Is there any reason why a two-pronged same-side-of-roller chain checker tool wouldn't work?

Question

I always saw two-prongs-from-a-bar and three-prongs-from-a-bar chain checker tools and thought that they're used the same way.

It turns out that the former measures from opposite sides of rollers, which would of course not account for the diameter of roller used.

Yet a two-prongs chain checker tool could be easily made with instructions to simply pull the chain, as the red arrow illustrates in the figure.

Is there any reason why a two-pronged same-side-of-rollers chain checker tool would be inaccurate, or would otherwise not work?

Answer 1

Broadly speaking yes, there are lots of ways to make this kind of design work. The Shimano chain wear tools are an example.

The logic of this kind of tool is that everything between the contact point with the roller and the pin is going to be roughly the same at the first contact point as the second, so they cancel each other out and leave a reading of the pin-to-pin distance, from which you can divide by the number of links to get the pitch.

That logic is better than integrating roller and bushing thickness into the measurement the way that tools like the Rohloff and Park do, but it does mean that there is still some approximation happening. It's not enough to matter in most cases, but what I will note, from working in an environment where the Rohloff tool is predominant whether I like it or not, is that sometimes that tool measures different at different spots of the chain, more than I would expect. Observing that has led me to suspect that total roller and bushing thickness isn't necessarily all that uniform always. It is most of the time though.

The main problems with measuring chains with a ruler is that it makes less trinkets for people to buy, it requires a steady hand and confidence, and since the various cute little tools have become popular, you can look like you're doing something unprofessional in a shop setting if you don't use one. As far as being accurate tools for measuring chains, rulers are unsurpassed. (Unless you count yardsticks, which are more accurate because the sample section is larger, but they're much slower and don't work with the chain on the bike).

Answer 2

The reason they are inaccurate is because of two inaccuracy sources:

• Firstly, chains do not have to have precise roller diameters. If you measure from same side of rollers, 12 half-links should be 6 inches. However, if you measure from different side of rollers, 12 half-links should be 6 inches plus whatever the roller diameter happens to be minus whatever the roller "slop" (roller wear plus half-bushing outside wear plus "slop" when the chain was new) is. Most chains have very similar roller diameters, so that different-side-of-rollers measuring tools can give a reasonable guess of chain wear. However, there are chains that have slightly different roller diameters than the most common diameter, meaning the tools won't work for such chains at all (they give so inaccurate readings that the reading is entirely worthless).
• Secondly, both rollers and half-bushings wear on the roller inside diameter and half-bushing outside diameter as the rollers are rolling around the half-bushings. These components do not make the chain out of pitch: the only wear components that make a chain out of pitch are pin wear and half-bushing inside wear. If you have a chain that is exactly half inches per half-link, but the rollers and half-bushing outsides are very worn, the chain is still half inches per half-link and doesn't cause accelerated sprocket wear despite the rollers being very sloppy around the half-bushings. How the different-side-of-rollers tools work is that they guess a particular wear rate for the pins, half-bushing insides, half-bushing outsides and roller insides. So for example if the wear rates are guessed to be 1:2:3:4, it means every 0.1mm of pin wear corresponds to 0.2mm of half-bushing inside wear which corresponds to 0.3mm of half-bushing outside wear which corresponds to 0.4mm of roller wear. In such a case, the wear that matters is 0.1mm+0.2mm = 0.3mm and the wear that doesn't matter is 0.3mm+0.4mm = 0.7mm. What makes the tools kind-of-work is that if they measure many links, the wear that matters is accumulating per every link, but the wear that doesn't matter occurs only once. So for example if you have ten links, you have 10*0.3mm = 3.0mm of wear that matters and 0.7mm of wear that doesn't matter. If the guess of wear rate ratios is correct, you would make such a tool report "barely worn" when they observe 3.7mm of total wear. However, the guess could be incorrect and you might have 3.3mm of wear that matters and 0.3mm of wear that doesn't matter, and in such case the total wear is 3.6mm which the tool would categorize as "good" (incorrectly).

I'd say the more links you measure with the bad kind of tool, the less the error matters, assuming the roller diameters are precisely the same across different chain brands. That's a big and possibly incorrect assumption. There may be chains with nonstandard roller outside diameter, and the tools give junk results for such chains.

Here's some food for thought: chain rollers can be 7.7mm or 7.64mm outside diameter (source). That's 0.06mm difference. If a chainwear tool of the incorrect type measures ten half-links, the 0.5% wear would be 0.635mm. That's already nearly ten percent error from the roller outside diameter difference. I'm sure there are pairs of chains where the difference would be more than 0.06mm, though. Also roller factory "slop" and roller inside diameter wear and half-bushing outside diameter wear can vary, so there probably is at least 0.06mm error from the wear components that don't matter and 0.06mm error from the differing factory "slop". Thus, I'd say there's at least 0.06mm + 0.06mm + 0.06mm = 0.18mm error. If a chainwear tool of the incorrect type measures 10 links, that's 0.18mm error vs 0.635mm wear measurement. The error is 28%. Thus, a chainwear tool of the incorrect type can tell 0.64% worn chain to be "good" and 0.36% worn chain to be "bad" even though the threshold is intended to be exactly 0.5%.