thanks / duh - reading is fundamental...

Shucks - I was hoping to start a collection of Sigma stem christmas tree ornaments.

It is strictly a volume game....and I'm guessing based on TriRig's size, the CNC option is by far the cheapest option....or he would have been amortizing his tooling costs well beyond the life expectancy of the product.

back in the day, we designed a frame that had an integrated, structural piece that was supposed to be forged....but our projected volumes were pretty low and we couldn't justify the tooling costs. We then looked at a modified design that could use an extrusion, but that wasn't feasible so we have to go with a CNC piece. While not as strong as a forged part, it was similar (or superior) to a cast part and was the best option across the board.

---

Chicago Cubs - 2016 WORLD SERIES Champions!!!!

"If ever the time should come, when vain and aspiring men shall possess the highest seats in government, our country will stand in need of its experienced patriots to prevent its ruin." - Samuel Adams

With all due respect, but in the case of loading on a stem (which is what is being discussed here), a small bit of time with an instrumented stem and a data acquistion system would get you all you need to know about the possible loadings to create an effective design. Apply a reasonable factor of safety, and the engineering is fairly routine...especially in this day and age of PC based analysis tools.

This is true of basically the entire bicycle as well. Some companies, such as Cervelo, have done such things for a long time now: http://www.cervelo.com/.../lab-vs-reality.html, and http://www.cervelo.com/...-reality-part-2.html

Those same CAD tools also make generating a CNC program fairly easy as well. Heck, there are online sources where all you do is upload your 3D CAD model and relatively cheaply (and quickly) have CNC machined parts returned to you. For example: http://www.protolabs.com/firstcut

Now then, your opinion on this may be colored by the fact that until just fairly recently, the level of true engineering in the bike industry was on a somewhat low scale (and in some pockets is still somewhat "seat of the pants")...so that's understandable. Even now though, and with large companies (cough...SRAM road discs...cough) I'm sometimes amazed at the lack of things common in other industries...such as not doing validation testing in thermal test chambers over the expected use environments, and instead relying on "test riders".

---

http://bikeblather.blogspot.com/

"With all due respect"

i don't think you do give respect, and i don't mean you disrespect the people actually involved in bike manufacture and development, but i don't think you respect what actually happens in the field. i don't know when i've seen as many recalls in the bike business (let alone the car business). do you think SRAM does not employ engineers? or specialized? or cervelo? do you think engineering just got discovered? yes, the tools engineers used have advanced greatly over the last decade and a half, but let me ask you this, as a thought experiment: why do fatigue testing machines exist? why are they needed at all? are they needed at all?

---

Dan Empfield
aka Slowman

> why do fatigue testing machines exist? why are they needed at all? are they needed at all?

Answer: unmodeled dynamics.

Even million dollar finite element analysis programs running week-long simulations on supercomputers only crudely approximate a subset of the conditions a part may see in the chaos of the real world.

I can almost guarantee that the TriRig dude used SolidWorks or whatever to simulate a lifetime of hitting potholes with a super-stiff aluminum bike while being ridden by Lampre-kit guy, and then added a huge safety margin. But that's just a starting point.

Dan- It looks like Nick is saying the original stem was die cast, and not forged or machined...The new one looks to be machined. Traditionally in terms of fatigue strength Die Casting < Machined < Forged

"Answer: unmodeled dynamics."

aka unanticipated shit that happens. which is replete in the world of bike manufacture, because there are so many unanticipated or unquantifiable or even qualifiable or identifiable forces subject to a bike during cycling. further, even testing machines don't catch everything. as a bike maker we encountered failures in the field we could not duplicate in the lab, with testing machines.

---

Dan Empfield
aka Slowman

"Traditionally in terms of fatigue strength Die Casting < Machined < Forged"

i am no expert. i am just an old ex maker of bikes and bike stuff. in my experience, just about everything is machined. nothing is JUST forged. that said, one big issue is what you machine it out of. i'm open to being educated here, but it seems to me nick is paying special attention to the underlying condition of the billet or block or thing that gets chucked up into the CNC mill, to a degree he previously did not. T6 condition of that 7000-series aluminum he's using is a stress relieved billet entering the machining process. starting with a forged blank and machining in the fine points, including threaded holes, precise holes for steerer and handlebar, is another typical and acceptable way.

in my experience, the parts of the bike that are subject to a lot of repeated twisting and loading in various axes, if they're made of aluminum, want to be forged. stems, cranks, derailleur bodies, hub flanges are good candidates. even then, some forged products are still not ready for end use. cranks, in the old squarehole days, needed to have their squareholes cut via a broach and coin process, to make the squarehole hard enough to withstand the pressure against the BB spindle. it was not really understood, during the 90s, why this was necessary, and a lot of up and coming crank makers found this out the hard way. along with their customers. me included.

nothing makes you an expert in the bike manufacturing business like failure. it also causes you to be very respectful of the methods and processes used by bike makers over the past century and more.

---

Dan Empfield
aka Slowman

Dan- I agree with everything you said about manufacturing and your understanding about how parts are made. The only thing I am unsure of is whether the sigma stems were cast (i.e. molton metal poured into a mold) and then machined to make them pretty and put in functional details like threads, or if they were cast (see previous description), then forged (heavy hammer coming down on cast part to align the grain in the metal), and then cnc'ed. I take it that you believe it was the second set of processes, but I don't see anything in the letter to indicate that there was a forging step. The forging step would add a lot of benefit for a part of this nature.

"I take it that you believe it was the second set of processes"

no, i guess i didn't explain myself well. my original post to this thread was to question whether the better remedy to replacing these stems with a machined piece was to replace these stems with a forged piece (a secondary machining process being necessary to cut threads, etc.). my question is the use of machining as the major way to create the piece, in this particular application. it just did not work out so well in years past with cranksets. however, nick is starting with a properly conditioned metal (T6 7075) which i think addresses this concern to a significant degree. he's also spreading out the bolts from each other, and perhaps adding material high stress areas.

---

Dan Empfield
aka Slowman

Gotcha...I think I understand a little bit more where you are coming from.

In regards to manufacturing process affecting the material properties there are a couple aspects that effect the overall 'strength' (using this term generally) they are
material (i.e. chemical composition)
porosity (is there any air in the part)
hardness (how hard is it)
grain structure (are the particles in the material all aligned)

Nick appears to be switching the material to 7075 from ADC12, I have never played with ADC12 but going to 7075 has rarely been a bad move. He appears to be switching from a casting where the porosity COULD be an issue to a billet where it will almost certainly not be a factor. The hardness was not specified before and with a casting, unless it was heat treated afterwards, the hardness would assumed to be pretty low, but switching to a 'T6' (T stands for temper and the 6 is a specification call out) it will be a known quantity and will almost be certainly hard enough. Which takes us to grain structure. This is where forging can help significantly over a billet piece. Most, but not all, billets will be cold forged before selling it. The grain structure will all point in one direction depending on how it was forged. When you machine the billet you will be cutting across the grain structure and weakening the part slightly. When you forge the part into a close to final shape you will align the grains to go with the curves of the part, thereby strengthening the part.

This is all a simplification, and I'm sure someone will point out where I glossed over some details but I think it adds to the discussion overall.

In regards to the CNC'ed cranks of the 90's they were usually putting a soft CNC'ed aluminum crank and bolting it to a hardened steel spindle, and then wondering why the crank lost.

"In regards to the CNC'ed cranks of the 90's they were usually putting a soft CNC'ed aluminum crank and bolting it to a hardened steel spindle, and then wondering why the crank lost."

the cranks of that era also lost the battle against hardened pedal spindles, and that is the battle they lost most often.

"When you forge the part into a close to final shape you will align the grains to go with the curves of the part, thereby strengthening the part."

i referred to grain size and orientation in my posts further up, which is why forging is, to me, the preferred process when using aluminum in bikes in areas of repeated bending and twisting. used to be jo klieber (syntace) was the king of stem testing. he built his own testing machines and even competing stem makers would send their stems to him and he'd fatigue test them. assuming this is still the case, if i were a stem maker i'd send my stem over to jo and have him test my stem.

in fact, here you go, the red monster:

---

Dan Empfield
aka Slowman

Sure forging has ideally the best material properties, but engineering is all about the compromises. First, actually getting those optimal properties from forging is not straightforward and there are many errors you can make in your forging process that may result in less than ideal properties. Second is obviously cost, forgings are just so expensive and CNCing aluminum is just so so much cheaper. Even in aerospace where we probably worry a bit less about cost than in cycling, there are some big machined aluminum pieces used, some even directly replacing a forged part. These parts are under all sorts of bending and twisting. Sometimes a forged part is replaced with a machined part that is even lighter, because of compromises in the design because it is forged. The speed that you can machine aluminum at is pretty spectacular. As long as it is designed properly a machined part would last as long as a forged part, assuming both are designed for the similar loads/lifetime.

I apologize if you misunderstood...I must not have been clear that my expression of respect was aimed at you and that I was responding to the portion of your statement above that I put in bold which, if I may paraphrase, stated that the loading on a stem is not easily determined or able to be modeled. Actually...it is.

Slowman wrote:

do you think SRAM does not employ engineers? or specialized? or cervelo?

Of course they do. In fact, I linked above to some of their output, didn't I?

Slowman wrote:

do you think engineering just got discovered?

Of course not. But, as I also said above, there are still "pockets" (and surprisingly many of them in the bike industry) where the design process is basically "let's just try this and see what happens".



They should exist as design validation and process verification tools, NOT as design tools. If something is going to fail under fatigue loading, the likelihood of that should be established well ahead of that prior to building any test models using analysis tools (be it hand calculations or computer modeling, or both), and that likelihood reduced by design. Testers such as that should be used (like the thermal chamber I described above) to validate the design and to catch things such as manufacturing process errors and the like prior to a product being released to production.

---

http://bikeblather.blogspot.com/

I don't think that's correct. If you don't know the load case, how do you set up a fatigue tester to apply the loading?

Like I said above, it would be fairly easy to develop a fairly comprehensive load case for a bike stem. If you know the load case to set up a fatigue tester, then you have the load case to do modeling on as well.



I think using HALT/HASS testing may be more along the lines of what you were thinking...

---

http://bikeblather.blogspot.com/

I'm not certain from where you are developing your thoughts on the state of engineering in the bike industry, but I can tell you from personal experience that it is wrong.

One of the companies I worked for tested EVERYTHING before we could spec it.....static load testing, fatigue load testing, etc. When we designed our own parts, they were "tested" digitally and then in our lab and then under riders.

Are there some companies who just spec stuff w/o testing them? Sure...same as any industry. But to say that engineering was something that was just recently discovered / employed by the bike biz is, quite frankly, laughable.

---

Chicago Cubs - 2016 WORLD SERIES Champions!!!!

"If ever the time should come, when vain and aspiring men shall possess the highest seats in government, our country will stand in need of its experienced patriots to prevent its ruin." - Samuel Adams

"I apologize if you misunderstood"

you apologize if i misunderstood, or you apologize because you described your point inartfully? ;-)

"They should exist as design validation and process verification tools, NOT as design tools."

i'd like to know whether the engineers in the bike business agree in principle with what you write above (i certainly do) but act at variance with this (i suspect they do).

the modifications made to this stem, both in material and in design (e.g., "wider overall profile, and a wider bolt stance") are examples of the very failure of a product used as data points for a redesign. you mentioned that cervelo uses predictive computer modeling and that "the engineering is fairly routine" but i don't know if anybody - including anybody at cervelo - thinks this is routine or straightforward or easy. that's why everybody tests. no reputable bike brand relies on its engineering alone.

there is no reason for this "validation and verification" unless the modeling of a product's fatigue life is inexact or incomplete. let us say you modeled the aerodynamics of a bike, took it to the wind tunnel, along with putty. isn't the very fact that you brought putty using testing as a design tool? let's say that you engineered a part, tested it and, tho it should have been sound, it failed in the same place numerous times. what do you do? not use that information? go back and discover your engineering flaw and re-engineer the part? or do you just add more beef to that area?

i suspect and suggest most bike makers rely on the putty and the beef, because of the gap between how we think a bike should work and how it really does work. that's not to say that we don't start with engineering. it's just to say that it's not routine, it's not easy, and fatigue failures during testing expose the shortcomings in our engineering.

bike companies can't simply use engineering to fix the fatigue failures discovered during testing, because the very engineering upon which it relied failed to expose this weakness. so it adds beef to the area. or redesigns to inoculate the area against the failure. or changes the material. which is what nick is doing.

i'm interested to see how the bike engineers on the forum characterize the way they deal with a testing failure.

---

Dan Empfield
aka Slowman

 

>I don't think that's correct. If you don't know the load case, how do you set up a fatigue tester to apply the loading?

Attitudes like yours are why it's a "best practice" to make the testing/QA team totally independent from the engineering team. :)

A load case is an approximation. There's no way to capture all the possible vibrations/forces, etc, that could happen in the real world. I'm involved in this for military parts. We come up with mission profiles, often based on instrumentation during actual missions. Then design off those. Then run them through extreme testing on 3-axis shock-and-vibe tables (fun to watch), set to exceed the mission profile by a large margin, usually.

Then after that, the testing starts. And it can all go back to the drawing board. Because of technology the likelihood of going back to the drawing board is much lower. And cheaper. But it's pure hubris to think that it's the be-all, end-all.

Depends on your definition of "recent", no? Seeing as how I've been playing around with bikes now for >30 years, "recent" to me is more along the lines of in the last 15 years or so ;-) My thoughts are developed from my own personal observations.

"When we designed our own parts, they were "tested" digitally and then in our lab and then under riders." In other words, the load cases were defined, they were modeled, and then they were validated using test machines. Sounds good.

---

http://bikeblather.blogspot.com/

Aaah...I see...it appears you were mistaking a fatigue tester (something like what Dan posted a pic of above) for something used for HALT/HASS testing.

I've participated in many random vibe/shock tests, so I'm well aware of how that is used ;-)

---

http://bikeblather.blogspot.com/

Yes. Both. I apologized if I caused you to misunderstand BECAUSE I wasn't clear. I guess I wasn't clear again...




I would submit that those changes indicate this was possibly something that could have been "caught" prior to even prototype build with more thorough load case definition and modeling.

---

http://bikeblather.blogspot.com/

Well, seeing as how the testing I described above was our SOP back in the mid-late 90's, and the guy driving the procedures was a notable, lifetime industry vet, I'd say that goes well past your definition of "recent."

And as a side note, no matter how much we modeled and tested, we always were hit w/ surprises. Sometimes it was caught before production, sometimes not......but unexpected developments were pretty common.

---

Chicago Cubs - 2016 WORLD SERIES Champions!!!!

"If ever the time should come, when vain and aspiring men shall possess the highest seats in government, our country will stand in need of its experienced patriots to prevent its ruin." - Samuel Adams

So...do you think those procedures were fairly common in the industry at the time, or do they stand out since they weren't? :-)

Were most of the surprises due to unexpected process issues, or were they actual "design flaws"?

---

http://bikeblather.blogspot.com/

ktm520, sorry to hear you had one that broke. How bad was it?

I got the last v2 Sigma three weeks ago. I wanted to see if it would let me convert my Shiv TT over to an aftermarket 31.8mm bar system to get more adjustability than the Shiv allows. It worked great for that. As far as I know it's the only stem out there that works with the Shiv TT - my Holy Grail.

• I dialed in my position;
• much tighter, lower and faster;
• convinced I bought some speed;
• raced it twice and beat guys that were beating me in my earlier races this season;
• then I got the recall email;
• took the stem off;
• put the stock Shiv front end back on, and
• wait for v3 to show up.

Nick's doing the right thing. I support that. As bummed as I am, he must be miserable.

Nick promised the new version will definitely work on the Shiv TT.