A quick experiment on CS Dimpling

[Daniel_Y]

Why?:

I never felt comfortable dimpling chainstays. Sure you can just crank on them with a round object, but it’s not repeatable and I worry it adds stress risers to the tube.

• If chainstays fail, they usually crack at the dimple.
• I wanted to come up with a mathematical-ish way to support the tube to help it ovalize and dimple.
• I wanted to test PCBway for quick turnaround on CNC parts

After months of thinking and a few 3D-printed dies, I finally gave in and decided 80% is good enough and ordered some machined aluminum dies to experiment.

The Results:

I am really happy with the results. The dimple squished the cross-section to 13mm wide (from 19mm) and added a 2.8deg bend. I estimate the dimple is 25mm in radius.

I cut the tube in half at the “sharpest point”. You can see how chill the radius of the bend is. I think a lot of that can be contributed to the sidewall support that die offers.

Now the big question is, how does it compare to squishing it against the flat vice jaws?

There is a subtle difference. The flat-plate squished chainstay (right) has a flat spot and sharper corners. Is it worth the price of the dies? Depends on your standards. On a titanium bike with a raw finish, flat spots are noticeable.

Regardless of the result, this project was a quick and dirty way to test the PCBway services, which was a great success.

The Setup:

We needed a road bike CS sub-assembly for:

• 19mm chainstays
• 700x32c tires
• 415CS
• BSA 68mm BB
• Road mid-compact 52-36

The die in the middle was a piece of scrap 2.5in diameter aluminum. Bigger might work better. It was all clamped together in a used $40 vice.

Squished in tandem:

The Die:

I wanted to use some mathematics to squish the tube in the most gentle way. I knew I needed a dimpled cross-section of 13-14mm to get the clearance required.

Using a numerical approximation of the perimeter of an ellipse (above), I estimated that it would bulge out to ~24mm. I set the depth of the ellipse to 19.05mm so that the walls would “constrain the bulge”

Did this math matter? Probably not. But having some mathematical model is better than guessing and checking.

Finally, I uploaded the .STEP file (no PDF drawing) to PCBway, and the parts arrived a week later. The cost was $120 total. On one hand, it’s expensive for a simple tool. On the other hand, for a two-of cnc’ed part delivered in a week, its pretty awesome.

Next Steps:

You may be thinking, what was the point, the result is marginally better than squishing it with flat plates! And you are right, but I learned a few things

• The mathematically derived ellipse die shape is a stepping stone for more complex dies
• PCBway workflow was great. No BS, parts arrived on time.

The more important application is Titanium chainstays

• They have a ton of spring back
• they are known to fail if you dimple them too much
• Imperfections are easily visible with raw finishes

Now I need a 22.2mm version to experiment with titanium!

[sikocycles]

[project12]

Interesting stuff as always! Can’t wait to see your results in titanium. Concerning the dies: I can recommend making them in PLA on your 3D printer. I have made a couple of dimplers like that for my steel chainstays and even after around 40 dimples they still work great. And if they ever break or wear down it will be a short print to get new ones.

New video by Michiel Burgerhout

[SlowCraftCycles]

Looks great! These dies look awesome and super useful.

[e-RICHIE]

Long ago I made a crimp tool for creating chainring and tire clearance. I haven’t used it in decades, but it’s still in the loft just in case. It probably too 30 minutes to fabricate. The chain stay is dropped into the well, and the entire thingy is put between vice jaws where I show the part who’s boss. It was repeatable and predictable, and cheap.

Here is a pretty lame Flickr album that shows some of the components.

[Album] dent

[photon]

I’ve recently been experimenting with CS dimpling and have gotten really good results using a 0.375" (9.5 mm) thick, 40A durometer rubber sheet underneath the tubes when I form them using an arbor press.

The leftmost image below is a 30x16 mm, 1 mm thick Dedacciai CS pressed directly on top of the plate of the arbor press (no rubber backing) and it flattened quite a bit in addition to receiving a modest dimple.

The center image is the same tube pressed with the rubber backing and the curvature of the outside wall was preserved almost perfectly. This tube and the flattened tube at left have nearly the same final thickness, but in this case the deformation was isolated to the dimple.

I had some scrap 0.75" (19 mm) 0.028" (0.7 mm) wall tube laying around, so I tried dimpling it using the rubber backing as well - the thought was that a thinner wall tube with greater curvature might flatten a bit anyway, but the curvature of the outside wall was also preserved almost perfectly. From memory I think this dimple reduced the cross section of the tube by about 6 mm.

I’m really pleased with the results for such a simple setup, the idea going into the experiment was that a relatively thick, relatively compliant rubber sheet would conform to the tubes while providing enough support to preserve their shape, and that’s basically how it worked in reality. The 0.375" thick, 40A durometer specs were a total guess and I haven’t tried any other thicknesses or durometers, but I bet others could work just as well.