Head Set sizing/buying

[Pebble]

I'm thinking the steering bearings on a push bike is its "Head-Set"?

Anyway it has been getting worse and worse this year and I've finally had enough, it goes from loose and no feeling from one cattle grid then at the next cattle grid it goes back to indexing, and this last week it has developed a deeply irritating rattle.

Dismantled it the day and it is going to need replacing. How do I size it to make sure what I am buying will fit, seems to be 1000s of different types.

The top bearing looks as good as new, do I need to replace both?

I'm back on the road sunday so it will need to go back on - could it be worth going cageless until I get a replacement.
the bearings in the broken cage are 3.79mm - should they be 4mm, I have enough 4mm to go cageless, they should work shouldn't they?

[Pebble]

And is the bit on the forks an interference fit - will it just tap off in the same way the cup on the frame just tapped off?

[richardfm]

This might help with sizing https://sheldonbrown.com/cribsheet-headsets.html

[tim-b]

Hi
A couple of articles that may help:
New threaded headset
New threadless headset
You might hear the term "brinelled" for the state of your headset, explained here... Brinelled headset
EDIT: you can make your own tools with threaded studding, large washers, nuts and bits of PVC waste pipe
HTH
Regards
tim-b

[Brucey]

your headset uses 5/32" balls. It is badly worn but if you fit loose balls and align the races so that the indents in the crown race and lower set race don't line up when the steering is straight, the headset will work OK for a while, possibly longer than that.

cheers

[tim-b]

Hi
A bike with a "brinelled" headset will never be better than just OK, but it's kept me going in the past until I could source a replacement
You won't be able to adjust the headset accurately to remove play which will increase the likelihood of juddering under front braking with some types of brake, and don't trust it while riding "no-handed". Ideally don't use it until the headset can be replaced
Regards
tim-b

[Brucey]

A bike with a "brinelled" headset will never be better than just OK, but it's kept me going in the past until I could source a replacement
You won't be able to adjust the headset accurately to remove play which will increase the likelihood of juddering under front braking with some types of brake, and don't trust it while riding "no-handed". Ideally don't use it until the headset can be replaced....

if you believe that, you might conclude that there is no point in even trying. However IME if you follow the procedure I described above and adjust the headset correctly most of these issues become insignificant.

Note that a major factor in the development of the problem in first place is usually that the headset was never correctly adjusted, i.e. the headset got wrecked because it had too much preload. If you adjust an old headset badly it will perform badly. If you adjust a new headset in the same way, it will soon be wrecked just like the old one was. Hence my italics above. I think the record is someone wrecking four (expensive, high quality) headsets in a row before they bothered to critically assess their adjustment technique.

Bottom line is that when you try it you will see how well it works -or not- and you will be able to act accordingly.

I would argue that it is far better to experiment with your headset adjustment technique on an old headset than a new one.

cheers

[tim-b]

Hi

snip...and adjust the headset correctly most of these issues become insignificant.

How can you adjust the headset correctly? The races are deeply pitted and some way beyond the tolerances that were originally intended. I'd agree that there's some mileage in honing mechanical skills, and reusing the headset in the way that you describe would keep you going at a pinch, but it'll never be better than OK
Regards
tim-b

[hoogerbooger]

Mine wasn't as indexed as yours, but I followed Brucie's advice (asabove) and so far it been good. Handling much better than before I realised my headset was overnight and brinneled. It's cheap to try anyway, so surely worth a go.

[Brucey]

Hi

snip...and adjust the headset correctly most of these issues become insignificant.

How can you adjust the headset correctly? The races are deeply pitted ….

Bearing races which are damaged in this way are still mostly smooth but with occasional pits. Loose balls are not spaced like clipped balls so regardless of the alignment most balls will be sitting between undamaged parts of the raceways. The remaining balls won't be under load, but these balls will start to bear load once the headset is moved. The net result is a bearing that adjusts like normal and works like normal, but at the 'nth' degree isn't quite as smooth as a new one. Because it has more balls in with the loose ones than with the original clipped balls, the bearing still might be stronger than when the headset was new.

If you turn the raceways so that the indents don't line up with one another in the straight ahead position, you don't get even one ball sitting between two indents until the steering is turned about ten degrees off line. Until that point the worst case is that you have two or three balls (out of ~20-odd) which are sitting between one indent and a section of undamaged raceway, leaving the remainder to do the work.

cheers

[Pebble]

Thanks everybody for your input, very helpful, Brucey seems to be on the money yet again.

Local LBS is ordering in a new headset, (providing I fit it myself, he's snowed under with work,think he's heading for his best year ever, which is good cause he's a great lad) Anyways, I took the forks in with me for him to measure, and he re-positioned the bottom crown race saying its worth a try, and I refitted the cup on the frame at a different rotation. I also went cageless, 32 ball bearings and it is like new. can't believe the difference, I had not realised just how awful the old one had become, was able to tighten it so there was no play or indexing, popped out for half a mile and it felt great.

I take it with such a headset it is like other bearings where you need to find that level of tightness at the point the play disappears?

Curious as to why it went so wrong. Started feeling odd last December, put up with it for a good few hundred mile until I had a proper look - broken cage - replaced it but it was still far from right, it was either loose or indexing, seems to have got steadily worse this last few thousand mile. So could the damage have set in riding it with a broken cage. Prior to that I may have not had the headset apart for 15k

[Brucey]

I take it with such a headset it is like other bearings where you need to find that level of tightness at the point the play disappears?...

correct. You should assess the headset for free play with the locknut tight only.

Starting with it 'too loose' increment tighter by tiny fractions until the free play disappears. 3.6 degree increments is about 10um in terms of bearing clearance, and even at that level of increment, you may only have a choice of two settings at which the adjustment might be deemed 'OK', between 'too much free play' and 'too much preload'.

A neat trick to get the starting position about right is to

a) tighten the adjusting race by hand
b) tighten the locknut by hand on top of the adjusting race
c) back the adjusting race into the locknut until everything is tight.

In c) you hold the locknut still and move the adjusting race ACW until it is tight against the locknut. IME it is unlikely that you will be more than a few small increments away from *perfect* if you start this way.

cheers

[tim-b]

Hi
There are the obvious bearing pits (the false brinelling) often caused by vibration and the far smaller pits, which possibly arose from either rust or corrosion and, less likely, fretting. The race surface has a smaller usable area as a result, with less area for usable grease
Both types of pitting are less likely with better lubrication and bearing pits are less likely with correct pre-loading, which is another term for correct adjustment
Clipped bearings in damaged races have few balls in correct contact with the races. Rotating damaged races and filling them with balls partially addresses this, but correct pre-load adjustment is difficult at best. While this might seem to be a good solution it's never better than OK but it will get you out of a proverbial hole, and might be serviceable for some time. It will be vulnerable to accelerated wear compared to new and "at the 'nth' degree isn't quite as smooth as a new one"
The best solution will always be to fit new parts, lubricated and pre-loaded correctly
Regards
tim-b

[Brucey]

Theory is one thing but reality is quite another. In practice I have put headsets back together using the method I have described (fully intending it to be only a temporary measure) and then ridden the bike quite happily for tens of thousands of miles; all whilst the new headset sits gathering dust in a box, forgotten even.

In the meantime other folk have destroyed multiple 'vital new headsets' on similar machines, or even wrecked the frame/fork trying to fit them. They have certainly wasted an awful lot of time and effort on it for no gain whatsoever.

There are a couple of bearings on bikes that don't need to be 'perfect' but they always benefit from being 'perfectly adjusted'. If you have bad hubs, BB, pedals etc then it is never good (although it might be tolerable under certain circumstances I suppose). But if you have a slightly imperfect freewheel bearing or headset bearing, but it is perfectly adjusted, it is actually better to ride on than a notionally smoother/better bearing that isn't (or doesn't stay) adjusted so well. In an ideal world you would have both of course but it is very easy to get completely fixated on entirely the wrong thing.

cheers

[mikeymo]

I thought I read that it was a good idea to have an odd number of balls in a loose ball headset. The idea being that they couldn't divide up into two sets of the same number, and so settle into a regular pattern.

Or did I dream it?

EDIT: In fact, looking at that race, there seem to be 16 pits. 32 new balls might line up one in each pit and one between. Whereas if you take one out, you've got 31, a nice prime number, so the balls might never "settle" in the pits. Just an idea.

Of course they might settle in pairs, with one gap.