Slartibartfast

Nice! Could you make the specifications available such that I could knock one together myself? I've actually got a 3D printer on it's way now so hopefully I'll be able to get a test rig up and running and finally start smashing out some code :thumbup:

Thanks Drakoz, that's all pretty much as I expected, other than the stick being required to calculate its own 'spring-force'. Presumably it can't be that hard (it's no doubt just a quadratic function of the difference between the stick's position and its 'centre') but still I thought it would be part of the simulator's job. As for aerodynamic forces, is the stick really required to determine these as well??? I really would have thought aerodynamic forces would have been part of the sim's job description. Again, we can derive a crude algorithm based on airspeed and possibly AoA but really, what is the simulator actually doing in order to 'support' FFB? In regard to being "beyond us" I was referring to implementing our own motor drive controller itself. As I said, if it's as simple as modulating a single voltage I'm sure we'll have no problem but I was trying to imagine what might cost the $500 Berniyh mentioned.

Thanks Drakoz, that is a very detailed and much appreciated post. I had been thinking that the force required to be exerted on the stick would essentially be calculated by the simulator itself and all we'd need to do would be to fulfill these requests. I had thought these requests would basically just be vectors (as in X by this much and Y by this much) along with a handful of pre-canned sequences such as saw-tooth waves and momentary spikes. That sort of thing. I take it this is not the case. In terms of implementing a motor drive controller if modulating a single voltage is all that is required I'm sure it wouldn't be too hard to do our selves. In fact in my tinkering over the past few days I've actually already done as much using a L293D ;) Here's a good tutorial I followed about using the L293D with an Arduino. Obviously if the amount of current drawn by the motor would fry the L293D chip we'd need something in between but surely we'd be able to find a way. However, if the motor controller is required to monitor the position of the rotor and interpolate between poles and God knows what else, then perhaps it will be beyond us.

Fair enough. So I've only worked with RC servos before (like the one pictured above) and the issue with those servos is that while you have good control over the desired deflection (that is where you want the motor to turn to) you don't have control over how hard it "tries" to get there. For use in the FFB system all we care about is the current deflection and we wish to dictate the amount of 'drive' the motor is to apply accordingly. We actually have no interest in instructing the servo to head to a given position as such. Will these motors allow us to instruct them to 'pull' with a given force regardless of their orientation? Also, will we be able to get a reading from them as to their current position? If they do give us such a reading we may be able to use this to actually read the joystick's position itself and save us needing to add hall effect sensors (or whatever) to the joystick. That would definitely be a win! I'm not so sure that is cost effective. We're looking to build the entire project 300€, and while the 300€ isn't a fixed limit spending 600€ on motors alone probably isn't feasible, though 20 Nm is a staggering amount of torque! I was imagining something in the order of 1/100th this much torque, so their may well be something in the acceptable range. What does the 'control unit' do? I take it that's something we can't build our self?

What do you mean by servo, just so we're all clear on what you're talking about. Do you just mean 'motor' or are you referring to something different? I take it you're not talking about one of these things: (which is what I think of when you say "servo") The reason gearing is necessary is because while electric motors can be surprisingly powerful generally speaking they're actually only capable of providing a remarkably small amount of 'native' torque. In fact it is surprising just how little torque is provided by most electric motors. They are however capable of spinning at truly impressive speeds (with virtually no loss in torque), which is what gives rise to their spectacular performance. I know electric car's are famous for the amount of torque they have available but it's not actually the amount of direct torque the motor produces that's impressive, it's more how much the torque can be multiplied (threw gearing) that is impressive and it is this same multiplication we are intending to leverage our selves. Given the motor is capable of spinning at many 1000's of RPM with virtually no loss in torque and we only need the driven element to rotate at say 100 RPM we may as well implement a 10:1 or 100:1 reduction and thus achieve a massive boost in torque (or rather only require a motor capable of producing but a fraction of the amount of torque we would otherwise need). You may think only a small amount of torque is necessary and that we are over delivering in this project, which may be the case for your needs, but I dare say doing away with the reduction gearing all together would reduce the torque at the stick so severely that even your meager requirements would fail to be satisfied. Now there are "Direct Drive Torque Motors" on the market and perhaps these are used in some sim wheels but generally speaking such motors are not really feasible to your average consumer. If you have access to such motors than I'm sure we would love to hear about it but unless you have the buying power of one of the world leading manufactures I wouldn't think such options would be viable –though I would love to be proven wrong. Do you know of any such direct drive products that would be suitable? I would be most keen to investigate the idea but I am not aware of any such motors that would be viable.

I think Berhiyh is talking about 'slop' or 'back-lash'. I don't think it will be much of a problem and if anything the reduction will reduce the 'cogging' felt from the motor itself but I think that's what Berhiyh is referring to.

Thanks for that Metal. I take it those instructions are for loading the program onto an ATMega32u4 Arduino by using a separate Arduino Uno in place of a boot-loader. Given I have a Teency 2.0 rather than a Pro Micro can't I just load the program directly onto the board using the built in boot loader? As it is I did manage to load something onto the Teency and once it started running a device called "Big Block" appeared as a game controller: However I have no way of interacting with it but I assume this is the program running as it should. So while it seems to be functioning fine I don't relay have a means to do anything with it at present. While I appreciate the final design of the joystick is still a ways off I was hoping for some means to assemble a test rig such that I could start hacking code. I assume I'll need a couple of potentiometers, a couple of motors and no doubt a fist full of resisters and things but being a code monkey I get all confuddeled when it comes to this 'meat-space' malarkey :doh: So what would be the best way to interact with this bad boy?

Yeah, I can't actually find these instructions :( Do you reckon you could put all the relevant information in the first post so we'll all know where to find it. It would make things a lot eaiser to follow. All the best, (: Edit: Oh, I see you're not the OP. Any chance you could incorporate the relevant information into your first post anyhow? As it is there is now a awful lot of information in this thread and while I've read the whole lot twice now I've obviously missed the critical bits ;).

Well, it has just arrived :thumbup: Now to work out what to do with it :( Is anyone able to give me guidance as to what I actually need to do to get started? :joystick: Basically I just want to set up a simple test rig so I can start attacking bits with a debugger :smartass:

I for one am keen to start working on the software but to be honest I'm kind of a bit stumped on how to get started. I've installed "Atmel Studio 7" as suggested by MetalGear_Honk and have attempted to write a simple "heart beat" program (the embedded equivalent of a "Hello World" program) but unfortunately Amtel don't have a 'ATMega32u4' simulator available so I'm not really able to run anything. From what I understand the 'ATMega32u4' is part of what is known as the "AVR 8-bit family" and is the main processor on the 'Teensy' board: adafruit have kindly made a "Breakout Board" featuring the ATMega32u4 available which, from what I understand of this page, is basically a slightly more developer friendly variant of the Teensy: Dispite the page saying: I ordered one anyway and am hoping I'll be able to figure out what to do with it when it arrives :thumbup:

*deleted* (didn't really know what I was asking)

Might it be possible to spec out multiple options? For me I think £300 for a 0.25 Nm system sounds fantastic! (in fact I would actually be quite surprised if I get out of this project for as little as £300) but for other enthusiasts even 1000 USD wouldn't be all that expensive. Most heli cyclic controls seem to start around the 1000 USD price (such as Flight Link's G-Stick III). Could a single design be fitted out with either mid-priced or heavy-duty components, or would the physical design itself need to me significantly different? If the same design can be fitted out with 0.25 Nm components for 500 USD or 0.5 Nm components for 1000 USD this would allow builders to construct either a 500mm flight stick or a 1000mm cyclic. It would also be good to offer a desk mount or side-stick option as well, essentially a MS Sidewinder FFB 2 replacement, though I understand this would require a significantly different design. For me though 0.25Nm sounds perfect and if it can be done for £300, that would be absolutely fantastic!

Oow, oow oow! That is absolutely gorgeous!! (you may need to leave me alone with that rendering for a moment :p) Seriously though, that relay looks absolutely perfect!

Wow, what a find, and it's even had an OCR tool run over it which is fantastic for keyword searches! Yes, I did, sorry for not responding. I've been busy of late but will hopefully be able to give it my full attention shortly. For my part when I went seeking these design limits I was quoted: 10° forward 25° backward 17° left and right The actual length of the stick from pivot point to tip of the grip is 300mm for pitch, and 250mm for roll (the pitch axis being 50mm lower than the roll). Also, the grip, though vertical, is also swept back 120mm: These specific values are from a glider but should be typical for most stick controlled aircraft. I don't recall exactly what the full limits in my design are but I did add adjustable bolts such that one can tailor individual limits for the amount of throw in each direction and I have the bolts set to achieve exactly the dimensions specified above. So if you design is 30° is from full lock to full lock it might actually be a little constricted (though probably fine), however if your talking about 30° from center to one side, then yeah, that's way toooo much :)

Interesting. I see they propose to use arrays of electromagnets in each axis rather than a single magnet which, to me at least, seems more viable: Interesting though it may be I didn't actually mean to subvert the thread with this discussion, so I'll let it go back to the actual project itself :thumbup: But thanks for the incite, it is indeed most interesting.

How would a "floor stick" or cyclic be different? I'm trying to build a floor stick/cyclic and I reckon this design looks great! What am I missing?

After watching VR_FlightSim's video I came across this promotional video by Iris Dynamics describing a "contact-less" FFB system: They also have a slightly more detailed video . While a neat idea I'm not completely sold on the concept itself. In particular I can't see how 5 magnets would be enough to give "hard-stops" as mentioned in the video. I appreciate that stepper and DC motors all work through magnetic forces but they have many more 'poles' or 'phases' (or whatever their called) that one could effectively set as a 'stop'. I just can't see how 5 magnets would provide enough resolution to give that fine level control, but perhaps I'm wrong. If it can be done with only 5 magnets that would be fantastic, but it sure would put a lot more onus on the programmer :D. Anyway, it's a neat idea. Not sure if it would be applicable to what we're working on here but it sure is neat ;) EDIT: BTW VR_FlightSim (and anyone else who cares), the way the YouTube tags on this forum seem to work is that you just need to place the video's serial number (that is the part in the address following the "v=...") between the [YOUTUBE] tags rather than the whole address. It took me a bit to work this out as well, but this seems to be how to do it ;)

I would certainly appreciate some help with gimbal design. As mentioned before the joystick I've been working on is intended to be used in a glider sim and a curious thing about gliders is that the X and Y axis often pivot at different heights. That is the pitch-axis is literally longer than the roll-axis. Like this: (but the other way round, ie. long pitch and shorter roll) I have done the best I can designing such a gimbal in SketchUp and even made a prototype but I'm afraid I really am at my limit as far as mechanical design goes :sly: Here's an image of what I have designed though: The bolts in the image are so you can set physical limits on the amount of travel (arc) each axis has available. The glider I'm modeling has a range of movement of: 10° forward 25° backward 17° side to side And here is a short video of the prototype I made: However I have no idea how I'm going to fit the FFB gear to this. Fortunately space isn't too much of a constraint as this is going to be bolted into a pit and the rest of the pit can be literally designed around whatever's needed. One thought I had was rather than connecting the motors to the axle's of the gimbal itself either notch or attach gears on these parts: And place the motors themselves on the 'inside' such that when activated they 'crawl' around the stationary gear and turn themselves along with the rest of the stick. Anyway, if you willing to offer any assistance on this design (or something else that can be modified to accommodate my unique requirements) I'd me most appreciative. P.S. If their of assistance I'll be more than happy to make my SketchUp models available but like MetalGear_Honk says, these files are far from professional :wacko: :doh:

Once I've worked out how to do it myself I would be willing to do as much for you but I really would suggest you'd be far better of learning how to do so yourself. The problem with sending pre-installed boards out is that you would then be stuck with whatever version was sent to you and I dare say there are likely to be several revisions to the code as this project progresses. Loading a program onto a PCB really isn't that hard and no real programming knowledge is required to do so. It really would be just a process of "following the instructions". So I would advice that getting your own board and learning how to upload code to it (even if you don't understand how that code actually works yourself) would serve you a lot better. In fact, if you have the know how to build the rest of the FFB system uploading the program to the PCB would be the easy part :smartass:. When I get on top of this myself I'll do my best to provide a set of easy to follow instructions for the community such that everyone can install/update their boards, regardless of their level of programming knowledge :baby:

Yeah, I'll be keen to, though I'm a little unsure of where to start. I am however between jobs at the moment so have plenty of time to devote to the venture. So, where do I start??? I've read through the entire thread and never came across a guide or even a particularly succinct parts list. While I've been a software engineer for many years now I've never worked with embedded code before so find the initial step a little daunting, though I'm most definitely keen to get into it! So I take it the first thing I'll need is an Arduino, yeah. But which should I order (DUE?, pro mini?, or even a Teensy?). When I was looking into building my own joystick I was leaning toward the Leo Bodnar boards but I take it the advantages of the Leo boards would never be realised in this project as we are essentially programming the whole thing ourselves, yeah? So what would be a suitable board to start off with? And what else should I order in order to get cracking? Cheers, Slarti.

Oow, oow. Pick me, pick me :clap:

Wow, I hadn't checked this thread for a month or two thinking it had all gone quiet but POW, it's just exploded!!! Great to see :thumbup: I made a brief mention in this thread, adding my voice as support back in June when I thought things were stalling out but am delighted to see this project picking up steam again :) :) That is pretty much exactly what I am after CrossWise. For my part I am working on designing a sim-pit to replicate a glider and accordingly was designing a joystick (control-stick) to use in the sim-pit but hit a road block when it came to designing the re-centering springs. It is actually surprisingly difficult to design a good centering mechanism, at least it is for a bloke like me :D. That is when I happened upon this thread and have been following its progress with great enthusiasm ever since. In particular what I am most hoping to achieve from this project is the ability to set a trim point that is off-center of the joysticks center. That is, if the trim is set nose down, have the stick 'return' to a point a little forward of center and if trimmed nose up have it 'return' aft of centre. This would also be most welcomed when flying helicopters that use a 'force-trim' (such as in the DCS Huey). The current system of re-mapping what the center of the joystick represents (rather than changing where the stick returns to) is cumbersome to say the least. As it is I have actually resorted to removing the spring entirely from my X56 HOTAS when flying the Huey as I find the simulated force-trim so disagreeable. While removing the spring works fine for flying it means you can never let go of the stick while in motion as once you do the stick just flops over leading to an inevitable crash. Most annoying! Additional 'effects' such as velocity dependent force (ie. sloppy controls at stall point) and 'canned effects' such as machine gun fire and landing gear jolts etc. are added benefits, but it really is the trim that I hunger for. That and a way to re-centre my homemade joystick without resorting to springs :D

Yup, auto-unhook was the key ;) Thanks fellers. I also found this guide most helpful: https://forums.eagle.ru/showpost.php?p=2035161&postcount=5

I'm having trouble with the cargo. I can hook the cargo by hovering somewhat above it. Admittedly some of my hovers have been less than stellar but I get the message that I am hooked, so I just raise collective and head off into the wide blue yonder. However when it comes to landing the load I am completely unable to do so without damaging the goods. I fly in over the green smoke ad 100' or so and hold a hover as best I can (usually a lot better than my takeoff hover) and gently try to lower the heli till the cargo is deployed. I do my best to hover directly over the green smoke and have the sink rate at less than "5" on the dial, but no matter how gently I bring the cargo down I always get a message that the cargo has been damaged and I fail the mission. What am I doing wrong?

Okay, I just realized I hadn't adjusted the compass. When I set the "gyro-compass" to the value shown on the "magnetic-compass" the arrow from the "LU" station does indeed point to about 120° and it does indeed take you straight to Batumi. (: