Proper torque on that hub is 88 in lbs for the rear using 2x5mm allen keys. At that torque, the hub should have a small amount of play (0.002", which extrapolates out to 0.010-0.015" at the rim), it is a sign that the balls are perfectly aligned as what you are feeling is the ability of the balls to rock laterally in the races as there is no lateral preload in the system.

I think in general there are a lot of misconceptions out there about hubs. We started making our own hubs 4 years ago now when we had supplier issues that warranted it. The first lesson you learn is that a hub is possibly the most difficult part on the bike to really make well. The first 18 months of our hub project saw over 400 design revisions, and in the years since then we have essentially redeveloped the thing every year.

From a machinability and tolerance perspective I don't think I'm going out on a limb to say that our 2006 stuff is the tightest toleranced hub product ever released into the bike industry. We've put new machines on the floor which can repeatably deliver +/-0.0002" tolerances in 3 axes...something almost no CNC machine on earth could maintain in production even 5 years ago. The internals are produced on a special lathe developed for manufacturing chronograph internals for companies like Chopard and TagHeur, so we're talking numbers that are almost unfathomable, but critical in an assembly that needs to run this smoothly.

From our testing the biggest issue with disc wheel hubs is stiffness and control of geometry from left to right as most all disc hubs are made from two halves bonded into the disc. This means that lengths, concentricities, etc, can only be controlled as well as you can bond the two parts together. This is why a lot of companies use loose balls on one side and cup/cone type arrangements. A cup/cone system just uses preload adjustment to compensate for tolerance issues in any of the mating components, if the outer race is too small the cone just doesn't screw in as much, if the hubshell is too short is just screws in further...these systems also allow for angular misalignment which is why an axle doesn't necessarily have to be straight in a cup/cone hub but has to be absolutely perfect in a cartridge bearing hub.

The stiffness is simply a result of design and geometry, but poor stiffness in the hubshell will result in higher rolling friction as the deformation of the hubshell ruins bearing alignment, we use a 15mm oversized axle and the 2006 product has been totally redesigned in such a way that the bearings are spaced further apart and the shell is slightly oversized relative to last year, which actually stiffens the assembly by about 8% while we also cut about 40 grams of weight out.

All in all with a perfectly toleranced and aligned hub with the highest precision bearings (we use 10 millionths toleranced Swiss balls/cartridges, these are more than 2.5x rounder than any other ball available and our cartridges have race diameters controlled to tighter than ABEC9 standards) can save 1-2 watts over cup and cone or adjustable cartridge bearing type hubs made by traditional CNC methodologies. Our 1 millionth toleranced ceramic bearings save an additional watt or so. To give you an idea of how critical these millionths of an inch are, we tested our hubs with some aftermarket 25millionths toleranced ceramic bearings and the wattage to spin increased over the stock steel balls, so the roundness and control is of paramount importance. We have a white paper on this that explains it slightly more in depth: http://www.zipp.com/...ents/Anoteonhubs.pdf

In terms of overall disc performance I would say that the hub is in the top three of importance, but the reality is that the construction and design of the disc itself is certainly primary as a poor shape, or aluminum rim lip or other surface discontinuity can cost you 3, 4, 5 watts of aerodynamic efficiency whereas the finest hubset in the world can only save you 2-3 watts over an average hub. The other issue would be weight, not for the sake of weight itself, but because so many people seem to not ride their disc when they really should due to the event having a climb or something in it, so if buying the lightest disc possible is what is necessary for you to want to ride it whenever and wherever, then weight would be the third factor on my list. (although technically stiffness trumps weight for most course profiles and with most modern tt/tri frames, but I rarely meet anybody who isn't riding their disc in an event due to stiffness concerns).

For more information on this see the papers in our Hub Technologies section: A Note on Hubs

A Note on Bearing Technology

SHIFT (Spoke Hole Impact Forming Technology) http://www.zipp.com/technologies/hub/shift.php

The answer of whether the Z series is worth the price really depends on the individual. I've heard comments about who should buy the Z series wheels ranging from 'anybody to whom 3-4 seconds per 40k is important' to 'anybody with $3000 to spend on wheels' and the answer is really all of the above. The entire point of the Z series wheels is just to do a cost-no-object wheelset that is exactly equal to what the top pro's are riding.

Although most companies are selling product that is very visually similar to what you can buy in the store, these parts are not the same, for instance most 'aluminum wheels' used in the TdF are actually limited run magnesium rims that the pro's get for key mountain stages and the like, but you can't actually buy. Most teams also now upgrade their hub and component bearings to ceramics through third party sources (in some cases that's us), or at least to higher grade steel bearings, and 90% of the pro teams out there will rebuild their pre-built sponsor wheels using spokes and building techniques that the mechanics trust. Of course, not everybody is doing this, but many, many of the parts that the top level pros are really riding have been 'tuned' relative to what you will find in the store.

Now we've never believed in that, so we wanted to integrate all of this into our wheelsets from the outset. So in 2000 we developed the original 280 rims around a wheel to be built for Jan Ullrich at the olympics, and ended up with 280 gram rims (future 303 tubular rim), Sapim spokes and Tune hubs with ceramic bearings that we helped spec (we didn't make our own hubs then) and Jan bought 3 sets of the wheels and loved them. So we just started trying to integrate all of these technologies into a stock wheelset.

When we started with CSC, we wanted to just be albe to ship them product off of the shelf so that they weren't getting custom everything, and in the first year we make like 200 changes to rims, hubs, bearing seals, wheelbuilding, etc. to hit the criteria that was demanded, and likewise the Z series underwent the same development, so that now any random customer will receive wheels built right alongside and no different than what CSC, Phonak or Peter Reid would get, and there is no 'cherry picking' of products for these guys, but of course the top level guys still need 'tuned' parts for ultra high level events, so the Z series fits that bill, and again is a product that we can have on the shelf that may be shipped to you or to Ivan Basso without hesitation.

As for ceramic bearings, just like steel bearings, they are all not created equally. Many affordable ceramic bearings are no better than low grade steel bearings and use the same low grade races. These are primarily used in high temperature industrial applications like on carts used to roll lots of aluminum in and out of ovens for heat treating, but have absolutely no benefit to performance. Our ceramic balls are made in the US to a grade 1 tolerance (round and diameter matched to 1 millionth of an inch) making them about 1000 times rounder than a Campy Record grade 25 ball and are assembled in Switzerland into ABEC 9 races which are cryogenically heat treated to refine grain structure and even use a super groovy ceramic specific grease which costs over $250 per kilogram.

Most of the ceramic bearings hitting the cycling industry right now use either grade 25 or even grade 40 balls (making them significantly less round than the bearings used in the standard Zipp hub (grade 10)) and use traditional ABEC1 or ABEC3 steel races and traditional greases, which leads to premature wear or damage to the races as they were not designed to handle the higher hardness of the ceramic balls. Like many things, you get what you pay for, matched grade 25 balls sell for about $1 per 100, while grade 10 balls sell for about $12 per 100 and grade 1 ceramic balls sell for about $7 per ball...with 15 balls per cartridge these get expensive very quickly.

Our ceramics are very expensive, but we actually sell these bearings to 3 non-Zipp affiliated ProTour teams as well as providing them to CSC and Phonak in Z series products. Shockingly enough, the ceramic bearings are one of our lowest profit margin components as applying standard profit margins would make them almost unaffordable.

Are they worth it? Well to a guy like Dave Z, their benefit was nearly equal to his winning margin over Lance in the Tour Prologue last year...he probably would have beaten him without the bearings, but they made for a reasonable cushion. We also have age group TT's world wide where the top 2 or 3 guys in a 40k are separated by less than the 3-4 second margin afforded by the 1 watt savings of our ceramic bearings. Don't get me wrong, I can no longer break an hour in the 40k so they aren't doing anything for me, but to anybody who's ever been beaten by a razor thin margin, they are probably worth it... if you have the money of course :-)

The Z-series wheels have now been replaced by our Zed-Tech custom wheels line. You can find more information here: http://www.zipp.com/zedtech/

I think that Jens' study is great and really shows just how far we have come with steel bearings in the last 4 or 5 years. First of all, we still see about a 0.78 watt difference between the two using highly sensitive bearing testing equipment, that difference is for one wheelset or 6 bearings, but we also see a larger difference than that in using different types of glue, or even two of the same tires as production tolerances for all those things can add slight variation, such as a slightly thicker/thinner tread or more latex on the sidewalls.

Remember we’re talking about some very small numbers here and wringing the absolute last bit of performance out of the wheelset – the standard Zipp steel bearing already uses a ball that is more than 2.5 times rounder than a Campy record ball, and uses a lot of the other technology in the ceramic cartridge. And our ceramic cartridge is the finest ever used in cycling, you cannot find any other with similar specifications. So my viewpoint on this is the same as it has been for a few years now...we have nearly acheived ceramic performance in our standard bearing, which I know sounds like a cop out, but our standard bearing is a Swiss made bearing that costs 15X more than a bearing from Asia, but look at the difference. There is a reason that I can buy a complete hubset from Asia for less money than the standard Zipp bearing set, and the ceramic pushes it up one more notch, but the differences are very small. If it were me I would try to do this test using the identical tires without glue, and you will need to control the pressure very tightly, we use a gauge from the motor racing industry with 0.1 psi resolution for wind tunnel and similar testing... but even with that you are looking for a very small difference in a very complicated system...

Here’s the engineering white paper on the current state of Zipp bearing tecnology if you’d like to know more:
A Note on Bearing Technology (PDF)

Like so many other issues in cycling, my answer has to start with “buyer beware”. There is a lot of misunderstanding about about ceramic upgrades that misses one critical point. Bearings are incredibly complicated to design and spec in a wheel and I would guarantee that the ceramic bearing problems most people are experiencing on the bike are due to using bearings in assemblies they were not designed for. In general, the ceramic ball is harder and lighter than steel. It isn't necessarily rounder or better unless you specifically spec it that way. As an example, the standard grade 10 Zipp steel ball is rounder than about 90% of the ceramic balls people are selling for bicycle wheels.

One huge market for ceramics is industry, where many of the uses are specifically for hot environments, or environments where lubricants aren't allowed or possibly the parts are subject to chemical exposure. So you have relatively low grade ceramic bearings that have no need for high precision (applications like furnace carts) and also don't need high speed, but need very high load capacity. It is common to see grade 40 and 50 balls for these applications. On the other end of the spectrum you have applications that are extremely high speed and sometimes also high temp, if you buy any car with a turbocharger chances are they are using ceramic bearings as speeds in a turbo can be well over 120,000rpm and get so hot that they glow, a perfect ceramic application. But you also see them in F1 and CART car wheel assemblies due to their light weight, low friction, and extremely high load capacity, or even in wind turbines where you have very high loads and relatively low speeds of only a few hundred rpms.

Clearly every one of these instances needs a very different bearing and the same is true with the bicycle application. Choosing an industrial grade bearing designed for high temp, or a marine grade bearing with tight seals and washout resistant grease designed for corrosion resistance are not going to get you any performance gain in your wheels or BB or whatever. Also to consider are internal clearance of the bearing, type of cage, grade of ball, ABEC or ISO rating of the races, race material, race surface finish and grind quality, and finally (and possibly most importantly) the fit of the hubshell into which the bearing is pressed.

I think the hubshell fit is a huge issue as manufacturers cut their bearing bores to specs that work best with their bearings. Since even high ABEC rated bearings have a relatively large tolerance for outer diameter, and since hubs from Asia can have very large variance, you can just have the wrong bearing in the wrong bore. I have seen bearing bore variance of as much as 0.002" between hubs and even high end hubs generally use a +0.0005/-0.0005" tolerance, combined with the ABEC3 outer diameter variance of 0.0003" you can have an interference fit of nearly 0.001” in high end hubs and over 0.001” in other hubs.

In machining terms these are huge numbers. You can fix that by using a bearing with large internal clearance, but then that bearing will be very loose in a well cut bore, which in some hubs may the the bore on the other side of the hubshell. Combine all this with variance in seal friction, grease drag, difference in ball cage design and materials, and you can easily end up with a ceramic bearing that it higher friction than steel. Granted, it will still last longer and have a higher load rating, but if it is higher friction, that really doesn't matter.

We have tested bearings 4 times now in 5 years with an independent lab in california the difference between our ceramic and steel bearing, as well as other grades of steel bearing, trying to find optimal grease type and quantity, optimal internal clearance of the bearing, etc., and when it is all designed to operate together (in this case our bearing bore tolerance of +/-0.0002”, the ABEC7 tolerance of +0.000/-0.0002” and our spec for internal clearance and grease, cryogenically tempered races polished to near optical standards, teflon cage, and high hardness bearing steel races (non-stainless)), the savings under load equate to 0.8 watt for 150lb rider compared to our standard grade 10/ABEC5 steel bearing.

Our standard bearing also shows between 0.5 and 1 watt better performance compared to a grade 25/ABEC1 with brass cage and standard seal, with that 0.5-1 watt savings being the result of about a 6 factors, mainly reduced seal drag, lower viscosity grease, teflon cage instead of brass, higher ball precision, higher race precision, and use of ultra hard bearing steel instead of stainless.

Anyway, no matter what wheels you have, you are almost always best replacing the stock bearings with bearings from the hub manufacturer as they will know what works and what doesn't, and there are so many small factors in bearing specification that it can be pretty easy to end up with a very expensive bearing that is slower than what you started out with.

You can find more information about bearings at:

Si3N4 (Silicon Nitride) Ceramic Bearings http://www.zipp.com/technologies/hub/si3n4.php

It is not the ceramic that makes most ceramic bearings so good, but the ultra-precision tolerances that can be held with ceramics. However, just like steel balls, you can get ceramics in low tolerance grades at lower prices. These are generally used in high temperature industrial applications like furnace carts where the ceramic has the temp resistance necessary but only needs to be low tolerance.

Just looking at what is out there right now, almost every ceramic bearing you can find is actually less round than our standard steel ball bearing (granted we are using the highest grade steel balls of any company in this industry), so it always makes me cringe that somebody wasted $100 or even $200 of their money to put grade 40 ceramic balls in one of our wheelsets when we already ship them with grade 10 steel balls, so in effect you are paying all that money for a less round, lower quality ball, and the extra hardness of the ceramic can in no way make up for poor roundness. The only company other than us widely detailing their spec is FSA, they are using a grade 3 ball. By comparison we are using a grade 1 ball, (3 millionths of an inch variance vs 1 millionth of an inch) but to give you an idea, a grade 3 ball costs about $5 and a grade 1 ball of the same size costs about $15, so with ball bearings in general and ceramics in particular you really get what you pay for. By comparison, most mail order ceramics that I see people buying these days are using at best a grade 25 ball and even more common a grade 40 ball (a grade 40 ceramic ball costs about $0.30) this is the only way to get the costs down to those levels.

All told, the difference between our standard bearing and our ceramic is roughly 0.75-1 watt for the entire hubset (6 bearing cartridges), and the difference between our grade 10 steel ball and a Campy record or DA spec grade 25 ball is about 0.5-0.75 watts for the entire hubset.

Here some more info on bearings from our engineering white papers: A Note on Bearing Technology (PDF)

The dust cap on the hub is a possible source of noise, primarily rattling, but occasionally a creaking if the cap is displaced. This issue was solved in mid-2005 model year with a spring loaded plastic clip that replaced the o-ring system and unfortunately is non-retrofitable. Skewers are generally the number 1 source of a creak, and we recommend lubing both the cam mechanism and the threads on the QR, as either side can be a source of creaking.

Lastly, the spokes can be a source of creaking either at the hub or rim. A drop of lube on each spoke hole at the hub, and on the spoke/nipple interface at the rim will be the solution for that. I'm not sure of the age of the wheels, but often a creak or spoke noise can be a sign of reduced spoke tension in older wheels. This is rare, but the spokes are stainless steel and will exhibit some creep over the years especially under heavier riders or particularly rough roads (though we pre-stress the spokes 4 times during wheel build pre-stretching them by over 1mm, they can still continue to stretch slightly). This creep (stretch) can be offset by re-tensioning the wheel which is something your LBS can do quite easily, and they can check the tension for you very quickly to know if the wheel needs tensioning. Standard tensions are 100kg front and rear drive side with 80-90 kgs rear non-drive.

Good eye on spotting the hubs. For the record, it was Tyler Hamilton who wanted us to work with Phonak when he went there from CSC, but our company is rather small and family owned, so we just couldn't afford the sponsorship of two teams, plus we make what is probably the most expensive to produce hub available right now, so DT and cyclops came on to supply hubs for the team and we became Phonak’s rim sponsor. However, for all critical stages, you will see the team on Z series wheels with our ceramic bearings and hubs.

First the noise issue. Our previous generation of 202 hubs had a heavier pawl spring than what we are using now, we have since replaced this in the new design with a Phynox watch spring, this is the same stuff they use in high end automatic watches as well as avionics equipment, and allows for the use of a spring that is roughly half the thickness and therefore less stiff. However, the reason the hubs seem loud is because the spoke tension telegraphs the hub noise out to a hollow rim which acts like a drum head.

The king hub is is actually slightly louder than ours unless you continually pack it with grease, and sounds more like a swarm of bees chasing you when laced into one of our wheels. Any hub will quiet down if you lube it, we recommend 10W-40 Mobil1 synthetic motor oil, this is the base oil of most every high end lubricant on the market, is quite cheap by comparison, and is compatible with the barium-hydrocarbon grease in the bearings.

I've said it here before, but hubs are a very difficult product to make as everybody has a perception of what a hub should be and should do. We have chosen to go for the lightest weight and most efficient spinning hubs on the market, and that leaves us with a rear hub weighing 182 grams and spinning with roughly 1-1.5 watts less resistance than anything else currently available. Our Grade1 ceramic bearings save another 0.75-1 watt. Now to get there, we are using Alcoa 7XXX series aluminum and making all parts in the US on Swiss type machines with the tightest machining tolerances ever used in this industry (even including King) and perform 100% dimensional inspection of all bearing bores and axial tolerances in order to match parts for the tightest possible final assembly. But you have to understand that these are ultra high end racing parts, sort of like the formula1 car of hubs, they can't be everything to everybody. The main complaint we get is that there is no bearing adjustment, but every hub is set from the factory with bearings aligned axially within 0.0005", which is tighter than most CNC machines can hold bearing bore tolerances (we hold those to 0.0002").

This gives the ultimate in low friction and precision, but is a number so small that no human could possibly adjust the hub within that tolerance band using adjustable cones, so we have to match hubshells at assembly and fix this dimension with machined parts. The new hubs have a higher tooth count on the ratchet (up to 30, same as Campagnolo and tighter than most everybody but King who has 72, industry standard is 24) and engage faster than old designs, but still require 12 degrees between engagement teeth compared to King's 5, although the reduced tooth count increases coasting efficiency as internal frictions are severely reduced in our design over theirs.

For those who haven't like our hubs in the past, you should really look at the 2006 models, they are significantly different from anything we've previously done, are extremely user friendly and have thus far proven to be unbelievably solid considering their weight (CSC has been racing and training on prototype and production models of this hub for nearly 2 years and found it to be nearly bulletproof)

Here’s a little more information on the desing and production of our hubs: A Note on Hubs (PDF)

SHIFT (Spoke Hole Impact Forming Technology) http://www.zipp.com/technologies/hub/shift.php

The reason that the millionths matter so much on the balls is that ball tolerances cover both sphericity and diameter. If you think of having balls in the race at the high and low end of the tolerance, you end up with the smaller balls not even making contact, thus concentrating the load on the larger balls. Since the sphericity of the balls is also covered in this number, you start to realize that the roundness of a 40millionths ball vs. a 10 millionths ball is dramatic, this is why Campy always had such a good reputation, they were the first to use true 25 millionth toleranced balls back when they were very expensive and hard to get (and everybody else was using 40-50 millionths balls), but as a result their hubs were always butter smooth.

There's a million other things to consider as well, like we were the first in the bike industry to use thermoplastic ball retainers instead of brass, this saves like 1gram per cartridge and can nearly double bearing life, and now numerous companies have gone in this direction. Grease is another huge factor, we work with a German company and we spent 4 months in testing to determine the optimal type and amount of grease in the cartridge. The stuff we ended up with costs something like $100 per kilogram, but as a result the bearings run with lower torque values and last longer than anything else we can find. After about 6 months of CSC being on the stuff we started getting calls from some of the other teams as the mechanics were hearing that CSC never replaced bearings, so we now sell our bearings to a number of other teams and their suppliers, so in our opinion those millionths of an inch are in fact a very big deal.

As a machinist you will appreciate why we feel that our American made hubs will never be supplanted by Asian made hubs, mainly air-conditioning. We temperature control our manufacturing facility and our cutting fluids within a 2 degree F window. This guarantess that parts made today are identical to those made tomorrow and so on. At the tolerances we work with a 10degF shift in temperature will take a part out of the tolerance band, but if all parts are made identically at the same temp, then they will all grow and shrink identically once they are assembled and out in the real world.

I often joke with our Taiwanese agent who is always trying to sell us hubs that they will never get there until they air condition their factories (this is mostly a joke as the super volume oriented Asian machine shops also shy away from ultra tolerance equipment for the sake of equipment that can produce larger volumes, whereas for us volume will never be an issue as we will never make enough to fill the machines we have, so we concentrate on making them perfectly).