Lum Chinese : some hemp cutting and commentary / discussion

[kbuzbee]

Thanx Cliff. We've got some CBN Triangles for the Sharpmaker that just came in. Should be shipping them shortly. I liked what I found when testing, but I wanted some additional "real world" testing.

Let me know if you want a set to test?

sal

Hey Sal, what is the equivalent grit?

Ken

[sal]

Hi Ken,

Should be about 400 mesh.

sal

[kbuzbee]

Excellent! Thanks Sal!

Ken

[Cliff Stamp]

Let me know if you want a set to test?

Sure.



This is the result of three runs, at this point the predictive power is about 15%. I was hoping it would be a little better, but based on the math that is about what I would expect there are just so much variances in hand cutting (speed, force, angle, material, environmental conditions, initial sharpness, steel consistency) .

I can probably get it under 10% but in order to do that I would have to increase the run count way up to 10, I might do that at some point just to check the math and verify that the variations are random (the systematic ones will not average out), but it likely won't be for awhile.

The problem is though that at +/- 15% this means you can only see coarse differences because this means that in order to see a difference in a steel it has to be bigger than twice this variation (as they add when you difference). This means that VG-10, ATS-34, AUS-8, S30V, etc. all measure the same.

I will add a couple of knives to show the extreme points which should be picked up, a ZDP-189 Spyderco and I will see if I can't get one of the ones in the 420 or similar steel as well to show where the low point falls.

[sal]

done.

sal

[Cliff Stamp]

This is mainly math/statistics and is likely not of interest to a lot of people, however if you are curious as to where those curves and the edge retention numbers come from and the work involved this might be interesting to you.

The equation which is used to generate the curves and the edge retention limits (5%, 10%, 15%) is based on the solution to differential equations which model wear, deformation, chipping, etc. . The basic form is :

S(x) = S_0/ (1+a * x ^ b)


S(x) is the sharpness after cutting x amount of material

S_0 is the initial sharpness

x is the amount of material cut

a and b are physical constants (they depend on the material being cut and the steel properties and how the cutting is done)

The rigorous solution of this requires nonlinear regression on S_0, a and b. I used to do this years ago using code I wrote in Fortran to do nonlinear regression and then feed that to GnuPlot for graphing, but recently switched to using Calc (OpenOffice) because of the ease at which it automates the process and produces graphs and results for tables and such.

However to do it in Calc I had to do an approximation. I set S_0 equal to the measured first sharpness point rather than fit it. This allowed me to use log transforms and then a linear regression. I was satisfied with this approach because while the first sharpness point is not measured well and thus setting S_0 to the first measurement is a very coarse approach, for most of the data (which takes place at low sharpness points) the low precision of the very first measurement doesn't have much effect on the reduced data.

(the reason this happens is that you end up with a term which looks like R-1 and R >> 1 and thus you can just approximate this as R)

However recently I found a nonlinear solver for Calc and I used it to recheck the hemp results. There was no significant change in the results they just moved around in the scatter. In short, the approximations I made were valid (maths is maths). That being said if you wanted to do it there is no need to use the linear approximation because NLPSolver is very easy to use in Calc. However I doubt I will rewrite all the sheets to use it.


--

As an aside, this exact same equation holds if you are cutting cardboard, whittling wood, or even using a dental scraper on teeth as they all have the same fundamental differential equations. The only thing this equation doesn't model is catastrophic damage which is based on different underlying physical equations.

[1965ford]

fascinating information Cliff, way more detailed analysis than I could ever do. Also interested in new options for the Sharpmaker, but I am not sharpening knives often enough to be a useful tester. Still putting off getting a set of the diamond stones to re-profile a couple of knives, just too many other things going on. Threds like this make the Spyderco forum a wonderful place to learn and expand our knowledge on all things knives, and having Sal actively participate makes it that much better.

[mattman]

Sorry to go off-topic, but are the CBN rods ready for prime time? Or still in testing phase?

Thanks for your efforts, Cliff.

I always read with interest!

[Cliff Stamp]

... having Sal actively participate makes it that much better.

Indeed, imagine if you could talk to the CEO/President of the company of any product that you had an issue with, and actually see them change products in direct response to feedback, and as well make an entire list of test products just for evaluation. The level of customer service from Spyderco is insanely high when you look at products in general, let alone knives.

[sal]

Sorry to go off-topic, but are the CBN rods ready for prime time? Or still in testing phase?

We should be shipping in a few days. We've done some in-house testing. Enough to want to make them. Now we'd like to know more from you. Sorry for the distraction.

sal

[Cliff Stamp]

As an update there are a few interesting observations which lead to a few hypotheses, but to be clear I need more data to know if they are true. A very interesting one is formed when you combine the hemp cutting with the previous cardboard cutting : http://www.cliffstamp.com/knives/reviews/cardboard.html .

In the cardboard cutting I found that the performance sort of peaked and that pushing carbide volumes very high not only didn't increase the ability of the knives to retain a cutting edge, it actually decreased it. Until recently the idea I had was that because the edge angle was low (14-16 dps) then the low edge stability of those steels were causing them to blunt by micro-chipping due to carbide instability. Thus when I started the hemp rope cutting I jumped the edge angle right up to 24-26 dps to check that hypothesis and see if the edge retention peak would shift to the extreme high carbide steels.

But it doesn't look like it will.

Again, I need more data to be certain - but I think now that 600 DMT might be the issue and that it is too coarse and the scratch patterns combined with the high carbide volume is making the edges unstable on the very high carbide steels and thus they are losing sharpness prematurely even though they have very high wear resistance because the edge isn't blunting by the same mechanism which is tested in regular wear resistance tests. Now in order to see if this is the case then I will repeat another trial but jump the edge finish up to MXF DMT which is much finer and see how that affects the results.

[tvenuto]

So the upshot of this is (assuming the numbers play out as you expect):

Unless you are willing to finish the edge to a certain degree of "grit," you may actually be better off going with a lower carbide steel. For example S30V may actually perform better than S90V below a certain grit level.

All this, of course, assumes all other things being equal, and with the caveat: "in general." Is the above statement what you mean by your post?

If so that's certainly an interesting observation. Even though the high carbide steels are very wear resistant, it no longer matters, because "wear" is no longer the main reason they dull.

[Cliff Stamp]

This has to be true in an absolute sense, as just try to use a ceramic knife with a very coarse finish, it likely will not work very well at all.

There are a few interesting questions because of how grit finish affects edge retention in general. For example if you are slicing materials then the best grit for edge retention isn't the highest one, a coarse finish will offer a performance advantage in general, the optimal finish depends on the material as well as the steel.

Lets say that at the MXF finish that S110V has better edge retention than Elmax (at the same finish) slicing hemp, this does not mean however that it has better edge retention than Elmax at 600 DMT slicing hemp.

The true optimal steel (for a given cutting task) may simply be a lower carbide steel because of the ability to retain both low angles and coarse edges without fracture.

Again though, what you are cutting and how influences it, wood carving for example is very different than slicing hemp.

[tvenuto]

Ah, yes. I guess I was just injecting my own bias because I can't see a "coarse finish" but I can see a jagged and irregular edge, I take them to be different. Actually they are just different degrees of the same feature.

It's interesting to imagine that, given a certain medium, the finish level required for optimal (or even acceptable) retention might never intersect the finish level required for optimal cutting performance. I guess low carbide steels find such favor with those who use them because they are less sensitive to this, and can thus be made to cut more things better with less worry about things like edge angle and finish. Thus the oft cited "how fast this thing takes a wicked edge."

[Cliff Stamp]

Yes, there is indeed a question to be asked about edge retention vs cutting ability.

Consider for example in the extreme if you do not care about cutting ability at all then the optimal finish for edge retention is just a 10 micron flat on the apex. This apex would have exceptional stability and strongly resist deformation, chipping and wear - however it would cut fairly poorly, but stay at that level basically forever.

That extreme point however is unlikely to be used by many (but some do) but in general (very general) the less you favor cutting ability the more that higher carbide steels will move towards optimal.

There is also another issues about wear/maintenance and just cost of replacement. If you use a knife frequently and often, and you are willing to wear it out, you can get exceptional cutting ability often with finishes/geometries which are a high rate of consumption of the knife, just try for example a 10 dps/X-coarse DMT edge on hemp rope on a very inexpensive knife. The cutting ability and edge retention are both very high, but the rate at which steel is removed when it is sharpened is very high. This makes the curious point that it might be the less expensive knives which are the optimal performers because most people will not be willing to wear out a high end knife and will thus go with a finish/geometry which is far less performance optomized but has a very slow rate of consumption.

[bearfacedkiller]

I can't get enough of this kind of information. Thanks again.

[Cliff Stamp]

If you want something interesting to consider, and again I don't have enough data to make even a weak conclusion at this point :

It looks like the cutting ability of edges as a function of carbide volume is nonlinear in force applied, specifically, if you apply more force you will get a stronger cutting response with higher carbide steels. To make a few clarifications :

-the cutting depth (or amount) always increases with more force, it just increases more with higher carbide steels

-this difference gets larger when the blade is duller

-it is really only significant when the blade is very dull, as in < 5% of optimal

I have graphs/calculations showing this, but the data is so scattered right now you can just see a pattern you can't say with confidence that it exists, it could still just be noise.

I should have enough data compiled in a month or so to see if it is real or not.

For me it isn't of practical value because it only really gets significant if you are using a really dull knife (the kind that you can easily drag over your skin and dent it and not cut yourself), but it is interesting from an academic perspective anyway -AND- it might shed some light on why a certain user group is so fond of high carbide steels.