Adventures in newer (optical) pc steering wheel hacking

[BadMouth]

EDIT:  This thread follows my trials and mostly errors.  Once everything is worked out, my findings will be organized into a more coherent thread.

          Completed Logitech MOMO Racing hack  information here: http://forum.arcadecontrols.com/index.php?topic=113040.0
          Skip down to the start of G27 tinkering: http://forum.arcadecontrols.com/index.php?topic=112330.msg1202251#msg1202251



Started tinkering with this last fall, but dropped the feedback motor and the tiny encoder wheel mounted to the back of it shattered.  
I only found two of pieces of it at the time.  
I've collected just about all the parts I need to revamp my driving cab with all arcade parts, so it's time to figure this out.

I'd given up on hacking a modern wheel that uses an optical encoder and ordered a Logitech Driving Force EX off fleabay.  
The seller sent me a GT Force.  It's still a pot-based wheel and I could have used it, but the cheap little thing only has 6 buttons.  
Never buy anything used with a stock photo and description.


So that sent me back to the Momo Racing wheel that I started with.
I searched under the couch and actually found all the parts of the encoder wheel, but one.  Superglued it back together.
It's not good enough to use, but it's good enough for testing.  Good thing I hadn't vacuumed under the couch.

Anyways, haven't found much info on hacking a modern wheel and that motivates me to make this work.
The PITA part is that the movement of the steering wheel passes through 5 gears before it reaches the encoder wheel, so you can't just attach the encoder wheel to the end of your steering shaft the way you could a pot.  I also want to get rid of the feel and sound of plastic gears.
I've still been too lazy to count the teeth, but did a little testing by counting the slots on the encoder wheel and how many revolutions it needed to make to turn the wheel lock to lock.
(I wouldn't rely on this being real precise, but in the ballpark)


So I figure if it needs 8 revolutions to move 270 degrees (which is 75% of 360 degrees),
it would need 10.66 revolutions to move 360 degrees.  There is a stop that was removed and the wheel was actually moving farther than 270 degrees, so I'm just calling it 10 revolutions until I decide to count teeth on gears.  So 10 revolutions of a 60 slot wheel=600 slots need to pass by the sensor per revolution if I want a 1:1 ratio with the steering wheel.

I tried printing some wheels on inkjet transparencies using available software, but it just couldn't get the resolution high enough.
Eventually, I ended up with a 7" diameter wheel that worked, but wasn't reliable or practical.  For anyone else that tries printing their own encoder wheels, a dremel mandrel makes a good shaft and you can put a cut-off wheel on each side to stiffen up the transparency.  

Now I'm thinking the next step is to find a self contained optical encoder to use in place of the logitech wheel and sensors.
I did find a company in Japan that is making wheels that use the original Logitech G27 board with one.  They even show the encoder, but I couldn't find it's specs anywhere.
http://www.frex.com/gp/
I'm having a hard time finding one with 600 pulses for a reasonable price.  I'm thinking I'd be better off to go slightly lower resolution than higher because I could make up for it by adjusting the saturation in logitech profiler.  But if the wheel on the screen reached its limit before the one in my hand, I'm not sure how to correct for that.

Would something like this work?
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=516-2023-ND

Here is a pic of the opto board from the logitech wheel.
Would someone be willing to label the wires for me?
I looked at diagrams, but I can't tell if I'm supposed to be looking at the sensor pins from the top or bottom.  


Anything else I need to know before dropping more money on parts to tinker with?
 

[lilshawn]

a laser printer can print a higher resolution image, and they can be had pretty cheaply now a days.

OR

print it out at 10x the size you need and take it to a copy place that has good digital copiers and reduce it down.

OR

take it to a place like Kinko's  and have them print it out on their fancy pants printers.

oh and i found this too...

http://www.bushytails.net/~randyg/encoder/encoderwheel.html

[BobA]

Don't try to use the encoder you listed as it is way too expensive and complicated for this job.   Can you show a top side photo of your encoder board.  It is a bit hard to tell exactly where the traces go just from the bottom pic.  Either my eyes are fuzzy or I am seeing shadows or something not traces. Maybe scratches on the bottom of the circuit board.

[BadMouth]

Don't try to use the encoder you listed as it is way too expensive and complicated for this job.   Can you show a top side photo of your encoder board.  It is a bit hard to tell exactly where the traces go just from the bottom pic.  Either my eyes are fuzzy or I am seeing shadows or something not traces. Maybe scratches on the bottom of the circuit board.

The pic above was made by placing the board on my scanner.  It's pretty small and hard to get pics of.
The original encoder wheel is about the same size as one from a rollerball mouse.
I couldn't get decent pics with a camera.  These were taken with a magnifier app on my phone.
Not much better, but here they are:



After lilshawn's comments, I tinkered around with the encoder wheel software some more because it wasn't my printer's resolution I was having problems with, it was the output of the encoder software.  I was exporting a photo of the encoder, so I could put several variations on one page to conserve transparency sheets.  Apparently, what it was exporting was much lower resolution that what it prints directly to the printer.
So, thinking that I'm not going the transparency route anyway, I printed one directly from the encoder wheel software and managed to get a decent 600 slot wheel that was only 3" accross.

It didn't work as I had expected though and it takes almost two full turns of the wheel to go lock to lock.  Far from a 1:1 ratio.  Not even a 2:1 ratio since I'm only turning it 270 degrees.  So I printed another wheel at a resolution of 1200, but it is not picked up by the opto board at all.  Not sure if it's because the lines are so small or there is overspray from the inkjets or what.  I think the lines are so small that the sensors on the opto board aren't even on the same line.
The wheel doesn't budge at all when using this disc.  Not even a twitch.

Am I wrong in thinking that 600 slots=600 resolution=600 pulses per revolution?
(You all are going to force me to read and learn again aren't you?)  
EDIT: It is 600 Cycles per revolution.  One cycle is a slot and a line.  Pulses would include all signals sent by both sensors.
 (there has to be two to determin which direction the wheel is moving)

I still think the self contained optical encoder is the way to go.  It's just knowing which one to use.
I emailed the place in Japan selling the wheels that work with G27 boards and asked them what the resolution was on the encoder they were using.
Longshot, but maybe they'll think I'm a potential customer and have the impression that it's a spec I think is important.

I don't mind paying for it if the end result is as easy as a pot swap & is going to be ultra reliable.

[lilshawn]

Am I wrong in thinking that 600 slots=600 resolution=600 pulses per revolution?

it sounds right to me. your math makes sense to me

It didn't work as I had expected though and it takes almost two full turns of the wheel to go lock to lock.

i'm wondering if there is some bleed by with the LED emitter that's causing the pickup to miss pulses. maybe your black stripes aren't wide enough.


maybe you could change the emitter/receiver to something like a  mouse one... it would likely be able to pickup the smaller spaces.

OR

you can print out a larger disk... maybe 5 or 6 inches in diameter... that way the slots are big enough for that particular encoder to see.

[BadMouth]

counted teeth on gears (twice and thrice to make sure I got the right numbers).
Coming up with  16:1 ratio.
This would mean I need a resolution of 960 for a 1:1 encoder wheel.
This doesn't fit with my first tests or the fact that I have to make almost two full turns with the 600 slot wheels to go lock to lock (one 3", one 7" printed using different programs....at least I'm getting the same result with both.
   I'll check back in after I've redone the math a few times.

Wish I just had a bunch of encoders so I could play the high/low game until I found the best fit.
I will probably end up doing that with transparencies, but I don't have too many left.
Done for the night, tomorrow I will make a 960 wheel and see if it seems to work 1:1.

Like I always say, I'm more persistent than good.

If anyone actually enjoys working out this type of thing and wants to check my math before I make any more huge mistakes....

(the 20 & 120 are 1:1 molded from the same piece of plastic and the 20 &14 are 1:1 on opposite ends of the feedback motor)

[SammyWI]

Your math looks good to me.  That is about a 16:1 gear ratio.  The encoder will rotate 16 times for every one rotation of the steering shaft.  (I design gears and gear boxes).

[lilshawn]

16.058823529411764705882352941176:1

[BadMouth]

Thanks.  I appreciate the help! 
This evening I will make a 960 slot wheel and try to get the spacing and size of the slots to match the original wheel.
(The slots are a lot wider than whatever you call the solid area between them)

Seems odd to me that it didn't come out a whole number.  It makes me doubt my counting, but maybe the gears on the feedback motor were something off the shelf instead of made specifically for the wheel and they just went with it.

[SammyWI]

It's common for a gear ratio to not be a nice integer value.  Close enough generally works fine. 

[lilshawn]

Ya, its kinda weird, but it looks like they had to gear it way down to use a cheap motor so they could get some torque then back up a bit for the encoder wheel to get some resolution back.

[leapinlew]

Not sure if it's much help, but you can see how I did my mod. Worked great.

http://forum.arcadecontrols.com/index.php?topic=54685.0

[BadMouth]

First off, I screwed up counting the gears on the encoder wheel (twice).  I counted them by clicking my fingernail across each one and skipped the missing one in the lower left both times (pic above).  After looking at the scanned image and counting it, it has 35 teeth.  So my final ratio is 1:15.6 and I would need 936 slots on my wheel, assuming I didn't screw up elsewhere which is more likely than not.  I'm beyond help.

But anyways.......

This evening I printed out a big wheel at a resolution of 960 and it tracked reliably but was still only moving the wheel half as much as it should.
So I looked online to make sure that I wasn't misunderstanding the resolution, then it dawned on me to make a wheel with a really low resolution and count the slots.  Sure enough, a resolution of 20 only = 10 slots.  So for a wheel with 960 slots, I would have to set the resolution to 1920.  (after correcting for the encoder wheel gears, it would be 1872...I'm going with 1800)

In summary, what I've taken a couple days and too many sheets of transparency paper to learn:
1. A cycle contains one slot and one line (so the cycles per revolution does = the number of slots in the wheel)
2. Resolution is the total number of slots and lines on the wheel (combined).

At any rate, I'm down to a few transparency sheets and there's no way I'm going to get that high of a resolution on a disc into a 3" space.
Going to print a couple more giant wheels to confirm what resolution encoder I need then start looking for a decent deal on a self contained one.
I still think that's going to be the most durable & accurate way to go.  

If it works for the MOMO wheel, I might start looking for a broken G27 to hack.  
I'm sure a lot of the sim guys would like to see a tutorial for that.  (I've been searching forever and haven't found squat)

EDIT: wheel at a resolution of 1800 is close to dead on 1:1 with the wheel on the screen.....and it's the size of a plate.

[lilshawn]

  good lord.... this is too complicated for me.

EDIT: wheel at a resolution of 1800 is close to dead on 1:1 with the wheel on the screen.....and it's the size of a plate.

if it's hidden behind a panel of some sort... who cares right?

[BadMouth]

if it's hidden behind a panel of some sort... who cares right?

It's gotta fit next to the feedback motor  

A smaller self contained unit with a 1/4" shaft should drop right in where the potentiometer was using the same bracket.
Sucks that they cost so much, but I believe it's the best way to go.

(sorry for all the confusion, just put the thread in a dark corner somewhere and I'll ramble to myself until I get it sorted)

[BadMouth]

Did some testing and figured out what each wire was. 



The lower voltage for the emitter will go unused on the replacement encoder,
as it will get power from the same 5v pin as the sensor.

I will order the encoder tonight.  Not sure how long it will take to come in.
I have PLENTY of other projects to work on in the meantime. 

[BadMouth]

Finally got around to hooking up the optical encoder from US Digital and the encoder itself works great!
The wheel shows centered as it should when first hooked up.  It hits the left limit on the screen at the same time as the physical limit.
I turn it back to center and it's still centered perfectly.  

The happ setup allows the wheel to rotate farther when turning it right though.  
This is causing problems.  The wheel on the screen reaches it's limit and stops moving, but the physical wheel can turn another inch or two.
When turning the wheel back, the one on screen starts moving immediately and the center point is all screwed up.
I guess the solution is to build up the bumper on the right limit until it corresponds with the on screen one.
I find it odd that happ would design the wheels this way.

EDIT:AFTER GETTING EVERYTHING SET UP WITH A HAPP WHEEL, WITH THE 900CPR ENCODER, I'M HITTING THE ON-SCREEN STOPS TOO SOON.  I CAN LIMIT THE ROTATION AND STILL USE IT, BUT I WON'T HAVE THE FULL TAVEL THE HAPP ORIGINALLY HAD.  A SLIGHTLY LOWER RESOLUTION ENCODER WOULD BE BETTER.  I CAN'T AFFORD TO BUY A SELECTION OF THEM TO FIND OUT.

EDIT AGAIN: I BOUGHT SOME TOOLS TO DO SOME TESTING AND FIGURED OUT THAT I NEEDED AN 800CPR ENCODER.  FOUND ONE ON EBAY FOR A DECENT PRICE AND IT WORKS PERFECTLY.  I DON'T EVEN HAVE TO SHIM THE STOPS ON THE HAPP WHEEL.


A lower resolution encoder would have allowed the physical wheel to hit both it's limits before the wheel on screen.
I'm not sure if calibration of the center point would still be messed up though since the wheel turns itself back and forth to each limit every time the computer starts.  (it may just choose the center between those two points as center, but the steering wheel would be crooked)

This answers another question I had.  I was wondering what would happen if I removed the stop and allowed the wheel to turn 360 degrees.
The answer is that it's going to mess up the calibration of the logitech wheel.  On the original MOMO wheel, there is another sensor that picks up a silver line on the center of the steering shaft.  I put some silver reflective tape on the steering shaft and held the sensor near it, but it didn't seem to have any effect.
I've read in forums that people have unhooked it and it makes no difference.  It would be nice if it did reset the center whenever it passed the sensor, but that doesn't seem to be the case.  I was hoping maybe I'd be able to run another set of leads from the encoder to a mouse hack and have it do double duty for 270 & 360 degree games.  Looks like I will have to find another way.

The optical encoder does drop right in where the potentiometer was and was easy to wire up.


It was pretty pricey.  By the time I piled up the options like ball bearings & a wiring harness, it was  $103 shipped.
They only had the resolution I needed with an index track and that was extra too.
(flame on)  I'm crazy and wanted to see if it would work.  

I was very pleased with their service and plan to order a couple cheaper,
lower resolution encoders from them when I get around to making rotary sticks.

[lilshawn]

crazy! but it looks like it will work for you though... so bravo!

do you have a datasheet on that encoder? how many cycles is it rated for? your not going to have to be replacing that thing every year are you?

[BadMouth]

Oh, it's total overkill.  Especially with the ball bearing option, but it was only around $6 extra.

The sleeve version is rated for >1,000,000 rotations.  The ball bearing version has some crazy formula, but I'm sure it's many times more.
The only thing that could possibly wear out is the bearings.  The other components never touch anything.
It did come with a 2 yr warranty (defects in workmanship, blah blah blah)

[newmanfamilyvlogs]

Oh, it's total overkill.  Especially with the ball bearing option, but it was only around $6 extra.

The sleeve version is rated for >1,000,000 rotations.  The ball bearing version has some crazy formula, but I'm sure it's many times more.
The only thing that could possibly wear out is the bearings.  The other components never touch anything.
It did come with a 2 yr warranty (defects in workmanship, blah blah blah)

Am I right reading that as

L10 = (19.3/Fr)^3

Where Fr is the shaft weight in pounds?? Is there even any load ON the shaft the way you've got it mounted?
Even if you had a half of a pound on the shaft that would be
((19.3/0.5)^3) * 10^6 = ((38.6)^3) * 10^6 = (57512.456) * 10^6 = 57,512,456,000 revolutions.

At the rated maximum revolution of 10,000 RPM that's almost 11 years of 24/7 spinning.

[drventure]

57,512,456,000 revolutions.

You're doin' it awesomely right 

Does a feedback motor like that have to be driven by specialty software, or can it work just like any other Forcefeedback wheel (ie via the MS apis, so that it works with your typical PC racing games?)

[BadMouth]

As long as the bracket is straight there should be zero pressure on the encoder shaft.

Does a feedback motor like that have to be driven by specialty software, or can it work just like any other Forcefeedback wheel (ie via the MS apis, so that it works with your typical PC racing games?)

I'm hacking a logitech MOMO wheel, so it's standard direct x stuff.  (any more detailed answer would be over my head)
The logitech profiler software is available to tweak, but isn't necessary for the wheel to work.
This wheel has worked on everything I've tried it with (all listed in the driving cab info thread).

I read that the ffb in Sega Rally Revo doesn't work with this one or any wheel older than a G25.  (this is the only game I've heard of where ffb doesn't work with a MOMO racing wheel)
I've ordered a copy and will find out next week.  This is one of the reasons I'm determined to figure out the best way to hack a newer wheel.
Developers might drop support for ffb on the old wheels, but new wheels seem to be backwards compatible with anything.
It's probably good that Howard is using Directx 8 and a basic wheel to tinker with ffb.  It's the lowest common denominator.
If he gets it figured out, it should work with everything but the oldest of wheels.  

The feedback motor itself will be controlled by an industrial servo controller amplifier that takes the 12v/low current that was going to the little logitech feedback motor and translates it into 24v or 36v high current to the happ motor.  I'll probably tinker with that tonight.

Since I'm still limited to 270 degrees, I haven't really gained much by hacking this wheel instead of a pot based one, except for knowledge and maybe a little more compatibility.
Eventually I hope to apply what I've learned to a G27.  I'd rather make the mistakes on a $40 MOMO wheel first.
(and it's what's in my driving cab already, so all the software is already configured for it)  The cool thing about hacking this wheel is that everything is on one small rectangular board and all the buttons are on wiring harnesses.  No soldering and no other parts from the pc wheel would be in my cab.  Just one small non-descript board.

I'm shopping for a broken G27, but don't want to pay much for it unless I'm sure that it only has physical problems like stripped gears or parts broken off.
There's one on fleabay now, but the seller claimed they had no way to test it when I asked if it was detected by windows when plugged in.
(aren't you sitting at a computer replying to my question?!?!!?!)  If it goes for next to nothing, which is doubtful, I may gamble on it.

[lilshawn]

As long as the bracket is straight there should be zero pressure on the encoder shaft.

I'M GETTING SOME KIND OF DIVIDE BY ZERO ERROR 

[newmanfamilyvlogs]

As long as the bracket is straight there should be zero pressure on the encoder shaft.

I'M GETTING SOME KIND OF DIVIDE BY ZERO ERROR 

You forgot to take into account the pull of gravity on the part past the joint! Gotta easily be a few thousandths of a pound (:

[TopJimmyCooks]

It was pretty pricey.  By the time I piled up the options like ball bearings & a wiring harness, it was  $103 shipped.

I don't feel so bad about forking out for $100 worth of aluminum sliding tracks and steel gears and racks now.   

[BadMouth]

On to the G27! (and a whole new set of problems!)

this thread is to ramble and keep my place.  If you're looking for full hacking instructions, I haven't figured them out yet.
If you have something to add, I'd be grateful.

Disassembly guide for the main unit here: http://jlvrh.de/G27_UK.htm
Shifter unit here: http://www.chilicoke.com/mods/ebrake.htm (solder points look like a PITA)

The stops for the onscreen wheel on this one are only at 180 degrees.
The wheel is adjustable up to 900 degrees.  If the travel is set higher, it just takes more turns of the wheel to get the on screen one to move to the stops.
The physical wheel movement is not 1:1 with the one on screen unless you reduce the travel enough to make it that way.


All the boards say SNOWBALL on them.  I guess it's an internal code-name. 
Hopefully it's not an indication of how they feel about their customers.


The opto board layout is similar in layout to the MOMO one, but instead of having 1.2v on the last pin to power the IR LED, it gets the power from the 5v pin and sends it through a resistor.  The last pin sends 5v to a NTC thermistor (I didn't know what it was, but the electronics gurus on BYOAC always come through).
After edumacating myself, I think it's there as a temperature sensor for the feedback motor.  The optical board is enclosed in a plastic housing that is snapped onto the back of one of the feedback motors.  There's no ventilation, so it would hold the heat if the motor were getting hot.  Not sure if that's right, but that's my theory.  The NTC measures 1.8ohms at room temperature and resistance goes down as it gets hot.


I hooked my US Digital optical encoder up, but the wheel will not work unless the calibration procedure goes correctly.
I was able to fudge it a couple times by moving the encoder back and forth while the motors were going through calibration, but I can't do it with any consistency.  When it does work, the wheel tracks ok.  The resolution isn't really important because the travel on this wheel is adjustable.  I set it to 150 degrees and the wheel seemed to move 1:1 with the one on screen.  When I reached the stop on screen, the feedback motor would kick on.  That's how travel is limited (when the wheel isn't disassembled).  During calibration though, the wheel rotates through the full 900 degrees, so I might not get away with my 900cpr encoder.  WARNING, BAD FUZZY MATH FOLLOWS.... My thinking is that if the 900cpr encoder only makes it through 150 degrees, then to make it through the entire 900 degrees, I'd need 6 times as much (5400cpr).  I only checked US Digital, but they didn't have an encoder that went that high and I probably wouldn't spend the money on another one anyway.  I hooked up the happ wheel & FFB that's limited to 270 degrees to see if it would just calibrate to that and it wouldn't.  The wheel goes back and forth, but the wheel on screen does not respond.

A really weird thing happened when I left one of the G27's FFB motors hooked up and had the servo amp/happ motor hooked in place of the other one.
Even though the original encoder board wasn't hooked up and the US Digital encoder wasn't being moved, if I turned the G27 wheel(and ffb motor), the happ ffb motor would turn the happ wheel to match (at the same speed even!).  The G27 FFB motors only have power and ground going to them, so I'm not sure where the position info would be coming from.

I remember coming across some info on the calibration in a forum, but don't remember much except them saying that if the gears were off by 1 tooth, the wheel wouldn't work.  Gonna have to look the conversation up again and read it.

That's about all I've figured out so far.  Gonna let that stew for a while and wait for an epiphany.




EDIT: Shifter hack for the G27 being worked on here: http://forum.arcadecontrols.com/index.php/topic,134988.0.html

[BadMouth]

WARNING, BAD FUZZY MATH FOLLOWS.... My thinking is that if the 900cpr encoder only makes it through 150 degrees, then to make it through the entire 900 degrees, I'd need 6 times as much (5400cpr).

There's a lot wrong with that, idiot.    (the physical wheel was moving 270 degrees, the wheel on screen was moving 180 degrees, the limit on the logitech software was set to 152 degrees.   )  I'm not even going to try to revisit it.

Anyways on one of the occasions that the wheel made it through the calibration process, I drew a line on the end of the US Digital optical encoder and saw how many turns it took to go from center to lock (with the wheel set to allow the full 900 degrees of travel).  It took 1.25 turns.  Lock to lock was 2.5 turns (900 degrees!).  In other words, the 900CPR encoder is the right one for the job if the wheel is allowed to turn a full 900 degrees.  

Going back to bad fuzzy math for a moment.......
In those 2.5 passes, 2250 slots will pass through the optical sensor.  So we need 2250 slots to pass through the optical sensor for the pc to see the wheel as moving lock to lock.  If the wheel were restricted to 270 degrees instead of 900 degrees, you'd need an encoder wheel that squeezed 2250 slots into that 270 degrees of movement.  270 degrees is 75% of 360 degrees, so 2250/75% = 3000CPR encoder needed if you wanted it to drop right into a happ wheel.

I don't have the money for such an encoder (blew it on the last one).  The US digital encoders look like they only go up to 2500CPR, so I guess gearing might have to be involved at some point to stick with the standard Happ 270 degrees.  I guess the 2500cpr would work if the stops allowed for it to rotate through 2025 cycles (324 degrees by my calculations which are always suspect).

The startup calibration issue.......
The calibration process fails when the wheel isn't allowed to turn the full 900 degrees (too few slots pass through the sensor).  I guess it figures that there is an obstruction and the wheel must be stuck against something, so it doesn't allow the wheel to function.  I didn't think it could be too specific since I could get past it sometimes by spinning the encoder in my hand, but I always had to make several turns.
Since I only have the 900cpr encoder and the happ wheel was limited to 270 degrees, I removed the stop to allow the wheel to spin freely.
During calibration, the wheel will keep spinning to each side until I stop it with my hands, then it will spin back the other way.  Of course, the calibration is off since I'm inconsistent about where I'm stopping it.  But so far it seems to make it through the calibration and allow the wheel to function as long as I allow it to turn at least 2.5 turns between stopping it (simulating the locks).

Deciding what to do next.  Ideally I'd install stops on the happ wheel that allow for 900 degrees of rotation, but I don't have the money (or desire) for machining.  If I figure out some cheap way to do it with off the shelf parts, I'll try it.

Part of me just wants to use the MOMO hack and get the happ wheel into my cab already.
I could probably put the G27 back together and sell it for enough to buy a comfy leather seat.  

I would like to see this through to a tested, working solution though.

EDIT: Decided just to put the G27 back together and use the MOMO board in my cab.  It's just a lot easier and makes more sense since I don't need many buttons and not many games support an analogue clutch anyway.  I do believe the G27 hack would work fine if the wheel was physically restricted to 900 degrees.

[axle182]

OK so after replying and not checking this thread thoroughly, I can report that the Logitech Force GT wheel has the exact same gearing as the Momo Wheel as outlined by Badmouth. So next phase, Badmouth, can you hook me up with the model number of the 800CPR you used?

I know you mentioned its much easier to hack a Pot wheel, but im wondering if the resolution and feel is better with the Optical wheel? I havent modded my GT wheel yet, and im still on the fence deciding whether its going to be smarter in the long run to find a EX etc to mod and sell the GT wheel. Your thoughts?

[BadMouth]

Badmouth, wondering if you could hook me up with some information? I understand if the wheel runs past the onscreen end stops, the wheel becomes uncalibrated, but wondered if the wheel hits its physical end stop before (or well before) the onscreen stops, does it still calibrate ok? Im doing this mod using a Force GT (I had it in the basement) and found the dreaded optical encoder. Its almost identical to the one used on your Momo wheel.

Do you still recall the gear ratio of your wheel? I have pictures of all the gears in the GT, and will do the ratio calc today. If its the same ratio as your momo (from steering wheel to FFB motor, the optical gearing looks identical) I should be able to purchase the same 900CPR Encoder. Which brand and where did you purchase it from? Cheers!

See first post in this thread.  800CPR turned out to be the right one, but is hard to find for a decent price.
I bought the 900CPR one from usdigital.  I had an 800CPR one, but it stopped working and I haven't been able to find another one.
Keep in mind that this is all experimental and might have many kinks to work out.

If your'e using a servo amp for the ffb, you'll get the dreaded wobble and have to turn centering feedback way down until it goes away.
My current theory is that it could be lessened by putting a larger gear on the feedback motor, but I have way too many other projects going on to mess with it.

[axle182]

Mine will be used on a Obutto seat and with Iracing etc, so hoping to not experience the dreaded centreing issue.

[NoDFs]

I felt compelled to share the fix that I came up with for my Black MOMO.  The problem I had was that when I turned the wheel too fast it seemed to slip, as if the physical wheel spun a bit on the input shaft.  So When I returned it to center my vehicle was still turning.  I found that I could jerk it back the other way to 'slip' it closer to where center was supposed to be.  I got pretty good at spinning it back the right amount to fix it during the game I was playing, though I had to do this multiple times each race.

(reference first picture in top post by BadMouth)

I got tired of waiting for a straight stretch of road to safely do this so I took it apart to investigate.  I already know a little something about how the wheel might work inside and found that I was dead-on.  The 'Encoder Wheel' did exist as the method of operation.  I surmised that the electrical system could not handle the frequency that the vanes would pass by the LED sensor when the wheel was turned at high speed.  I sacrificed steering resolution for dependability by cutting every other vane out VERY CAREFULLY with a brand new, super sharp exacto-knife blade.  I knew that I had to cut out these puppies perfectly or risk destroying the whole encoder wheel.  I sat on a solid chair hunkered over my table and with the encoder wheel held in a hobbyist vise I got the job done.  It was like cutting thru butter, I literally could not sense the cuts.  I held my breath to help stay steady.  Since the vane is so thin and delicate that when you use a new blade you have to know that it is cut by sight, not feel.  If you were to mistakenly use a duller blade then the force used to cut one vane might just cause you to jump through the next one or two before you can catch it.  You might also think about using a magnifier of some sort like an eye loop.

The whole operation took maybe 2 hours and for the last 2 years I have not experienced the 'slip' nor have I ever noticed any precision loss.  I hope this helps anybody as I joined this forum just for this post.

[crazytestpilot]

Hi guys, wow i'm really impressed by what you guys are doing here. I'm also kind of excited hoping i can share some of my wheel hack-a-mojo tales with others who understand what i'm talking about and also may even be interested in hearing about it....i hope. Ok, quick scoop here: Oldschool...incredible pots upgrade. My g27 is ok but i always liked the quick response off center i got from my old momo force red. Problem was that the steering was always inconsistent and jerky...bad.  Ok, bad pot np. Began experimenting with pots of all shapes, sizes and values getting a wide range of good and bad results but never finding that sweet setup. Until, i came across a heli-pot. This 10 turn military grade pot required a couple of cogs to get in the game but ...OMG !!! Precision...Ultra!!!!  G27 ain't got nuthin on this. After further r-n-d i settled on a 3 turn(allowed for less extreme gearing) 1k  heli pot mated to the momo gearbox where the motor used to be so i could use  the internal gears to mesh. Now this old wheel is just amazing and so much fun, it's better than my momo black and g27. Ok, i hope somebody besides me can also have some fun with this. Thx.

[BadMouth]

Somebody is selling Logitech Driving Force GT pcbs on ebay for $11.
http://www.ebay.com/itm/logic-board-for-Logitech-Driving-Force-GT-Racing-Wheel-/131016814899?pt=US_Other_Video_Game_Accessories&hash=item1e8135f133

These would be a steal if I'd worked out the gearing/optical encoder combination for the G27 (this should be the same).
The big thing is getting it to see the correct number of pulses to pass calibration. 
Might be easier with adjustable stops on a screw drive or something.

Ordered one to play with. 

[bberger]

Encoder on the Frex SimWHEEL TypeG V3 (the japanese product you mentioned using the G27 electrical board) should be a

Copal/Nidec RE30E-500-213-1 (Datasheet: http://www.toho-inc.cn/toho2011/pdf/COPAL/English/RE30E_E.pdf)
(a internet "friend" opened the enclosure and sent me the spec)

The wheel is physically limited to 900deg lock-to-lock and the ratio from wheel axis to encoder axis is 1:1 (900deg wheel turn = 900deg encoder turn).

I actually have the Mk1 version of the wheel (some years old) that uses that crapload of Immersion FFB Controller so I want to swap it for the G27 pcb. The motor turns just fine, but the calibration wouldn't complete because the encoder on the first version is a Copal RE12D-300 (300ppr) and doesn't have enough resolution for this.

My problem is: I really want to upgrade the PCB as my G27 PCB is actually a 48V mod (double speed) @ 4.4A which also allows me to use a way stronger motor. I would like to order the RE30E-500 ppr encoder (~120Eur from China), but I don't really get how this would match your story with the 800ppr encoder..

I _REALLY_ don't want to waste that much money on something that might not be enough as I have no way to return this. I'm not entirely sure that the 500ppr are correct. He might have misread it or something..

Can you share any thoughts on this?

[BadMouth]

Can you share any thoughts on this?

It's not just having a count high enough, the count needs to be perfect..or at least close to it.
I've read if the rack and pinion setup on the original G27 is off by one gear, calibration will fail.

I'm not going to post any guesses for the risk of someone wasting their money based on my guess.

Next time I get around to selling some stuff for fun money, I plan to spend it on some encoders (for rotary joysticks) and one of these:
http://www.usdigital.com/products/interfaces/pc/usb/QSB
If I understand it correctly, I should be able to hook it up to the encoder on the G27 and see exactly how many pulses it posts lock to lock.
It will probably be at least a month though, and no guarantee that I'll get to it.

I've seen youtube videos of someone using an arduino to fake the pulses on startup to pass calibration.
I'll try to find it when I have more time.  If you find it, maybe you could ass that person for the number of pulses they are sending.

[bberger]

Can you share any thoughts on this?

I've seen youtube videos of someone using an arduino to fake the pulses on startup to pass calibration.
I'll try to find it when I have more time.  If you find it, maybe you could ass that person for the number of pulses they are sending.

Genious! I'm actually working on a FFB controller based on a Teensy 3.0, but I've got a couple of spare Teensy 2.0s idling around here for wheel electronics - seems like one of them has to be abused for a fake quadrature encoder!

I'll make a library out of it and host the code on GitHub once I've got it working so other people can benefit from it.

Thanks for the idea

[BadMouth]

Spent the evening figuring out the Driving Force GT board that I bought off ebay.  The process is a lot harder without any connectors.
Steering input wouldn't register until it passed the startup test.  It's nowhere near as picky as the G27.  Moving the encoder shaft back and forth without paying any attention to how much seemed good enough every time.

I thought the Driving Force GT had much in common with the G27 internally, but it looks like I was wrong.    I thought it was a 900 degree wheel; must be confusing it with another model. (EDIT: It is 900 degree per Logitech's website)  As someone mentioned above, the gearing seems to match the MOMO.  The encoder I was using for testing was 900cpr and it hit the onscreen stops just a little to early.  An 800cpr one is probably correct.  I'll be able to confirm once I make a bracket to attach my new 800cpr encoder to the happ/servo amp hack that I have.
(ordered the encoder from this seller a while back, but haven't tested it yet. http://www.ebay.com/itm/330824661021?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1438.l2649 It's either an Omron or a knockoff with Omron written on it.  Way nice for the price)



I could not make any sense of the pins to the steering wheel buttons.  The MOMO used a pretty simple matrix, but this one just wouldn't register any button presses no matter which pins I jumped.  One combo made all the buttons register simultaneously.

Overall, it's a pretty useable interface for $11.  Not sure the cost savings are worth the hassle of not having the connectors.
I'm going to email the seller tomorrow and see if I can get any harnesses if I order a second one.

Driving Force GT Pinout:

[Howard_Casto]

I've been following this one, but I didn't have anything to add.

Those connectors... what is the pitch on them?  You might want to see if a computer jumper will fit on them because you can buy jumper wires for a little bit of nothing.  I bought a pack of 10 5 pin jumper wire sets for like 5 dollars a while back.  They won't snap into place like the originals did, but some hot glue would fix that. 

What did these wheels originally support?  PC only or the consoles as well?

It seems like it would be a good option for converting a arcade panel without having to hack up a pc wheel.

[BadMouth]

I've been following this one, but I didn't have anything to add.

Those connectors... what is the pitch on them?  You might want to see if a computer jumper will fit on them because you can buy jumper wires for a little bit of nothing.  I bought a pack of 10 5 pin jumper wire sets for like 5 dollars a while back.  They won't snap into place like the originals did, but some hot glue would fix that. 

What did these wheels originally support?  PC only or the consoles as well?

It seems like it would be a good option for converting a arcade panel without having to hack up a pc wheel.

I had some .1" jumpers that I used.  The pins and connector for the optical encoder are smaller.  They were definitely oversized for that.
Luckily the USB connector from my momo wheel worked, so I didn't have to mess around with that one. 

The seller of this board has them and some other driving force GT parts in bulk.  My logic is that they were either refurbing them and had leftover replacement parts or they are parting out returns.  If they were parting out returns, they should have the cables.  I'll post when I hear from them.

According to Logitech, the Driving Force GT works on PC, PS2, & PS3 (and it's 900 degree).

If it works with my 800cpr encoder, I'll probably go this route for my current twin project.  The only advantage it has over the MOMO is that it shakes the wheel in games that only had rumble ffb and the MOMO doesn't.

[Howard_Casto]

Well that sucks.... it still isn't the end of the world, but that makes the whole thing a bit more inconvenient. 

Unfortunately it could be the other way around.   A company could be recycling the hardware of old wheels onto new pcbs.  Like perhaps they are coming out with an upgraded version of this wheel that will work on the ps4.  If that is the case then they wouldn't have any cables to spare.