You may be having the same kind of problem I had when I installed mine (BTW - this is the BEST, QUIETEST opener I've ever had. I wish we had these when I had my old house with the bedrooms over the garage!) Mine problems were manifested in the inability to consistently lift the door from floor to top of door, but the same thing can cause an imbalance that causes the cable come off the drums.
The big question is: do you have your rails all the way up at the ceiling, or are they mounted at the height of the garage door opening? If the former, read on - if not, skip this post.
I have a 3-car garage with 12' ceilings and 8' X 8' doors. When I first moved in, there was a standard rail system at 8' high with long expansion springs and a screw-type center-beam garage door opener. Later, to accommodate two-car parking with a 4-post lift in one of the bays, I had a specialty firm come out and raise the side rails to the near the ceiling and add two panels to the door so the standard screw-drive door opener had something to attach to and pull back to open the door. As part of the install, they converted to a torsion bar spring system too.
Even later, I decided to install the 3800, on my own - and remove the screw-type door opener. No longer needing the two extra panels to connect to old opener to, I removed them so the remaining panels just covered the 8' high door opening. The 3800 install went smoothly, and everything SEEMED alright. I was able to set the top stop, bottom stop, and lift/fall pressure settings.
But the next morning the side lock was stuck. I pushed down on the door to remove the pressure and it released - and the door slowly inched up about 3 inches and held steady there. I cycled the door a few times and it went up fine, but going down, it would hit bottom, and within a few seconds (often before the side lock could engage) it would drift up.
It took a lot of figuring, but here is what was happening:
With the original extra 2 panels on the door, going up nearly to the ceiling, the door in its closed position used to weigh more. Without those two panels, it weighed less. The 3800, since it works on the torsion bar itself, could out-muscle the torsion springs and turn the bar to lower the door all the way to the ground. But once the motor disengaged, the tightness of the torsion springs (without the weight of the two extra panels) would slowly pull it up again. If I disengaged the 3800, the door would bounce up at least 12", and it took all my weight plus a bit to push it all the way to the ground (Which says something for how strong the motor in the 3800 is).
OK, so I'll just take a few turns off the torsion springs, right? Yes, I used the proper rods, was very careful, and stayed on the proper side of each spring as I released the lock nut and made my quarter turns. I got it to the point where the door would go all the way down, and stay down.
But now it was IMPOSSIBLE to lift the door above about 2.5 feet (Insulated doors are heavy!) What the devil? I took two panels off - even with the 1 or 2 less turns on the torsion springs, I can't believe it is this hard to lift! how can this be?
Well, here's how I figured the physics of the new situation:
With the original two extra panels, the torsion springs were sized exactly right to lift the weight of all 7 panels from the ground to the ceiling. When the door was all the way closed, the torsion springs were wound their tightest and able to lift the most weight. But as the door used to lift, the panel near the ceiling would make the bend and go horizontal - taking it's weight off the to springs - just as the spring was unwinding itself and it's ability to lift was being reduced. As long as the panels go "over the bend" and reduce the amount being lifted, a properly sized torsion spring's reduced lifting as it unwinds is equalized - pretty much the whole way up the doorway.
Now, without the 2 extra panels, the highest panel doesn't start going horizontal, and removing its weight, until the door has opened nearly 3 feet. But the spring has unwound by a number of turns by the time it hits 3 feet (and actually started with fewer turns since I released a few turns) - but it's still lifting the same weight as it was at the bottom until that first panel gets to the ceiling. The higher you lift the door, the fewer turns on the spring to help you - and you STILL haven't gotten any of the panels over the curve to start taking weight off! If you're strong enough to get the door up high enough to start the panels going over the curve, the effort remains constant to keep lifting it the whole way up - generally - unless you've taken off enough spring turns that it's basically useless through the last foot or so of lift.
Now your circumstance is quite probably a bit different, especially since you are starting with a fresh install and not trying to adapt something was was originally installed for a different configuration. But if there is more than 6" or so between the top of the door panels and the curve near the ceiling, you will likely have much the same experience as I did. The fact that you mentioned that the door does seem to stay in any position means that what I'm describing may not apply to your situation. But hopefully this post will help someone else who stumbles on this post looking for an answer to their problem.
So what's the solution?
A new, properly sized set of torsion springs and "High Lift" cable drums. The circumference of the drum that spools the cable changes through the first few wraps changing the torque (lift) ability until the first panel gets to the ceiling bend, and then the spool continues at the same rate as a normal spool. Installing them means completely removing the torsion bar and re-installing it - something most people prefer to let a pro do. But you can get the parts individually, or as a kit, from a few places - including a place call
diygaragerepair dot com (look for the hi-lift links on the left). It's not impossible for a careful do it yourselfer.
See this garagejournaler's post for an example of a 1' extra lift:
http://www.garagejournal.com/forum/showthread.php?t=23837&highlight=lift+door+ceiling
In my case my increase in clearance was over 3 feet.
Another, very temporary, and very dangerous approach is to do the following (Note - I am sure that pros will strongly dissuade you from taking this approach, and I'm not endorsing it - it's very dangerous and you can injure or kill yourself while doing it because it involves over-tightening a spring. It will also significantly reduce the life of your spring, and it may end up snapping some day on you and your door may come crashing down on you or your car. Know what you're getting into):
Completely release one torsion spring so it turns freely (again, safety first - if you don't know how to do this, don't even try). Then tighten the other one 2 or 3 extra turns. You are overtightening the spring, so the potential for explosive failure increases with each turn.
Tighten and test your work until you get to the point where the door will go down and just stay down - but not creep up when you let go. If you can't get to this point with 4 extra turns, you should probably give up - more than 4 and you are risking imminent failure.
Actually, the number of turns you can safely put on a spring depends on its length and wire thickness. There is a good tutorial on torsion springs at ddmgaragedoors dot com under the "How to measure" link.
If you can get the door to the balance at bottom, you may be lucky enough that it also provides enough lift to let you open the door fully with 40 or less pounds of lifting effort. This is enough that the 3800 can fairly safely lift it and hold it open.
Why does this work?
Well, with two springs doing the work originally, by the time the door got to the 2 foot point, both springs had unwound enough that even together they didn't have enough oomph to keep lifting.
But if you have one, overtightened spring, those 3 or 4 extra twists (which just about equal the strength of the 2 springs together at the normal number of twists) are enough to lift the door off the ground, and even as the door goes up without any of the panels reaching the curve, there are still enough twists left in the spring to continue lifting. Once that first panel gets to the curve, the remaining twists in the spring are equalized by the reduced door weight.
It's a hack, and a dangerous one, but it can get you by in pinch.
Hope this helps!
