Sedan seat belt hack: tilt sensor delete

[Buzz Skyline]

No matter how I fiddled with them, I couldn't get my sedan donor seat belt reels oriented so the tilt sensors wouldn't lock up. So I removed the sensors. Now the belts work great. The lock up feature due to sudden jerks on the belt is unchanged.

I took the driver's side reel off the car to delete the sensor, but I did it on the passenger side with the reel still attached.

To open the the reel, pop the three plastic pins that secure the transparent cover

35771

The cover should pop off, I pried it gently with a flat head screw driver. Don't turn it over or the gears will fall out. They're easy enough to drop back in though.

35767

35768

The tilt sensor is the orange part in this pic, although it's green on the 818 passenger side (formerly WRX driver side) belt

35769 35770

It easily slips out.

Make sure the gear is in the right position . . .

3577235773

. . . before you replace the cover.

35774

Test the belt a few times to make sure everything is replaced properly, then pop the plastic pins back in and you're done. Now the belt will work with the reel in any orientation

[Bob_n_Cincy]

I don't think I would have done that.
The tilt sensor locks the belt before driver movement is detected by sensing g force on the frame.
I originally had stock wagon belts in my 818 and they worked fine.
If it works in the donor, it should work in the 818.
Bob

[Buzz Skyline]

There are three types of seat belt tensioners that I know of (not counting the explosive tensioners in front seats, of course): centrifugal clutch(feed out sensor), g-sensing (aka tilt sensors), and dual centrifugal clutch/g-sensing. Subarus use the dual technology, but centrifugal clutch should be good enough. I'd prefer both, but deleting one so that the belts can be used at all is a reasonable solution, considering many cars over the years have used only centrifugal clutch designs. Would quicker lock up be better? Maybe, but the centrifugal clutch system locks up with less than an inch of belt fed out. I can't see that it would help much to be even quicker.

Besides, I use a lap belt pretensioner for autocross, and would probably use one all the time in this car. And there's nothing faster than a pretensioner. ;)

[Buzz Skyline]

BTW, you can spend a lot for a latch to pretension your lap belt http://www.amazon.com/CG-Lock-CG001-Seatbelt-Stabilizer/dp/B004NY1DHS

Or you can spend a little http://www.ebay.com/itm/New-Help-Car-Seat-Locking-Clip-Seat-Belt-Adjusters-Safe-for-Children-Adults-/201017366115

I spend a little.

[Bob_n_Cincy]

Hi Buzz,
Are you using the rear seat belts?
Do you know why the OEM belts are not working?

I have all the belts out of my 2 donors. (both wagons). I'll play with them to see if I can figure out why yours aren't working. Wish I had sedan belts like you to test.
Bob

[Buzz Skyline]

Hi Bob,

Yep. I'm using the rear sedan belts from my donor (2002 WRX). I believe they didn't work because the orientation of the reel when mounted on the 818S is off enough that the tilt sensor was tripped. When I removed the driver side one, I could find an orientation that worked, but every time I mounted it on the frame the tilt was in the wrong orientation and kept the belt locked. That's why deleting the tilt sensor works.

It's easy enough to pop back in, and I may tinker with it more on the passenger side. But I will probably leave the driver side as is and rely on the centrifugal clutch, along with a locking clip to keep the belt tight on my lap. I wouldn't expect most passengers to want to use the clip, so a tilt sensor is more important there as a redundant safety measure.

I believe the wagon belts have a white tilt sensor in them that lets the belt feed out provided the reel is level. The ones for the tilted sedan reels are orange (WRX passenger side) and green (WRX driver side), at least for the 2002. Potentially, you could use the sedan belts in the wagon position if you swapped in some white sensors. I may do that because I have a pair of ugly and worn beige wagon belts that I would rather not put in my 818.

-Buzz

[freds]

The rear seat belts are designed to lock solidly when installing a child safety seat. When used for that purpose obviously the belt should not unlatch like the normal inertia reel

That device is able to be released by allowing the belt to retract "a certain amount" and then pulling it out rapidly. Its an iffy thing to get to work and I had a lot of trouble achieving that on mine, so I too removed that locking device. It does not interfere with the normal inertia reel function.

There is a a page of SUBARU instructions for locking/unlocking that safety seat securing function...should be in the owners manual, but certainly available from Subaru. I'm sorry, but I cant find the copy I had when I was facing exactly this problem. If I find it I will post it.

[freds]

"ALR" is the always locked condition. "ELR" is the normal condition.

3580235803

[Buzz Skyline]

That's it! Thanks, Freds.

The problem must be that in installing the reels I pulled the strap all the way out, putting it in the always locked position. I'm guessing that because of the location of the reel, I could not let the strap retract all the way, so there was no way to get out of the always locked condition.

Of course that may also mean that I will have the problem again, if anyone ever pulls the belt all the way out. I'll check the next time I'm home.

If that's the case, then the sedan location of the reel is just plain unworkable - eventually someone is going to put it in the always locked condition.

Not good news, but good to know before I commit to the sedan location. I may have to drill out all those rivets - again. Sigh.

Thanks again, Freds.

[Buzz Skyline]

You can see in this diagram that the stock location of the reels on a WRX sedan allows the belt to retract much farther than for the 818.

http://parts.subaru.com/a/Subaru_2002_Impreza205-5MT-4WDWRX-SEDAN/_54103_6028970/REAR-SEAT-BELT/G11-646-01.html

If you need to let the belt fully retract to get out of the always locked condition, it would never be possible in the 818. Dang it.

So the message seems to be that you simply can't use sedan belts in the 818 unless you're willing to unbolt your belt (which would require me removing my seat) every time it gets pulled all the way out.

On the bright side, now that I know how to delete or swap tilt sensors, I can mount my sedan belts to the wagon reel mount locations.

[RPGs818SNA]

Thanks for your posts, Buzz and Fred! Over 7 years later, I ran into the same sedan seat belt problem where, once fully extended, the rear belts will only retract until fully retracted. But once the belt end is attached to the frame near the floor, it can’t retract far enough to release the mechanism that lets it extend again. As Fred noted, this is so child seats can be held securely in the back seat.

Taking Buzz’s lead, I decided to try to remove that unneeded feature while keeping the others. Fortunately, the reel mechanism cover is transparent, so I was able to see how it worked.

165158

The white lever in the bottom left corner remains out of the way until the belt is fully extended, after which it prevents any extension until it is pushed back out of the way upon full retraction.

To remove that lever and its function, starting on the passenger side, just pry off the very top transparent cover with a small screwdriver. The following photos are of the same process on the driver’s side.

165159

Note where the notch in the top gear is positioned near the top.

165160
165161

Turn the gear until its notch is over the lever and pull the lever and its tiny spring out.

165162

Turn the gear notch back where it was and replace the cover. Repeat on the driver’s side.

Now the inertia function still works, and maybe the tilt, but otherwise the belt moves in and out freely.

I suspect many other builders have discovered this procedure over the last 7 years and documented it in their build threads or elsewhere, but this was the only thread I could find using a google search.

RPG

[Goldwing]

Oh man, thank you for showing all this. I pulled my 818 out of the garage for the first time in 6 years (a csf leak sidelined me) and this issue popped right up. I couldn’t get the seat belt out. I looked over my spare set of seat belts, and made the child seat modification. I noticed my left and right spare belts had different g/tilt sensors. One used a ball bearing inside the sensor, the other used a similarly made cylinder. The ball bearing rolls around freely in the plastic sensor while the cylinder pushes against a calibrated spring. I am wondering, at least for the ball bearing model, if a slightly smaller bearing would increase the tilt envelope enough to be more reliable in the 818? The bearing inside the sensor was 12.56mm. I’m tempted to see if I could swap a 12mm, or 11.5mm bearing in there instead to keep the tilt function, but make it more reliable in the 818 mounted location? I’m really hesitant to remove that feature altogether. A hillside rollover type situation without a jerking impact might allow a bunch of seat belt out before locking and leave us unsecured when the impact happens. A few washers to shim the seat belt reel mount angle might be a better approach, especially for the cylinder style tilt sensor. That one seems to use a calibrated spring that only compresses with the full weight of the cylinder bearing down on it. Looks a bit tougher to modify for our purposes with tight spring calibration a tough thing to source though adding a small amount of weight might work. But, if that weight dislodged over time, could disable the system. Feels sketchy. The bearing style looks more straightforward to modify safely. Well, input the disclaimer that all of this is at your and your passengers own risk. FFR seems to try to get that tilt sensor close, so shims might be all that’s needed. Maybe there’s a small variation in the frame jigs over time? One benefit of the cylinder style tilt sensor reels is that the spring side of the reel had a square drive core such that I was easily able to add a few extra winds to the spring to stiffen the belt a touch. The seat belt path just has a little too much resistance making the spring feel weak. The belt sometimes doesn’t feed itself back into the reel. The reels with the ball bearing style tilt sensors didn’t appear to have a spring holding mechanism to wind it further. There are two rectangle lat slots near the spring center, maybe there’s a way to engage the spring using those slits, but I couldn’t make it out.

I also note that when removing those covers on the reel ends, there are little center cores in the cover clips to push out before prying loose. It makes the process far easier on the clips if you remove those cores first using a tiny screwdriver. They’re like body panel clips where a center cores engages the side clips, just smaller. Push those cores out before prying the lid off.

This whole seat belt issue was almost welcome when the engine, cooling, fuel, and oil systems seem to be in perfect order despite the long 6 year pause in my build. I expected more problems. Just the battery (battery tender showed appropriate green lights but lied to me and did absolutely nothing) and the seatbelt as new problems after siphoning out the old gas. Everything else is just waiting for me to get back to work.

[Goldwing]

For what it’s worth, as a follow up, since the child seat lock deleted spare was still not installed yet, swapping a 12mm stainless ball bearing into the tilt sensor, opened the free spooling window by about 3 degrees in each direction. The axis the spool spins on (what would translate to the roll axis of the car) has the narrowest window of about 22 degrees, so that window receives a larger percentage increase or 6 degree bigger roll window before locking up the belt. New window 28 degrees of roll. The pitch axis has a larger window to start with of (~38 degrees), no doubt for maintaining freespool in the kind of terrain I grew up driving in (Pittsburgh PA) as well as starting and stopping forces. The window increase in degrees was the same, but a smaller percentage increase because of the larger range pitch already had. A larger operable window could mean the spool not locking up in a specific case crash, but it may mean an operable seatbelt where there wasn’t one which might encourage one to choose to forego the belt altogether. Do what you like with that information and decide for yourself if such a mod is worth the risk and benefits.

The original bearing measured 12.64mm
The tested bearing measured 11.94mm
Using a harbor freight caliper. Hundredths accuracy absolute isn’t guaranteed, but I’d trust the comparative difference of .60mm. The purchase was for a 12mm ball bearing.