Steel Edge Retention testing

Any Cal. said:
I believe he went with median to reduce the effect that any fluctuations to an extreme would have on the testing. Using your numbers, theonew, in each of 2 tests, (1) 40g and (1) 60g cut is an extreme measurement, a fluke, statistically speaking. Using the median reduces the weight that the fluke(s) carries on the measurement as a whole. I am just guessing, but you have to remember that Vassili is a statistician, so tries to reduce the variables with numbers.
Click to expand...

My example was weighted to the extreme, so change the number of cuts on the upper and lower ends a bit and you are still getting an unnecessary degree of inaccuracy I would think. A fluke in my mind would be if you had one or two cuts which were either 20 grams or 80 grams, using the example numbers I gave and this may be what Vassili is seeing. But from the few examples he's given it looks like the range is constricted to a 20 gram variance.
 
This is all based on my experience - I did quite thread cutting before came to conclusion to have median. In your example - yes mean will be better, but you example is synthetic, in real life if for example edge has cheap it will time to time shows result much over average. Same as in housing market.

So to me median is better represent overall sharpness.

But thing you mentioned happening all the time - if you see my latest CPM 440V (S60v) testing at 50 cuts I have 60 g then at 60 cuts - 70 and then for 70, 80, 90 and 100 - again 60g.

this is happen because of border jumps of median as in your example:

50 cuts

40 +
50 ++++
60 +++++X++++
70 +++++
80 +

60 cuts

50 ++
60 ++++++++
70 X++++
80 ++++
90 ++

Median right on the border

70 cuts

40 +
50 +++++
60 ++++X++
70 ++++++
80 ++

Median move just three measurement back and cross border back.

So because of this I post full results here (this page linked form the beginning):
http://playground.sun.com/~vasya/Manila-Rope-Testing.html

So for each cut you may see for yourself each measurement and you may do your own math.

Median is just some more close then other descriptor, more stable for this random by nature results. It also does not show you range and many other thing, it is just convenient to put in the table single value.

Thanks, Vassili.
 
nozh2002 said:
So because of this I post full results here (this page linked form the beginning):
http://playground.sun.com/~vasya/Manila-Rope-Testing.html

So for each cut you may see for yourself each measurement and you may do your own math.

Thanks, Vassili.
Click to expand...

Vassili, sorry, I have visited your full results page before but didn't realize that you are actually listing the results for each sharpness test. Wow, thank you. :D:D
 
Vassili I looked through your sharpness test results and didn't expect such a variance between the highs and lows, often around 40 - 50 grams and can better understand why you are using a median. Nevertheless I don't see many instances of 12 million dollar homes being sold alongside mostly 500 thousand dollar homes (to use your housing market analogy) :) I'll work on computing the results using a mean.

It is still a bit disturbing to me that the same knife can make both an 80 gram cut as well as a 150 gram cut during the same test. Thread tension seems like it may be rearing its ugly head.

Thanks for your hard work and full disclosure. :)
 
Without marking the spot where I cut, I just did a little check on tension. I was getting around 55g with no tension in the thread (sagging in the middle) and 30 with it taut between the screws. It's a big difference, but so was the tension. The numbers were consistent with the taut thread, even though I have no idea how tight it is from one cut to the next.

I also know that where you cut can have an effect, I was getting a 40g difference 1/8" from the mark in previous edge checking.

I got a spread of 50g - 240g on a really coarse edge (120 grit).

I think that's why Vassili does so much cutting.
 
Yes, I did the same check and found the same thing. No surprise really as the tension itself is a force that "weaken" the thread. It is actually fairly sensitive, I've found. A "just so taut" tread will require more force, than one that is pulled "really taut". So I try to avoid a taut thread. On a slack thread, the variation is much smaller. I found that on lots of slack (when you really let the thread dangle to the bottom of the gap) it picks up again. I assume, that is because the thread can wrap around the blade some and it is difficult to keep the blade vertical to the thread. But I found that whether you give it a 1/4" slack or 1/2" slack makes very little difference in the results. This is why I attach the tread on one side, pull it taut over the gap and then give it about on pinky wide slack before fastening the other side.
 
OK how bout this for a quick and clean test. Each knife is held vertically in a vise edge up, a ceramic rod is stroked across the whole edge, held as perpendicularly as possible and the force applied simply the weight of the rod. Sharpness tests are then run. This would test one of the key components of edge retention, i.e, abrasion resistance, and it shouldn't take many strokes before the edge is too dull to test. It would also minimize the effect of the edge profile on testing, minimize the effect of chipping or rolling, and wouldn't make a big mess. It would be mostly just a test of abrasion resistance.
 
I tried something like this - I only use vodka bottle. But without this clip you use to hand bottle on thread I found it to complicated... Good idea. I do not see how do you measure force. What I did - I put knife on scale instead of thread.

But anyway - I think set of measurement will be nessesary any way. Thread tension not an only factor in random nature of this test - edge is also wears out not perfectly even.

Thanks, Vassili.
 
IIRC, sodak was slicing and measuring length of edge, not force needed to cut. To try and keep the force consistent, he lifted the thread by the blade the same amount for each slice.
 
hardheart said:
IIRC, sodak was slicing and measuring length of edge, not force needed to cut. To try and keep the force consistent, he lifted the thread by the blade the same amount for each slice.
Click to expand...

I believe you are right about him measuring the length of edge to slice the twine, hence the marks on the blade for a reference. That looks like a better and more consistent way to measure the slicing distance (thus slicing sharpness) than I used in my testing (http://www.cutleryscience.com/evaluations/d2_sorg_vs_cpm_d2_military/michael_c.pdf), when I tied off the twine to the water bottle and held the other end of the twine by hand in front of me. In the future I will try using that method. I also want to build a scale set up like Vasilli's for thread cutting to measure the push cutting sharpness, as it is accurate and repeatable. That way I would be able to measure both the push cutting and slicing sharpness, and see how each degrades with different types of cutting. In my future testing I also plan on building myself a block to cut the rope on like Vasilli has, as in my test a lot of the edge wear was from hitting the cutting board, to the point where I think it greatly decreased the edge retention of both knives.

Mike
 
Let's not forget...

Happy New Year!

HappyNew2008Year.jpg


Thanks, Vassili.
 
Vassili, I love your decorations! With a 3 year old around my house they may not be the best choice, especially how sharp you like them.

Happy New Year.

Mike
 
gunmike1 said:
Vassili, I love your decorations! With a 3 year old around my house they may not be the best choice, especially how sharp you like them.
Click to expand...

LOL, I'm not sure they'd be safe around me even. :D

Right back at ya with the Happy New Year
 
Sorry for the late response, but yes, I measure blade length on the slice. I want to get set up for measuring push cuts also, I think the difference would be interesting. Like Gunmike1, I'm probably dulling the blade shoving it into plywood while I cut the rope, so I'm going to work on that also....

Vassili, you must have been VERY good throughout the year! :D
 
Median is always a better descriptor when the underlying population is non-normal. It is usually common practice to first due a normalcy test to the data to determine whether a switch to non-parametrics is necessary. Or given the intuitiveness of past experiences one can generally assume non-parametric of most non-manufacturing controlled processes and lessen one's error statistic by using, again, the median.

Vodka bottle would nomalize string tension when used throughout depending on whether one was decrementing the contents to lessen the pain of repetitive motion.

Good job and Happy New Year Vass.

ZdrasvetyeAugs
 
HoB said:
Yes, I did the same check and found the same thing. No surprise really as the tension itself is a force that "weaken" the thread. It is actually fairly sensitive, I've found. A "just so taut" tread will require more force, than one that is pulled "really taut". So I try to avoid a taut thread. On a slack thread, the variation is much smaller. I found that on lots of slack (when you really let the thread dangle to the bottom of the gap) it picks up again. I assume, that is because the thread can wrap around the blade some and it is difficult to keep the blade vertical to the thread. But I found that whether you give it a 1/4" slack or 1/2" slack makes very little difference in the results. This is why I attach the tread on one side, pull it taut over the gap and then give it about on pinky wide slack before fastening the other side.
Click to expand...

This is similar to what I do. I was surprised to see so much variation in sharpness testing even when using the same part of the edge for each test. But the tension is a big factor. I can take a dulled knife scoring over 200G and make it score in the 50's or 60's by giving the string a small amount of tension. Slackness seems to help but the results still aren't as consistent as I'd like, e.g.:

---Junkyard Dawg---

+Run 1+

Initial sharpness: 42 55 23 40 50 56
10 feet cut: 182 186 156 140 169 163
25 feet cut: 158 102 189 187 154 134
50 feet cut: 183 198 192 203 187 162
100 feet cut: 280 179 238 215 198 189


For initial sharpness alone there is a huge variation, even though I throw out any test that shows an extremely abnormal reading if I think anything went wrong with that test run, like me shaking my hands a little too much. I wonder how well a constructed mechanism for holding the knife and slowly pushing it straight into the thread would improve the consistency of the results? I'd like to try something like that when I have the time.
 
I wonder how consistent is the thread itself. Just by nature of how the lengths of thread are made wouldn't there will be places where the fibers meet in a random way?
 
Well, edge itself is also quite inconsistent - I think sharpening makes it more or less consistent, but who really knows how 10 manila rope cuts affects it - I doubt that it will be as even and consistent as after sharpening.

Again this test is random by nature.

I think my arms get used to hold knife tighter and now I can fix it at certain scale point and wait for a sec and then proceed to next point. I found that I can "cheat" moving blade to fast. Scale has some inertia and thread need some time to be cut. So moving blade fast you get bigger result. So now I do thread cutting step by step, giving blade some time on each point about 1 sec until move to next.

This is why I start retesting Busse - it was first blade I tested and results are not as stable as I can get now. Also I got some skills in sharpening and able to get it to 20g.

Thanks, Vassili.
 
I wonder if He-Who-Shall-Not-Be-Spoken's idea of using a force plate would reduce the amount of variation in measured values for a given set of thread cuts. I have to imagine it's hard to watch exactly where on the scale you are when the thread breaks. If a force plate has a fast-enough sample rate then you could measure the force applied over time (while in the process of cutting) as well as the final maximum amount of force applied just before the thread gave way.
 
You must log in or register to reply here.
Back
Top