G27 belt drive mod

[sesselpupser]

Long story short, I've been mentally designing a G27 belt drive mod over the past couple of days. I don't have much source material other than this:



So, like, I've identified the need for some things that I can't find a source for:

- Large diameter hollow shaft (I'd guess that 'steering column' is about 25-30mm?)
- Timing pulley for said large diameter shaft
- A high power motor.

The rest I can find here, here or here. RS and Farnell's prices are so high, though, that although they almost certainly stock the motor I need (not that I know what to look for yet), I'd be better off buying something like an electric drill to take the motor from...

From what I know from taking a look inside a Driving Force Pro, the motors are driven by four MOSFETS - two P- and two N-channel - in a H-bridge configuration, like this:

So in theory, I should be able to just take the gate voltages that usually tell the FETs what to do and feed that into bigger FETs that can handle the power and current to drive a bigger motor.

Then it's 'just' a case of finding the right mechanical components to connect the motor to a series of pulleys, the pulleys to a bigger shaft for the steering wheel itself and then I have to connect the steering wheel to the shaft via a quick-release boss, because having a removable wheel was the entire goal of this mental design project in the first place.


Edit: Oops, pressed 'post' too soon.

Anyway, what do you think? Have you any comments, suggestions or advice?

 

[AussieStig]

Interesting mod, good luck with the build mate. Just wondering though, are you going to try and make a housing for it or have it all spread out as per the video?

Cheers, AussieStig

 

[LeMansAid]

There was someone (in Germany I think) doing similar ultra-modified G27s, and had them for sale. I did a bit of a search but can't find the site though. They had everything in a clear perspex enclosure, so there may have been some visible hints on offer.

 

[sesselpupser]

Interesting mod, good luck with the build mate. Just wondering though, are you going to try and make a housing for it or have it all spread out as per the video?

Cheers, AussieStig

I'll make a case for it eventually, but given how long it has taken me to do basic things like permanently mount my G27 to my rig if it doesn't need to be in case, it'll probably take a few years...

There was someone (in Germany I think) doing similar ultra-modified G27s, and had them for sale. I did a bit of a search but can't find the site though. They had everything in a clear perspex enclosure, so there may have been some visible hints on offer.

I think I remember seeing that too, I can't remember for the life of me where I saw that, though. I think they used to do a MOSFET upgrade kit for the G25 or 27 too but they disappeared and I have no idea how to find them again. Their name might've had 'Club' in it? I have no idea.

Edit: Might've found it, was it this one?

The site is mostly broken so I can't find anything too useful on this particular mod, but this helpful page still works: http://jlvrh.de/G27.htm

 

[LeMansAid]

I think I remember seeing that too, I can't remember for the life of me where I saw that, though. I think they used to do a MOSFET upgrade kit for the G25 or 27 too but they disappeared and I have no idea how to find them again. Their name might've had 'Club' in it? I have no idea.

Yeah, pretty sure that you could send your wheel to them, or buy the full shebang. It was belt drive, powerful, and expensive. May well have been G25 at that stage actually. Can't for the life of me find anything on it though.

 

[Team THRT Drift]

Yeah, pretty sure that you could send your wheel to them, or buy the full shebang. It was belt drive, powerful, and expensive. May well have been G25 at that stage actually. Can't for the life of me find anything on it though.

That's because the site has been taken down. You get a 404 message.

 

[LeMansAid]

Oops

 

[LeMansAid]

Edit: Might've found it, was it this one?

The site is mostly broken so I can't find anything too useful on this particular mod, but this helpful page still works: http://jlvrh.de/G27.htm

That must be the one. Opposite side.......

Might be worth checking out some forums to see if there's any useful tidbits. May be some more pics lurking around in there too.

 

[sesselpupser]

Thanks @LeMansAid, I'll see what I can glean from these images and the internet tomorrow with fresh eyes, I think I kind of have to open up my G27 though and see about testing the rpm ratios of the various bits before reading up on gears, pulleys and all that to see if I can replicate what the G27 does in maths and modelling, then maybe I'll order some pulleys or something.

 

[mrbasher]

If you can order from here, they have all kinds of different things that might be useful to you including the belts, etc. Cool project!

http://uk.misumi-ec.com/

 

[sesselpupser]

So I opened up my G27, uncovered the encoder and tried to count the rotations by eye - that's just not happening, it's far too fast. Shocking, right? So I've ordered a few bits to try and make a test rig, namely a toy wheel, shaft and shaft coupling to make the motor hand-crankable (I could use the encoder to crank it but I don't want to snap it!) and a laser diode. The idea is to screw a protractor to the output shaft (i.e. 'steering column'), point the laser at it, turn the motor by hand and see what angle the protractor gives. That way I can work out the ratio of motor turns to wheel turns, then I can start working out which pulleys I need.

While I wait for that stuff to arrive I'm using a rough target ratio range of between 17:1 and 18:1 (sadly it's not a round number, though) to see what configurations of pulleys on a 3-shaft system (like the SRS and JLVRH mods) will get me close.

Edit: In case it wasn't obvious that I have no idea what I'm doing, I decided to open the G27's gearbox and count the teeth, since the pitch would be the same that should give me the ratio, right? Well, I counted 11 teeth on the motors and 180 on the main gear, giving a ratio of 16.364:1 so I guess that's what I'm aiming for now... Duh.

 

[eKretz]

Nice! I'm rooting for you man! Looks like fun. I think that other mod was by ARC Team maybe?

 

[sesselpupser]

@eKretz Thanks! And yeah, the ARC Team G25/27 with the fans cut into the top case look familiar.

I've got a shopping list worked out but I have to wait until pay day to place the order. There are three hings that are undecided at this point, though:

- Do I need a mechanism to limit the wheel's rotation? I can't see one in the SRS video or the JLVRH photos, but what happens if you turn a G27 beyond the hard stops? Does it just ignore the fact that you've exceeded what it's normally designed to do, or does it recalibrate as you drive? I don't think it would be too hard to come up with a system to limit the rotation, if I had infinite money I'd do something sexy like Frex did with the SimWheel (that ball screw, mmm) but, well, I don't so I could maybe do a smaller version with a threaded rod. I'd like to avoid gears though since part of the point of a belt conversion is to reduce the noise from gear lash.

- I can either get T2.5 pulleys or T5 pulleys (the number refers to the tooth pitch). The former are less than half the price, exactly half the diameter, the pulleys will have less inertia - but they'll be less robust, they might slip under increased torque (in case I do ever upgrade the motors) and both of the mods featured in this thread so far seem to use at least T5 pulleys. Should I assume that's for a reason and go for the bigger option, or go for the cheaper option and try to make something more compact than the original G27?

- Belt lengths. I know it's possible to have a belt that's long enough to work but not long enough to provide sufficient lap on the smaller pulley to deal with torque, so how can I tell how long the belt should be? Should I use tensioners, I wonder?

As for a chassis, initially I'll just use MDF as I have some scrap lying around waiting to either be used or thrown away. If I can get that to work I'll either graduate to a proper thing (for some reason, even though it's likely to be really expensive, I'm drawn to a laser cut thing kind of like a PiBow case (pictured below) but much bigger) or an aluminium profile thing. My rig is already made of that stuff so it makes sense, I guess, and again I have a lot of scrap off-cuts that I might be able to use.

 

[TakumiFuji01]

I just want more lock to lock. Maybe 3.5 turns lock to lock compared to 900/2.5 turns lock to lock.

 

[sesselpupser]

I just want more lock to lock. Maybe 3.25 turns lock to lock compared to 900/2.5 turns lock to lock.

(I think) I've worked out the stock G27 ratio to be 16.364:1, meaning that for every 16.364 rotations of the motor shafts, the G27's wheel turns once. That means if you multiply 16.364 by 2.5 (the maximum rotation of the wheel), the motors are only allowed to turn 40.91 times between stops, so it's safe to assume the G27 is programmed only to read that many rotations of the encoder wheel before it fails to calibrate properly, right?

So what you need to do is alter the ratio of encoder wheel turns to steering wheel turns (40.91 encoder wheel turns / 3.25 wheel turns = a ratio of 12.587:1). You have two ways of doing this:

- Add another shaft just for the encoder wheel. This means an additional 4mm shaft, two 4mm bearings, two bearing blocks, two pulleys (one for the 4mm shaft and one for the other shaft that drives it) and additional space to mount it to the rest of the system. You might also now need to reconsider how you've mounted the G27's main board or get/make a longer cable to connect the encoder to it because it'll presumably be further away.

- Adjust the pulley ratio of the entire system. This is cheaper than maintaining the stock ratio because you don't need the same large, expensive pulleys you need for 16.364:1 but you do lose feedback torque and rotation speed - about 25%, which is a significant chunk.

In addition, either method you use will reduce the sensitivity of the encoder which may be noticeable while playing. That's because the encoder has 60 slots, the G27 will ignore rotation beyond 40.91 turns (or thereabouts) so you effectively have 60*40.91 pulses to play with at the most. At the stock rotation that figure accounts for 900 degrees so you have that much sensitivity, adding more degrees to the rotation means that same figure is stretched further which means less precision because of the lower resolution (like sampling potentiometers at 8 bit vs. 10 bit), whether or not it would be noticeable is a completely different story though.

 

[Clevohead]

Now to throw a cat amongst the pigeons!

There are two types of encoder for the G27 the older 60 slot and the later 30 slot...

I don't know what that does to you calculations???

 

[sesselpupser]

Now to throw a cat amongst the pigeons!

There are two types of encoder for the G27 the older 60 slot and the later 30 slot...

I don't know what that does to you calculations???

I have no idea, I'd have to have a look inside a new G27! No matter what, though, I'm quite sure the new G27 will work the same way in that it counts a predefined number of pulses and anything over that number will stop it from calibrating, so the theory is the same.

Speaking of which, I tried my G27 without the rotation limiter in it yesterday - it doesn't like that at all. The calibration routine does just turn the wheel one way until it stops, turns it back the other way until it stops and then divides the number of pulses between the two as a centre point. Without the limiter it just keeps turning, but if you grab the wheel it'll stop and turn back the other way again. I experimented with trying to get it to turn further than 900 degrees but if you do that the calibration fails, the shifter LED just blinks constantly and it doesn't know what to do. So, yeah, I do need a mechanism to limit the rotation, I guess I'll have to think about how to do that. The easy way would be a ball screw like Frex have in their SimWheel:

That also allows for on-the-fly lock configuration which is nice (you can see the adjustable rubber stoppers on long threaded rods). But that would push the cost up above £200, so... If I can't keep the cost below that I'll call the whole thing off!


Edit: The cheaper option is to copy what Logitech have done; a rack and pinion with blocks on the rack and shaft to stop it from rotating at either end. Pretty simple, I just hope it's quiet!

 

[sesselpupser]

I was going to use a bolt-on hub for a 25mm shaft to mount the wheel to the rest of the thing, but having seen what they look like, how big they are and how much they weigh (it'd account for a third of the weight of the entire thing!), I decided it'd make a lot more sense in every way to try and either machine something out of aluminium or maybe even use steel and use it as an opportunity to learn how to weld. I'd have to make an adaptor for a hub anyway, after all.

I also made a drawing on Sketchup to try and work out a few things and, well, visualise how it'll look. I think I might also scrap the massive 25mm bearing blocks I'd picked out, they too are pretty hefty and seem like huge overkill for my purposes, but that depends on me being able to find a suitable replacement - I can get bearings no problem, but they've got 47mm outer diameters and are quite thick, machining aluminium that thick at home would be difficult and 47mm is quite large, I don't know if you can get reamers that big and I would assume that if you can, they won't be cheap.

Also I found a place that sells 25mm carbon fibre pipes in both roll-wrapped (i.e. not obviously CF) and woven finish (what most people imagine when you say 'carbon fibre'), I have no functional reason to use them besides the fact that I won't need to paint the visible portion of the shaft to make it match the wheel but if I can cut enough out of the rest of the budget I might be tempted.

Other than that, I'm just waiting for pay day so I can start ordering!

 

[sesselpupser]

I've just ordered the first batch of components, mainly because the bearings I was looking at weren't being restocked and there were only a few left at 25% off. I still need to order the pulleys, belts, 25mm shaft and the rotation limiting rack and pinion, though (still waiting for pay day), so nothing exciting just yet.

 

[AussieStig]

This is going to be very interesting mate, I am following your posts, so keep em coming. I hope it all works out for you.

Cheers AussieStig

 

[Jet Badger]

Awesome as always, neema! I've been toying with the idea of making a robust sim wheel few years ago too. Never felt comfortable watching that G27 flex downwards once you start resting the weight of your arms on it. Really looking forward to seeing your project fold out. 👍

Just a thought about the encoder and the mechanical stop. You could use a 3rd party rotary optical encoder, especially if you go with an uprated motor. They're available in various step counts on ebay and such. Using a microcontroller between the encoder and the main board could also be a possibility. Just a thought, but you could have a second optical sensor that would read when the wheel's pointing straight then process the data from both sensors and feed it converted into what the G27 would understand as it's own encoder signal. Since, I reckon, the G27 has FFB locks too, that would let you get rid of the mechanical stop as you could just program it to "stop" after a certain amount of steps off-center. And also that would possibly let you set the rotation degrees independently.

Easier said than done I guess.

 

[sesselpupser]

Awesome as always, neema! I've been toying with the idea of making a robust sim wheel few years ago too. Never felt comfortable watching that G27 flex downwards once you start resting the weight of your arms on it. Really looking forward to seeing your project fold out. 👍

Just a thought about the encoder and the mechanical stop. You could use a 3rd party rotary optical encoder, especially if you go with an uprated motor. They're available in various step counts on ebay and such. Using a microcontroller between the encoder and the main board could also be a possibility. Just a thought, but you could have a second optical sensor that would read when the wheel's pointing straight then process the data from both sensors and feed it converted into what the G27 would understand as it's own encoder signal. Since, I reckon, the G27 has FFB locks too, that would let you get rid of the mechanical stop as you could just program it to "stop" after a certain amount of steps off-center. And also that would possibly let you set the rotation degrees independently.

Easier said than done I guess.

First of all: Thanks!

I have thought about possibly adding a custom microcontroller-based system to do... Things, but I can't really think of a purpose. The main problem is that I think if you exceed 2.5 turns after the wheel is calibrated you'll see some weird stuff and potentially crash the wheel, it counts something like 2455 pulses from lock to lock (16.364 ratio * 2.5 turns = 40.91 encoder rotations * 60 pulses per rotation) so if you significantly exceed that you might crash the microcontroller on the wheel by making it lose count, who knows what the result of that would be? So that kind of precludes having anything but an unbreakable hardware-based rotation limiter that will definitely limit it to nothing more than 2455 pulses, as much as I'd like to be able to keep the moving part count (and therefore noise) down.

But I'll keep it in mind for sure, for now the goal for this first step of this project all I want to do is replicate how the G27 works, except instead of using gears it'll be using a belt instead. Almost all of my other projects have suffered from scope creep and feature creep and then I end up so overwhelmed by trying to think of everything at once that nothing gets done! I've also got ideas for a quick detach system, MOSFET and motor upgrades, the case and other features like adding resistance to the paddle shifters to make them clunk in a satisfying manner, a WRC-style push/pull paddle so you can have a free hand for a handbrake, a little mount for a smartphone/tablet/other display for a digital dashboard, maybe case LEDs that react to the G27's wheel LEDs for a bit of novelty... But they're secondary and tertiary concerns right now.

 

[AussieStig]

First of all: Thanks!

I have thought about possibly adding a custom microcontroller-based system to do... Things, but I can't really think of a purpose. The main problem is that I think if you exceed 2.5 turns after the wheel is calibrated you'll see some weird stuff and potentially crash the wheel, it counts something like 2455 pulses from lock to lock (16.364 ratio * 2.5 turns = 40.91 encoder rotations * 60 pulses per rotation) so if you significantly exceed that you might crash the microcontroller on the wheel by making it lose count, who knows what the result of that would be? So that kind of precludes having anything but an unbreakable hardware-based rotation limiter that will definitely limit it to nothing more than 2455 pulses, as much as I'd like to be able to keep the moving part count (and therefore noise) down.

But I'll keep it in mind for sure, for now the goal for this first step of this project all I want to do is replicate how the G27 works, except instead of using gears it'll be using a belt instead. Almost all of my other projects have suffered from scope creep and feature creep and then I end up so overwhelmed by trying to think of everything at once that nothing gets done! I've also got ideas for a quick detach system, MOSFET and motor upgrades, the case and other features like adding resistance to the paddle shifters to make them clunk in a satisfying manner, a WRC-style push/pull paddle so you can have a free hand for a handbrake, a little mount for a smartphone/tablet/other display for a digital dashboard, maybe case LEDs that react to the G27's wheel LEDs for a bit of novelty... But they're secondary and tertiary concerns right now.

Maybe you could mount a phone as I have done on my G27: Cheers AussieStig

 

[sesselpupser]

@AussieStig Yeah, that's pretty much what I want to achieve, I just can't decide where to put the phone. I'll probably get a new phone on contract soon so my old 4S could even be mounted permanently... Anyway, I'll think more about it when the time comes, thanks for sharing.

I spent maybe two hours designing a wheel mount adaptor to fit the stock wheel on a hollow 25mm shaft today, and then I had a great idea... 25mm is a pretty standard size for a pipe, so surely someone already makes something to attach pipes to flat surfaces? They do, of course:

So I've ordered one of these at just £1.75 with free shipping from Ebay. Hopefully it's like the one in the photo on the listing - thick walled with a grub screw to fix the pipe in place rather than one of the flimsy ones designed for a clothing rail - so I can just screw it to my home-made adaptor.

About that, well, I could explain but here are a couple of photos that I think explain the idea adequately:

The thing from the photo before will screw onto the back of this and then clamp onto the pipe... I think it'll work, but if it doesn't then at least I've only wasted £1.75 and maybe an hour of my time. And I know, I know, pine isn't a great engineering material (hence the bit I sanded flat, a couple of large chunks splintered and fell out when I fit the adaptor the first time...) so I want to get some Delrin and some proper router bits for my Dremel - this was 'routed' quickly using engraver bits, hence why the edges are awful - when I make the actual piece, I've heard you can work it a lot like aluminium but without most of the, um, disaster. Ideally I'd be able to melt some brass M4 taps into it too so I could use machine screws instead of self tappers, that'd be nice.

Otherwise I'm still just waiting for pay day to order the pulleys, pipe and belts and for the stuff I ordered yesterday to arrive.


Oh yeah, before I forget, I'd like to thank @eKretz and @mrbasher for their indirect help. I've been Googling for as much information as I can get on the electronics of the G27, things like motor datasheets, MOSFET part numbers and that sort of thing. I've found two threads here that @eKretz either started or posted in a lot and one at ISR that a member by the name of 'mrbasherman' from Redmond, WA posted some very helpful information in, so thanks for the help!

 

[eKretz]

Neema, if you do get Delrin, I would make a suggestion for your threaded holes: buy or make some threaded inserts to use. This will let you drill and tap the Delrin with a bigger diameter thread for strength, then you screw the inserts in that have internal 4mm threads. Trying to melt something in will be a disaster I think.

 

[sesselpupser]

Ok, so I can't decide between a brief recap or a long post detailing everything... I think I'll go for a quick one.

- The steering lock mechanism is the only part I'm not sure will work - it all depends on me being able to make the sliding block part accurately enough - but I'll give it a try anyway, the gear and rack aren't cheap at about £13 in total but I can't think of a better solution.
- I think the first prototype will be made out of 15 or 18mm MDF, I think I have a mixture of both lying around so it makes sense to just use that for now. I guess that means self-tapping screws, too, because that's just easier.
- I'm also not completely confident the 25mm bearings I've got will do the job, they're quite stiff and with the two of them... I don't know, the pulley is pretty significant in diameter so there'll be a fair bit of torque fighting them, maybe the extra resistance will even be a good thing to help prevent oscillation across the deadzone (you know, the way it gets if you let go of the wheel when driving straight in GT4/5/6?), but I guess I'll have to try it and see because it's not something I can easily calculate. I'll order 25mm bearings with the next batch of stuff but milling a 47mm hole, accurately, twice, in something to make blocks for them with what I have to hand is not going to go well!

Apart from those points, I think I'm ready so I'm going to go ahead and order the pulleys, rack and pinion, belts, shaft and electronic components (2mm JST receptacles, crimp pins for them, wire, heatshrink, ferrite cores) when I've confirmed that the 3.175mm to 4mm shaft coupler will do the job as that's quite a critical part. Oh and then I need to order two more of those, too, because 3.175 to 3.175mm couplers don't exist and I need to couple the motors together.

 

[eKretz]

Make sure you don't have sealed bearings with the rubber seals - those add a lot of friction. Use bearings with steel shields. Wash them out with acetone or similar solvent, then use a light synthetic oil to relube them. That will get them spinning as free as possible. Take your block that needs to be bored for the bearings to a local machine shop with a 6-pack, I bet one of the machinists will hook you up!

 

[sesselpupser]

Make sure you don't have sealed bearings with the rubber seals - those add a lot of friction. Use bearings with steel shields. Wash them out with acetone or similar solvent, then use a light synthetic oil to relube them. That will get them spinning as free as possible. Take your block that needs to be bored for the bearings to a local machine shop with a 6-pack, I bet one of the machinists will hook you up!

The bearings I'm using for the 25mm shaft are these UCF205s:

I've ummed and ahhed over keeping them for so long that I can't actually return them now, so even though I'm pretty sure they're imperial (so 25.4mm inner diameter and who knows what size mounting holes, although they're sold as 25mm and M12) and quite stiff, I think I should really try these first. I don't think I can wash them out, though. The place I'm ordering most of my stuff from also sells these 25mm deep groove bearings that claim to be ok for high speed operation (no idea if that means they're less stiff or not, though) but they'd require the aforementioned custom-made blocks because I can't find off-the-shelf parts like I could for the 4 and 8mm bearings I've got, so I think I'll see how it goes with the heavy-duty, stiff, imperial bearings I've got and if it's just not viable I'll find something else for them to do (maybe some sort of mini Dremel lathe thing or something) and look harder for bearings and mounts. As for a local machine shop, I have absolutely no idea if there is such a thing near here and I wouldn't know where to start looking...

 

[eKretz]

Yeah those rubber seals add a serious amount of drag to something like a Sim steering wheel. If you're stuck with them you're stuck with them though. Good luck!

Where are you, exactly? Machine shops are everywhere, they just aren't usually something open to the general public so people don't even know they're there in most cases.

 

[sesselpupser]

Yeah those rubber seals add a serious amount of drag to something like a Sim steering wheel. If you're stuck with them you're stuck with them though. Good luck!

Where are you, exactly? Machine shops are everywhere, they just aren't usually something open to the general public so people don't even know they're there in most cases.

The seals aren't rubber, they're metal, but they're still stiff anyway - the bearings are clearly meant for some pretty heavy machinery but they were cheap-ish and seemed ideal (until I got them, realised they were as stiff as they are and probably imperial). I don't want to write them off just yet because I don't think I'll feel anywhere near as much resistance with a 140mm long lever like the G27's wheel and, come to think of it, the torque required to turn that bearing is actually less than the torque required to turn the stock G27's steering shaft without power and with no wheel on it, but that's including the torque of the two motors as well so not a great comparison. It also doesn't include the other 25mm bearing, three shafts, four pulleys, two belts and two bearings that'll be between that bearing and the motors... Hm! In any case, if I'm not happy with them in the end I do still have options. Well, an option.

I'm in Hemel Hempstead, I've tried searching for machine shops but I just get loads of job sites and a place in Wales that claims they're in Hemel... And some other places with no websites that aren't exactly local.