Which way up do the seals go in a Shimano threaded headset?

[rogerzilla]

Specifically, HP-7400.  The seal has an "L" cross section.  Does the top of the "L" face the balls, or the bottom of the "L"?  My headset binds when tightened although all surfaces are faced properly.

[rogerzilla]

Update: tried them the other way up, no difference.  The balls are new but are grade 10 so should be ok.  Next thing is to try without seals, I suppose, just to eliminate them as the source of the problem.  The headset doesn't bind as it turns; there's just no sweet spot between too tight and too loose.  I could just whack a Tange Passage on there but that would be replacing a light £90 headset with a heavy £10 one.  Still, if it works...

[robgul]

This   http://web.fc2.com/jump/?url=http://dt6110.web.fc2.com/parts/7400/HP-7400GT.jpg   looks as if the balls sit inside the lip on the seal.

Rob

[Brucey]

IIRC this headset uses 3/16" ball bearings.  I prefer to build this sort of headset up with loose balls but a fair number of them have crown races with  no 'blend' between the bearing surface radius and the flat surface that the seal sits on, and have a shoulder instead.

 This means that loose balls (or balls in the wrong type of clip, or balls that are loose in the clip) can sit in the wrong place and jam against the shoulder. If the headset is assembled in this way the result is a headset that does not adjust correctly; it is either too loose or binding, with a tiny movement of the adjusting race between the two settings. IIRC the L shaped seal helps to prevent this, but does not mean it is impossible. [The same kind of sensation occurs if the clips are the wrong way up too, but you surely won't have done that.]

Headsets that behave this way include old stronglight steel headsets (that superficially resemble campag record steel ones) and steel headsets of a type (that I can't remember the make of) that was fitted on many Dawes bikes in the mid-1980s. [edit; I remembered; Hatta Swan road...]

If you have this problem the easy way out is to use the original clipped balls.  The less easy way is to fit the balls into the cups with grease, and then to let the cup and the cone touch once (and once only) during assembly, so that the bearings never get a chance to turn when they are not making good and well aligned contact.

  So for example with the frame upside down lower the forks into the head tube (but not all the way) and then thread the adjusting race on. Raise the forks so that the upper bearing is in contact, and then screw the adjusting race on further, maintaining good contact all the time. The lower race should make contact without turning, and this means the balls won't go walkabout.  If you have succeeded, it ought to be possible to hand tighten the adjusting race and the bearing will still feel like a bearing. if you have failed, it will be all bindy and horrid when hand-tightened.

   If the adjusting race is backed off more than half a turn there is sometimes enough room to let the balls escape and start to sit against the shoulder...... if this happens start again.  I have known folk give up and fit new headsets, they have found it so difficult to assemble ones like this.

cheers

[rogerzilla]

Thanks.  These are new grade 10 balls in the original retainers, fitted so the open side of the cage faces the cone (as usual).

I don't think the headset was well adjusted when I got the bike secondhand.

Turns out the stack height of these is more than I'd thought, so I'll fit a Tange Falcon.  That's a far-superior headset anyway.

[rogerzilla]

The Falcon went on like a dream (the DA crown race was a bit loose) and is silky smooth with no play.  Weird.