Anatomy of a test

Earlier on, when we initially announced the Rail, some chucklehead on a forum somewhere threw the BS flag at us, saying essentially "let me get this straight - the company that makes no bones about selling other people's products that they haven't developed, and does no testing, is going to develop and launch their 'own' product. No f-ing way." It was easy to shoot the guy down, simply because we had always been completely up front about how we did what we did. When we did open mold stuff, we were unique in that we didn't BS about "we had design input here" or "this is a product we developed in house" like other people were doing. Mostly that's called lying, and we're not fans.  

Anyway, that one encounter/accusation, though misplaced since we actually had done a bunch of testing, had a big impact on us. We've always wanted to know as much as possible about how everything we work with works, and to share that knowledge wherever possible. A lot of our testing is iterative, as it should be - the real world is always going to tell you things that the bench doesn't, and what you see in the real world makes your bench tests better. Sometimes the best test is a hybrid - after paying a lot of attention to tubeless cross tire setups for the last couple of months, I've got a test track set up that is designed to produce the dreaded burp if the dreaded burp is there to be produced. 

This was a pain in the butt but worth itThis rig is one of our most important test fixtures. With it, we can measure the stiffness of any wheel we build. We use it to isolate variables like spoke count, spoke lacing, spoke choice, hub choice, and rim stiffness within a build. Benchmark a known entity, then test a controlled variant of it. Our accuracy is to within a couple thousandths of an inch. By changing the configuration, we're able to measure rim stiffness in isolation, and we've measured several samples of every rim we've build with, along with several that we don't. Testing components both as parts and then within assemblies gives good insight. For example, some rims might not be very stiff on their own, but can shine as a component of a properly spec'd assembly. We wouldn't sell you a 20h Stan's 340 front build, but put that rim in a 28 build and it becomes the backbone of a great wheel - light, stiff, strong, durable, and ready for tubeless. If you want an awesome wheel to do your heavy mileage on all winter without worry, it's a great choice.  

What gets measured gets managedEveryone's got a scale, so everyone pays attention to what things weigh. Weight is also an important part of inbound QC - we don't like to see a lot of variance at all. Alloy rims will move around a bit more than carbons, simply because as the tooling wears, the parts get heavier.  You have to account for that in stiffness testing, too - weight and stiffness impact each other, so you have to normalize for component weight when checking stiffness. "Claimed weight" is a large stack of BS - things weigh what they weigh. We fully expect that everyone who's concerned about weight will weigh what we deliver to them. Anyone who makes a purchase decision based on what something's supposed to weigh, and then doesn't weigh it? Yeah...

This is our newest test calamity, lovingly named Frank the Tank. With Frank, we can test brake heat, rolling resistance, brake track durability, bearing effectiveness, lube effectiveness, and probably a bunch of other stuff we haven't even figured out that we can test yet. This one was a pain in the butt to sort out, but will give us a huge depth of insight into a world of different stuff.  

As we develop products, we'll always try and quantify any aspect of their performance that we can. We're right now on the hunt for some great hubs that offer a good compromise between the top end hub selection that we've got now and the OEM stuff that's around. We've got some candidates identified, so we're testing the crap out of them in every dimension to see if any are going to make the team.

Mike has jokingly asked a few times whether it's possible to build a wind tunnel from the McMaster catalog. He should not joke about such things. For the time being, that remains something we're happy to outsource, but don't think I don't have a few pages bookmarked for when the time comes.


Back to blog


Thanks for the explanation of the bench test you had pictured on Instagram today. I really am enamored with your carbon wheels as well as the Pacenti builds you do. I'm hesitant jumping on the carbon hoops bandwagon but your Rail 34 model intrigues me. Keep up the interesting blog topics and I'll keep reading them. Hope to soon make a purchase as I've got the need for more speed.


I have described the McMaster catalog to folks unfamiliar with it as porn for engineers. Hubba hubba.


Definitely going to look forward to the home built wind tunnel from the mcmaster catalog. :D


Mike – Yeah. Dial indicators don't grow on trees (every "I" needs to support an "R") but we're on that one too. Some do the cowboy hat, some do the gangster lean. There's a right amount of spokes for every rim – 28 on a Rail, quite frankly, doesn't do anything better than 24 does. Rim is stiff as hell and there are enough spokes to keep it perpendicular to the hub axis. Soft rims need spokes to keep them in plane (cowboy hat), stiff rims need spokes to keep them in line with the hub (gangster lean). – Dave

Dave K

Dave, great to see someone doing some modern testing of this. Go ahead and add a dial indicator at 90 and 180 deg to the load. Then borrow a 20 spoke Z*** rear wheel and you can show the world why 28 spoke rear wheels are so much better.


Leave a comment

Please note, comments need to be approved before they are published.