Trackball via Ball Transfer Units *3D Mock-up Made*

[jacobolus]

At Dorkvader’s suggestion, I got one of these for a $25 best offer www.ebay.com/itm/330620442341?orig_cvip=true which should arrive sometime next week, Synnöve is sending me 3 of those ball transfer units, and I already have a couple of those Kicklighter ADNS-9800 sensors, so I should also be testing some stuff out starting in about 2 weeks (I have some keyboard stuff I want to try to build before KeyCon, so the next 2 weeks are filled up).

I’m going to start by trying to CNC cut enclosure parts out of wood, since I have access to a ShopBot at TechShop, and that’ll be faster turnaround for prototyping and cheaper than ordering 3D-printed parts from Shapeways or whatever. (I don’t think one of the cheap extruder 3d-printers will make parts that are as dimensionally accurate or sturdy or nicely finished as I’d like, but I’d be curious if someone else wants to try with those.)

[dorkvader]

That eBay trackball uses a very similar unit internally (higher CPI and a slightly diiferent PCB seem to be the only differences) as this P&G unit in this trackball HaaTa has:
https://plus.google.com/photos/113845661925823397356/albums/5788605520602625569


now that's what a trackball is supposed to look like inside!

Keep in mind it's a 50mm (2") ball.

Keep in mind it's a 50cm (2") ball.

mm*

What's an order of magnitude between friends?

[jacobolus]

Keep in mind it's a 50cm (2") ball.

For prototyping, I’m not too worried about the precise ball size, as long as the ball is a nice weight, smooth, and very spherical. Any design I make shouldn’t be too hard to adjust to fit a slightly bigger or smaller ball.

[jakkdl]

Following this thread with great anticipation, I'm in for buying a better finger trackball than my kurrent kensington orbit.

Keep in mind it's a 50cm (2") ball.

mm*

[jacobolus]

I got the P&G trackball from ebay and the ball transfer units from Synnöve, and I’m going to try to make a little tetrahedral cardboard housing to see how well they spin.

The ball transfer units are *tiny* – I guess I knew that from the stated dimensions, but it didn’t sink in until I had them in my hands. I think we could probably get away with substantially bigger ones and still have them fit just fine, so if there’s any price difference for buying in bulk, we should just go with whatever is cheapest. Alternately, if we can the trackball balls, we probably e.g. make some with a 1 inch ball and these tiny ball transfer units, and integrate it a tiny portable keyboard (though at some size fitting the sensor(s) starts to be tricky).

The P&G trackball spins really well in the direction diagonally toward/away from the corner where the two rollers meet, fairly well in the other diagonal direction, and basically not at all about the axis perpendicular to the table (because it has to scrape against the rollers to go that way, instead of rolling). But the ball is currently very smooth and spherical (it’s brand new), not scratched.

I think the ball transfer units will make it spin much better in pretty much every direction, because we should get mostly rolling and very little scraping, regardless of the direction it turns, and hopefully trackball longevity will be pretty good, because it shouldn’t pick up scratches like existing trackball designs facilitate.

If you roll one of the ball transfer units along a wooden surface, it’s definitely loud, but I don’t think in practice screwed into a housing and spinning against a typical trackball it will be especially loud, or at any rate not any louder than the P&G trackball’s ball bearings. Figuring out how to plan a few little pieces of sorbothane into the design might help with any noise though.

I’m still not sure whether I want to use 2 sensors or 3, or exactly what shape I want to make the final housing. I’ll probably make some plasticine models in a week or two (sometime after KeyCon), but I’ll try to bring some kind of simple cardboard mockup to KeyCon if anyone who’s going wants to get a feel for the ball transfer units.

Overall, I’m pretty optimistic about the project.

[Synnöve]

I think we could probably get away with substantially bigger ones and still have them fit just fine, so if there’s any price difference for buying in bulk, we should just go with whatever is cheapest.

This was my reaction as well, though I'd clarify that it's important to make sure the support balls and internals are stainless steel for longevity's sake.

If you roll one of the ball transfer units along a wooden surface, it’s definitely loud, but I don’t think in practice screwed into a housing and spinning against a typical trackball it will be especially loud, or at any rate not any louder than the P&G trackball’s ball bearings. Figuring out how to plan a few little pieces of sorbothane into the design might help with any noise though.

I agree that if it's screwed in to a sturdy housing with some dampening material installed, it should yield acceptable results.

I’m still not sure whether I want to use 2 sensors or 3, or exactly what shape I want to make the final housing. I’ll probably make some plasticine models in a week or two (sometime after KeyCon), but I’ll try to bring some kind of simple cardboard mockup to KeyCon if anyone who’s going wants to get a feel for the ball transfer units. Overall, I’m pretty optimistic about the project.

Look forward to seeing your mockup and hearing feedback from others concerning the feel of the units. Why would you need more than two sensors?

Here is a little mockup I made using my pathetic 3D modeling skills (am more of a rigger/animator). Idea with it is to be symmetrical, having a good amount of room for the circuitry, large amount of buttons (should be six currently), and be very ergonomic/supportive of the palm. Thoughts?

[jacobolus]

This was my reaction as well, though I'd clarify that it's important to make sure the support balls and internals are stainless steel for longevity's sake.

I’m pretty sure the Delrin ones would work too. Remember, the trackball is much lighter than the loads these ball transfer units are designed to support, the motion mostly rolling, not scraping, so there should be a lot less wear than with any existing trackballs.

[Synnöve]

I agree that the weight shouldn't pose a problem, but I would say plastic bearings would be more prone to pitting, dents, and would be rated for less revolutions vs. SS. The scraping notwithstanding, every plastic bearing I've seen in a trackball is rendered somewhat useless after several years of use so I don't tend to trust it for longevity.

In any case, at least we have the choice of which sort we can get! : D

[bueller]

Watching this closely, love older trackballs but all the current models on the market are uncomfortable for me to use.

What is uncomfortable about them?

@Synnöve - I have two older (off white) CST Trackballs if you would like them for testing and building your model. PM me if you are interested.

Will do!

Its different on every one really, I usually feel like the housing is not ergonomic for my hands. Some of the old Microsoft ones were awesome but the sensors in them are really showing their age now. Eventually I'd love to build my own or mod a Microsoft with new ball and sensor, finally getting my hands dirty with some 3D modelling so I'll get to it one day.

[jacobolus]

i think your mockup doesn’t leave as much of the ball exposed as I’d like. For a 2-d trackball it’s probably fine, but for a 3-d trackball it’s a huge advantage to be able to grip as much of the ball as possible.

* * *

Anyway, I’ve been playing with putting the ball transfer units into little pieces of cardboard and spinning them, and I have to say I’m very impressed.

The trackball spins in pretty much any direction as well as (or perhaps slightly better than) the best axis on the Penny & Giles. If I spin it fast, it will spin for several seconds before stopping (for comparison, a logitech trackball I have stops spinning almost instantly after I remove my finger). Very fine movements do seem to be just slightly “scratchier” (scratchy like a modern Cherry MX linear keyswitch), but I think it’ll be a bit smoother if the ball transfer units are mounted in a rigid housing with a very precise shape – currently they support the big ball slightly off axis, and cardboard is not the stiffest material out there. :-) The sound produced by this scratchiness is transmitted pretty well through cardboard, but I think in a housing made of a different material, possibly with a bit of dampening rubber carefully placed to prevent the vibration from traveling, they can be made quieter than a standard roller trackball.

There’s basically no scraping at all in any direction between the main trackball and the ball transfer units. I don’t know what kind of wear the ball units get between their main balls and the little helper ball bearings, but I think the wear between the ball units and the trackball ball is going to be negligible, regardless of the material used for the trackball or ball unit ball. This potentially opens up a lot of other trackball materials to be used, though I think whatever this P&G trackball is made of, or any random billiards/snooker/whatever ball, or whatever, is going to be great in practice. Also, I’m curious whether we can try some of the plastic (Delrin?) ball units, to see how they compare friction-wise to the stainless steel type.

* * *

Along with maybe being able to use larger ball units and/or plastic ones, I also think we should try the other model type of Alwayse ball units. That is, try 11-MI-05-13 or 11-MI-05-15 (is carbon steel vs stainless steel going to matter for us?) instead of the 11-MI-05-17. The few millimeters of extra overall size aren’t going to be a dealbreaker for us, and because they seem to use larger helper bearings, I suspect they might spin even smoother. Also, the longer metal bolt might come in really handy. I don’t think there’s any disadvantage for us in picking the flat top instead of chamfered-top kind.

* * *

After KeyCon I’m going to try to get the laser sensors hooked up to a Teensy 3.0 and see if they will usefully track the surface of the ball (it’s a smooth glossy black ball; not sure if that’s a good surface to track on or not). If I can get that working okay with one sensor, then I’ll start trying to design a housing I like. I’ll probably try to CNC cut it out of hardwood, and at some point integrate it with a keyboard.

[Synnöve]

i think your mockup doesn’t leave as much of the ball exposed as I’d like. For a 2-d trackball it’s probably fine, but for a 3-d trackball it’s a huge advantage to be able to grip as much of the ball as possible.

When you say 3D I assume you're talking about the up down & side to side axis but with the addition of twist? The mockup exposes as much of the ball as possible while still using Hanya's design for the bearing and sensor arrangement (and a little extra due to the width of the plastic). The only part that's obscured is the rear but that's there to support the palm so one doesn't have to hold their hand above the ball.

Along with maybe being able to use larger ball units and/or plastic ones, I also think we should try the other model type of Alwayse ball units. That is, try 11-MI-05-13 or 11-MI-05-15 (is carbon steel vs stainless steel going to matter for us?) instead of the 11-MI-05-17. The few millimeters of extra overall size aren’t going to be a dealbreaker for us, and because they seem to use larger helper bearings, I suspect they might spin even smoother. Also, the longer metal bolt might come in really handy. I don’t think there’s any disadvantage for us in picking the flat top instead of chamfered-top kind.

I wouldn't go with carbon steel due to it's weakness when it comes to oxidation; once the oil coating the balls is gone they will be vulnerable to humidity. I agree that they might spin smoother (though at the expense of loudness I'd assume), and that flattop won't make any difference.

After KeyCon I’m going to try to get the laser sensors hooked up to a Teensy 3.0 and see if they will usefully track the surface of the ball (it’s a smooth glossy black ball; not sure if that’s a good surface to track on or not). If I can get that working okay with one sensor, then I’ll start trying to design a housing I like. I’ll probably try to CNC cut it out of hardwood, and at some point integrate it with a keyboard.

The CST trackballs are black and glossy and are read via laser just fine; I don't think it'll present a problem. I look forward to seeing your mockups!

[jacobolus]

i think your mockup doesn’t leave as much of the ball exposed as I’d like. For a 2-d trackball it’s probably fine, but for a 3-d trackball it’s a huge advantage to be able to grip as much of the ball as possible.

When you say 3D I assume you're talking about the up down & side to side axis but with the addition of twist?

That’s right, I want to measure the full 3-dimensional rotation of the ball. Internally to the trackball firmware, I want to at some point in the logic have the last time increment’s rotation stored as a quaternion. Then it’s possible to take this quaternion and process it to back out the amount of rotation about any arbitrary axes.

This means you can (a) on the fly pick which axes to use for x/y mouse movement, for instance if your hand has easier access to the side of the ball while you’re using a nearby keyboard, or if you naturally rotate about different axes when you use your thumb vs. using your fingers, (b) you can back out useful quantities like the magnitude of the rotation irrespective of direction (this could be useful if you want to use the ball to scroll a list or perform some repetitive action... you can just put your finger on the surface of the ball and move it in any pattern you like and do your scrolling: e.g. in circles, think original iPod wheel), (c) you can send the full 3-d rotation to the computer and use it for rotating 3-d models or cameras e.g. in CAD or visualization software, (d) you can interpret various gestures performed using two fingers on the ball, somewhat similar to the kinds of gestures you get with a multitouch trackpad/touchscreen (though the distance between fingers is going to necessarily remain constant on the ball surface), etc. etc.

Anyway, I think it’s possible to fit everything with about 70% of the surface of the ball exposed, maybe more.

[jacobolus]

I brought the ball transfer units and trackball in my super classy cardboard housing to KeyCon, and everyone agreed: super smooth!

[politie]

I brought the ball transfer units and trackball in my super classy cardboard housing to KeyCon, and everyone agreed: super smooth!

this it? looks nice

[jacobolus]

Yup, that’s it!

[dorkvader]

I brought the ball transfer units and trackball in my super classy cardboard housing to KeyCon, and everyone agreed: super smooth!

this it? looks nice

what a great hand model.

I'm glad to have been a part of this!

[politie]

so is there no sticking/skipping on really small movement? and do the balls stay a little smooth after longer use? no dust collecting inside the units or anything?

[jacobolus]

so is there no sticking/skipping on really small movement?

There is a bit of “scratchiness” for small slow movements. I’m not sure if this is inherent, or just something to do with my flimsy cardboard mockup, or perhaps something that could be improved by using different ball transfer units.

and do the balls stay a little smooth after longer use? no dust collecting inside the units or anything?

This is impossible to tell before building a real housing and using the trackball more extensively. I’m guessing they’ll last pretty well.

[dorkvader]

so is there no sticking/skipping on really small movement? and do the balls stay a little smooth after longer use? no dust collecting inside the units or anything?

it's already far superior to a consumer trackball on small movements.

we suspect it'll last longer, but it also depends on the ball you use and how hard it is. The ball we used was a NIB Penny&Giles one which was very nice. I suspect a more well worn one will be worse, but still better than a consumer trackball.

In other news, A kensington trackball I used was "not bad". I still don't like ti's lack of bearings though.

[politie]

thanks, sounds good im looking forward to seeing more on it. L-Trac i find very well on small movements, if its better thats great.

[jacobolus]

I think it’s roughly comparable to the L-Trac on small movements in the X/Y directions and obviously much much better for rotation about the Z axis and in general much smoother for larger movements with a dramatically longer free spin time (but I think there might still be room for further improvement beyond that). I’ll have to try it with a real housing and a sensor actually installed to be totally sure.

[lkong]

I opened L-trac to see if there's any room to fit in 3 ball bearings with 3d printed frame.
The answer is no the bearing on the right side will overlap with trackball right key.

[Synnöve]

There is a bit of “scratchiness” for small slow movements. I’m not sure if this is inherent, or just something to do with my flimsy cardboard mockup, or perhaps something that could be improved by using different ball transfer units.

To elaborate on this and answer some questions other's have had: very small movements send a little vibration in to the trackball which makes it feel slightly scratchy, however the ball still rolls incredibly smooth, smoother than any solution that currently exists. The reason I say that is other solutions have friction that must be overcome when initiating a small movement, causing the trackball to jump slightly; rollers cause friction the most when the trackball must grind against all rollers at once, the Teflon bearings cause friction equally which makes motions more uniform but at the expense of micro movements of the trackball.

The transfer solution has none of those problems, rather, the only problems are noise and a small amount of vibration sent to the trackball (it's incredibly slight). Who knows, these problems might evaporate once a housing comes in to the equation.

[politie]

the scratchyness doesnt sound to bothering in the video.

[dorkvader]

I opened L-trac to see if there's any room to fit in 3 ball bearings with 3d printed frame.
The answer is no the bearing on the right side will overlap with trackball right key.

I was thinking about that, and decided it would be a lot easier to just move that microswitch elsewhere in the case.

That said, the CST trackballs are not the best targets (though it would improve their most obvious flaw: the bearings)

the scratchyness doesnt sound to bothering in the video.

It was barely noticeable. I really really liked it and I'm a "trackball-bearing snob".

[lkong]

I opened L-trac to see if there's any room to fit in 3 ball bearings with 3d printed frame.
The answer is no the bearing on the right side will overlap with trackball right key.

I was thinking about that, and decided it would be a lot easier to just move that microswitch elsewhere in the case.

That said, the CST trackballs are not the best targets (though it would improve their most obvious flaw: the bearings)

on a second thought, if i can find ball bearing units small enough, it is possible to put down three of them and secure them with sugru.

another concern would be cleaning, how hard it is to clean/maintain ball bearing units?

[MrJohnK]

I've been searching for a place to purchase the ball transfer bearings in 3mm or 5mm.  If anyone finds a way to buy them, please let me know.  I plan on revising my 3D printed trackball design to incorporate these.  It looks like much good work has been done so far and I can share my models when/if they get done and work.  It would probably take a 5-axis CNC to mill the socket properly, so 3D printing makes sense.  I only have a 3 axis CNC and milling the pockets for the bearings in the sides at an angle would be the challenge...or just work those by hand?

The only thing I can't quite figure out is why the need for two of my sensors verses one located right at the bottom of the ball.  What does twisting the ball do for you?  Maybe one located at the bottom for "normal" movements and one along the side to detect the twist?

I think it makes sense to incorporate some adjustment via set screws backing each bearing to achieve the optimal 2.4mm distance from sensor lens to ball.  I know it was a booger to get all three of my bearing lined up to that distance when I made my trackballs.  This also allows for centering the ball in the socket for any inconsistencies.  There is only about 0.2mm of tolerance to this measurement from ball to sensor, so getting exact dialed in distance is key to a solid motion reading.  I can imagine that getting two sensors along the side of the ball lined up at that distance would be challenging for sure.

I'm also giving serious thought to a Kickstarter campaign if I can get trackball design that I'm happy with that I can make more of in a reasonable assembly line (like casting the ball sockets and bodies).  This will help us all get the benefits of group buying.  I'm a fan of the thumb ball design, but not ruling out a center mounted option as well.

-John

[dorkvader]

I've been searching for a place to purchase the ball transfer bearings in 3mm or 5mm.  If anyone finds a way to buy them, please let me know.  I plan on revising my 3D printed trackball design to incorporate these.  It looks like much good work has been done so far and I can share my models when/if they get done and work.  It would probably take a 5-axis CNC to mill the socket properly, so 3D printing makes sense.  I only have a 3 axis CNC and milling the pockets for the bearings in the sides at an angle would be the challenge...or just work those by hand?

The only thing I can't quite figure out is why the need for two of my sensors verses one located right at the bottom of the ball.  What does twisting the ball do for you?  Maybe one located at the bottom for "normal" movements and one along the side to detect the twist?

I think it makes sense to incorporate some adjustment via set screws backing each bearing to achieve the optimal 2.4mm distance from sensor lens to ball.  I know it was a booger to get all three of my bearing lined up to that distance when I made my trackballs.  This also allows for centering the ball in the socket for any inconsistencies.  There is only about 0.2mm of tolerance to this measurement from ball to sensor, so getting exact dialed in distance is key to a solid motion reading.  I can imagine that getting two sensors along the side of the ball lined up at that distance would be challenging for sure.

I'm also giving serious thought to a Kickstarter campaign if I can get trackball design that I'm happy with that I can make more of in a reasonable assembly line (like casting the ball sockets and bodies).  This will help us all get the benefits of group buying.  I'm a fan of the thumb ball design, but not ruling out a center mounted option as well.

-John

Hey John! Good to see you here again.

The second sensor for "twist" is to give you a three-axis trackball instead of a normal two-axis one. This is for if you want a third input axis for whatever reason. It'd be hard to use in daily life, like for a volume control or scrollwheel or something, but if you can "twist" then you can manipulate objects in 3D a lot easier. Heck with a trackball, you could set it up to follow the movement of the ball 1:1 (and then use a keyboard or whatever for zoom / pan)

There has been a significant amount of debate as to where to adequately locate the second sensor. My instinct is the same as yours. To have one on the bottom, and one on the "side" in the space between ball transfer units.

The setscrews idea is a really good one.

[jacobolus]

I've been searching for a place to purchase the ball transfer bearings in 3mm or 5mm.  If anyone finds a way to buy them, please let me know.  I plan on revising my 3D printed trackball design to incorporate these.  It looks like much good work has been done so far and I can share my models when/if they get done and work.  It would probably take a 5-axis CNC to mill the socket properly, so 3D printing makes sense.  I only have a 3 axis CNC and milling the pockets for the bearings in the sides at an angle would be the challenge...or just work those by hand?

I’ve been very busy with random other stuff for the past month, but I’m going to try to start in on this again in earnest ASAP (and I’m sure emailing you with sensor protocol questions). I want to try to CNC cut it out of wood, on a 3-axis CNC router. I think I can do it by building some little jigs to hold the workpiece at 2-3 different angles so I can properly mill out the little pockets to hold the bearings and sensor. It’s going to take more elaborate setup than it would with a 5-axis mill, but I think it should be doable.

The only thing I can't quite figure out is why the need for two of my sensors verses one located right at the bottom of the ball.  What does twisting the ball do for you?  Maybe one located at the bottom for "normal" movements and one along the side to detect the twist?

Personally I want to record the full 3-dimensional rotation of the ball, so I can use it for rotating objects in a 3-d environment (like a cad program or a data visualization or whatever).

If you just want a regular trackball, one sensor is sufficient.

[MrJohnK]

Thanks for clarifying the dual sensor design.  Moving something in 3D like that sounds cool.  I imagine that twisting could be a unique way to access scrolling too.

I have made a new version of the ADNS-9800 sensor board as related to a custom project I was contract for.  It is much smaller and only works with 3.3v signalling (Teensy 3.0/3.1, Arduino Due or other similar 3.3v ARM based systems).  The circuit board is rectangle shaped with measurements of 30mm x 17.5mm.  It has mounting holes that are 25mm apart for use with standard 2-56 screws.  Since this was a custom project, I don't have stock on the circuit board, but at least the design is done and it would fit in tighter spaces, such as along the side of a ball socket.  The lens is actually slightly wider than the circuit board itself.  I have not decided if I'll make this one available in addition to the round board.  But, if you guys think you need a smaller version of the ADNS-9800 sensor board, it can be done.

Feel free to send me questions about the sensors.

I noticed many vendors selling 8mm load transfer bearings on Ebay and AliExpress.  If I cannot get a source on the smaller ones, I may give these a shot.  They are about $1 each or less.

-Thanks

John

[zyxxiforr]

It'd be hard to use in daily life, like for a volume control or scrollwheel or something

Kensington slimblade has twisting working as scroll wheel and it works exceptionally well IMO.

[dorkvader]

It'd be hard to use in daily life, like for a volume control or scrollwheel or something

Kensington slimblade has twisting working as scroll wheel and it works exceptionally well IMO.

The slimblades use a scroll ring around the side of the trackball: a much different design (that does indeed work nicely).

The idea on this one is to twist the trackball itself. You could use it for scrolling, but it'd also happen a lot when you're just moving the cursor around, making it a bad solution, imo.

[jacobolus]

The slimblades use a scroll ring around the side of the trackball: a much different design (that does indeed work nicely).

The idea on this one is to twist the trackball itself. You could use it for scrolling, but it'd also happen a lot when you're just moving the cursor around, making it a bad solution, imo.

Actually new Kensington Slimblades include 2 sensors and interpret twists of the ball as scrolling.

http://brightcove.vo.llnwd.net/e1/uds/pd/1414329509001/1414329509001_1544858659001_819f15a3-3f7f-4362-9249-be454ab48957.mp4

[zyxxiforr]

The slimblades use a scroll ring around the side of the trackball: a much different design (that does indeed work nicely).

The idea on this one is to twist the trackball itself. You could use it for scrolling, but it'd also happen a lot when you're just moving the cursor around, making it a bad solution, imo.

Actually new Kensington Slimblades include 2 sensors and interpret twists of the ball as scrolling.

http://brightcove.vo.llnwd.net/e1/uds/pd/1414329509001/1414329509001_1544858659001_819f15a3-3f7f-4362-9249-be454ab48957.mp4

Exactly. And in the 9 months of using it, it never happened when I didn't want it to, so it's definitely possible to implement. I think it just checks if there's any rotation (with some margin of error) around any other axis than Z, and if there is, it's interpreted as cursor movement only. (You can't move the cursor and scroll at the same time)

[MrJohnK]

Does anyone had the skills to write an operating system driver (MS Windows and Mac for instance) that would allow the settings of a custom trackball to be adjusted?  The trackballs I've made have been Arduino powered, so to adjust them, you have to change the source code and upload new firmware.  This isn't how the average gaming mouse is done since they usually have a fancy little program running in the system tray allowing the settings to be adjusted on the fly.  I'm still working on the "commercial design" of my existing trackball for mass production.  It has a mini keyboard of 15 Chery MX switches that can be programmed for any mouse button, scroll or keyboard stroke, assuming you want to recompile the Arduino code running on the Teensy 2.0 inside. 

Thanks.

-John

[jacobolus]

Does anyone had the skills to write an operating system driver (MS Windows and Mac for instance) that would allow the settings of a custom trackball to be adjusted?

What kind of settings are you talking about? I was planning on building a trackball into a keyboard and making the firmware support some amount of custom behavior via typing the appropriate keystrokes (e.g. hold down a specific modifier to give the trackball the behavior of repeating the previous keypress for each unit of rotation / repeat the 'delete' key for rotation the other way, or hold down another modifier key to turn the trackball into a volume knob).

What would a custom operating system driver do exactly?

[MrJohnK]

I think I've got a solution.  My trackball has 15 programmable buttons, so am looking for a graphical interface to program button profiles.  It would also be nice to set the default CPI and save all this in EEPROM memory on the MCU so that it boots up this way even after being powered down. 

Since I posted about it, I've found Mono (http://www.mono-project.com/).  I plan to learn how to make a graphical interface with that and communicate over the virtual serial port with the Teensy running the trackball.  This is the initial structure for mine that will eventually be used in the one I plan to sell.  Communicating over the serial port shouldn't be too hard.  I just have to learn the ins-and-outs of programming a GUI with Mono (see for more: http://www.mono-project.com/docs/gui/gui-toolkits/).  Using Mono in this way should make for a cross-platform compatible trackball programming interface for Windows, Mac and Linux.

Now, on to load transfer bearings...  Those one that you all are testing with from Alwayse are almost $50 each. 

I'm planning on making a modular ball socket that can be mounted for either thumb or middle finger(s) usage.  Then, I just need to make hand rests for each application that will accept the modular ball socket.  This approach also allows the socket to be turned to match the X & Y axis to a comfortable position for your individual application.  I can also engineer in a mounting for a second sensor for the twist detection, an optional add-on.  Any thoughts on this approach?  Good bad or ugly?

-John

[Grim Fandango]

There aren't enough trackballs on the market.

Also, forefinger operated trackballs are evil. I feel sad that there are only two thumb-operated models on the market, and they're both mediocre.

Somebody said Logitech hates thumb-trackballs, and sics their lawyers on anyone who markets one. If true, the entire patent system and anything tangentially related needs to die a fiery, painful death.

I feel the same way. I really love thumb trackballs. I am using one at work and have done so for years. if you have to go across multiple screens and constantly select and drag things, then a trackball is arguably faster and easier to use (for me at least). People might also be surprised with the speed and accuracy with which you can use it.

Anyway, I really wish there were more options to choose from. Such a shame that there is barely anything else than the logitech M570. Which is surprising given that it is actually fairly popular. The only thing like it I have found is a rather unimpressive Japanese one.

[jacobolus]

Can someone who likes thumb trackballs explain exactly where they want to position their thumb, and what thumb motions they’d ideally want to use to move the ball in each direction? Do current thumb trackballs get the housing shape right? What about the sensor position or interpretation of rotations into X/Y movements? Do they put the finger buttons in a good place? How many finger buttons could be supported on a trackball actually shaped like a hand? maybe 1-2 per other finger? (Or should the pinky not get buttons?) How big should the ball be?

With a totally custom-designed trackball, any arbitrary rotational axes could be interpreted as X/Y (would probably work best if the two axes were orthogonal, but that's not strictly necessary), and the amount of rotation required to move one pixel in each direction could also be different, if e.g. the thumb has an easier time moving along one axis than another.

I’ve only briefly used a thumb trackball, and didn’t really like it, but if it were optimized for the actual strength/agility of the thumb, along with a housing that fit the rest of the hand properly, I could imagine it being nice to use. (I still think the thumb moves really well along one horizontal-ish direction, and quite poorly in the vertical-ish direction unless you start moving the whole hand around, but maybe that’s just me.)

[Grim Fandango]

Can someone who likes thumb trackballs explain exactly where they want to position their thumb, and what thumb motions they’d ideally want to use to move the ball in each direction? Do current thumb trackballs get the housing shape right? What about the sensor position or interpretation of rotations into X/Y movements? Do they put the finger buttons in a good place? How many finger buttons could be supported on a trackball actually shaped like a hand? maybe 1-2 per other finger? (Or should the pinky not get buttons?) How big should the ball be?

With a totally custom-designed trackball, any arbitrary rotational axes could be interpreted as X/Y (would probably work best if the two axes were orthogonal, but that's not strictly necessary), and the amount of rotation required to move one pixel in each direction could also be different, if e.g. the thumb has an easier time moving along one axis than another.

I’ve only briefly used a thumb trackball, and didn’t really like it, but if it were optimized for the actual strength/agility of the thumb, along with a housing that fit the rest of the hand properly, I could imagine it being nice to use. (I still think the thumb moves really well along one horizontal-ish direction, and quite poorly in the vertical-ish direction unless you start moving the whole hand around, but maybe that’s just me.)

I have used thumb trackballs for a long time. I use them at work where I often have to select, drag, and move the cursor across multiple screens. Even though I also love normal mice, and use those at home and for gaming, I really like the comfort and ease of use of thumb-trackballs. Some people question whether people using trackballs at work can be as fast and productive as people using a mouse. It depends on the type of work. It is faster for some actions that you might need to perform, slower for others. For general use (browsing , working in text editors and so on) the difference in speed is negligible for someone who is used to trackball mice. It is not at all clumsy, and you can have the cursor stop at an exact location in the screen with about as much ease as a mouse once you adjust to it.

However, and this goes for any peripheral, there is no guarantee that you will like them. Something that works for me might not work for you and vice versa. We all have different habits, preferences and anatomies. For example, I have used larger finger trackball, and though I thought they were ok they were not as comfortable for me as a thumb trackball. Others swear by them.

We can limit the discussion to Logitech thumb trackballs, since there are few others on the market, and most of those are either no longer in production or would have to be important from Asia. Personally, I find them intuitive to use. Moving the cursor vertically as well as horizontally works well for me. I would not want to change the simple intuitive control of the scroll wheel as it is now. You might be able to do something that causes less fatigue (for example, different sensitivities for left-right and up-down), but I doubt it would make a big difference, and you lose some of the ease and intuitiveness of the trackball.

The best thing I can say about them is that I can keep my hand perfectly relaxed when using the trackball. My thumb does not get tired because I can control the trackball with enough accuracy that I can crank up the sensitivity, meaning that the amount of thumb movement needed to manipulate the cursor is really minimal. I like the button placement on the logitech thumb trackballs. The only thing I do not like is that in the newest version (the m570) they decided to make the part where you rest you hand a little smalller, where I would have like it to have been a little bigger instead. Still, the shape works well for me.

I noticed something when I saw others using my trackball. Some people put their hand on it and use it without a problem (someone sat down at my desk and wrote an e-mail, and she used it as though she uses one at her own desk), and others are extremely clumsy with it and unable to move the cursor the way they want. I have seen this a couple of times now, and think that trackballs are just more intuitive to some than they are for others. One person even kept moving the trackball (the same way you would move a mouse) when the cursor did not go where she wanted it to.

[MrJohnK]

I like the way these from Holo Pack look: http://www.holo-pack.com.tw/en/profile.jspx?no=A24003

They would allow for height adjustment and alignment of the ball in the socket.  Overall size is 8mm.  I think I'll try to get a few samples or a small quantity if the price is reasonable. 

-John

[JackMills]

There aren't enough trackballs on the market.

Also, forefinger operated trackballs are evil. I feel sad that there are only two thumb-operated models on the market, and they're both mediocre.

Somebody said Logitech hates thumb-trackballs, and sics their lawyers on anyone who markets one. If true, the entire patent system and anything tangentially related needs to die a fiery, painful death.

I feel the same way. I really love thumb trackballs. I am using one at work and have done so for years. if you have to go across multiple screens and constantly select and drag things, then a trackball is arguably faster and easier to use (for me at least). People might also be surprised with the speed and accuracy with which you can use it.

Anyway, I really wish there were more options to choose from. Such a shame that there is barely anything else than the logitech M570. Which is surprising given that it is actually fairly popular. The only thing like it I have found is a rather unimpressive Japanese one.

If anybody in this thread wants to get into the thumb trackball design, please don't overlook the left-handed users. I'm not left handed but I do use my left hand to control my mouse, it is a habit that I don't want to give up. But after trying out the logitech M570 I want to switch to using a trackball, but this will force me back to using my right hand for moving the cursor. I know I could get a regular trackball, but I haven(t found one with ergonomics that I like or the size to be easy transportable.

[Binge]

Gotta say this has been a fascinating read, and I am inspired by passion of fellow hackers looking to play with some balls.

 

Needless to say, subbed.

[CPTBadAss]

I am inspired by passion of fellow hackers looking to play with some balls.

You're so sophomoric Binge!

....I too eagerly await the progression of this project.

[AKmalamute]

I had mostly forgotten about this project because it looked like everyone was ignoring the thumb-use cases.

I can say that I used to use a mouse reluctantly except for gaming ... but that was still awkward because you needed so much space to play ... then (and this would have been in the early '90s) I stayed at my uncle's house, who had an early logitech thumb-trackball, and I played through Wolfenstein (he had the full walkthrough) and fell in love.

I have since found that thumb trackballs let me game with the pointer in my lap, closer to my body than the keyboard which is also my lap. As to what direction I move the ball ... that's decided by firmware, right? Unless I'm missing the question, moving the cursor requires a certain motion from the ball, and that's done by pushing or pulling with the thumb.

As to left-handed designs ... It's a valid consideration. Do the shops that 3D print, offer to invert the image? Same firmware, controller, parts, you'd just need to make everything be flipped left-to-right, and you'd have a left handed housing.

[damorgue]

As to left-handed designs ... It's a valid consideration. Do the shops that 3D print, offer to invert the image? Same firmware, controller, parts, you'd just need to make everything be flipped left-to-right, and you'd have a left handed housing.

That can be done in about 5 seconds in software. It can be done either by mirroring the entire thing or scaling it to -100% in one direction depending on whether it is parametric or in a polygon format. It is trivial either way.

[dorkvader]

As to left-handed designs ... It's a valid consideration. Do the shops that 3D print, offer to invert the image? Same firmware, controller, parts, you'd just need to make everything be flipped left-to-right, and you'd have a left handed housing.

That can be done in about 5 seconds in software. It can be done either by mirroring the entire thing or scaling it to -100% in one direction depending on whether it is parametric or in a polygon format. It is trivial either way.

You can even make one design for both cases, like what IBM did with the palmrest on the spaceball.

Just make a "module" with buttons and a trackball on one edge, and a connector palmrest that can fit on either side. SOmethign to think about, but if we are 3D printing them anyway it won't save time/cost.

[MrJohnK]

Flipping an axis in code is a simple as multiplying the results by -1.  This can be done for X or Y. 

I'm planning to make a modular ball socket that can be dropped into a custom shell for left, right or middle control.  The shell would need buttons, but that isn't too hard. 

I also completed a Java app this weekend that allows for a GUI to control settings of my Teensy based trackball.  It uses the virtual USB serial port to communicate with the Teensy.  It stores the profiles in the microcontroller EEPROM, so they travel inside the trackball and survive the power being cut.  At this point, looking at 10 custom profiles.  Java App Screenshot: https://www.flickr.com/photos/8465641@N07/14937404399/

It looks like those Holo Pack bearings I linked previously cost about $4-$8 each in small quantities.  The lower cost being all plastic and the higher cost being all stainless.  I'll have to order them from Taiwan directly with a $100 minimum order amount and direct bank account transfer only (another $50 fee).  I'll probably hold off ordering these to prototype until funding on the Kickstarter project.  For now, I'll just stick with the metal nub bearings I'm using now to make the newer trackball version.

John

[dorkvader]

Flipping an axis in code is a simple as multiplying the results by -1.  This can be done for X or Y. 

I'm planning to make a modular ball socket that can be dropped into a custom shell for left, right or middle control.  The shell would need buttons, but that isn't too hard. 

I also completed a Java app this weekend that allows for a GUI to control settings of my Teensy based trackball.  It uses the virtual USB serial port to communicate with the Teensy.  It stores the profiles in the microcontroller EEPROM, so they travel inside the trackball and survive the power being cut.  At this point, looking at 10 custom profiles.  Java App Screenshot: https://www.flickr.com/photos/8465641@N07/14937404399/

It looks like those Holo Pack bearings I linked previously cost about $4-$8 each in small quantities.  The lower cost being all plastic and the higher cost being all stainless.  I'll have to order them from Taiwan directly with a $100 minimum order amount and direct bank account transfer only (another $50 fee).  I'll probably hold off ordering these to prototype until funding on the Kickstarter project.  For now, I'll just stick with the metal nub bearings I'm using now to make the newer trackball version.

John

I wonder if you can get a sample before you commit fully to an order.

This is truly excellent: I am glad I didn't use up my sensor(s) on a "normal" trackball build already. I really enjoy trackballs, but the "comfortable" ones have bad bearings, and the well-bearing'd ones are not comfortable. It would be nice to kill two birds with one stone here for the "optimal" input device.

[MrJohnK]

I did ask for samples of the Holo Pack bearings, but the response from the company representative was that they do not provide samples.

-John