layout on the OS side), they're far better than nothing.

I'm sure I'll still jump at the chance to grab an affordable USB Tipro, should the opportunity arise, but PS/2 is pretty much off the table because of the requirement of being programmed in a 32-bit Windows OS. Program once and use forever? Yeah, right... I'm still changing up the G86 layout (not the alphas, but the parts that were meant to be programmed by design) on a nearly daily basis. This is pretty simple now. It wouldn't be if I had to fire up a different OS to do it.

EDIT: POM is Delrin, huh? Yeah, these DO feel a lot like Delrin...

I modified the `~ to get it out of the bottom row (couldn't stand it there). To do so, I removed both locking tabs, then cut a slit into the stem just like the ones I cut into the straws for the donor keys, and inserted a folded piece of straw (again just like the donor keys).
(It didn't work.)

In my case, I chose to re-stem my Print Screen key and stick it in the spot vacated by `~ simply by breaking the glue, cleaning out the key, and gluing it back on. Since I attach the key with the stem already in place, the exact location of the locking tab is not particularly important.

Last, I took the clear tops off F1 to F4, and F9 to F12. This is to make them easy to identify by touch. The raised rims on the relegendable keys (which are there to keep the paper centered) are quite noticeable, even more than the height difference is. I miss the gutters between F4/F5 and F8/F9, and this serves as an alternate method of finding my groups of four by touch. I have also gone back to a normal relegendable key for "DON'T PANIC", the relabeled ABS key was too ugly for me. This didn't make it into the picture, but since it's not an actual remapping, I didn't think it was necessary.

Mal-2 wrote: ↑I also believe that if someone wished to put relegendable keys in the bottom row, they could. Snip off the TOP locking tab instead of the bottom, and rotate the entire key 180 degrees. This then would leave them with a bottom row key which would have to be modified as above.

I tried it. It doesn't work. Fortunately I am still able to use the relegendable key on most of the keyboard even though the upper (straight) rotation lock is missing. I just thought I'd add this for anyone reading this as a reference, possibly years from now.

Addition: Colemak works better than Dvorak on this keyboard. No key moves more than one row, so the difference in contour is acceptable.

Good news everybody!

I have decided to attempt the 3D printing of stems that fit the G86, but with MX compatible mounts on the top.

The bad news? It's my first time trying to do anything in CAD, ever. Good news? I'll be the first guinea pig if it doesn't work, and nobody else will have to deal with it.

More bad news: it's not my 3D printer, and the friend that owns it isn't interested in going commercial. There is still good news, which is that I would open up the plans to everyone — appropriate license to be determined. That's where I want to hear from others. What is a good open hardware license that will ensure that someone else can't tweak it, patent it, and freeze everyone else out?

I can see the basic principle working for any rubber dome keyboard except buckling spring (which would still be possible but require short springs that would probably be bad for feel).

This design will be done in inches because it is obvious that's how it was designed. The stem diameter is exactly ¹⁄₄", which was my first sign. Then the distance from the bottom of the stem to the bottom of the skirt was ³⁄₈". I was starting to sense a pattern. Sure enough, the length of the anti-rotation key is ¹⁄₁₆", and the thickness of both it and the stem walls is ¹⁄₃₂". The design must date back to Cherry's American origins.

More news -- I can't CAD my way out of a wet paper bag. I just proved it by trying. The MX-stem project is dead unless someone wants to translate drawings into a CAD model. (I can provide measurements and all that.)

Also, I've moved to a G86-63401, the matrix model with trackpad. I like the matrix part. I wish it didn't have the trackpad (I'd much rather have 12 more keys, and it's a low-res trackpad), and I found it impossible to touch-type on the relegendable keys because they're all so broad as to feel like one solid wall of keys. It's almost as if there's no space between them. I had to steal the "normal" keys off the 61400, which means it in turn is reduced to being less than useful. (I'll probably end up buying a 62410 just to steal the keys from it, so I can turn around and sell the 61400 which is worth quite a bit more.)

So what's this new hybrid board look like? I thought you'd never ask. Oh wait, you didn't. I'm going to show you anyhow.



I tried using 2U keys as a "spacebar", but they have a tendency to get stuck down when pressed in the wrong spot. This never happens with the four 1U keys, and even if I press more than one at a time, I guess the de-bouncing algorithm perceives them as part of the same event because I don't get two spaces.

If, by chance, someone actually wants a G86-61400 that is filled with nothing but relegendable keys (except for the non-1U keys, and all but one of the bottom row modifiers), I'd be happy to post pictures. It works just fine, it just doesn't feel right under my fingers.

Update: De-bouncing must occur at the application level, because while Word and Firefox see just one press of Space when I hit two buttons, Encore and Cubase detect both. Both use the spacebar to start and stop playback, and if I catch two buttons, both programs respond by starting and instantly stopping playback (or conversely, stopping and instantly re-starting).
The 2U buttons work alright after all, if I use them upside down ("real" stem on the bottom). My thumbs never hit the spots that make them jam. I've moved the typing area as far to the right as possible, and placed the keypad on the left. The object between the keypad and alpha area is just a fragment of a membership card, and it's just floating in there to make it a bit easier to see where my hands go. I can move it to lie between the YIX and FDB columns when I catch myself crossing the mid-line with my hands and want to remind myself to stop.

The "rabbit teeth" space buttons work reasonably well, but occasionally I find myself typinglikethis because I'm tapping AltGr instead of space, and not realizing it. It happened a lot at first, but not so much any more. I still need a left thumb Space for WASD-style keying in games, that's why it hasn't been reassigned to Backspace or anything. The downside of the rabbit teeth is that they need 3U of space around each of them, one each in the outside, downward, and diagonal directions. If I were able to lower the alpha area by one row (if I had no trackpad), I wouldn't have to sacrifice 4U of space below them, only 1U to each side.

The trackpad itself I finally disabled in Device Manager. I was inadvertently clicking the buttons or tapping the trackpad, moving my cursor at inopportune moments. It also has a quite poor "DPI" or whatever the trackpad equivalent is making it not very worthwhile anyhow. It's probably adequate for the target market (point of sale systems) but it's next to useless if a real mouse is present. If offered the choice, save a few bucks AND gain twelve keys: buy the non-trackpad version.

Also, on a Skype conference call today, I was asked what kind of mechanical keyboard I have because everyone could hear me typing. The microphone base was actually in contact with the keyboard so I figured it might be direct transmission and moved the mic stand. It didn't help. So apparently, according to my cow-orkers, my keyboard is mechanical-grade loud. I was typing in Los Angeles, and they could hear me in Estonia and Sweden. It's not Buckling Spring loud, or even MX Blues loud, but it's pretty "thocky".

Another small update: POM confirmed as the material for the keys. It's printed in raised letters on the underside of the 2U keys. Also on exceptionally close inspection, the pad printed nature of the keys is confirmed. The legends sit infinitesimally higher than the surface of the key body.

Using the method I just described here, I have determined the activation force of the G86-63401 to be somewhere upward of 55g, but probably less than 60g (with the normal alpha keys, not relegendables which might weigh slightly more or less). The 61410 is slightly less snappy, but that could be due to use (the 63401 was purchased new).