Saturday 12 June 2021

UltraTorque BBs - the actual truth (I'm not trying to get a rise out of you, unlike some) ...

Some will know, we at Velotech and I in particular, have a very long-term association with Campagnolo - you only have to visit our website at www.velotech-cycling.ltd.uk to realise that, as a factory-appointed, trained and accredited Service Centre, we have a close working relationship with Campagnolo.

I have to say, though, that just because we have the association with Campagnolo that we do and just because I, as lead technician and MD of Velotech Cycling Ltd have worked with them in various capacities since 1987, I don't pretend that they've never made a mistake, nor that everything they make is perfect.

We are independent of Campagnolo, just as we are of the other companies with whom we work closely.

Which leads us to discussions of UltraTorque.

I've been having an "interesting" discussion with Hambini over on his (in my view, unnecessarily toxic and confrontational) YouTube channel about UltraTorque, in which he makes a number of, as far as I can see, unsupported assertions about the way that UT is designed.

When UT was introduced in late 2006 as part of the 2007 model year range, I wasn't actually working directly with Campagnolo - I saw it as a prototype product at the factory whilst there on a visit to discuss technical training in 2004 (I'm amazed to this day, that I was allowed to see it) - but by the time it was introduced into the market, the technical training business that I was a director of had foundered and I was "back on the tools" courtesy of a good friend with a generous heart, who made a job for me at this shop (he didn't really need another mechanic but having worked for him before, he had enough confidence in my skill and knowledge to take me on to look after some of the high-end jobs that came into his store).

UT interested me. At the factory, I had been privileged to speak to the design engineering team and to be given quite a lot of advance information that I found interesting - they'd adopted some fairly novel solutions to a range of problems, some of them a little unusual in the cycle industry.

The problems that they needed to solve were that in oversize axle, external cup bearing situations, a means needs to be found of compensating for bottom bracket shell widths (why that should be, we'll come to later), a set of strong patents held by Shimano (who had launched Hollowtech II a couple of years before, though their patents had been filed for some time) and a desire to keep the stance width of the rider (sometimes called Q factor) the same as that offered by their existing cranksets, as well as to improve, if possible, the ankle clearance.

Cost, of course, was a factor, as was ease of maintenance, the fact that they wanted most if not all the tooling used in assembly to be available freely in the market or from non-cycle specific sources, the design needed to be durable and it needed to be capable of modification to a variety of upcoming scenarios that were already on the horizon, in terms of frame interfaces.

In brief, the solution they devised was two cranks, each carrying a semi-axle, with a Hirth joint centrally placed. The bearings would be interference-mounted to the base of each semi-axle, with the right hand one retained by a circlip. The fact that the BB shell and therefore the cups that the assembly would be fitted to, might vary in width as a result of machining the mating surfaces of the BB shell parallel (in some materials, this parallel can drift as a result of the frame manufacturing process, as well as simple poor QC) was accommodated by the use of a sprung pre-load washer under the left-hand bearing, with the pre-load held within the tolerance required by the bearing manufacturer (SKF) by carefully controlled specification of that pre-load washer.

The BB cups themselves were to be simple recipients for the bearings, themselves a tight slip-fit into the cups. This allowed for future development of the cups for shells that might not be threaded ...

Any tendency for side-to-side motion of the crankset was controlled by an oddly-named "safety clip", which is basically a sprung semi-circular clip installed on the drive side, with two ends that penetrate the RH bearing cup and limit side-to-side movement. The placement of the drillings that take the clip allows for small variations in manufacturing tolerance, so there can be a tiny amount (+/-0.125mm) side-to-side play in the position of the chainrings relative to the BB shell and so the front mech.

I queried the name "safety clip" and the engineers looked a little puzzed and said - "Yes, safe - so, we don't really need it most of the time because most riders don't introduce a big lateral component when they pedal but for those who do, it's safe - it guarantees front derailleur function". Light dawned ... 

So - can this general design be improved upon? Well, yes, probably - although it's noticeable that others have used some of the solutions that Campagnolo did - BB30 uses a preload washer to control the preload on the bearings, Specialized OSBB is actually a close copy in mant respects, using both a Hirth joint and a spring washer (but no safety clip, because they, like Trek in BB90, maintain a close control over their BB shell widths).

The question is, can it be improved upon and the cost and patent considerations be accommodated?
Can weight be pared off it? Would a different solution carry a weight penalty? What about tooling?

Of course, it's a maxim in engineering that even I, with no formal engineering training know, that TIAMTOWTDI ... There Is Always More Than One Way To Do It ... (apparently, though, not being an engineer, I am not bright enough to read the books, talk to engineers or to have gained a lot of very direct, practical experience, so have no right to be commenting on such matters).

Despite what certain individuals see as my educational deficiencies, I was pretty intrigued by two central bits of the technology - the wavy washer and the safety clip. I've never been one to take too much for granted, so when the Rogue Bicycle Mechanic popped his nose over the horizon, proposing the use of shims to prevent creaking in UT bracket sets, I decided to brush up a bit on my knowledge of bearings, pre-load and wave washers and examine his solution to what I'd found to be a very rare problem indeed.

I contacted two suppliers that had nothing to do with Campagnolo - EZO, to gain as much information as I could about deep-groove cartridge bearings and Smalley, one of the world's leading makers of spring washers used to control bearing pre-load in a variety of situations from automotive, to aerospace, the food industry to earth-moving. I did, I must admit, also speak to SKF, as Campagnolo, for reasons of IP, didn't want to tell me the pre-load spec for their bearings ... but I spoke with a design engineer at SKF and gave him a general picture of the bearing use and the forces it was likely to be subject to - not a difficult calculation for anyone with a passing competence in geometry and resolution-of-vector calculations - and he was kind enough to give me a figure and tolerance range.

This had all gained a lot of relevance for me because by 2008, I'd been asked by the Campagnolo factory to take on their technical education role in the UK and to set up a new Service Centre for the UK. I'd been to the factory and I'd been put through my paces, along with my business partner, Jeff Beech, on a variety of their then-current technologies, slightly ironically, by the same (albeit extended by some new members) engineering team that had first shown me UT in 2004.

So having a deeper understanding of the system and how it worked or might fail had become of some importance to me.

Given access to the parts in quantity, I set about making my own tests of the spring rate and consistency of the wave washer and the dimensional tolerance of the spring clip placement as both of these factors were being questioned by the Rogue Bicycle Mechanic.

I tested 100 wave washers from 100 mixed BSC and Italian threaded BB cup sets and measured the same BBs floor-to-safety-clip-drilling dimensions, to test the physical dimensions of the wave washer and the spring rate in it's critical compression zone (i.e. the pre-load it would exert on the bearings when constrained within the cup, if the BB shell was within the 1.6mm overall width error envelope allowed by Campagnolo, for screw-in cups). I was interested in how much lateral movement the spring clip might allow.

I found that the pre-load exterted by the wave washers in all cases fell within the range recommended for the 3 types of 6805N deep-groove cartridge bearings supplied by SKF to Campagnolo for the application.

With new clips, I found that the side-to-side play in the cranks would fall within the +/-0.125mm range set by Campagnolo.

Field experience subsequently showed a small amount of wear could occur on the clip and the hole through which it passed could ovalise but that annual replacement of the spring clip removed significant problems from this area.

I also thought a bit about other ways the same effects could be achieved - could a fixed method of pre-load be used to both control any lateral movement, to preload the bearing? Of course,the answer is "yes", there are other ways.

Methods that immediately spring to mind are:

  • Two threaded rings, one inside the other, could be used to take up variations in BB shell width (as Campagnolo later did in the OverTorque design), but measuring the inter-ring torque in order to generate a known pre-load would be awkward and the actual pre-load generated would change as the bearings were subject to wear.
  • The bearings could be accommodated inside a casing with screwed-on covers, one of which could double as a pre-load cap, as Mavic did in their 631BB design in the 1990s but that would add significantly to weight and cost.
  • A series of shims at very fine gradations of width could be used to pre-load the bearings against the compression generated by the Hirth joint but that would require a level of measurement which would be beyond a good many mechanics and most consumers (even 0.1mm is a country mile in shimming-a-bearing of these dimensions, I was told by a senior design engineer at EZO, one of SKF's main competitors) ...

So in all, although a more complex solution to hit a very specific set of pre-load requirements and lateral play limitation "can" be devised, given that the system has now been in trouble free use for 17 years, one does have to ask "why" ...

If anyone is interested I still have my lab books (of course ...) and am quite happy to share my test data and my test protocol. I will remove the actual pre-load values for the bearings as that is subject to IPR that I don't own - but the graphs I was able to create for each spring, based on five loading data points for each spring clearly show a much higher consistency than the +/-25% that some commentators, who don't appear to have done any n>1 testing, suggest.