Re-Create the DataHand - Thumb cluster under development. Project 75% done.

[Turbinia]

Did a quick mock up of the new case concept I have been working with.

Didn't see the thumb switches up on the git hub so didn't include them (besides wip). The idea is to have 1/8" aluminum plates (6" stock will work) or 3mm acrylic for the base plate and palm plate. those plates sandwich spacers and the outer case walls. Should reduce cost over a ton of 3d printing at only minimal loss of ergonomics. Also anodized aluminum would look pretty slick and professional.

The tilt and tenting plates are optional. The thumb screw knob closest to the finger keys is for tilt and the other is for the tenting (would probably have a second thumb skrew for tilt on the other side, no room really for the tenting one). I did not model the hinges, but it should be clear where they need to be.

The palm rest could be 3d printed out of plastic or perhaps even glazed ceramic. Found they make a black cue ball http://www.amazon.com/Action-Black-Cue-Ball/dp/B009H6ZN1O which looks pretty cool, assuming it works with the laser pointer. Designed the cup for the ball much in the same way as hweller's prototype with the sensor under the palm. There is a cut out in the base plate to let the ball get lower and drain all the crap that inevitably falls in track ball black holes...

The casing between the two plates could be 3d printed, but that would probably be at least $100, so perhaps just print one prototype and make a cast for resin. Might even be appropriate to machine out of delrin or something like that as it is less than 1" thick.

As a design note for the finger PCB there is room to route wires for the thumb and track ball around the finger switch guard and the edge of the case. No need to put the plugs on the south side of the finger switches.

*edited to add the thumb switches. The scale can be estimated by measuring the distance from the center of the ring finger keyswitch to the left edge of the case, this is 52mm.

[OldDataHands]

How _awesome_ is that!

I want that on my desk. Like now.

FWIW: the thumb switch work is all on a branch named "thumb_work"
so that I don't have untested garbage on the master branch.

[gbjk]

I've been thinking about this quite a bit from an incremental development point of view.

Mass-marketing would be as hard now as it was originally. Nothing's changed.
There's still a big learning curve for a datahand.

The first phase, to me, has always been to secure a working replacement for existing datahand users.
That's certainly what's driving olddatahands, and others.

Well the case for the existing datahands isn't horrifc, and we all already have one.

So phase 1 feels like it's a "able to replace finger switches and electronics in existing datahands".
Anyone else looking at this form that perspective?

Phase 2 could be a new case. And lots of people want to go straight with a second pointing device, too.
( I think it'd be good. But right now I have a trackpad between the datahands, and it serves me well enough )

However I really don't view wireless + battery as that unattainable.
And if *is* attainable, then it would have an impact on the casing design, right?

Bluetooth 4 obviously lowers the target for how much battery power you'd need,
and arduino's popularity means there's already chips kicking around we can use.

I have a Myo here. If that's able to charge for a short while, and then work for most of a day,
then I think a datahand should be able to too.

Don't know if I've just gone properly pie-in-the-sky with my thinking, but I really feel like
wireless + battery powered datahands in existing case could be a viable phase 2.

And then phase 3 becomes a new package that makes it all smaller and more accessible,
also potentially marketable as a new product.

Just thought I'd share that.

I have a bluetooth 4 chip coming, and I'll run some tests off an 32u4 and a li-on battery.
I have an eye on http://proto-pic.co.uk/polymer-lithium-ion-battery-400mah/ as a first thought.

[hweller]

Found they make a black cue ball http://www.amazon.com/Action-Black-Cue-Ball/dp/B009H6ZN1O which looks pretty cool, assuming it works with the laser pointer.

Excellent case design!  I look forward to putting together a complete DodoHand.

On the issue of ball choice I tried several and found that the more features there are on the surface the better the laser is able to track it.  So old scratched balls work better than brand new clean and shiny ones.  After playing for a while and trying to get a response from Aramith about getting balls with a metalic finish as used in the slimblade I settled for a "golden 8" http://www.aramithpoolballs.com/bbgold8.html which works REALLY well.  It is possible that the problems I experienced with new shiny balls relates to the inaccuracy of the positioning  of the laser relative to the ball (I used a belt-sander on the cup) and if the cup and laser sensor mount were manufactured more accurately e.g. by 3D printing it would work better.  However, using a "golden 8" ball is the safe and reliable readily available option.

[vvp]

Isn't the black number 8 problem when it gets aligned with the sensor?

[hweller]

Isn't the black number 8 problem when it gets aligned with the sensor?

I expected that it might but so far I have not noticed a problem.  It would certainly be better if we could get hold of 2 1/4" balls with the fine metallic finish used in the Slimblade balls.

[dorkvader]

However I really don't view wireless + battery as that unattainable.
And if *is* attainable, then it would have an impact on the casing design, right?

Bluetooth 4 obviously lowers the target for how much battery power you'd need,
and arduino's popularity means there's already chips kicking around we can use.

There is probably room in the back of the case for a few AAA batteries. If you are able to send HID over BT4.0 it'll be pretty awesome for the KB community as a whole, and would also maybe work out for the DH project as well.

The next issue you'll have is wireless communication between the halves. In most OSs I think you can just set them up as separate keyboards and it'll still work with split shifts and things (so you press shift on one and then type on the other it'll work) but maybe not. Some testing needs to be done.

Maybe you can set up the BT chip as a "master" and a second BT chip on the other hand as a "slave" ? I don't know as much about the BT protocol to see if this is possible but it's an idea.

Isn't the black number 8 problem when it gets aligned with the sensor?

I haven't had any issues with the pool balls I've used in my CST trackball (black or otherwise) I imagine the adns9800 that everyone got from mr kicklighter won't either.

[OldDataHands]

A few cut traces, and a few jumpers later, and the
double-action thumb switch has been proven to work!
This makes it 6/6 for the thumb switches functioning
as switches! Woot!

I've already corrected the PCB layout for the thumb
cluster so that I won't have to do that again.

I've also realized that my footprint for a PNP transistor
on the finger PCB which was to support evaluation of
the EasyPoint switch is wrong. I hope that this is what
is preventing Col4 from registering any switches on
the right side where I populated Q1. The thumb cluster
is working completely on the left side though, which is
great.

One of the details which needs to be corrected in the
thumb side switches bit me today: The magnets are
too loose, and one jumped out all the way. This will
have to be addressed in the re-design for the stronger
magnet. It was an irritating fight with a magnet that wants
to stick to everything to get it back into its hole with all
the parts already mounted on the board.

[Wildcard]

Found they make a black cue ball http://www.amazon.com/Action-Black-Cue-Ball/dp/B009H6ZN1O which looks pretty cool, assuming it works with the laser pointer.

Excellent case design!  I look forward to putting together a complete DodoHand.

On the issue of ball choice I tried several and found that the more features there are on the surface the better the laser is able to track it.  So old scratched balls work better than brand new clean and shiny ones.  After playing for a while and trying to get a response from Aramith about getting balls with a metalic finish as used in the slimblade I settled for a "golden 8" http://www.aramithpoolballs.com/bbgold8.html which works REALLY well.  It is possible that the problems I experienced with new shiny balls relates to the inaccuracy of the positioning  of the laser relative to the ball (I used a belt-sander on the cup) and if the cup and laser sensor mount were manufactured more accurately e.g. by 3D printing it would work better.  However, using a "golden 8" ball is the safe and reliable readily available option.

Wow good find. I've personally been using solid black cue balls in my CSTs which have all worked well. I'm not sure if its been mentioned yet but I strongly recommend using mini bearings similar to what's used in mouse-trak.

[hweller]

I'm not sure if its been mentioned yet but I strongly recommend using mini bearings similar to what's used in mouse-trak.

I am using 2mm ruby balls as used in the Slimblade.  I have an old Marconi trackball which uses ball races and steel axles but I am not happy with the slightly biased drag and prefer the completely regular behavior of the trackballs with ruby bearings.  I have read many reviews of the CST LaserTrac and it certainly looks interesting and I considered purchasing one to compare with the Slimblade but it is rather expensive to buy just for testing.  Do you have access to a Slimblade?  If so could you comment on the difference in motion of the ball, drag, smoothness etc.?

[dorkvader]

I'm not sure if its been mentioned yet but I strongly recommend using mini bearings similar to what's used in mouse-trak.

I would opt for ball transfer units. I tried them oat keycon once and they were amazing. almost no friction at all.

But, barring that, then I strongly recommend recommend ball bearings like in the mouse-trak, traxsys, assimilation and older kensington trackballs over the greased POM bearings that CST uses.

having used the slimblade it is a compromise. A ball bearing will roll really well in one direction but be a little more frictiony in the other. Overall I much prefer ball bearings but ball transfer units are better in every way (more uniform like the slimblade (which isn't a bearing at all), less friction than ball bearings).

[Turbinia]

Any recommendations as to sourcing those components? The smallest ball transfer unit I could find was 8mm, in a 16mm diameter package on Ebay. Going to be a tight fit in there.

Also curious about the smaller package for the ADNS-9800, any idea as to those dimensions?

The image should illustrate the fit issues, grid is 1cm square. In thumb cluster I got rid of the screw hole in the rear of the switches and moved the pivot 2.5mm inboard. Not sure if moving the pivot is possible, but that screw hole shouldn't be needed. Also rounded the backs of the keys instead of leaving them square, lets the case be smoother and tighter to the keys.



There are some parts of the shape of the thumb cluster that don't make sense to me from an appearance point of view. I know you are replicating the original, but some of the gaps and spacing in my mind should be tweaked. Like the slide switch has the indent that makes the gap for no reason that I can see. Also why are the upright switches in two parts, seems like a waste of material and another point of failure. From a travel and durability point of view that vertical switch is really unprotected, could possibly have a switch like the others there. also the switches are 2cm tall, could they be 1cm? Not having a datahand myself it is hard to say.

Hopefully for the next round of prototypes I can have a case design ready to play with.

[OldDataHands]

Hi Turbinia,

I put that third screw into the model to help handle
accidental loads. It wouldn't normally do anything...
only coming into play if the switch was bumped in
the opposite direction of normal actuation.

While I do want to preserve that, I think the screw
could be brought in, and on one side without losing
that benefit completely. Might get the same effect
by moving the two screws to a more central position
so that they can handle loads from both sides, and
then do away with the third...

One of the things that certainly needs to be done is
to reduce the mass of the side thumb keys. your
solution is very nice in thinning them out.

They're currently two parts each just to make them fit
into a smaller volume for 3D printing. I agree that there
are benefits to making the one piece, and probably, for
my purposes, I'll have a final set printed that way.

The thumb side switche "keycaps" all do need to be
made shorter, perhaps by 4mm or so, and I intend to
try and reduce their travel at least a little. This conflicts
a bit with moving the pivot inward in the current design,
as the taller and longer I make the base of these
switches, the less travel I can get away with where the
thumb contacts the "keycap".

I think that there is an opportunity to move the thumb
pad switch body further towards the tip, reducing the
length of the "keycap" in order to make more room for
the ball. I only seem to touch this "keycap" with the
very tip of my thumb anyhow.

It might also be a decent compromise to take the in/up
switch and move it so that it comes in from the tip of the
thumb. It would be a little bit different motion, but would
make the integration with the trackball much less of a
fit problem. It would require a bit of a change to the
keycap shape, but that's the easy part.

The funny gap towards the tip is just cosmetic laziness
on my part. That isn't representative of the original.

[dorkvader]

Any recommendations as to sourcing those components? The smallest ball transfer unit I could find was 8mm, in a 16mm diameter package on Ebay. Going to be a tight fit in there.

The one I got to play with at keycon has very small units, the balls were probably under 2mm in diameter.

more info here:

http://deskthority.net/workshop-f7/frictionless-trackball-via-ball-transfer-units-t8286.html
https://geekhack.org/index.php?topic=59444.0

There is also a youtube video of me playing with jacobolus's prototype at keycon, but I dont have the link handy.

[gbjk]

Turbinia,

I notice you've modeled the overhead thumb lever to come right out across the top of the thumb unit.

I think we tend to drop our thumbs in through that gap, and I don't feel like you need it to come up and over like that.

The direction of travel for it is very much across, more than up.

HTH

G

[hweller]

Also curious about the smaller package for the ADNS-9800, any idea as to those dimensions?

I received two of the smaller ADNS-9800 breakout boards from John Kicklighter last week.  The PCB is rectangular 30x18mm with two mounting holes on the centre-line, 25mm between centres and the lense centred.  The other dimensions are the same as the larger circular board.  I have attached two pictures, sorry they are out of focus but my digital camera had trouble focussing on it.

[Turbinia]

Ah that is a much more compact package, much easier to fit.

Does John Kicklighter have the IGES drawing for the ADNS-6190-002 lens? It looks like a deprecated product, so didn't see it on Avago's website. Have the data sheet http://www.mouser.com/ds/2/38/V02-1725EN+DS+ADNS-6190-002+16Aug2011-75622.pdf but that doesn't give the sort of dimensions needed for the base plate (which in this case would be the ball cup).

Does the mounting angle need to be something in particular?

I will continue to refine the case design as I get more information on the various components.

[hweller]

Does John Kicklighter have the IGES drawing for the ADNS-6190-002 lens? It looks like a deprecated product, so didn't see it on Avago's website.

The ADNS-9800 is now manufactured by Pixart:
http://www.pixart.com.tw/product_data.asp?product_id=76&productclassify_id=1&productclassify2_id=3

and they supply the datasheets but I haven't seen CAD for the lens or the sensor.

[wolfv]

The DodoHand firmware is coming along nicely.
I don't have the DodoHand hardware, so I will test the firmware on a keyboard simulator, and then test it on a breadboard.
The simulator and bread board contain a subset of a keyboard's layout.
10 keys is to enough to test the layer and control keys.

Here are some expected milestone dates:
   12/02/14 DodoHand subset tested on keyboard simulator.
   12/07/14 DodoHand subset tested on breadboard with MCP23018 I/O expander.
   12/14/14 release Keyboard library and breadboard firmware.
   12/21/14 DodoHand subset tested on breadboard with PCA9655E-D I/O expander.
   12/28/14 DodoHand firmware with all keys mapped to DataHand layout, so someone with DodoHand hardware can test it.

[nandop]

This really appeals to me, but the case makes it look gigantic. Can it be just the wrist rest, thumb and finger buttons as the whole package? I can print the legends and have them with me as I get used to the design so as to not have that extra casing in the front of the finger buttons. That would be so perfect. I only say this because I really want it, and the only thing keeping me from buying it is the thickness of the case and the bulginess. Only way I would compromise a fix around that would be to make holes in my desk to fit them in.

[Turbinia]

@nandop, the case is really as compact as it can be. In the previous image and the following it can be seen that the pcb fills the whole of that upper portion. The overall dimensions are 15cm by 20cm by 3cm.



This shows one concept I have been working on to be milled out of billet aluminum. The bottom plate would be 3mm acrylic or 1/8in aluminum plate. Looking at the cost of current keyboard cases the block may be about $100-150 per hand. I don't think the adjustably of tenting and tilting are really feasible for the first round. Like the ergodox I think people will find ways to get the tenting and tilting that they need.



In tandem I am also designing an acrylic case to target a lower cost. It may not be quite as low as you might expect with things like the ball cup and switch guards needing to be 3d printed that would be milled in the CNC aluminum design. Currently designed for 3mm acrylic. Bit tricky to go back and fourth between both designs to make sure that all screw holes and standoffs work together.



Certain aspects such as the thumb keys and such are still in flux so the design is subject to change.

[dorkvader]

If you put cutouts in the bottom like wall mounting points for routers and the like, then it would be really really easy for people to come up with their own tent angle and just clip the KB to it.

y'know these guys with a big opening for the screw head to go through and a smaller one to hold it in place

[nandop]

@nandop, the case is really as compact as it can be. In the previous image and the following it can be seen that the pcb fills the whole of that upper portion. The overall dimensions are 15cm by 20cm by 3cm.

Show Image

This shows one concept I have been working on to be milled out of billet aluminum. The bottom plate would be 3mm acrylic or 1/8in aluminum plate. Looking at the cost of current keyboard cases the block may be about $100-150 per hand. I don't think the adjustably of tenting and tilting are really feasible for the first round. Like the ergodox I think people will find ways to get the tenting and tilting that they need.

Show Image

Show Image

In tandem I am also designing an acrylic case to target a lower cost. It may not be quite as low as you might expect with things like the ball cup and switch guards needing to be 3d printed that would be milled in the CNC aluminum design. Currently designed for 3mm acrylic. Bit tricky to go back and fourth between both designs to make sure that all screw holes and standoffs work together.

Show Image

Show Image

Certain aspects such as the thumb keys and such are still in flux so the design is subject to change.

I take that back. The acrylic's purpose to lower the cost would be great for me as I would like to customize it at some point. ^^

Great design and a great project. I would like to purchase one, even if in prototyping stages! :D

[tuft]

I recently switched to a DataHand and am benefitting enormously. It does feel tenuous to stake my livelihood on an old piece of electronics that I can't readily replace, though! The "p" key sticks, which I've worked around by rebinding "Enter" in software, and occasionally the unit spazzes out and needs a factory reset. I recently lost an eBay auction to secure a backup unit at $2500!

Anyway, I'm very excited for this project, and just want to lodge my support as another willing buyer. =)

[hweller]

I have been using a DataHand DH200 with the trackball for a few weeks now and I am not entirely satisfied with the behavior of the opto/magnetic switches; some are a bit sticky, they require different force to actuate (the RH pinky switch is particularly difficult to press) and the tactile feedback is poor.  Even before I obtained a DataHand I have been considering the possibility of creating an equivalent device using Cherry MX down switches and Panasonic micro-switches for NSEW.  I have just completed the first prototype which proves the principle but it difficult to get all the switches in suitable positions so that the centres are located in the same positions an in the original DataHand.  All the switches feel good and I am now building the micro-controller board so that I can practice typing on it before constructing the left-hand unit.  Here is a picture showing the general layout, I can post more details if there is interest.

[tilmann]

Wow, awesome prototype!

I have been using a DataHand DH200 with the trackball for a few weeks now and I am not entirely satisfied with the behavior of the opto/magnetic switches; some are a bit sticky, they require different force to actuate (the RH pinky switch is particularly difficult to press) and the tactile feedback is poor.  Even before I obtained a DataHand I have been considering the possibility of creating an equivalent device using Cherry MX down switches and Panasonic micro-switches for NSEW.

Yeah, I've experienced the same, on my DataHand the LH pinky is particularly difficult to press. I also noticed that between different DataHand models the activation force seems to be slightly different (I assume DataHand Inc. has been tweaking the switches over time). I tried to build a finger board with micro-switches myself a couple of months ago and got good initial results, I was using the smallest 20g micro-switches I could find (I'm happy to dig out the exact part number). The switches also have a really nice tactile click sound.

Which micro-switches are you using right now?

I have just completed the first prototype which proves the principle but it difficult to get all the switches in suitable positions so that the centres are located in the same positions an in the original DataHand.  All the switches feel good and I am now building the micro-controller board so that I can practice typing on it before constructing the left-hand unit.  Here is a picture showing the general layout, I can post more details if there is interest.

Please post more details

I'm curious, what material did you use for the wrist rest and how did you get it into shape?

[hweller]

I started playing with microswitch types after I bought a vintage Microwriter and was very impressed with the feel of the switches.  I had to take the Microwriter apart to try to fix/upgrade the electonics and found that the microswitches used are standard Cherry switches which are still available.  I purchased some of these switches and one of every equivalent from all the other manufacturers including Omron and Panasonic and found that the feel of the Panasonic switches are better than all the others and readily available.  The issue with the VERY small microswitches is the lifetime and the durability of the lever so I tried to create a design which uses the standard switches which are about 10mm high.   Here are my notes-to-self on the choice of switches:

Down-switches
    Low force Cherry MX, either the linear "red" or tactile "brown".  One switch
    is required for each finger and thumb.  Obtaining Cherry MX switches in the
    UK from The Keyboard Company:
    + 10x http://www.keyboardco.com/product/cherry-key-switch-module-brown-tactile.asp
    + 10x http://www.keyboardco.com/product/cherry-key-switch-module-red-soft-linear.asp
    The keycaps should have spherical rather than cylindrical indentations.
    Signature plastics supply DSA series Cherry MX keycaps with various
    lengths:
    + Fingers: 8x 4INHEAKS78  DSA Extra Blank Keys  1 Space Deep Dish-Black (NDY)
    + Thumbs: 2x 4INHEAKS78  DSA Extra Blank Keys  2 Space Deep Dish-Black (NDY)
Side-switches
    There is not enough space to use Cherry MX for the side switches and small
    tactile switches generally require too much force and do not have a good
    feel.  In the DecaKeyer https://github.com/Henry/DecaKeyer
    I played with both Cherry and Panasonic micro-switches and found that the
    Panasonics to have a slightly better feel
    and are more readily available in the UK.  RS-Components stock a large range
    of Panasonic micro-switches, the AVT and AVL types being most suitable as
    the leaver is more robust and has less lateral play.  The "L" versions are
    longer-life (at least mechanical) but the actuation force on the plunger is
    1.47N compared with 0.98N for the standard "T" versions (lower force
    versions are also available, 0.49N and 0.25N).  The long-leaver http://uk.rs-online.com/web/p/microswitches/6990511/?searchTerm=699-0511&relevancy-data=636F3D3126696E3D4931384E525353746F636B4E756D6265724D504E266C753D656E266D6D3D6D61746368616C6C26706D3D5E5C647B337D5B5C732D2F255C2E2C5D5C647B332C347D2426706F3D313426736E3D592673743D52535F53544F434B5F4E554D4245522677633D4E4F4E45267573743D3639392D3035313126
    has an actuation force of 0.25N whereas the long-life long-leaver http://uk.rs-online.com/web/p/microswitches/6994664/?searchTerm=699-4664&relevancy-data=636F3D3126696E3D4931384E525353746F636B4E756D6265724D504E266C753D656E266D6D3D6D61746368616C6C26706D3D5E5C647B337D5B5C732D2F255C2E2C5D5C647B332C347D2426706F3D313426736E3D592673743D52535F53544F434B5F4E554D4245522677633D4E4F4E45267573743D3639392D3436363426
    has an actuation force of 0.44N which may be too high for finger
    switches but maybe good for thumb switches.

I built a finger prototypes with the standard and long-lifetime Panasonic swtiches and although the lower force standard switches feel better in some respects the long-lifetime switches require the same force as the Cherry MX down switch and this consistency feels good.  Also the click is even more positive on the long-lifetime switches so I decided to complete the first prototype with these switches to see if the force is acceptable for typing.

The switches are mounted (glued) on to cross-shaped ABS plates glued to the heads of M6 bolts and bolted to the ABS base-plate to allow changes in height and orientation.  The hand-rest it hand-carved out of a piece of laminated EPE packing-foam which is quite easy to work with a modeling saw and coarse glasspaper and has a very nice warm comfortable feel.  It is so good that I am not planning to change the material in subsequent versions as this is easy to replace when it gets dirty and also washable and different shapes can easily be made for other hand shapes and sizes.

[Turbinia]

Very interesting effort hweller. When I was looking at recreating the switches myself I was looking at a similar setup. Instead of using microswitchs I looked at using tactile switches and 3d printing a lever to actuate. You can get everything from .5 to 1.6N in SMD or through hole to tune certain fingers or keypresses and mount to pcb instead of lots of gluing to screws. One of the advantages of the DataHand however is the force fall off through actuation instead of increasing to actuation like a traditional keyswitch.

You may not be satisfied with the switch feel anymore, but how is the trackball feel, still good?

[hweller]

I also investigated tactile switches and the best I found were the Omron B3M but the force is a bit high and the travel too short so as you suggest some kind of lever arrangement would be required which is already provided with microswitches which simplifies the build -- but it wasn't easy!

I am OK with the DataHand switches but I am always looking for the next improvement, the next step in keyboard design and a DataHand with microswitches is something I have wanted to have a go at for sometime.

I am VERY happy with the trackball location and feel, and will fit one to the microswitch DataHand if I take it further than just a prototype which is likely as it is already a promising design.

[OldDataHands]

Epic.

[norcalli]

I was about to consider spending a whole lot of time doing what you guys have already done. I'm glad I found this thread and will be eagerly awaiting updates. As a 21 year old programmer who's wrist is already starting to feel sad, I'm intensely curious. I may even take a stab at creating my own microswitch prototype! I thought Cherry MXs would have worked on the sides with a lever mechanism, but microswitches look good.

[Spanjer]

Keep it up i'm 22 and in desperate need of a pair of data hands, i'm currently looking into making myself a pair but i'd love to buy a quality build.

[DrudgeDance]

 

[wolfv]

I am testing a DodoHand firmware, but the manual doesn't tell all, and don't have access to a DataHand for comparison.
Someone please test the following key strokes on a DataHand and tell me what you get.

LEGEND:
"press NAS" means press right thumb down lightly (not NAS Lock)
"press MF" means lift right thumb (Mouse Function)
Letter keys K, H, and K assume QWERTY locations

Test 1:
 press MF, release MF, press H key
Does it print "h" or move pointer left (mouse)?

Test 2:
 press MF, press K key, release MF, press H key
Does it print "h" or move cursor left (arrow)?

Test 3:
 press MF, press D key, release MF, press H key
Does it print "h" or move ponter left (mouse)?

Test 4:
 press NAS, press MF, release MF, press H key, release NAS
Does it print "h" or "6"?

Test 5:
 press MF, press NAS, release NAS, press H key, release MF
Does it print "h" or move pointer left?
Does it matter since right thumb can not hold NAS and MF simultaneously?

Thank you.

[wejn]

Full disclosure: DataHand Personal edition.

Well, I'd say that your test forgets one important thing -- depending on the state my DH boots up I can have mouse on or cursor keys off. Turning mouse off = in "mouse mode" (or, rather, I'd call it F-keys mode) you tap "D". To turn off you tap "K".

But to your tests:

Say I have the mouse on indicator (orange diode on "D" key) lighting up; so normally in mouse I move pointer around and click with L/R index fingers down (F, H keys).

Then:
Test1: mouse pointer moves
Test2: mouse mode deactivated,cursor goes to left
Test3: mouse mode activated (diode ligths up and stays up), mouse pointer moves
Test4: pointer moves (mouse mode stays active because as soon as you get mouse up [while holding down NAS] the mode gets activated)
Test5: pointer moves, because NAS gets activated only for the duration of you holding it down

I think the key misunderstanding here is that you don't assume the mouse/f-key mode to go "sticky". Basically Normal and Mouse is a flip switch, NAS is just a push button (unless you go NAS-lock). Analogy: same way you hold shift [=NAS] or caps lock [=NAS lock, normal, mouse], but the Normal vs Mouse/F-keys is a XOR.

Btw, I have official DH docs lying around dh-docs.zip (1957 kB - downloaded 187 times.), in case that's of any value. I'm saying this because there are many more things configurable that's not readily apparent. For instance, I have 10-key mode completely disabled (because it sucks). Also, I've noticed that your NAS and my NAS is slightly different (left ring down in NAS = x, right ring right = enter or Win key or other). In the attachment I'm including the layout my DH uses.

[wolfv]

wejn,

Thanks for the test results and the DataHand Personal User's Guide (I only had the DataHand
 Professional II User’s Guide).
It does not mention the "cursor keys off" option.  How is that useful?

[wejn]

wolv,

Sorry, that was a typo. All I wanted to say was that depending on in what mood my DH wakes up in I either have "mouse" or "function" enabled when I switch to Mouse Function. It was all to outline the exact setting I started your tests in; I just failed to edit the message properly (I rewrote the post couple of times; both before and after posting).

Btw, to clarify -- when I said "completely off" for 10key I, of course, meant that I've set "protection" for those keys (meaning that I have to hold down NAS lock to turn 10key on.

It's actually quite hard to remember the exact setup I've made, because it's a long time since ~2006 when I bought the thing... and I basically haven't touched the settings since then.

Btw, I'm thrilled at the prospect of being (finally) able to modify the layout with DodoHand and your fw (without fearing I'll brick this irreplaceable DH of mine).

[invultri]

Hello all,

First off having tried to design this myself you all have my respect for pressing on and making such amazing progress. Since I am also one of them programmer people that is having a long track record (over a decade) of having the associated issues I have been looking for a datahand myself. However I tend to get outbid and they do not show up that often, this project actually gives me hope that I can realistically get the datahand typing experience. The recent addition of the trackball and the pretty looking design of the case closed the deal and I am totally hooked.

Hence I am very excited and will assist in any group buy / crowd funding effort! As an alternative I will work with the github sources and go into a mad sourcing campaign . Judging from the thread (I read it all..) I think an early pair sourced individually including fancy case should be obtainable for less than 1000EUR.

Once the wired version gets done and an iteration towards wireless is made the following kickstarter could be interesting. A small fuel cell 200grams full delivering 2W of power at 5volt that runs of lighter gas. It is actually asking for money to do a production run so they are past the prototyping stage, it will take about a year if all goes well to get them available to the public. This could be an interesting option to ditch the wires (even though I find that wireless devices can have some annoying lag but I have never used BT3 or above) without the need for actual batteries. Link: https://www.kickstarter.com/projects/265641170/kraftwerk-highly-innovative-portable-power-plant . This of course is only possible if a cavity can be made of 3.94 x 2.95 x 1.18 inch or maybe less if its case can be removed.

[wolfv]

The a 10-key version of DataHand is running and tested on a breadboard.
It has a Teensy2.0, MCP23018 IO expander, and 10 tactile switches for these keys:
   D, R, K, Shift, tenKeyOff, tenKeyOn, Normal, NAS, NASLock, MouseFunction
   (the D, R, K keys are for testing mouse, arrow, and layer keys)

Next I will add features to the firmware in this order:
   Full layout for the DodoHand hardware
   Document and upload to GitHub
   PCA9655E-D I/O expander
   Protected Mode
   NUM LOCK
   Sticky Mouse Buttons
   Double Strike
   LEDs

I have some questions about dodohand-master2014.1.16\documentation\switch_matrix.svg,
What rows map to what keys?  For example:
   Row0 = north
   Row1 = east
   Row2 = south
   Row3 = west
   Row4 = well
Are the row-switch mappings different for left and right hands?
What pin number do they go to?
What is the symbol to the right of the diodes?

Thank you.

[OldDataHands]

I have some questions about dodohand-master2014.1.16\documentation\switch_matrix.svg,
What rows map to what keys?  For example:
   Row0 = north
   Row1 = east
   Row2 = south
   Row3 = west
   Row4 = well
Are the row-switch mappings different for left and right hands?
What pin number do they go to?
What is the symbol to the right of the diodes?

Have a look at the same document in the thumb_work branch.
I already partially addressed the north/east/south question but haven't
pulled it back into the main yet. Note that it reflects the left hand.
There is only one PCB layout, reversible, so the switch-to-finger
position mapping is mirrored on one hand vs. the other. (west on
one hand is east on the other).

Switch: Left        : Right
Matrix : Teensy : PCA9655E
---------------------------------------
ROW0 : F0       :  IO1_0
ROW1 : F1       :  IO1_1
ROW2 : F4       :  IO1_2
ROW3 : F5       :  IO1_3
ROW4 : F6       :  IO1_4
COL0  :  B0       :  IO0_5
COL1  :  B1       :  IO0_4
COL2  :  B2       :  IO0_3
COL3  :  B3       :  IO0_2
COL4  :  D2       :  IO0_1
COL5  :  D3       :  IO0_0

The symbol to the right of the LEDs is my version of the normally
closed mechanical/optical switch between the LED and the
phototransistor (i.e. the symbol (diode included) is my attempt to
represent the dodohand switch, LED, phototransistor and all).

Excited to be able to try out your firmware!

[randomhead]

Hello!

I've come here to post my interest for this project. Unlike a lot of people here, I found out about the DataHand too late in the day to own one. It was only after hopelessly typing "open source datahand" or something of the like into the search bar that I found this thread. Thank you internet, you really do have everything!

I understand (and respect) that the reason that more talk of group buying/sourcing hasn't come up more often/hasn't been organised yet is probably because it would be premature at this stage, is this correct? I'm trying to work out if I want to chip in to a group buy effort (or even if it would be feasible for me - uk based here), or whether I should just jump in and get a hold of the parts that have already been developed and build a full set up slowly as it's perfected.

OldDataHands, are there any plans to create another revision of the fingers or thumb modules, or would you say they are good to go? If it's the latter, I would be sorely tempted to just jump in and buy the supplies I need to get my own set built up. My only concern would be the exotic parts (magnets and the custom made clips). Also, would you recommend my approach? I certainly wouldn't mind being a beta tester, but what I build up would probably end up being 'it' for me, so if any further revisions are required I would be out of luck. Another concern would be if the case is required for the keys to work properly (or would the black plastic be enough by itself to block out light pollution)?

Edit: Oh, and a question to anybody who might've used the combination: What's vi like on the datahands? It's made to be logical on a qwerty (can't speak for dvorak), it's my text editor of choice, I wonder how people's experiences have been with it on a different physical 'interface'?

Thanks

[arisian]

I've used vi (or ViM) on a datahand for years.  The default keybindings do have minor inconveniences, but it's nothing you can't fix with some re-mapping in your .vimrc.  The main thing you'll likely want to change is the movement keys; hjkl are not the most natural way to do movement on the datahand.

I'll give you my personal mappings below, but I type dvorak (on a datahand), so unless you also type dvorak, you'll probably need to modify them.  I changed the movement keys to be N/S/E/W using my right forefinger, which for me is g/m/d/[ (would be u/m/h/' on a qwerty mapping).  I then had to move the functions usually performed by those keys to alternate locations.  What I wound up with was:

Code: [Select]

nnoremap l g
map g k
noremap h d
noremap , m
noremap d h
noremap m j
noremap [ l


[OldDataHands]

I understand (and respect) that the reason that more talk of group buying/sourcing hasn't come up more often/hasn't been organised yet is probably because it would be premature at this stage, is this correct?

Certainly not ready for a group buy, and a group buy will barely make sense as the bulk of the
expense will be for the 3D-printing, which doesn't change price with quantity.  There might be
some gains with the case, or PCB, but nothing that'll make or break your decision.

OldDataHands, are there any plans to create another revision of the fingers or thumb modules, or would you say they are good to go?

The finger modules are functional, though care is required during assembly, and the extremely
small number of assembled units (only mine to the best of my knowledge) and the low hours
(i.e. 0 as the thumbs aren't done) leave me unable to say that they're reliable with an acceptable
longevity.

The thumb switch design is not yet useful, particularly requiring some re-design of the magnet
retention for the side switches.

My only concern would be the exotic parts (magnets and the custom made clips).

The magnets are commercially available, so maybe they aren't that exotic. The clips are an
irritant, though I was able to make the first prototypes into a useful shape with only pliers and
the bench shear for cutting the starting strips. You probably won't have perfectly identical feel
from switch to switch, but would function fine.

Another concern would be if the case is required for the keys to work properly (or would the black plastic be enough by itself to block out light pollution)?

You'll need a case of some sort to provide the palm rest and to hold the thumb switches in the
right spot relative to the finger switches. Ambient light in a room doesn't seem to be a problem.
You'll have to tell us how it goes if you try to operate them in direct sunlight, but I would expect
direct sunlight to be a problem.

[Turbinia]

Have you considered using a second magnet as a target instead of the clips?

[OldDataHands]

Have you considered using a second magnet as a target instead of the clips?

I have. The challenge is in achieving contact between the two magnets when
the switch is closed.  You need to have positive retention of each magnet, which
means material that sticks out in front of each magnet, which interferes with the
other magnet getting close enough. The clip works 1) because it is retained from
only the one end, behind the face, and  2) because it is much thinner than the
magnet, so can fit between the two magnet retention lips and make contact with
the magnet surface. Contact between the two is important for achieving the most
"click" or "snap" out of this switch design. If you don't have this it feels mushy.

[eviltobz]

... What's vi like on <non-standard qwerty boards> ...

i'm currently using a kinesis at work and an ergodox at home, both with colemak layouts instead of qwerty. i decided to stick with the standard vim mappings rather than trying to go with something more customised to my layout since all those resources out on the internet presume a normal setup. i decided to try to embrace the advice of using hjkl as little as possible, more for off-by-one errors rather than main navigation, so their shuffling wasn't a big problem. in colemak's case h stays the same but j, k & l change to n, e & o respectively, all of which are common, important vim keys with good mnemonics, so i didn't want to swap their meanings. i get on with it just fine, so i'd recommend that you try to just work with it as is before resorting to remappings.

[invultri]

The magnets are commercially available, so maybe they aren't that exotic. The clips are an
irritant, though I was able to make the first prototypes into a useful shape with only pliers and
the bench shear for cutting the starting strips. You probably won't have perfectly identical feel
from switch to switch, but would function fine.

Silly question, would it be feasible to 3d print the clips? If we do not consider price, would the clips be more consistent?

[OldDataHands]

Silly question, would it be feasible to 3d print the clips? If we do not consider price, would the clips be more consistent?

Are you aware of a 3D printing service or technique that makes ferrous objects? I've never heard of one, but would be cool!

[Turbinia]

http://www.shapeways.com/materials/steel

Not sure as to the alloy, if there is too much nickel it will not be ferromagnetic.

As an alternative the magnets can always be glued in for magnet on magnet. Not ideal, but not really that bad to loose some pocket change for a key replacement magnet.

[OldDataHands]

http://www.shapeways.com/materials/steel

Not sure as to the alloy, if there is too much nickel it will not be ferromagnetic.

As an alternative the magnets can always be glued in for magnet on magnet. Not ideal, but not really that bad to loose some pocket change for a key replacement magnet.

Seems as though it is 420 stainless and brass, with the 420 part being magnetic at least to some degree.
Unfortunately, the min bounding box, min separation, and min wire thickness make it unsuitable for the
current clip design.

I'm certain that it would be possible to design in such a way that you could get the right feel with gluing
in magnets, or bits of steel, but I think it would take quite a bit of either insight into designing with magnets,
or numerous prototypes...