Now that I’ve knocked down the big sources of noise, I can hear the smaller ones. Given the unexplained offset in the derailleur adjustment, it should come as no surprise that I was now contending with derailleur noise.
And yet this was a surprise, as Di2 on my HDQ, after working through the initial chain sway problem, was nearly silent. Slight noise when shifting, none while pedaling in any particular gear. With no mechanical cable nonsense, Di2 shifting is very precise and absolutely repeatable. The only explanation was improper calibration. In fact, the –16 Di2 offset (at the absolute limit of adjustment) that I’ve been running is incorrect. After recalibrating, I’m sitting at –12. Though still near the limit, –12 at least allows for the official Shimano calibration process that involves adjusting left (more negative) until hearing chain noise, and then adjusting right by four steps. Now the derailleur is silent under power and much quieter when shifting. I have no explanation for the bad initial calibration other than the fact that I was in a hurry. My first calibration of the HDQ was similarly messed up. Go figure.
Still, why the big change from the HDQ offset (+4) to the Sprint (-12)? A close comparison of the respective cassettes revealed about a 1/2 mm difference in the distance between the shoulder and the outboard (smallest) cog. Pretty close; I don’t think anything has changed in the cassette spacing of the Di2 wheel since it was running on the HDQ. The proof would be to throw the wheel back on the HDQ and make sure I get the same +4 offset after recalibrating the derailleur, but that didn’t end up being necessary.
The only remaining explanation lies with the dropout and derailleur hangar. After some quick measurements, my suspicion was confirmed: the Sprint derailleur hanger, combined with a significant ridge around the inside of the axle hole in the dropout, is about 2.1mm wider than the equivalent distance on the HDQ. That represents more than half of the range of adjustment built into the Di2 derailleur and almost exactly accounts for the calibration difference. The hangar width is about the same between the Sprint and HDQ– 8.1-8.2mm. The difference is 2mm of ridge around the axle hole that doesn’t exist on the HDQ.
Ultegra on left, Capreo on right. Shoulder-to-outer-cog distance not that different.
Sprint on left with 2mm inner ridge (in black), HDQ on right with no ridge
Here’s how the horizontal spacing works out with the Shimano road 11-speed cassette, chain, and the Sprint dropout/hangar. I got the cassette dimensions from here. From these measurements and a little fiddling with Di2, I estimate the movement of a single click of offset to be about 0.125 mm. So when backing off the offset by 4 clicks in the final calibration step, the chain guide is moving about 0.5mm, and the total range of adjustment is +/- 2mm. During calibration, I think that when shifting the chain to the left until it makes noise rubbing against the fourth sprocket, the chain is hitting the edge of the teeth. Because the teeth are narrower than the body of the cog, the chain is further left than the picture below implies.
Ultegra Di2 Cassette Spacing
At 0.125mm per click, the 16-click difference between the HDQ and Sprint offsets represents 2mm. This closely matches the measured 2.1mm difference in hangar widths. Mystery explained– all is well with Di2 on the Sprint. I prefer the dropout design of the Q but they both seem to get the job done. As a kid, I enjoyed the Encyclopedia Brown books. Now that I’m old enough to be his grandfather, I’m happy to be able to crack a case worthy of his adolescent reasoning skills.
I’m really glad to hear you got that particular noise fixed, and I’m impressed with the process and analysis. Isn’t it great when you can not only fix a problem but also be completely certain you know what caused it? I often spend the better part of my work day trying to troubleshoot obscure, even bizarre problems in electronic controls systems. There are several ways these things turn out:
– The problem vanishes inexplicably (often re-surfacing when I just need it to WORK and don’t have time to troubleshoot).
– The problem is intermittent and won’t stay broken long enough to test and observe.
– I find a fix that seems to work, but can’t find the cause of the problem or why the fix worked.
– I think I know the cause but don’t have the skill or resources to prove it and find a solution.
– After some detailed observation and experimenting I pinpoint the cause and find a workable fix.
The last one seems to happen infrequently, maybe rarely. But, I’ll admit, the process is much of what makes the job fun. Except when it’s a screaming pain in the…
Yeah, that was exactly my reaction. I anticipated the troubleshooting going south on me in some way so was very relieved when it all snapped into place. Di2 is an absolute dream again, it still surprises me how much better my ride is with that system.