ErgoDox - Custom split ergo keyboard.

[ped]

No. Only the right side may be used independently of the other, or at least that is the way it is being designed right now. For the left side to be active, it must be connected to the right side via TRRS cable. The reason for this is that it was decided early on by Dox, I think, that only one controller would be used, which means that one of the halves had to be chosen as the 'host' half, so to speak.

[OrangeJewce]

In that case, I would like to register my interest as well!
Speaking of, which time frame are we looking at at this point? A couple of months, a year ... ?

I wouldn't be able to afford in the next two or three months, but later in the year, I'd most likely have the finances to cover it.

I've been thinking, not sure whether this has been discussed previously, as I haven't followed the thread the whole way through, will we be able to use either side of the board independently of the other? I could see myself / others wanting to use it as a gaming pad, without having to necessarily have both sides on the desk.

Dox had suggested that the case prototype should be arriving shortly. I am not sure if there will be a second round of PCB prototypes (move thumb keys), after that there isn't really a huge amount of work left from the hardware side. I would say it is looking like it will be in the next 2-3 months as opposed to 6-8. Maybe Dox can chime in here.

Cheers,

[Zifle]

No. Only the right side may be used independently of the other, or at least that is the way it is being designed right now. For the left side to be active, it must be connected to the right side via TRRS cable. The reason for this is that it was decided early on by Dox, I think, that only one controller would be used, which means that one of the halves had to be chosen as the 'host' half, so to speak.

That makes sense, although I'm slightly sad to hear the outcome, given my previous reasoning. Regardless, that alone should not be a reason to pass on something like this ^^

Dox had suggested that the case prototype should be arriving shortly. I am not sure if there will be a second round of PCB prototypes (move thumb keys), after that there isn't really a huge amount of work left from the hardware side. I would say it is looking like it will be in the next 2-3 months as opposed to 6-8. Maybe Dox can chime in here.

Cheers,

Sounds good, looking forward to seeing upcoming pics of a more finished device!

[alaricljs]

Don't forget that with sufficient desk space and a properly sized cable you just put the left half where you want it and tuck the right half out of the way (or vice versa).  If you choose to do your own case you could even devise a scheme to stand that half on edge so even less desk space is needed.

[bpiphany]

I think would be possible to make both halves identical using the same PCB for both the left and right half. It looks to me like there would be enough room to fit both the Teensy as well as the other IC there on top. If the PCB was done this way it would be possible to choose to use any or both halves, and also which half the USB cable would be running to.

The controller parts from the Teensy can be bought elsewhere and surface mounted, this saves some space. The ATmega chips can be run on their internal 8MHz oscillator as well, cutting the number of required components down. I've been experimenting successfully with this today =) This requires reprogramming some fuse bits though, which in turn requires reloading the bootloader, both which requires a programmer unit... And the stock bootloader as well as the LUFA one are both 4kB.

[OrangeJewce]

I think would be possible to make both halves identical using the same PCB for both the left and right half. It looks to me like there would be enough room to fit both the Teensy as well as the other IC there on top. If the PCB was done this way it would be possible to choose to use any or both halves, and also which half the USB cable would be running to.

The controller parts from the Teensy can be bought elsewhere and surface mounted, this saves some space. The ATmega chips can be run on their internal 8MHz oscillator as well, cutting the number of required components down. I've been experimenting successfully with this today =) This requires reprogramming some fuse bits though, which in turn requires reloading the bootloader, both which requires a programmer unit... And the stock bootloader as well as the LUFA one are both 4kB.

This topic was discussed a long time ago and it was determined that only one controller was going to be used. But everyone is more than welcome to customize their Dox however they Feel! (I think your mod would be sick, if you ever undertake it as a project :thumb:)

Cheers,

[bpiphany]

This would really not change anything (other than the PCB layout). There could be separate USB controllers on either side, or one controller and a slave side (either side being the slave/master), or a single PCB could be used as a one handed keyboard of left or right hand type. This would also double the number of same design PCBs which may or may not have an impact on price. If there are huge amounts already this doesn't make too big a difference. The main problem is the possible need to use a serial programmer to set up the controller chips. This would likely have to be done by someone from here in bulk before shipping them out to people.

I'm of course more than willing to help out with what I know about chip programming, bootloaders, PCB design and all.

[dorkvader]

This would really not change anything (other than the PCB layout). There could be separate USB controllers on either side, or one controller and a slave side (either side being the slave/master), or a single PCB could be used as a one handed keyboard of left or right hand type. This would also double the number of same design PCBs which may or may not have an impact on price. If there are huge amounts already this doesn't make too big a difference. The main problem is the possible need to use a serial programmer to set up the controller chips. This would likely have to be done by someone from here in bulk before shipping them out to people.

I'm of course more than willing to help out with what I know about chip programming, bootloaders, PCB design and all.

I may have brought up having identical PCB's. The issue is that it still has to work with it "flipped". You could maybe do this with a lot of settable jumpers (like: Set to "A" for LH and "B" for RH). Also, Trace routing becomes difficult, and the cherry keyswitches are not symmetrical, so you'd need an extra hole per switch.

Then, having the possibility of chosen controller per side was discussed. This idea had much more going for it, as it's actually somewhat feasible. I don't think anyone came up with a decent plan to implement this (I think HW implementation is the best, so the controller doesn't "know" any better. Having a "master and slave" controller setup was rejected as well (Is it even possible with teensy?)

Anyway, if you come up with a good implementation for this, I don't see any reason to not include the capability.
---

So other than ugly shapeways plastic, has anyone come up with a good place to make the case?

[bpiphany]

I don't think you need to do almost anything at all to get it to work flipped. But I just woke up and may not have thought this through all the way. Just mount the USB controller on one side and let the slave IC location be empty, And vise versa on the other side.

The switch is unidirectional by construction. They would be the only components needing to switch side. Through hole diodes are of course "flip symmetric" as as well (Internal diodes could still be used if they are mounted "backwards").

The flippable footprint should be no problem. Internal diode and LED holes also line up when it is flipped.

[ Attachment Invalid Or Does Not Exist ] 52840[/ATTACH]

[REVENGE]

Having the ability to build either side as the "main" board would be fantastic.

[uberben]

I don't think you need to do almost anything at all to get it to work flipped. But I just woke up and may not have thought this through all the way. Just mount the USB controller on one side and let the slave IC location be empty, And vise versa on the other side.

The switch is unidirectional by construction. They would be the only components needing to switch side. Through hole diodes are of course "flip symmetric" as as well (Internal diodes could still be used if they are mounted "backwards").

The flippable footprint should be no problem. Internal diode and LED holes also line up when it is flipped.

(Attachment Link) 52840[/ATTACH]

This sounds doable to me. It has a lot of advantages, but will require much more careful layout.

[Salisen]

You could have a 10x10 matrix for the switches, perhaps, and seperate it into two 5x10 halves.  Divert one of the matrices through the connector (would have to be 16 way) to the other side.

From each switch on the board, have traces running to the IC, and to the connector, total of 4 traces for each.

For each pin on the connector, have one trace running to the IC, and one running to the switches through an array of jumpers (you'd need 16 total, I think).

For the slave board, the jumpers would be present and an IC would not, so all IC connections would be redundant, leaving only the traces to the connector.

For the master, the jumpers would be left empty, traces would run from the IC to the set of switches on that board, and from the IC to the connector.

I don't think you'd have the option of putting the IC on either board though.  On one of the two it would have to be underneath as opposed to on top, or the pins would be the wrong. The Teensy is quite tall, so you'd make the keypad unnecessarily thick.

[bpiphany]

I was thinking of using SMT components to rebuild the "Teensy". And I am pretty confident the slave IC comes in a smaller SMT package as well.. I know DOX wrote the part number somewhere but I can't seem to find it again =P

You would only need to do a regular key matrix and trace rows and columns to both the serial slave IC as well as the USB controller. Then choose which side to solder the controller/slave IC to. The traces to the other would only end up unconnected. It will look funny that the traces from the slave IC will lead to the connector, and from there down to the controller again =D There should be no need for jumpers anywhere, I think when trying to figure this out in my head...

[OrangeJewce]

While I think the discussion about multi-controller and symmetrical PCBs is fantastic, and these ideas sound awesome, short of Dox completely renegging on this major design decision all of these proposals are going to stay in the realm of vaporware. I just don't see him having invested in several PCB prototypes just to do an overhaul on the PCB design, and because work has begun on the case prototype, it wouldn't make sense to push the project back further. TBH I really want to see the final product (pics!), not to mention have it in my hands in 3 months as opposed to another year for a redesign.

Cheers,

[bpiphany]

Quit being so negative.. geesh, and by the way

Otherwise, everything went well. With the thumb keys a bit closer, it will be perfect!

Looks like there may be a PCB redesign whether you like it or not.. It is pretty quick by the way. And I do think DOX had meant to do more work on the PCB as it was. Those prototypes don't look like a finished product at all...

[OrangeJewce]

Quit being so negative.. geesh, and by the way

Looks like there may be a PCB redesign whether you like it or not.. It is pretty quick by the way. And I do think DOX had meant to do more work on the PCB as it was. Those prototypes don't look like a finished product at all...

Firstly, let's not make this personal. I wasn't attacking you, or your ideas. I was simply stating that the decision regarding these most recent proposals had already been made a long time ago, and that bringing them back from the dead had very little chance to change anything. I'm not going to argue about what Dox may or may not do in the future, because I am not him, nor can I control what he does. That being said and based on what I read earlier, when Dox makes a decision he tends to stand by it. You're welcome to critique me as much as you want, but in this case I feel pretty confident in saying that neither multiple controllers, in any configuration, nor flippable PCBs are going to make it into the feature set of the ErgoDox hardware.

Secondly, shortening some traces on a PCB does not count as a total redesign. In fact, Dox was going to do the mod himself on the PCBs by simply cutting the thumb section and doing a little bit of hand-wiring...

I think that the thumb keys are a bit too far out. The 3 furthest keys are hard to reach and my thumb is a bit too stretched in the standard (thumb on space) position.
I will put them closer for the case prototype. I will cut the thumb from the PCB and hand wire it to test it.

I implore you to start seeing what is there, as opposed to what your mind wants you to see.


Until the man himself intervenes here, I'm done discussing hardware. Firmware and UI, I'm all ears, because very few decisions have been made. ic07 and I have expressed that we are open to new ideas and features for it, and thus encourage discussion along those lines.

Cheers,

[dorkvader]

Could be some excellent ideas for the ergoDOX V2

With the flippable PCB, how does it work out what with that extra chip? Also, one could have two locations for the Teensy, depending on what hand/side you have, you could use different holes across the top.

[bpiphany]

I PMed DOX to ask if I could have a look at the schematics. I'll wait and see what he says. I don't want to interfere with his work if he doesn't want me to. But really, I think a flippable PCB is quite simple to put together. That being said without me having given it a decent try.

The Teensy can be mounted on either side really. Since it is through hole it only needs to be mounted upside down if changing side. The only slight annoyance this comes with is reaching the reset button on it. That is easily solved with a hole in the PCB. This I have done before =) Bending the legs backwards on that other IC is probably not a good idea though...

[ic07]

Don't forget that with sufficient desk space and a properly sized cable you just put the left half where you want it and tuck the right half out of the way (or vice versa).  If you choose to do your own case you could even devise a scheme to stand that half on edge so even less desk space is needed.

Or for an extra ~$25 and a little work, you could put a Teensy and a TRRS connector in an altoids can, and use that for the right hand side .

[dorkvader]

Or for an extra ~$25 and a little work, you could put a Teensy and a TRRS connector in an altoids can, and use that for the right hand side .

Whoah, that's brilliant, I hadn't considered that.

Now I want like 5 more of those teensy's.

[mSSM]

As regards the switches:
7bit is currently doing a group buy of cherry switches (both plate and PCB mounted) over at deskthority:
http://deskthority.net/marketplace-f11/cherry-mx-taking-pre-orders-t2760.html

Just thought it's worth mentioning. Of course getting in on that group buy depends on the time frame we are looking at as regards the final production of the dox. :-)


How many keys we are actually going to have?


EDIT: Silly me, he has a thread here at Geekhack as well, in the group buys section.

[dorkvader]

I count 36 keyswitches per hand from the images DOX posted here:
http://geekhack.org/showthread.php?22780-Interest-Check-Custom-split-ergo-keyboard&p=602321&viewfull=1#post602321

So that's 72 total, right?

[ic07]

I count 36 keyswitches per hand from the images DOX posted here:
http://geekhack.org/showthread.php?22780-Interest-Check-Custom-split-ergo-keyboard&p=602321&viewfull=1#post602321

So that's 72 total, right?

Pretty sure it's 38 per hand = 76 total.  (I cheated a little though, since I looked at how many were mapped in the firmware before I looked at the picture again to make sure).

[mSSM]

Well, now I am certainly tempted to justs join that group buy. Which switches do you think would fit the ErgoDox best? I am thinking either linear MX Reds, or alternatively clicky MX Blues or MX Whites (I like their sound better).

[OrangeJewce]

Well, not I am certainly tempted to justs join that group buy. Which switches do you think would fit the ErgoDox best? I am thinking either linear MX Reds, or alternatively clicky MX Blues or MX Whites (I like their sound better).

Answer: Get two! One with reds and one with whites!

Cheers,

[ironman31]

If it doesn't end up being too much, I might get two sets of PCBs and one case to just change them out with different switches.

[Surger]

I've just scanned this thread and am most definitely interested in the keyboard that is being developed. Glad to see all the knowhow on geekhack is being put into practice! Add me to the list.

I've been on a quest for a better keyboard for many years and have tried *all* the usual suspects and failures, only to get more dissapointed. I'm currently compromising on a Kinesis Freestyle and concluded that separation is more important to me than angling/tenting, etc. If only it had better quality keys...

While I tried to grasp as much of the current status/development as I could, I may have missed some info, since obviously I couldn't read every single message in this thread. So forgive me if one of my comments has already been dealt with. Maybe it would be an idea to add some more FAQs on the website?

First, I agree with most of the development decisions, layout, functionality, etc. Congrats on your efforts of merging the best aspects of all current mediocre keyboards ;-)

Costs: I consider 300 dollars or more perfectly acceptable. I'm at the keyboard over 10 hours a day. I pay top money for the best ergonomic chair I can find and use high quality monitors. Why save on what I consider the most important aspect of computing: a comfortable input device. Honestly, I'd pay a lot more for any keyboard that met my needs. Quality, durability, flexibility outweigh costs.

Build: Can't really figure out how this will work. Are we supposed to order certain parts ourselves and finish the keyboard? Or will certain community members do this? My technical/engineering knowhow is too limited to do this. I can imagine this will be a mayor obstacle for people who are interested in the final product.

Functionality: While I understand the choice for compactness and minimalistic design, I find the lack of Function keys a serious flaw. Before you jump on me, let me explain. To me, a keyboard geared at heavy PC users should focus on being ergonomic in all aspects, including minimizing the usage of a mouse or pointing device. I use Alt+ Control + Function key combination hundreds of time a day in all sorts of programs and can't imagine using a keyboard that doesn't have these. It would drastically slow me down. (Does everybody seriously use a mouse to close a program when Alt+F4 is within reach?)

Adding these to a different layer or reassigning them doesn't really solve the problem: If I have to switch layers, use a combination, switch back, I'm doubling the amount of keys I have to use. If I'd assign them to the number keys, other shortcuts (such as Control+1 through 4) in Outlook are gone.

I fully understand that this contradicts some of your initial design/functionality choices, but still wanted to share my thoughts, mostly because I'm surprised this hasn't been discussed in more detail. ;-)

Maybe adding a X-keys device (key stick, XK-24) for Function keys... ?
Still interested in the keyboard though.... Hell, I'd buy a few just to keep you guys motivated!

[mSSM]

Functionality: While I understand the choice for compactness and minimalistic design, I find the lack of Function keys a serious flaw. Before you jump on me, let me explain. To me, a keyboard geared at heavy PC users should focus on being ergonomic in all aspects, including minimizing the usage of a mouse or pointing device. I use Alt+ Control + Function key combination hundreds of time a day in all sorts of programs and can't imagine using a keyboard that doesn't have these. It would drastically slow me down. (Does everybody seriously use a mouse to close a program when Alt+F4 is within reach?)

Something like that heavily depends on the operating system you are using, does it not?

For my day to day tasks I mainly use mutt and vim in a terminal multiplexer (tmux), Firefox with the Pentadactyl plugin, and PDF readers such as llpp or mupdf, where all of the mentioned programs have vim-like keybindings. My window manager (Xmonad) also has vim-like keybindings.

I _never_ use the mouse, and I do not even rely heavily on Alt, Control or function keys (CTRL is my most heavily used key).

While I certainly understand your need to cut back on using the mouse, your operating system (Windows of some flavour, I figure), is not really meant for that kind of work. But even so, if you use appropriate programs, you have don't have to have special functions keys.

I don't really understand what more keys you need on that keyboard. What's missing?

[alaricljs]

While I certainly understand your need to cut back on using the mouse, your operating system (Windows of some flavour, I figure), is not really meant for that kind of work. But even so, if you use appropriate programs, you have don't have to have special functions keys.

I don't really understand what more keys you need on that keyboard. What's missing?

So you're saying change his entire workflow and the programs he uses in order to suit the keyboard?  That's a bit silly.  

I'm interested in this project because of the engineering involved, so my opinion is not one coming from a potential user...  That said, I wouldn't buy a keyboard without F-keys.  I've been pondering designing my own board and am sorely tempted to include a 2x5 set on the left (ala Sun type-5) to have more keys.  This way I can have all the f-key combos for the programs I use and then all my own combos for other uses.

[OrangeJewce]

So you're saying change his entire workflow and the programs he uses in order to suit the keyboard?  That's a bit silly.  

I'm interested in this project because of the engineering involved, so my opinion is not one coming from a potential user...  That said, I wouldn't buy a keyboard without F-keys.  I've been pondering designing my own board and am sorely tempted to include a 2x5 set on the left (ala Sun type-5) to have more keys.  This way I can have all the f-key combos for the programs I use and then all my own combos for other uses.

Even as a Windows user and developer, I never use the F'n keys. Maybe I'm just not pro enough for that kind of thing, but if I really needed a certain set, I don't think it's unruly to add a few keystrokes to switch to a different layer which, say, has remapped the numrow to f'n keys, and then when finished jump back to the base layer. I understand your hesitance about not having the F'n keys, and while custom layers may not be the perfect solution, it's a heck of a lot better than nothing. This board, tbh, is probably not the most ideal for a Windows user, but it can definitely work. Still getting 2 if I can!

ic07, YGPM.

Cheers,