Technical FAQ: Oversized Pulley Wheels, long chains, and rear racks

Chain friction in systems with oversized pulley wheels, and more solutions for mounting a rear rack on a carbon frame.

Have a question for Lennard? Please email him at to be included in Technical FAQ.

Dear Lennard,
I just read a recent tech column regarding gear ratios, and I have a few thoughts I would like to share.

First, there seem to be a lot of folks who are striving to have certain gear ratios for their particular needs. One thing I don’t read much of is having non-traditional chainring combinations. After reading years ago about Rabobank using it, I tried 53-36 rings on my Hampsten road bike. I have had it in place for several years, and it works perfectly, perhaps attributable to the amazing performance of Di2 front shifting. If I am in the 36, I need to avoid the lower two cogs (I have avoided Di2 updates so I can still control my own destiny), as the chain will drag on the chainstay, since I have a standard rear derailleur and an 11-25 cassette. Based upon how that worked, I set up my gravel Trail Donkey with 52-34 and 11-34, and it, too, has been working perfectly for over a year. Obviously, neither of these combinations is officially recommended by Shimano, but they work to my satisfaction, so I think for some folks, the answer might be to consider different chainring pairings.

Secondly, a few weeks ago, I asked you about chain length and its effect on drivetrain drag, and your response was that the bulk of the drag happens on the tensioned area of the drivetrain, so chain length doesn’t have a drag effect. Given that, aren’t the derailleur pulleys on the non-tensioned side of things, and, if so, how do the oversized pulleys do their supposed watt-reducing magic? These things are certainly not cheap, and CeramicSpeed seems to be a data-based company.
— Scott

Dear Scott,
Regarding gearing, yes, you are correct that non-standard chainring combinations can widen the gear range and can often shift well enough to race on. In this article, I did following stage 15 of the 2010 Giro d’Italia won by Ivan Basso atop Monte Zoncolan, I mention how Alexander Vinokourov and those of his Astana teammates who had endured this late into the third week of the race were riding 53-34 chainrings for the Zoncolan climb. It clearly had shifted well enough throughout the rest of the stage that Vino was with the protagonists on this final climb.

That said, things were simpler 11 years ago in this regard. It was easier to mix and match chainring sizes when crank spiders were more standardized, like the five-arm, 110mm bolt-circle-diameter (BCD) SRAM cranks that Vino and Astana were using in 2010. Now, with four-arm road spiders unique to each manufacturer having unique bolt circle diameters, arms not uniformly spaced around the crank circle, and hidden-bolt chainrings, the barriers to finding the chainrings needed to set up non-standard chainring combinations are higher.

As for your questions regarding friction on the lower chain span and increasing chain length to reduce it, I asked Jason Smith, CeramicSpeed’s chief technology officer to explain the reasoning behind the company’s oversized pulley wheels. Smith set up the test that produced the data that went into this article on drivetrain friction, and this is what he said:

“You are correct that the bulk of the drivetrain friction is created on the top [chain] span, which is the high-tension span transferring power from the front ring to the rear cog. However, the friction created in the lower three spans, even though these spans are lower in tension, should not be discounted.
How much friction is created on the lower span? And how do oversized pulley wheels (PWs) help?

A CeramicSpeed pulley may help save crucial watts. Photo: Caley Fretz |

I’ll start with a paragraph from a document I produced for Friction Facts on chain friction:

From a simplified standpoint, the friction created (wasted energy) by a roller chain is generally proportional to: [chain tension] x [sine of lateral deflection] x [link articulation angle at a given engagement or disengagement point] x [number of link articulations per unit time at that point] x [total engagement and disengagement points in the complete drivetrain].

Essentially, friction is created each time an individual link, under a finite level of tension, bends (articulates) as it engages or disengages the teeth of a cog, pulley, or chainring.

Total chain friction can be viewed as the sum of the friction created by all of the links’ engagements/disengagements as the chain snakes through the front chainring, derailleur pulleys, and rear cog. Tension, link articulation angle, link articulation rate, and lateral deflection (aka cross chaining) are directly proportional to the total friction losses of a chain.

So, looking at this formula, and comparing the top span to the bottom three spans (comparative vs. absolute friction), lets 1) remove the cross-chaining variable to keep the formula easier, and 2) remove the ‘number of links per unit time’ since both top and bottom spans have identical articulation rates when compared to each other.

Then we have to look at tensions and articulation angles at each of the 8 engagement/disengagement points (EDPs) on a drivetrain. The friction at the two EDPs on the top span is relatively high because of the tension. A 53T front ring is a relatively mild articulation angle, but the rear cogs, especially a 10T, have large articulation angles. The articulation angle can vary, though, between a 10T and, say, a 50T MTB cog. Yet, the huge driver for friction creation in the top span is the tension.

Now, when looking at the three bottom spans, here’s why those spans need to be considered. 6 EDPs exist (vs. 2 EDPs on the top span), and the articulation angles are relatively large, whether a 11T stock PW, or say, an oversized 19T PW. BUT, BUT, the chain tension on the lower three spans is pretty low. So even though there are 6 EDPs, and a good amount of articulation, the very light tension keeps the relative friction in the lower spans pretty low.

For rough numbers, at 250W rider output steady-state, with a 53T ring, the top spans sees an average of about 50-60lbs of tension. The lower spans have about 3-4lbs of tension with a stock rear derailleur (RD) cage, and between 1.5 and 2lbs with an oversized one. Top span tension can be 15-30 times higher than the bottom spans But again, there is a lot going on in the bottom 3 spans.

This is where the oversized pulley wheel (OSPW) comes in to decrease the friction created in the lower span. The OSPW decreases the chain tension by decreasing the spring force applied at the cage. AND, by using larger PWs, the articulation angles decrease of the four EDPs associated with the PWs.

Let’s plug some numbers in. Say, for example, the top span creates 6 watts of friction, and the three lower spans create 3 watts of friction, given a stock drivetrain with stock rear derailleur. Now, regardless of what RD is on the drivetrain (stock or OSPW), the RD will never affect the top span numbers. So, the 6 watts stays the same. But we can manipulate the lower span friction. We know lower span friction, based on the formula, is proportional to chain tension and articulation angles. If the OSPW decreases the three lower spans’ chain tensions by 50 percent (note, the cage force creates the lower span tensions, and all three lower span tensions will always be the same because the free-to-rotate PWs create an automatic equilibrium), and the larger PWs further decrease the articulation angles of 4 EDPs by, say, 40 percent (for an 11T stock to 19T OSPW), then this 50 percent and 40 percent reduction of two variables in the formula drastically reduce the overall friction in the lower spans. In this example, the 3 watts of friction created in the lower span might be reduced to somewhere between 1 and 1.5 watts.

That’s how the OSPW works. But then onto the reader’s question about chain length.

All things equal, a longer chain actually could reduce friction, ever so slightly. A longer chain can affect the lower spans’ friction, because a longer chain allows the cage to retract further, and a retracted cage produces slightly less chain tension. This is based on the spring rate of the cage. But again, the delta in cage tension from, say, a two-link longer chain might not even be measurable. It is more important to properly size the chain, based on cage orientation (as you mention), and get an OSPW which is designed to decrease cage spring force. The OSPW cage alters the spring attachment point so the torsional spring applies less force on the cage. It is not good practice to increase chain length solely for the reason of decreasing (slightly) cage force/chain tension. Let the OSPW cage design do that.
 — Jason Smith, chief technology officer, CeramicSpeed

― Lennard

Regarding using a rear rack on a carbon frame:

Dear Lennard,
Howard might try the Tailfin rack. I use it, and it’s both fascinating in its design and stable as a rack that does not connect to the frame at all.
— Andy

Dear Lennard,
Howard asked about attaching a rear rack to his carbon bike for credit card touring. Instead of using clamps to attach to the seatstays, Bontrager makes a great rack for any road bike using rim brakes. It uses an extra-long skewer in place of the original to slide through mounting brackets on the rack. It is light and works great, and that is saying a lot from someone that is not a Trek fan. I have even used it on my S-Works Tarmac to commute to work! It is called the Bontrager Backrack Lightweight MIK.
— Todd

Dear Andy and Todd,
Nice solutions. Thanks.
― Lennard

Lennard Zinn, our longtime technical writer, joined VeloNews in 1987. He is also a custom frame builder ( and purveyor of non-custom huge bikes (, a former U.S. national team rider, co-author of The Haywire Heart,” and author of many bicycle books including Zinn and the Art of Road Bike Maintenance,”DVD, as well as Zinn and the Art of Triathlon Bikesand Zinn’s Cycling Primer: Maintenance Tips and Skill Building for Cyclists.” He holds a bachelor’s in physics from Colorado College.

Follow @lennardzinn on Twitter.