Left to read about the Blue Harry Perry
Although the Blue Harry Perry is still a fine bike, and probably as good a bike as one could ever have for long-distance touring, it is not exactly sporty by today's standards. The wheelbase is long (contributing to the excellent ride) and it is heavy. So since I started riding more seriously (due to Cycle Oregon XI in the summer of 1998), I've been thinking of getting a new bike.
My resolve was strengthened this summer by the difficulty that I've been having getting replacement parts for the Harry Perry, particularly for the CycloTouriste chainrings and the bottom bracket set, which are the heart of the bike.
Largely for budgetary reasons, I decided to stop dreaming about one of the Trek OCLV laid-up all carbon fiber frames, and purchased instead a 1994 vintage Trek 2010. This bike has carbon frame tubes but aluminum stays and forks. This particular model came with Shimano RX100 triple chainset and 7-speed gearing with bar-end shifters.
|
|
The bike was rideable when I bought it, but was clearly in need of some TLC. The chainrings were very worn, particularly the middle ring on the 30-40-52 setup, which had some teeth completely ground down.
My options were to do the minimum that would make the Trek 2010 smooth, reliable, and meet my needs, or a more extensive set of modifications.
Smooth and reliable meant replacing the chainrings and cluster. Meeting my needs meant adding a saddle that fits me and some lower gears. This would have been possible simply by hunting down a wide ratio (mountain-bike) 7-speed cluster, replacing the middle and small chainrings, and the chain, and adding a longer arm rear derailleur. This could all have been done for about $270. The problem with this is that I would once again have an obsolete bike, and in another couple of years, when these parts need replacing again, I'll be back on the Internet looking for sources for Shimano 7-speed SIS equipment. I would also not have the advantage of STI levers, which everyone tells me will change my life for ever. (I am skeptical of this).
So instead, I decided to go the full distance and upgrade to a 9-speed transmission with STI levers. The main additional items necessary for this are the levers themselves ($200) and the 9-speed ratchet body. Of course, I also have the additional labor required to attach the levers and cables, re-dish the rear wheel so that it is centered with the wider cassette, etc. With the FlightDeck cycle computer, lights, paint, and some cosmetic changes, the total bill is more than I spent on the frame. But for this I will have what is essentially a completely new bike, save the chainset, headset and brakes, which are all quite satisfactory.
I also considered replacing the aluminum forks with carbon forks. This is another $160 if one takes advantage of the Performance sale. Interestingly, the current Trek 2000 series bikes have aluminum frames and carbon forks, while this one has a carbon frame and aluminum forks. The ride seems quite satisfactory right now, so I decided to leave the existing forks in place; I would love to hear from someone who has done the fork upgrade as to whether the change in ride characteristics make this upgrade worthwhile.
Given the decision to replace all of the drive train, I have a blank slate to design my gearing. Well, not quite. Shimano has really limited my options by the decision to use a 130 mm bolt circle diameter (BCD) for the middle chainring on all of their road cranks. With 170 mm BCD, it is not possible to have a middle chainring smaller than 38T. Using 110 mm BCD , as on many mountain bike chainsets, would allow a lot more flexibility.
Here is my first shot at working within these constraints:
|
rings |
48 |
38 |
26 |
ring ratio |
1.263 |
1.462 |
mean |
1.359 |
overall |
1.846 |
|||||
|
small sprkt |
12 |
wheel size |
27 |
||||||||||||
|
outer-middle overlap |
7 |
desired sprkt ratio 1 |
1.124 |
||||||||||||
|
middle-inner overlap |
6 |
desired sprkt ratio 2 |
1.135 |
||||||||||||
|
Nr sprockets |
9 |
mean sprkt ratio |
1.129 |
actual mean |
1.107 |
||||||||||
|
sprkts ideal |
12.00 |
13.28 |
14.70 |
16.26 |
18.00 |
19.92 |
22.05 |
24.40 |
27.00 |
||||||
|
sprkts integer |
12 |
13 |
14 |
15 |
17 |
19 |
21 |
24 |
27 |
||||||
|
delta squared |
0% |
7.9% |
48.6% |
160% |
100% |
84.7% |
109% |
16% |
0% |
error |
2.0 |
||||
|
gears |
108.0 |
99.7 |
92.6 |
86.4 |
76.2 |
68.2 |
61.7 |
54.0 |
48.0 |
||||||
|
85.5 |
78.9 |
73.3 |
68.4 |
60.4 |
54.0 |
48.9 |
42.8 |
38.0 |
|||||||
|
58.5 |
54.0 |
50.1 |
46.8 |
41.3 |
36.9 |
33.4 |
29.3 |
26.0 |
The 12-27 cassette is the widest ratio non-MTB cassette in the Shimano lineup; lets take a close look at it.
Unfortunately, what we see is not nice. The green "sprkts ideal" row in the chart gives the ideal intermediate cog sizes -- the cogs that divide the 12-27 range equally. Whoever selected the sprockets for this cluster must have had too much Sake, or perhaps he was trying to use up overstock components. The 14T ,15T, 17T, 19T and 21T cogs should all be one tooth bigger! There is nothing that we can do about this, however, because the 17-27 group of sprockets are a single spider module, and cannot be changed out. It is tempting to think about throwing away the 14T sprocket and adding a 16T sprocket, but there is no 16T sprocket to add; Shimano does not make one that is not part of a spider module. We are stuck with a cumulative sprocket unevenness error of 2.0.
What about chainrings to go with this cluster? As noted above, the smallest middle chainring that can be used with a Shimano road crankset is a 38T. The largest inner chainring that I would consider with a 27T rear cog is a 26T ring. 38/26 is 1.462; if we apply that same ratio to the middle and outer chainrings, we would obtain an outer chainring with 55 teeth! This is clearly absurd with the 12T sprocket, even if it were obtainable. So we may as well give up on spacing the rings evenly, and instead choose the largest ring that does not lead to an absurd top gear, say a 48T ring.
The gears that result from this set of compromises is shown in blue. Although I have attempted to line up the rows, because of the uneven cog ratios, this isn't possible. For example, although the 54" gear can be obtained on any of the chainrings, the steps above and below this are of different sizes on each ring. Similarly, the middle chainring gives us two gears between 68" and 86"; the outer ring gives us only one in essentially the same interval.
One last set of number to look at are the ones that determine what equipment to buy.
|
chainring difference |
22 |
|
biggest run |
69 |
|
smallest run |
40 |
|
overall diff. |
29 |
The chainring difference is the difference in sizes (measured in number of teeth) between the largest and smallest rings. This translates into the difference in diameters of the chainrings. The cage of the front derailleur must be long enough to clear the large ring and have the chain run freely when it is on the small ring. Fortunately, 22T is within the listed capacity of most standard derailleurs, since it is exactly the difference between a 30T inner and a 52T outer.
The biggest run is the sum of the sizes of the largest chainring and the largest sprocket that will be used with that chainring. In our case this is 48+21=69, since I assume that the two smallest sprockets will not be used with the largest ring. Conversely, the smallest run is the sum of the sizes of the smallest chainring and the smallest sprockets with which it will be used, in our case 26+14=40. The difference between these numbers, 69-40=29, is the amount of chain that the rear derailleur will have to take up. Once again, this is well within the capacity of of one the long-arm road bike changes, like the Ultegra GS. In fact, I expect that some of those crossed combinations will actually be useable.
If we reject the 11-27 cluster as having too uneven or too close a set of cogs, we are left with the MTB clusters, of which the 12-34 seems most promising. (All of the others have an 11 tooth small sprocket, which is pushing the envelope a bit too much for my taste. 11 tooth sprockets must be used with a special Hyperglide "C" ratchet body, because the cutouts for the body's splines don't go all the way through the sprocket. If they did, there would not be enough metal left to hold the sprocket together! Lets take a look at this.
|
rings |
48 |
42 |
28 |
ring ratio |
1.143 |
1.500 |
mean |
1.309 |
overall |
1.714 |
|||
|
small sprkt |
12 |
wheel size |
27 |
||||||||||
|
outer-middle overlap |
8 |
desired sprkt ratio 1 |
1.143 |
||||||||||
|
middle-inner overlap |
6 |
desired sprkt ratio 2 |
1.145 |
||||||||||
|
Nr sprockets |
9 |
mean sprkt ratio |
1.144 |
actual mean |
1.139 |
||||||||
|
sprkts ideal |
12.00 |
13.67 |
15.57 |
17.73 |
20.20 |
23.01 |
26.21 |
29.85 |
34.00 |
||||
|
sprkts integer |
12 |
14 |
16 |
18 |
20 |
23 |
26 |
30 |
34 |
||||
|
delta squared |
0.0% |
11.0% |
18.6% |
7% |
4.0% |
0.0% |
4.3% |
2% |
0.0% |
error |
0.638 |
||
|
gears |
108.0 |
92.6 |
81.0 |
72.0 |
64.8 |
56.3 |
49.8 |
43.2 |
38.1 |
||||
|
94.5 |
81.0 |
70.9 |
63.0 |
56.7 |
49.3 |
43.6 |
37.8 |
33.4 |
|||||
|
63.0 |
54.0 |
47.3 |
42.0 |
37.8 |
32.9 |
29.1 |
25.2 |
22.2 |
The cogs selected for the 12-34 cluster are the best possible choices to span this range with even ratios, indeed, the cumulative deviation from the ideal (0.638) is quite small. I selected a 48T large chainring, because with the 12T sprocket one really doesn't need anything larger; a top gear of 108" is more than I ever need to push. (At a cadence of 80 revolutions/minute, this equates to 25.7 miles/hour; I stop pedaling and start drafting before that!). Similarly, I selected a 28T small chainring because a22.2" gear seems low enough.
Having selected a cassette, especially one with nice even spacing like the 12-34, one has effectively chosen a "step size" for the gap between adjacent gears. In this case, it is approximately 1.139 (the 8th root of 34/12). The gears obtainable on different chainrings will line up if the gap between the chainrings is a multiple of this step size. If we start with 48T, 49T and 50T large rings, the possibilities are (reading across the chart):
|
48 |
42.14 |
37.00 |
32.48 |
28.52 |
25.04 |
|
49 |
43.02 |
37.77 |
33.16 |
29.11 |
25.56 |
|
50 |
43.90 |
38.54 |
33.83 |
29.70 |
26.08 |
So 48-34-28 looks pretty good; 48-34-29 would be better, but 29T inner chainrings are hard to find. 48-37-29 would be even better, but here one again we run up against the tyranny of the 130 mm BCD fixing for that middle ring: you can't get a 37T ring on those centers. 49-38-26 looks pretty good too. 50-39-26 might be another good choice, except that the 112.5" top gear that results from the 50T ring and the 12T cog is rather high. Remember, all of these combinations are specific to the 12-34T MTB cassette; change to another cassette, and the ring ratios need to be recalculated.
Going back to the main gear table above, selecting the chainrings and sprockets in yellow gives us the gears in blue. It is pretty clear that the middle chainring is not buying us anything at all, except extra weight. Looking at it another way, the outer ring is giving us a single high gear, the 108", which will only every be used downhill with a following wind. In fact, I've been happy for the last 24 years with a top gear of 96", so I surely won't miss the 108".
Where I'm heading is to ask: why use a triple chainring at all? Why not just settle for a double? Of course, the double that we want will have an inner ring of 28T and an outer of 42T or 48T, so this isn't a double that you can buy from any of Shimano's road series cranksets. But, since the outer and middle chainrings fix onto the same bolts, we could just omit the outer ring and use shorter stack bolts on a "triple", and presto, we have a "double".
Lets look at the vital statistics:
|
chainring difference |
20 |
|
biggest run |
78 |
|
smallest run |
42 |
|
overall difference |
36 |
The 34T large sprocket means that we must choose a mountain bike rear derailleur, such as the Shimano Deore XT, but that should cope with the 36T overall difference without any difficulties. Note that this time, in computing the biggest and smallest runs, we have excluded only the single most crossed ring-cog combinations, rather than the two most crossed combinations as we did previously. This is because with the double chainring setup, we are more likely to want to use these combinations, and because we can increase the separation between the chainrings slightly to make it more likely that the chain will run without rubbing.
Lets call the first setup described above, with the 12-27 cluster and the 48-38-26 triple rings, the close ratio triple. We'll refer to the second setup, with 48-28T rings and the 12-34 cassette, the wide ratio double. How should we choose between them? Let's do some comparisons.
In the end, it comes down to a toss-up. I ordered the components for the "close ratio triple". However, 12 days and five 'phone calls later, I'm still waiting for Performance to deliver the 12-27 cassette. If it doesn't come tomorrow, I may give up waiting and try the "wide ratio double" instead. Of course, if readers want to contribute money to the cause of scientific research, I might be persuaded to implement both setups and compare them in the field.
So far. so good! I rode about 78 miles of the Torture 10 000 on it and enjoyed (almost) every minute.
The things that gave me trouble were re-dishing the rear wheel to accommodate the 9-speed cassette: some of the spokes were frozen, and the nipples would round off and have to be cut out just before I got the wheel perfectly round, so I had to start over ... This happened about four times :-(
Getting the front derailleur to work took a bit of tinkering. The chainset being from a 7 speed setup means that the rings are just a bit too far apart. The front derailleur requires all of the cable that the STI lever can pull to cover the range. Don Bolton is letting me try a newer Shimano 105 chainset, which is about 1.5 mm narrower, and this should help.
The handlebars have too deep a reach for me -- about 11 cm rather than the 8 cm that I have on my other bikes. I noticed that I was too stretched out, especially since the STI levers are large, and require riding on the tops a lot. I have ordered some 8 cm reach bars from Nashbar.