|
|
|
|
blondgecko
Moderator
Jun 16, 2009, 10:28 PM
Post #101 of 190
(9969 views)
Shortcut
Registered: Jul 2, 2004
Posts: 7666
|
adatesman wrote: cracklover wrote: adatesman wrote: cracklover wrote: What do you think? I think we're entering the land of ill-informed conjecture Roy is complaining about.   Seriously though, I think there's a definite problem with Aliens there somewhere given that some of them didn't slip and other brands never seem to have a problem. What I mean is that neither the degree of softness of the lobes, nor the degree of high effective angle due to mis-drilled axle holes, seem to explain the Aliens slipping from the fixture. So since you're making the argument for those being the primary reason why Aliens are slipping out of the fixture, I thought perhaps you'd like to defend that idea in light of your data. GO Gotcha... Um, well, you've looked at that part of the data more than I have and frankly I think you're right about the data not supporting my suspicions. Given that, I guess I don't have a theory that explains why they're pulling out. But that's certainly what they're doing, so there's a reason for it there somewhere. For argument's sake I will be retexturing my fixture again and testing a couple more and will report back with those results (it'll probably be next week sometime). In the meantime I'll see about finding somewhere to send the axles for hardness testing.
The wildly varying numbers for hardness say there's something going on with the alloy, but shear strength is the more applicable measurement here. Unfortunately, it's rather more difficult to measure.
|
|
|
|
|
jamincan
Jun 16, 2009, 11:55 PM
Post #102 of 190
(9934 views)
Shortcut
Registered: Nov 1, 2007
Posts: 207
|
jt512 wrote: qtm and cracklover, I realize that several people have done enough work to answer some of my concerns informally, but no one has yet done a systematic, well-controlled study of the validity of the procedure, in which the methodology of the study and the results have been rigorously documented. The procedure of estimating the cam center relies on the the subjective judgment of the operator to position the camera properly and to pick accurate sample points. For me to be convinced that this procedure produces valid, reproducible results, I would have to have been able to produce them myself (which I couldn't), or I'd have to see convincing data, presented in a manner that I could analyze. The effect of human error in this procedure needs to be studied more thoroughly and better understood. You may have informally examined this issue sufficiently to convince yourselves of the validity of your data, but to convince a skeptical third party, a more rigorous study is needed. Jay
I don't think positioning the camera is necessarily subjective. The alignment of the camera with the axle would be best done with a tripod with a level, and then similarly ensuring that the cam lobe is level (fairly easy if taking the cam apart, harder if leaving it assembled, but certainly not impossible). The centring can be confirmed later by examining the centre of the axle with the centre of the photograph. If they coincide, you're good to go.
If confirmation is necessary, you could print of a logarithmic spiral, photograph it, and then compare it with either the original digital version of the spiral, or with the software Aric linked us to.
|
|
|
|
|
hafilax
Jun 17, 2009, 12:09 AM
Post #103 of 190
(9925 views)
Shortcut
Registered: Dec 12, 2007
Posts: 3025
|
It seems to me that the easiest way of getting at the cam angle at each spot is to measure the angle between a line drawn from the center of the axle and tangent at the intersection of the surface. That's how the logarithmic spiral was defined in the first place. You can do it with a protractor and dispense with all of this camera nonsense.
This whole fitting thing makes for a pretty image showing an off center axle but it's not really what people are interested in.
|
|
|
|
|
jt512
Jun 17, 2009, 1:14 AM
Post #104 of 190
(9900 views)
Shortcut
Registered: Apr 12, 2001
Posts: 21904
|
cracklover wrote: adatesman wrote: When you were going through your analysis did you notice anything in regards to which axles bent most? I've been busy publishing all the photos to the web today so haven't been able to look at any of this yet. -a. Here's the data: axle sample rating failure angle rating average average avg %
bent number kn kn % angle fail kN rating
no 3 12 9.3 14 77.50% 18.4 9.4 77%
no 9 12 11.3 17 94.17%
no 8 12 11.1 18 92.50%
no 20 12 9.9 21 82.50%
no 7 8.25 5.25 22 63.64%
no 24 8.3 4.5 54.22%
slight 1 12 10.6 16 88.33% 18.2 11.8 99%
slight 14 12 12.1 17 100.83%
slight 21 12 12.6 19 105.00%
slight 12 12 12 19 100.00%
slight 17 12 11.8 20 98.33%
yes 23 9.8 6.8 15 69.39% 19.7 11.1 89%
yes 2 15.5 11.4 18.5 73.55%
yes 19 12 12.6 21 105.00%
yes 16 12 14 21 116.67%
yes 5 15.5 13.8 23 89.03%
yes 22 9.8 8.1 82.65%
very 15 12 13.5 16 112.50% 18.0 13.4 95%
very 6 15.5 14 16 90.32%
very 4 15.5 12.6 22 81.29% This time I broke it into quartiles based on the amount of bending I saw in your pics.
Those groups are not quartiles.
In reply to: I think it's pretty obvious that there's an excellent correlation both to the amount of force the cam saw, and to the relative strength of the cam (which is really just saying the same thing, but factoring in the size).
Those correlations are neither obvious nor excellent. Visually, I see a reasonable correlation between the amount of bending and the absolute failure load, and no correlation at all between the amount of bending and the failure load as a percentage of rated strength. A quick logistic regression analysis backs this up. For the relation with absolute failure load, pseudo-R˛ = 0.28 (this statistic in some sense measures the strength of the relationship, ranges from 0 to 1—bigger is better), and the p-vaule is 0.035. For the relation with failure load relative to strength rating, pseudo-R˛ = 0.08, and the p-value is 0.21.
In reply to: The only conclusion to draw here is pretty simple - the more force the cam felt, both in absolute kN, and relative to the size of the cam, the more the axle bent.
That's only true for the absolute force, as noted above.
Jay
(This post was edited by jt512 on Jun 17, 2009, 1:16 AM)
|
|
|
|
|
jt512
Jun 17, 2009, 2:29 AM
Post #105 of 190
(9875 views)
Shortcut
Registered: Apr 12, 2001
Posts: 21904
|
cracklover wrote: Aric, the trouble I'm having is that I don't see any correlation between effective cam angle and the point at which the cam slipped.
We wouldn't expect to see a correlation between cam angle and the applied force at which the cam slipped. Accoding to Vaino Kodas's equations, there is no relationship between the applied force at which a cam would slip cam angle. If the cam angle is less than a critical value, the cam will not slip at any applied force; if the cam angle is greater than the critical value, the cam will slip at any applied force.
In reply to: For example, here are the cams for which I was able to determine an effective cam angle, ordered by angle. I've broken them down into thirds for the purpose of getting averages: sample rating failure failure angle RATING Soft avg failure avg rating
Mode kN % Lobes kN %
-----------------------------------------------------------------------------
3 12 braze 9.3 14 77.50% Y (not counted due to braze failure)
23 9.8 pulled 6.8 15 69.39% Y 11.24 90.94%
1 12 pulled 10.6 16 88.33% Y
15 12 pulled 13.5 16 112.50% Y/N
6 15.5 pulled 14 16 90.32% Y/N
9 12 pulled 11.3 17 94.17% N
14 12 broke cable 12.1 17 100.83% Y 11.86 94.38%
8 12 pulled 11.1 18 92.50% Y/N
18 12 broke cable 12 18 100.00%
2 15.5 pulled 11.4 18.5 73.55% N
21 12 pulled 12.6 19 105.00% N
12 12 broke cable 12 19 100.00% Y
17 12 pulled 11.8 20 98.33% Y
20 12 pulled 9.9 21 82.50% N 11.36 89.82%
19 12 pulled 12.6 21 105.00% Y
16 12 broke cable 14 21 116.67% Y/N
7 8.25 pulled 5.25 22 63.64% Y
4 15.5 pulled 12.6 22 81.29% N
5 15.5 pulled 13.8 23 89.03% N Now you could make the argument that the cam angle in and of itself is not high enough to cause these cams to rip, but in conjunction with the very soft lobes on some of the cams, the effective angle quickly grows until it gets too high and the cam slips out of the fixture.
I wonder if it isn't so much a function of the flat spots on the cam increasing the cam angle as it the flat spots making the cam essentially no longer a cam. Would you expect something like this...
...to really work like a cam? Or maybe it would. After all hexes hold in parallel cracks.
Jay
(This post was edited by jt512 on Jun 17, 2009, 3:07 AM)
|
|
|
|
|
cracklover
Jun 17, 2009, 4:04 AM
Post #106 of 190
(9828 views)
Shortcut
Registered: Nov 14, 2002
Posts: 10162
|
jt512 wrote: cracklover wrote: Aric, the trouble I'm having is that I don't see any correlation between effective cam angle and the point at which the cam slipped. We wouldn't expect to see a correlation between cam angle and the applied force at which the cam slipped. Accoding to Vaino Kodas's equations, there is no relationship between the applied force at which a cam would slip cam angle. If the cam angle is less than a critical value, the cam will not slip at any applied force; if the cam angle is greater than the critical value, the cam will slip at any applied force.
Let's say the point at which the angle is too great is 30 degrees. If the effective cam angle starts at 29, a little deformation, and presto, you're over the line. So the closer to that magic number, the less force should be required to push you over it. But... that's not exactly what we're seeing.
Now if the relative hardness of the cam lobes is a much bigger factor, then it might erase any correlation based on the starting effective angle. But... I don't see any correlation based on either a combination of the two, or on either individually.
In reply to: I wonder if it isn't so much a function of the flat spots on the cam increasing the cam angle as it the flat spots making the cam essentially no longer a cam. Would you expect something like this... [img]http://www.shariconglobal.com/misc/pulltesting/RSaliens/Sample09/sample09_broken_left.JPG[/img] ...to really work like a cam? Or maybe it would. After all hexes hold in parallel cracks.
How exactly that flat spot affects the "friction" is waaaay beyond my paygrade. I'm really just a DBA/developer with a halfway decent grasp of basic physics and a good head for problem-solving.
I agree with Aric, though. Something is allowing the cams to slip, and it certainly seems to vary from cam to cam. I just don't think we've put our finger on what that key variable, or variables set is, yet. And it's distinctly possible that with such a small dataset, it may not be possible to find it for certain.
GO
|
|
|
|
|
jt512
Jun 17, 2009, 4:40 AM
Post #107 of 190
(9818 views)
Shortcut
Registered: Apr 12, 2001
Posts: 21904
|
cracklover wrote: jt512 wrote: cracklover wrote: Aric, the trouble I'm having is that I don't see any correlation between effective cam angle and the point at which the cam slipped. We wouldn't expect to see a correlation between cam angle and the applied force at which the cam slipped. Accoding to Vaino Kodas's equations, there is no relationship between the applied force at which a cam would slip cam angle. If the cam angle is less than a critical value, the cam will not slip at any applied force; if the cam angle is greater than the critical value, the cam will slip at any applied force. Let's say the point at which the angle is too great is 30 degrees. If the effective cam angle starts at 29, a little deformation, and presto, you're over the line. So the closer to that magic number, the less force should be required to push you over it. But... that's not exactly what we're seeing. Now if the relative hardness of the cam lobes is a much bigger factor, then it might erase any correlation based on the starting effective angle. But... I don't see any correlation based on either a combination of the two, or on either individually. In reply to: I wonder if it isn't so much a function of the flat spots on the cam increasing the cam angle as it the flat spots making the cam essentially no longer a cam. Would you expect something like this... [img]http://www.shariconglobal.com/misc/pulltesting/RSaliens/Sample09/sample09_broken_left.JPG[/img] ...to really work like a cam? Or maybe it would. After all hexes hold in parallel cracks. How exactly that flat spot affects the "friction" is waaaay beyond my paygrade. I'm really just a DBA/developer with a halfway decent grasp of basic physics and a good head for problem-solving. I agree with Aric, though. Something is allowing the cams to slip, and it certainly seems to vary from cam to cam. I just don't think we've put our finger on what that key variable, or variables set is, yet. And it's distinctly possible that with such a small dataset, it may not be possible to find it for certain. GO
Well if it's cam angle, then, according to theory, you should be able to put the cam back in the jig, with the flat spots on the cam lobes against the test surfaces, and the cam should pull out under any force.
A note on the cam angles in your table: they probably are unreliable. Consider all the sources of error in them: error in camera angle, error in point selection, error in estimating the % retraction of the cams in the jig, error in estimating the test % retraction point in the photo, and error in estimating the cam angle at the estimated % retraction point. These errors attenuate the correlation between the true tested cam angle and the estimated tested cam angle; and hence, would attenuate any observed correlation between estimated cam angle and maximum force held. So even if there were actually a correlation between the tested cam angle and maximum force held, it could be obliterated by the cumulative effects of sources of error in the determination of the tested cam angle.
Jay
(This post was edited by jt512 on Jun 17, 2009, 4:41 AM)
|
|
|
|
|
irregularpanda
Jun 17, 2009, 5:42 AM
Post #108 of 190
(9803 views)
Shortcut
Registered: Mar 13, 2007
Posts: 1364
|
Can I chime in here with a question or two?
First off, this sort of hread isn't really my bag, normally I'm all about poopin' and fartin' jokes, on top of the occasional medical tangent.
So, with all this hubbub and thorough testing of aliens going on, I had a couple questions for Aric. I have several, and so does a friend. I don't have the time to spend analyzing my aliens, looking at the centering on the drill holes, or pull testing them. Is it possible to send some to you (about 10 or more) to have them looked at but not destroyed? Is it possible to make any sort of a prediction based on a pull test (just a pull, not a destructo-pull) combined with an analysis of the centering of the lobes?
I know that there is certain data you're still gathering, and certain conclusions can't be made without fully destroying the cam, and then sawing the head in half to inspect it. I guess I wanna know whether the distance from center has a direct correlation to failure, and want to have someone look at my cams. Would you be interested / have the time to do that Aric?
If you do, I'd send you a bunch of cams to play with, with the hopes that they not be utterly destroyed..... and perhaps a better understanding of which ones could be more dangerous than the others.
Thanks
|
|
|
|
|
philbox
Moderator
Jun 17, 2009, 8:21 AM
Post #109 of 190
(9785 views)
Shortcut
Registered: Jun 27, 2002
Posts: 13105
|
jt512 wrote: ...to really work like a cam? Or maybe it would. After all hexes hold in parallel cracks. Jay
The difference being though that a hex would apply a torque to the sides of the crack whereas the pictured cam would in effect be pulling without that same torque. It would in effect be a parallel slug of alloy being pulled in an entirely centered manner.
That might be an interesting pull test, how well would a hex hold in the crack test fixture.
|
|
|
|
|
hansundfritz
Jun 17, 2009, 1:08 PM
Post #110 of 190
(9747 views)
Shortcut
Registered: Sep 18, 2008
Posts: 139
|
cracklover wrote: I agree with Aric, though. Something is allowing the cams to slip, and it certainly seems to vary from cam to cam. I just don't think we've put our finger on what that key variable, or variables set is, yet. And it's distinctly possible that with such a small dataset, it may not be possible to find it for certain.
Only the manufacturer has the resources to generate a robust dataset!
The irony here is that so much is apparently wrong that it is difficult to determine which of the problems accounts for the slippage. I'm no engineer or stats guy, but the hardness of the axles seems to be a leading candidate.
Aric: what about the suggestions made up-thread to flatten the axles so that your hardness tester will work? Does the finished surface matter? Would any heating from the flattening process change the properties of the metal? In any case, you can always send them out later to see if your results were accurate.
|
|
|
|
|
cracklover
Jun 17, 2009, 3:09 PM
Post #111 of 190
(9705 views)
Shortcut
Registered: Nov 14, 2002
Posts: 10162
|
jt512 wrote: cracklover wrote: jt512 wrote: cracklover wrote: Aric, the trouble I'm having is that I don't see any correlation between effective cam angle and the point at which the cam slipped. We wouldn't expect to see a correlation between cam angle and the applied force at which the cam slipped. Accoding to Vaino Kodas's equations, there is no relationship between the applied force at which a cam would slip cam angle. If the cam angle is less than a critical value, the cam will not slip at any applied force; if the cam angle is greater than the critical value, the cam will slip at any applied force. Let's say the point at which the angle is too great is 30 degrees. If the effective cam angle starts at 29, a little deformation, and presto, you're over the line. So the closer to that magic number, the less force should be required to push you over it. But... that's not exactly what we're seeing. Now if the relative hardness of the cam lobes is a much bigger factor, then it might erase any correlation based on the starting effective angle. But... I don't see any correlation based on either a combination of the two, or on either individually. In reply to: I wonder if it isn't so much a function of the flat spots on the cam increasing the cam angle as it the flat spots making the cam essentially no longer a cam. Would you expect something like this... [img]http://www.shariconglobal.com/misc/pulltesting/RSaliens/Sample09/sample09_broken_left.JPG[/img] ...to really work like a cam? Or maybe it would. After all hexes hold in parallel cracks. How exactly that flat spot affects the "friction" is waaaay beyond my paygrade. I'm really just a DBA/developer with a halfway decent grasp of basic physics and a good head for problem-solving. I agree with Aric, though. Something is allowing the cams to slip, and it certainly seems to vary from cam to cam. I just don't think we've put our finger on what that key variable, or variables set is, yet. And it's distinctly possible that with such a small dataset, it may not be possible to find it for certain. GO Well if it's cam angle, then, according to theory, you should be able to put the cam back in the jig, with the flat spots on the cam lobes against the test surfaces, and the cam should pull out under any force.
That's an excellent point. Aric, is this something you tried at any point with the cams that slipped out? And are there any that slipped out that are in good enough shape that they could be put in the tester, or were they all too dismantled afterwards for other testing purposes?
In reply to: A note on the cam angles in your table: they probably are unreliable. Consider all the sources of error in them: error in camera angle, error in point selection, error in estimating the % retraction of the cams in the jig, error in estimating the test % retraction point in the photo, and error in estimating the cam angle at the estimated % retraction point. These errors attenuate the correlation between the true tested cam angle and the estimated tested cam angle; and hence, would attenuate any observed correlation between estimated cam angle and maximum force held. So even if there were actually a correlation between the tested cam angle and maximum force held, it could be obliterated by the cumulative effects of sources of error in the determination of the tested cam angle. Jay
I'll agree with you to a point - the angle I'm estimating has a wide margin for error. But not so much so (IMO - I know our opinions differ on this) that it should wipe out any strong correlation. I'd estimate a margin of error no greater than 5 to 10 degrees.
Of course if there's a slim correlation between pullout force and effective cam angle, and not much differentiation between the best and worst effective angles (in this case, from 14 to 23 degrees, or not much more than the margin of error) then you'd need a big sample for the average of the errors to disappear.
I'd suggest that in any future tests you do, Aric, if it's possible to get a photo of the cam in the rig - shot straight on to the axle - before pulling, that would give us an actual measurable effective cam angle.
But if the theory holds true that the cams are slipping out when they get flat enough to go past the Kodas point, then any cams that start out with an angle close to that should fail at a much lower force. It's not necessary to get good precision on that one. Unfortunately, I don't know what that Kodas point is for Aric's rig. The number 30 degrees was being bandied around - but was that for cam on rock, or cam on steel?
GO
|
|
|
|
|
jt512
Jun 17, 2009, 3:58 PM
Post #112 of 190
(9686 views)
Shortcut
Registered: Apr 12, 2001
Posts: 21904
|
cracklover wrote: jt512 wrote: cracklover wrote: jt512 wrote: cracklover wrote: Aric, the trouble I'm having is that I don't see any correlation between effective cam angle and the point at which the cam slipped. We wouldn't expect to see a correlation between cam angle and the applied force at which the cam slipped. Accoding to Vaino Kodas's equations, there is no relationship between the applied force at which a cam would slip cam angle. If the cam angle is less than a critical value, the cam will not slip at any applied force; if the cam angle is greater than the critical value, the cam will slip at any applied force. Let's say the point at which the angle is too great is 30 degrees. If the effective cam angle starts at 29, a little deformation, and presto, you're over the line. So the closer to that magic number, the less force should be required to push you over it. But... that's not exactly what we're seeing. Now if the relative hardness of the cam lobes is a much bigger factor, then it might erase any correlation based on the starting effective angle. But... I don't see any correlation based on either a combination of the two, or on either individually. In reply to: I wonder if it isn't so much a function of the flat spots on the cam increasing the cam angle as it the flat spots making the cam essentially no longer a cam. Would you expect something like this... [img]http://www.shariconglobal.com/misc/pulltesting/RSaliens/Sample09/sample09_broken_left.JPG[/img] ...to really work like a cam? Or maybe it would. After all hexes hold in parallel cracks. How exactly that flat spot affects the "friction" is waaaay beyond my paygrade. I'm really just a DBA/developer with a halfway decent grasp of basic physics and a good head for problem-solving. I agree with Aric, though. Something is allowing the cams to slip, and it certainly seems to vary from cam to cam. I just don't think we've put our finger on what that key variable, or variables set is, yet. And it's distinctly possible that with such a small dataset, it may not be possible to find it for certain. GO Well if it's cam angle, then, according to theory, you should be able to put the cam back in the jig, with the flat spots on the cam lobes against the test surfaces, and the cam should pull out under any force. That's an excellent point. Aric, is this something you tried at any point with the cams that slipped out? And are there any that slipped out that are in good enough shape that they could be put in the tester, or were they all too dismantled afterwards for other testing purposes? In reply to: A note on the cam angles in your table: they probably are unreliable. Consider all the sources of error in them: error in camera angle, error in point selection, error in estimating the % retraction of the cams in the jig, error in estimating the test % retraction point in the photo, and error in estimating the cam angle at the estimated % retraction point. These errors attenuate the correlation between the true tested cam angle and the estimated tested cam angle; and hence, would attenuate any observed correlation between estimated cam angle and maximum force held. So even if there were actually a correlation between the tested cam angle and maximum force held, it could be obliterated by the cumulative effects of sources of error in the determination of the tested cam angle. Jay I'll agree with you to a point - the angle I'm estimating has a wide margin for error. But not so much so (IMO - I know our opinions differ on this) that it should wipe out any strong correlation. I'd estimate a margin of error no greater than 5 to 10 degrees.
Since the measured angles ranged from 15 to 23 degrees, then errors ranging from 5 to 10 degrees would be huge. For example, it would be possible for the cams with the largest measured angles in the sample to actually have the smallest angles. Random error of this magnitude would seriously attenuate correlation coefficients.
In reply to: Of course if there's a slim correlation between pullout force and effective cam angle, and not much differentiation between the best and worst effective angles (in this case, from 14 to 23 degrees, or not much more than the margin of error) then you'd need a big sample for the average of the errors to disappear.
Random error always attenuates correlation. The errors don't average out with larger sample size.
In reply to: The number 30 degrees was being bandied around - but was that for cam on rock, or cam on steel?
31 degrees for aluminum/steel, based on Kodas.
Jay
|
|
|
|
|
cracklover
Jun 17, 2009, 6:10 PM
Post #113 of 190
(9643 views)
Shortcut
Registered: Nov 14, 2002
Posts: 10162
|
jt512 wrote: Since the measured angles ranged from 15 to 23 degrees, then errors ranging from 5 to 10 degrees would be huge. For example, it would be possible for the cams with the largest measured angles in the sample to actually have the smallest angles. Random error of this magnitude would seriously attenuate correlation coefficients.
I agree, as I said in my that post (I'm afraid I threw you off with that erroneous bit about a larger sample size.)
Really, you can see quite well from Aric's photos the point at which the outer edge of the cam touches the fixture. Being off by a small amount in my estimate on where the cam meets the fixture would typically only result in a difference of 1 degree cam angle. So the only issue is how well the photos he used to generating the cam angles centered on the axle.
That's where I'm getting my estimate of 5-10 degrees of error from, but I'll let Aric speak to that. Putting them on a scanner would, of course, solve the problem, but it's a little too late for that!
GO
|
|
|
|
|
adatesman
Jun 17, 2009, 8:15 PM
Post #114 of 190
(9613 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
adatesman
Jun 17, 2009, 8:27 PM
Post #115 of 190
(9600 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
cracklover
Jun 17, 2009, 8:41 PM
Post #116 of 190
(9591 views)
Shortcut
Registered: Nov 14, 2002
Posts: 10162
|
adatesman wrote: cracklover wrote: That's an excellent point. Aric, is this something you tried at any point with the cams that slipped out? And are there any that slipped out that are in good enough shape that they could be put in the tester, or were they all too dismantled afterwards for other testing purposes? No, I haven't tried it. It would be easy enough to reassemble one of the not-so-mangled ones to see what it does, so I'll give it a try. I doubt it will hold much of anything so don't think having half the head cut away will make a difference.
Cool.
In reply to: In reply to: I'd suggest that in any future tests you do, Aric, if it's possible to get a photo of the cam in the rig - shot straight on to the axle - before pulling, that would give us an actual measurable effective cam angle. Unfortunately the bolts in the side of the fixture are a bit too close together to get a good pic that way. That said I'll give it a try next time anyway.
I see. Not an insoluble problem though. The cam doesn't have to be positioned as it will be when being pulled. Turn it sideways and the cam angle will still be the same.
In reply to: In reply to: Unfortunately, I don't know what that Kodas point is for Aric's rig. The number 30 degrees was being bandied around - but was that for cam on rock, or cam on steel? 31 degrees is for aluminum on steel, so directly applicable to the fixture. Aluminum on rock is ~17 degrees.
Hmm, that gives me an idea...
GO
|
|
|
|
|
adatesman
Jun 18, 2009, 12:35 AM
Post #117 of 190
(9549 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
angry
Jun 18, 2009, 1:37 AM
Post #118 of 190
(9527 views)
Shortcut
Registered: Jul 22, 2003
Posts: 8405
|
The dark vs light axle issue is just one of sandblasting if I remember correctly. Axles were sandblasted then they stopped. Probably makes it easier to tell whether they'd been hardened or not.
|
|
|
|
|
blondgecko
Moderator
Jun 18, 2009, 12:12 PM
Post #120 of 190
(9807 views)
Shortcut
Registered: Jul 2, 2004
Posts: 7666
|
I just had an idea about the relationship between Rockwell hardness and strength. In particular, the question occurred to me: is it a linear relationship? So I went looking here - in particular at aluminium alloys 6061-T4, -T6 and -T8. Here's what I found:
6061-T4
Hardness: Vickers 75 (~HRB 9-10)
Shear strength: 165 MPa
6061-T6
Hardness: HRB60
Shear strength: 207 MPa
6061-T8
Hardness: HRB75
Shear strength: 185 MPa
Interesting, no?
|
|
|
|
|
adatesman
Jun 18, 2009, 2:45 PM
Post #122 of 190
(9767 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
cracklover
Jun 18, 2009, 3:24 PM
Post #123 of 190
(9745 views)
Shortcut
Registered: Nov 14, 2002
Posts: 10162
|
Okay, now we're getting somewhere. So I think I've had an "ah ha!" moment.
Please bear with me, this post is kinda long...
If we assume that the surface texture has the effect of increasing the friction required before the cam will slip (I think we all can agree on that, even if we don't agree on whether to call it "friction" or "shearing" or "abrasion") then I think the pics I posted above tell a pretty good story.
It would appear that the effective cam angle at which the cam lobes will slip under load in Aric's fixture is somewhere from 31 to 39 (with a tight grouping around 35-37 degrees).
This points to a good candidate for why Aliens slip out while other cams break first:
1 - The other cams tested had consistently harder lobes, meaning that as the cam lobes deformed, it would have taken more force to get the point where the effective cam angle reached 35 degrees.
Despite the fact that there was a good deal of variance in the softness of Alien lobes, *all* were significantly softer than those of the other cams (correct me if I'm wrong here, Aric).
2 - For the most part, the other cams tested were larger. A larger cam requires a larger volume of metal be deformed before the flat section of the lobe is big enough to create the effective cam angle of 35 degrees. Deforming more metal takes more force.
The cams that pulled out at the lowest forces were all among the smallest cams Aric tested: All the blacks and blues pulled at the lowest ratings, and then many of the yellow Aliens did too.
3 - All of the Aliens tested had an effective cam angle greater than the angle of the other cams. This means that before deformation of the cam lobe even begins, the cam is closer to the point of reaching the critical angle of 35 degrees. The higher the starting angle, the less metal needs to be deformed to reach the critical point. Which again, means less force required.
Even if the method of measuring the affect of the misdrilled axles on effective cam angle is imprecise, there's little doubt that in almost all of these samples, it resulted in a larger starting cam angle. Perhaps we don't have enough precision to predict the variance from cam to cam, without laying the lobes flat on a scanner bed, but on average, they were pretty consistently off in the direction of too high a cam angle.
In summary, I believe that the three factors I listed above combined to create the "perfect storm" required to allow the cam lobes to deform enough to reach the "Kodas point" of around 35 degrees. Whereas the other brands of cams tested were not subject to those three factors, and so they failed due to another means before their lobes could be deformed enough for them to pull out.
GO
|
|
|
|
|
adatesman
Jun 18, 2009, 3:37 PM
Post #124 of 190
(9731 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
cracklover
Jun 18, 2009, 3:52 PM
Post #125 of 190
(9722 views)
Shortcut
Registered: Nov 14, 2002
Posts: 10162
|
If I am right, then the method you're using to test hardness is exactly the right one. It's a measure of how much force is required to deform a set amount of metal. This is exactly the question that determines how much force is required to flatten the lobes enough.
Shear force could enter the equation, though, if the metal is so soft that it begins to smear off at a force lower than the force required to deform enough metal to reach the "Kodas point".
GO
|
|
|
|
|
|
FrictionModel.gif
