Honey, I developed FFB joystick (DIY)

[walmis]

9 hours ago, Bard_the_Bowman said:

No I haven’t calibrated, but parts I have printed have been remarkably close to spec.  I got very lucky with my Ender 3 Pro and it seems to be remarkably accurate overall without calibration.  For example the main bearings fit on the gimbal arms perfectly with no major issues at all.  I think that my impact-heavy pressing strategy was probably the biggest issue.  This filament has incredible tensile strength but based on my testing does seem to be fairly vulnerable to strong impact force.  So I think that if I carefully use a press next time and make sure that any imperfections in the print are sanded away, it should be fine.  I also printed some reinforcement parts* that I superglued onto the cracking blocks, and I think I might add those straightaway this time as an additional measure.  
 

*Note that the paper-board reinforcement on the right side lower block in the picture is not what I’m referring to here, that was something I slapped on the lower block real quick as a preventative measure.  

 

 

The thing is that the "inside" holes are usually smaller than the "outside" holes. That explains why main bearings went in OK, but the smaller ones had a too tight fit. I usually add a 0.4mm tolerance for the inside holes. ABS has the worst shrinkage so makes the problem worse. PETG is best on the shrinkage aspects so parts are more dimensionally accurate. Not sure about your material.

[Thadiun Okona]

22 hours ago, Bard_the_Bowman said:

No I haven’t calibrated, but parts I have printed have been remarkably close to spec.  I got very lucky with my Ender 3 Pro and it seems to be remarkably accurate overall without calibration.  For example the main bearings fit on the gimbal arms perfectly with no major issues at all.  I think that my impact-heavy pressing strategy was probably the biggest issue.  This filament has incredible tensile strength but based on my testing does seem to be fairly vulnerable to strong impact force.  So I think that if I carefully use a press next time and make sure that any imperfections in the print are sanded away, it should be fine.  I also printed some reinforcement parts* that I superglued onto the cracking blocks, and I think I might add those straightaway this time as an additional measure.  
 

*Note that the paper-board reinforcement on the right side lower block in the picture is not what I’m referring to here, that was something I slapped on the lower block real quick as a preventative measure.  

 

 

Print a 20mm cube. If it measures 20.00 in x/y and z you are fine.  If it's off at all you should calibrate your printer. If this is not relevant to you it may be to others.  

 

It's a pain but well worth doing, even it's only a tiny bit off.. stock settings are usually just off enough to make bearings have trouble.  You will need to flash the settings on your printer, so first you need a 3rd party program that can do it.  I use Pronterface.. lightweight, free and does what you need it to.

 

https://www.pronterface.com/

 

Start by calibrating the extruder, everything downstream is affected by that so ensure it's proper first.

 

https://mattshub.com/blogs/blog/extruder-calibration 

 

Next follow this guide (has links to test objects in guide..)  I found 20mm cubes to be great but also used 100mm xyz thing too

 

https://all3dp.com/2/how-to-calibrate-a-3d-printer-simply-explained/

 

Mention was made of oversizeing holes when designing, same applies to undersizing bosses for the same reason.. interpolated circles in modeling programs either inscribe or circumscribe a series of straight line segments because curves don't exist in digital format.

 

Another thing that might be causing issue is the 'elephant foot' that may be forming a tighter lip on your hole than the nominal id. That said, you can also heat the part/bearing to make it more compliant to pressing it in

[TomVR]

On 10/4/2021 at 7:05 AM, Bard_the_Bowman said:

Lying face down on bed, printed with Polymax PC (polycarbonate hybrid).  I think the problem was that I pounded bearings into the blocks with rubber hammer instead of pressing.  Also maybe should have sanded inside of blocks before installing bearings.

PC is great at impact strength but not so great at cycling and fatigue, a PETG/NYLON blend might be better suited

Also Im trying to figure out how to compactly compound two pulleys onto the design to try and get a 25:1 ratio (closer to the MSFFB2 gearing) instead of the current 5:1. I am semi tempted to just reduce the gearing already in the MSFFB2 design but I guess it'll still feel pretty clunky like the original stick does.

edit:

I wonder if a page can be taken out of the hobbyist robotics community with a cycloidal drive for the motor output, very chungus though

 

 

 

Edited October 5, 2021 by TomVR

[Bard_the_Bowman]

2 hours ago, TomVR said:

Also Im trying to figure out how to compactly compound two pulleys onto the design

Onto what design?  How to do that depends entirely on the overall form factor of your build I suppose.  For me with my housing I built, it will be a matter of using narrower belts and using every tiny little bit of space I have lmao. Might have to space the power supply out a little further.  
 

And I’ve decided to leave pitch alone at this point, my pitch motor is an absolute unit and after spending some actual flight time with it I’m content with pitch force.  But the roll motor either needs replaced or reduced

[TomVR]

I am still thinking of using the existing MSFBB2 motors themselves instead of doing a full swap to new motors, but they are pretty whimpy 

 

edit: Starting to put my all 3d printed housing together based on Walmis's laser cut files.

Also thinking about how I can stack the pulleys, not sure where to put the motor input now though as there isn't really a lot of space, and want to avoid gearing and reintroducing the grinding feeling into the stick. 

 

Also more reading on Cycloidal drives seems that they would produce a similar "clunk clunk clunk" effect while being backdriven so other than their compactness they aren't really suitable. 

 

Travelling again for work so don't have a chance to actually open up my MSFFB and see what the as designed gear reduction is, reading conflicting posts that its either 25:1 or 16:1

Edited October 6, 2021 by TomVR

[Bard_the_Bowman]

Turns out I think I just didn’t have my bed and nozzle heat totally tuned right before.  Printed again and the blocks are basically dimensionally perfect.  Bearings pressed in very nicely.  This filament is difficult and needs just the right settings apparently.  

[98abaile]

So I'm trying to figure out how to even order the PCB (because they sure as hell don't make it clear) and I managed to get a 3D view of the PCB (because I also can't figure out how to view the backside of it in 2D). That's when I noticed that there are components on the board (and I don't know what they are because I can't figure out how to find out.....) that only appear to be connected to the ground plane. Is the board on the github working or is it unfinished?

https://oshwlab.com/olukelo/btn8982-three-phase-controller_copy_copy_copy#

https://easyeda.com/editor#id=961f5e5f9c364d169f6ebd1324297b1b

These are both sides of the same area, all three components are connected to the ground plane in the left image, but all the top vias are connected to nothing. If they're not required, can they just be removed?

[propeler]

10 hours ago, 98abaile said:

Is the board on the github working or is it unfinished?

It is finished, it is working.

 

10 hours ago, 98abaile said:

but all the top vias are connected to nothing

No they don't. All components are conected.

 

10 hours ago, 98abaile said:

how to even order the PCB

 

[propeler]

[Bard_the_Bowman]

4 hours ago, propeler said:

Oh myyyyy.  That is a thing of beauty!   Have you tested it?  How much force are you getting?  Are the motors smooth enough?

That is just beautiful

[Mabond96]

 

17 hours ago, propeler said:

That looks ace! What power supply are you using? 

[jpdesvals]

hello Walmis , 

You did the software ( FFB tool )on the pictures by yourself ? That is amazing ! 

I have no knowelge in programing but is there a way to use this kind of sofware to modify the default spring , damp etc force of my sidewinder force feedback 2 as on windows 10 there is no support anymore fore the sofware side winder  game manager 4.0 and even if the force feedback work correctly you can not adjust theses effect before playing dcs or  il2 for exemple.

if sSIMFFB is used then you loose all of the ingame effects .

anyway thanks for your work and i hope you will be able to sell your kits soon !! 

On 9/30/2021 at 10:15 PM, walmis said:

Hey guys, an update on my FFB development. 

Ordered FFB controller boards and boards for the motor control, redesigned the gimbal from scratch, so not to be bound by copyrights. Also spent some time on optimizing the design, so it could be built rather quickly. Finding sources of parts is also a challenge during these times of various shortages.

I'm hoping I will be able to start selling these electronics kits for anyone that wants to build own FFB joysticks. And later some finished builds, similarily how Prusa started with 3D printers

 

Added some prototype photos

 

 

[walmis]

7 hours ago, jpdesvals said:

hello Walmis , 

You did the software ( FFB tool )on the pictures by yourself ? That is amazing ! 

I have no knowelge in programing but is there a way to use this kind of sofware to modify the default spring , damp etc force of my sidewinder force feedback 2 as on windows 10 there is no support anymore fore the sofware side winder  game manager 4.0 and even if the force feedback work correctly you can not adjust theses effect before playing dcs or  il2 for exemple.

if sSIMFFB is used then you loose all of the ingame effects .

anyway thanks for your work and i hope you will be able to sell your kits soon !! 

 

Hello there,

Yes but sadly the app does not use any standard communications, it interfaces directly with my FFB joystick's firmware, so no way to affect MSFFB2.

[Bard_the_Bowman]

I really want to know if propeler has tested that latest build!  Such a tease to waltz in here, drop that pic and leave lmao

[Bard_the_Bowman]

Here's some photos of my propeler-gimbal FFB2 frankenstick, I realized I've never actually posted any here. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

More at : https://imgur.com/a/jKs44Ma

I still haven't got around to doing a write-up.  But long story short I improved upon existing FFB2 mods by locating some very good mosfets that can handle a lot of power, and on pitch axis which is stronger I'm pulling about 8 amps vs. about 0.7 for the factory motors.  I used Glentek servomotors which are excellent and have virtually no cogging. 

Edited October 20, 2021 by Bard_the_Bowman

[Bard_the_Bowman]

Has anyone considered trying to design a direct-drive FFB stick? Now that I'm mostly done with Frankenstick, naturally I am thinking toward the future.    I'm not sure whether motors with high enough size to performance ratio are easily available or not, but it seems like a direct-drive setup would allow for a drastically simpler gimbal design that could be made with all-metal components with relatively simple fabrication skills.  basically you'd only have two main components, a pitch axis driven by a stationary motor, and on that pitch axis would be mounted the stick and a roll motor, which would be free to move with the pitch axis.  Would probably be easier to draw it than explain it.  But I think that theoretically that could be a very simple design with very few moving parts compared to belt driven designs that need a complicated gimbal.  

And obviously the pitch motor would be doing more work since it has to move not only the stick but also the roll motor, but I think that could be easily overcome, because since the pitch motor is stationary you can use one of arbitrary size.

Also I think such a design could result in very good axis separation with barely any cross-talk.

Edited October 20, 2021 by Bard_the_Bowman

[walmis]

Here is my first prototype design. Dual pulley reduction, lots of torque  Some parts are missing, didn't maintain this project for a while. With anticogged motors it produces very smooth operation. There are a few issues with belt slipping. I initially I used HTD3M belts, but HTD5M with better belt tensioning should solve that.

I think the biggest problem with FFB with fancy motors is the huge price tag and sourcing. I did some tradeoffs and went for more ubiquitous motors.

 

 

[propeler]

12 hours ago, Bard_the_Bowman said:

Has anyone considered trying to design a direct-drive FFB stick?

 

It is not so crazy idea, but I did not found on the market any affordable motor to build such direct drive joystick based on it.

Edited October 21, 2021 by propeler

[TomVR]

19 hours ago, Bard_the_Bowman said:

Has anyone considered trying to design a direct-drive FFB stick? Now that I'm mostly done with Frankenstick, naturally I am thinking toward the future.    I'm not sure whether motors with high enough size to performance ratio are easily available or not, but it seems like a direct-drive setup would allow for a drastically simpler gimbal design that could be made with all-metal components with relatively simple fabrication skills.  basically you'd only have two main components, a pitch axis driven by a stationary motor, and on that pitch axis would be mounted the stick and a roll motor, which would be free to move with the pitch axis.  Would probably be easier to draw it than explain it.  But I think that theoretically that could be a very simple design with very few moving parts compared to belt driven designs that need a complicated gimbal.  

And obviously the pitch motor would be doing more work since it has to move not only the stick but also the roll motor, but I think that could be easily overcome, because since the pitch motor is stationary you can use one of arbitrary size.

Also I think such a design could result in very good axis separation with barely any cross-talk.

 

I think direct drive is also a safety issue, a belt will slip before your shoulder with dislocate for example.

Also what model of MOSFETs did you replace on your pcb?

[Bard_the_Bowman]

8 hours ago, propeler said:

 

It is not so crazy idea, but I did not found on the market any affordable motor to build such direct drive joystick based on it.

 

Where did you find those images?  Those look like very interesting designs.  But yes, I imagine suitable motors are probably pretty expensive because generally between (1) cheap (2) powerful and (3) small you can only pick two choices.  Although like I said earlier, I think only ONE of them would necessarily need to be expensive.  Because the stationary one could be arbitrarily big assuming you design the case around it lol

[propeler]

32 minutes ago, Bard_the_Bowman said:

Where did you find those images?

Somawhere  in internet space  Do not remember. I scanned unmeasurable quantity of resources looking for good gimbal scheme

[foobolt]

https://motion-control.wittenstein.de/en-en/expertise/applications/t-50/

The F-16 type stick can be found on this site.
Maybe it's not a product for general users...

[Bard_the_Bowman]

15 hours ago, TomVR said:

I think direct drive is also a safety issue, a belt will slip before your shoulder with dislocate for example.

Also what model of MOSFETs did you replace on your pcb?

I think you’d have to spend a quite large amount of money for it to have enough force to be a significant safety issue.  But yes you would probably want to have an easily accessible emergency kill switch like most direct drive racing wheels do.

 

this is the mosfet I used:

https://www.mouser.com/ProductDetail/vishay/sqj500aep-t1_ge3/?qs=jHkklCh7amiGIJk8viCi1A%3D%3D&countrycode=US&currencycode=USD

did not end up needing Schottky diodes like the traditional quad-power mod.  I did add some capacitors as per that mod tho.

That mosfet is probably about as good as it gets for surface mount P-N arrays. If a person wanted to use even more powerful motors than my Glenteks you would need to do a breakout board with discrete P and N channel mosfets.  Would actually be fairly easy to do I think.  And there’s at least one guy who apparently has tried the breakout board approach successfully, albeit probably unnecessarily.  Cause if he’d used my mosfets he’d have been fine as his motors draw less current than mine.

Edited October 22, 2021 by Bard_the_Bowman

[Skopro_PL]

Sign me up when you start making completed production units. Might be interested in a kit, I just don't build stuff anymore. I've been through 2 ffb sticks and although I have a virpil stick it's just been sitting here for over a year. 

[slimheli]

Okay. So I just want to summarize where things are at here. @walmis is working on building a kit that is basically a full re-design of @propeler's original design. I've been working on my own FFB design using ODrive and the MS FFB gimbal as inspiration for way too long and the designs I'm seeing here blow mine out of the water.

So if I wanted to try building this, which repository would hold the best CAD's for me? Walmis' repository just has a STEP file for the box. Propeler's repository has the first version of his design, but it looks like he moved way past that with the latest iteration.

Anyone have some input as to where to begin? I already have an ODrive and two of the D5065 motors with a good power supply. And I've got a 3D printer.

Thanks for all the effort you guys have put into this!