Playing the frame game

by Simon Hargreaves

Frames have got uglier and cheaper, but they’re straighter than ever. Suspension, on the other hand...

Last week’s Techwatch talked about how current production engines have close tolerances and matched components, and how automated mass-production reduces variability. If a modern engine makes 100bhp today, in 12 months’ time it’s likely the same model will also make 100bhp (although early BMW S1000RRs were an exception – bikes varied by up to 15bhp, and no-one knows why. The parts catalogue lists three different engine blocks and gaskets, which permits ‘room for error’).

Chassis production has mirrored engine production. As late as the mid-90s there were significant variations in frame geometry from bike to bike, and even alignment within one bike. Steel-framed CBR600s often arrived with the front and rear wheels out of line by up to 15mm front-to-rear, and with back wheels also running skewed in the swingarm because the chain adjuster marks didn’t line up. This wasn’t an isolated case; they rarely did on any bike (if you’ve ever got anal about lining adjuster and swingarm marks up after tensioning a chain, you probably wasted your time). And rear wheels were often out of line in another plane: both arms of a swingarm (and the hub of a single-sider) would hold the wheel axle a degree or two off horizontal, cocking it to one side.

Most riders mightn’t notice such wonky wheels because the effects are masked by camber, bumps, worn tyres and the echoes of last night’s chicken phall. But ridden back-to-back with a ‘straight’ bike or at extremes on a track, the out-of-line bike will exhibit awkward handling characteristics – more tendency to slide or shimmy under power, more likely to lose grip at turn-in or under power, and a preference to steer in one direction and reluctance in the other (the bike has to roll across the centre of the tyre to steer). The tyre won’t wear evenly either.
Rake and trail figures were also variable: manufacturers’ numbers bore little resemblance to reality. Sometimes errors were simply clerical: one manufacturer recently admitted publishing the wrong trail figure for a significant sportsbike for years, and no-one noticed. And sometimes errors were introduced in production: in the 1980s Honda frames were welded by ten or so workers at a turntable. Each performed their welds within the same amount of time before the table was turned, literally, and everyone did the same job on the next frame.

In the middle of the last decade the introduction of automated assembly processes, such as one-piece, flow-cast frames, reduced man-made variations.

Modern chassis dimensions are more repeatable and although not perfect, the wheels of most new bikes aren’t usually out of line by more than 5mm and steering geometry will be optimised and consistent from bike to bike.

Suspension isn’t quite as straightforward. While factories now usually get springing and set-up right (Yamaha FZ8 excepted), damping quality is prey to cost-cutting. You can build a cheap frame that has the same rake and trail as the next one. But if you pay £50 for a rear shock or forks from a supplier, they’ll have cheap internals and damping will be compromised. They’ll be consistent. Consistently bad.

shuggiemac's picture
Joined: 23/11/2008

Interesting article and I am absolutely sure that Simon is 100% correct in his figures. The thing that is surprising however is that it was still happening as late as the mid nineties as there simply was no excuse for it at such a late period.

The ability to check and control the relevant dimensions to a far greater accuracy than is indicated here, was widely available and used. The same can be said for manufacturing methods. It is hard to comprehend that such shoddy manufacturing methods were still in use at that time, as there simply was no excuse.

In todays manufacturing environment the 5mm figure should be seen as a Monday morning or Friday afternoon job !

pittsy's picture
Joined: 06/08/2011

Executing a series of welds in a different order can yield surprisingly different results to accuracy of shape. Jigs are one thing, but understanding how stresses are built up whilst I'm the jig are every bit as important. The workpiece will spring into a different shape once released from the shackles of the jig. I don't know whether frames are stress relieved post jig?

Then add to that the tendency of the telescopic fork design to magnify errors at the headstock. Oo eck.

Navy Boy
Joined: 12/02/2009

And just think all that time I spent getting my chain adjustment marks into line...

shuggiemac's picture
Joined: 23/11/2008

All good points Pittsy.

In a mass production vehicle there would have been a production run up phase where processes are nailed down and distortion effect from spring back etc can be tracked, analysed and adjustments made to the process to bring the effect under control. Of course it is not 100% repeatable but it can be fairly well predicted. Jigs can then be modified to compensate in initial location.

The wheel locations regarding square and parallel relationships should be easier to control by measuring the frame after its assembly and then machining mounting points accordingly.

I'm going home tonight and shall drag the old Bantam out the garage to measure it. I bet everything is within 15 thou on that and it was all done with a range of precision hammers.