Edge Retention Testing and Analysis

[DirkDiggler]

Cliff,, your behavior on this forum gets more reprehensible by the day. Why is it that whenever you are asked a simple direct question all you do is side step it then throw out a personal attack which is buried within a long paragraph of mumbo jumbo? My G-d, man, can you not honestly respond to a simple direct question with a simple direct answer for a change? Perhaps I should know better because we have been asking this of you for years and it has not happened yet.

Cliff,, I truly feel you should seek the help of a competent psychiatrist or other mental health professional. A person who acts as you do may have serious mental issues (hint: bipolar disorder) that should be treated before you harm yourself or others. You need not be embarrassed. There is no shame in mental illness. There is only shame in not getting proper treatment.

-DD

 

[hardheart]

 

[Cliff Stamp]

What kind of difference would you hope to see in your model to claim a difference in performance?

If some of the model parameters go undefined, the uncertainties are so large that the function is meaningless, then this is clearly an inability to prove a difference. Again as noted, this isn't the fault of the model, it works fine on the raw data. It was just the calculations peformed on it which blew up the noise which made it meaningless.

Unless the knives start with the same sharpness (Ci), it doesn't make much sense to look at sharpness values later.

Yes, I noted years back when I was correcting old reviews.

In regards to noise, generally you can compensate for with just lots of data which is why I take more intervals rather than more sets. I would be very leary of using just two data points to estimate a spread though. However the raw catra data isn't that noise as I showed in the second analysis, it was just the calculations performed on it (difference) which blew up the noise as the differences were not large compared to the noise.

-Cliff

 

[gator68]

If some of the model parameters go undefined, the uncertainties are so large that the function is meaningless, then this is clearly an inability to prove a difference. Again as noted, this isn't the fault of the model, it works fine on the raw data. It was just the calculations peformed on it which blew up the noise which made it meaningless.

...

In regards to noise, generally you can compensate for with just lots of data which is why I take more intervals rather than more sets. I would be very leary of using just two data points to estimate a spread though. However the raw catra data isn't that noise as I showed in the second analysis, it was just the calculations performed on it (difference) which blew up the noise as the differences were not large compared to the noise.

It doesn't look like the problem is "noise"; it is that data was not taken over a large enough range (intervals) to show the shape of the curve to fit your model.

You might try fixing the exponent b to something reasonable for these types of steels and refitting for the a parameter. DiamondBlade obviously did not design their experiment to fit your method of data analysis. This doesn't mean you can't try to find some meaningful way to analyze the data in a similar manner to compare to other knives you have tested.

 

[Broos]

Unless the knives start with the same sharpness (Ci), it doesn't make much sense to look at sharpness values later.

Thanks! - one comment - I understand Ci as *slicing "cutting ability", not sharpness.

I dislike defining a measure of slicing performance as "sharpness". I can design a test where a hacksaw blade will win a slicing comparison, and I also prefer to keep the common and technical definition of "sharpness" intact and unmolested.

To elaborate; Cliff has contended that the blades were not the same sharpness. As far as I can determine he thinks that two blades of equal sharpness/finish, out of different materials, will slice equally. On this basis and maybe others he throws out their methodology that was designed to start the test with all blades being identical, and wants to determine test baselines with varying geometry and sharpening optimized for each steel (according to an unexplained quantification of Kleff & Kligelhoeffer's work). The BYU Professors think that the blades were the same sharpness as measured by REST, but that they had varying cutting abilities due to factors not understood fully or measured by REST (microstructure that allows a sharper edge?). That is how I have understood the posts, anyway.

When I think about slicing, it does make sense to me that different steels would slice differently being otherwise the same. Wouldn't the steel that has harder and more resilient "teeth" slice better with other factors equal?

REST is the best measure we have available for measuring sharpness short of measuring edge width with a SEM. It is certainly more accurate and repeatable than doing the paper push cut test often mentioned. Nevertheless it would have been interesting to compare the edges under magnification initially and when they were cutting equally per stroke.

Also, since there is no way to measure this slicing "cutting ability" directly or otherwise independently of whatever slicing test we run, you are either going to have to run trial runs of the test to determine when the test blades have equal cutting ability (on the chosen media), or compare the blades at varying slice counts when slicing "cutting ability" are equal.

To directly compare slicing edge retention for S90V from slice #2 - #32 versus FFD2 at slice #146 - #176, after a baseline with the edges prepped equally, is not significant to me as a user and inherently biased from a human performance perspective. This seems inherently favorable to the blade that had low initial cutting ability after equal prep of the blades. Unless of course you agree with Cliff that despite the same geometries, same machine sharpening, and same CATRA REST values, the knives were not the same sharpness (or the test was biased). Shouldn't qualities of a knife steel that will make it better for all users (easy to sharpen to a point where it will outslice others for hundreds of slices) be considered?

And differences in how a steel performs based on finish and sharpening could be demystified and even quantified in relation to the test *(done by the Professors) by varying baseline geometry and sharpening of the blades.

 

[hardheart]

I think anyone who has tried to sharpen a 420HC 110 and one in BG42 with the same arkansas stones and number of strokes can tell you that using the same abrasive and technique for varying steels will not give the same sharpness.

Who can say if the S90V or the FFD2 were sharpened optimally? It certainly doesn't seem that way to me just reading about how they were sharpened. The FFD2 may see significant gains in performance with other sharpening media used.

A significant difference in push cutting ability and slicing aggression can be seen among steels when sharpened the same way, if only because one steel just won't get 'sharp' as quickly as another when abraded in the same manner. I don't much care for measuring push cutting and slicing at the same time and equating them, I know I sharpen to greatly different finishes when I want a knife to do one better than the other. Who hasn't posted about the 'bite' an edge will exhibit, or how another takes a fine polish easily? How many complaints about AUS6 going smooth too quickly, or descriptions of D2 as a steel that 'takes a lousy edge and keeps it' have we seen?

My biggest problem with the advertising is the hard sell of differential hardening/tempering (two different things, of course) which doesn't actually do much for knife performance, and leaves you with a useless blade at lower forces than one that is fully hardened.

 

[Broos]

I think anyone who has tried to sharpen a 420HC 110 and one in BG42 with the same arkansas stones and number of strokes can tell you that using the same abrasive and technique for varying steels will not give the same sharpness.

I agree when using a given number of strokes, but the test blades were sharpened by a machine at specified angles resulting in the same edge geometry (all edges fully developed). And the same REST results after machine sharpening.

 

[hardheart]

Was the time on the machine different for each blade? I'm still not clear on the process. They said it was a 600 grit diamond belt followed by 3 passes on CrO loaded cardboard wheels, with .020" thickness behind the edges. I'd like to know what the burr formation was like for the blades, the results seem pretty good, but just think if they could be better with different sharpening. I would like to see that explored, as well as performing the FF process on all the steels they tested the D2 against. It seems we're getting a bit of tunnel vision, some viewing it as FFD2 vs everything else, when it should be FF vs current treatment. I'm not as interested in what happens specifically to D2 as I am in what FF does across the board and how it affects use & maintenance.

I'm guessing it can't be combined with a regular heat treat to add strength to the spine of a blade, which is too bad if that's the case.

 

[hardheart]

d) sharpen blades with 600 grit diamond belt, using fixture to maintain edge geometry. Same number of passes, same pressure.
e) Remove wire edge (burr) with cardboard wheel and CrO compound

Ah, there it is, same pressure and number of passes. Again, if there is a significant difference in sharpening response, different sharpenings would have to be evaluated to see where FFD2 really stood out, or perhaps even fell behind. Minimizing burr formation would be something I would personally try to do when checking sharpening response, and see how well the FF steel does. Verhoeven has already shown benefit, and Ben Dale of Edge Pro espouses a burr free sharpening method as the best for edge longevity.

 

[Cliff Stamp]

It doesn't look like the problem is "noise";

The actual measured data as I showed was perfectly fine and fitted by the model with well defined parameters. It was only when the calculations were performed on it which exploded the noise that the curve became undefined. It still fit the data fine as measured by the residuals, the noise was just so high that the curve was undefined, meaning too large of a range of parameters could satisfy the spread of the noise in in the data.

I think anyone who has tried to sharpen a 420HC 110 and one in BG42 with the same arkansas stones and number of strokes can tell you that using the same abrasive and technique for varying steels will not give the same sharpness.

Indeed, you have a number of problems here, first off all grindability is a huge issue and has a number of influences on edge formation.

My biggest problem with the advertising is the hard sell of differential hardening/tempering (two different things, of course) which doesn't actually do much for knife performance, and leaves you with a useless blade at lower forces than one that is fully hardened.

This is however a huge selling point in general to the public with guy like Fowler really promoting such blades for the extreme ability to bend. Of course what has been pointed out clearly is that such blades are also much weaker.

I'd like to know what the burr formation was like for the blades...

This can be a big problem with high vanadium steels (high carbide steels in general) when they are left soft. You would expect obviously for D2 to form a much more crisp edge when it is so much harder. I talked about this when I did work on AUS-4 awhile ago and how to get very high performance out of it. It takes very careful sharpening to get a high quality sharpness with it because it is so soft (CRK&T) and thus difficult to get a truely crisp edge.

It would be nice to see more steels with this process and a "what is gained" report and even a precise reason for why D2 was chosen as the flagship steel. It takes time and money though to do this on a bunch of steels and then determine the optimal hardening methods so first you would expect a strong promotion to actually get some money to continue the research. In time you would expect other steels to recieve the treatment and be offered. For similar reasons it makes sense to compare to the two CPM steels as one is very popular now and the other is referenced as a high end benchmark. This as noted (the PDF file) is just a promotional tool not a scientific investigation an they are goaled differently obviously.

-Cliff

 

[Kohai999]

This is however a huge selling point in general to the public with guy like Fowler really promoting such blades for the extreme ability to bend. Of course what has been pointed out clearly is that such blades are also much weaker.

The "lesser" hardened material was not left as soft as Fowler's, Cliff, and you know this by your own testing, comparitively. The hardness on the back of the FFD2 blade would not be "much" weaker, it was left in the "40's" which is hardly "weak".

That being said, Fowler's knives are quite thick and advertised as dependable through thick and thin, the FFD2 knives are marketed as hunting knives, designed to skin game. A "weaker" spine would not matter much in this application, as prying should not even be attempted, or am I missing something?

Best Regards,

STeven Garsson

 

[hardheart]

I don't know the specific's of Cliff's testing of Fowler's knives, but I find this to be untrue from the reading I've done

A knife that has
a fine-grained, hard, tough edge for long-lasting sharpness
yet the remainder of the blade is softer and slightly “springy”
and very tough is one of the trademarks of a superior blade.
The technique was, and is still used by the finest Japanese
Samurai sword makers to build battle swords that employ
hard, long-lasting sharp edges, with a softer, springy spine
that will not break due to their differential heat-treatment.
This technique is also used by some premier custom knife
makers. For the first time, differential hardening can be
achieved in a commercial blade with Friction Forging®.

Blades that are uniformly hard throughout are more brittle and will
break rather than flex when significant transverse loads are exerted.
Extreme examples are flint, ceramic and glass. If sharpened, all would
have extremely sharp, hard edges but will not stand much transverse
force. The same holds true for a steel blade. If it is uniformly hard,
the blade is not very forgiving and will break more easily than a blade
with a softer spine but hard edge. Thus another measure of a high
performance blade is one that has a softer tough spine and a very hard
edge that is not so brittle as to chip in normal use.

Comparing a hunting knife to a sword is inappropriate, imo, but beyond that, it has been shown by others, one of the more notable being Tim Zowada, that a harder blade requires more force to break, and it takes less to permanently bend a softer one. It doesn't matter how hard a blade is within the plastic region, but once you go beyond that a harder blade will return to true while a soft one will take a permanent bend. By the time you have bent and then straightened a soft blade, you have fatigued the metal and screwed with the edge while a harder one would have suffered no effect.

I'm not saying that the FF blade would not perform well in it's role as a hunting knife, but I see no reason to call a diff-hardened hunting blade 'superior'. If only the edge can be hardened with FF, that isn't necessarily a bad thing (depending on knife application), but it isn't necessarily a plus either, imo.

 

[Kohai999]

I'm not saying that the FF blade would not perform well in it's role as a hunting knife, but I see no reason to call a diff-hardened hunting blade 'superior'. If only the edge can be hardened with FF, that isn't necessarily a bad thing (depending on knife application), but it isn't necessarily a plus either, imo.

Hardheart, I am not about to engage in a debate with you, however, if you can't see the POTENTIAL superiority in this process/edge....than don't buy it, pretend it does not exist, and go on about your business.

STeven Garsson

 

[hardheart]

I don't have a problem with what it does at the edge, it's the trying to sell the soft spine that gets left behind that bugs me. The extremely fine grain size at the edge is awesome for D2, all I'm asking is just don't tell me the unhardened remainder is better than fully HT'd steel. I would also like them to not mix hardening and tempering when talking about doing it differentially.

Though this is getting off the subject of the edge testing, and I am still really looking forward to seeing where they can go with the testing to show the limits of this process and where it can take steel knives.

 

[Broos]

For any who are stuck on the fact that different steels perform differently at varying geometries, finishes, and sharpening, and think that the baseline for the testing used, re. Equal geometry, sharpening, and measured initial sharpness - is wrong.

Let's say we proceed with a methodology comparing the steels at their "optimum" blade and edge parameters for slicing cutting ability, Ci(o).

First a preliminary question - Will the different steels have different Ci(o)'s with optimal geometry and sharpening? This is something that needs to be considered. If yes, should the steel get credit for a high Ci(o) in the test? If no, then you are contending that all steel's Ci(o) are equal *(or you may not care one way or the other).

Will you start the test & make comparisons from each steels Ci(o)? Then you will start each blade from its optimum until a set stop point C "ending" and compare total material cut?

Or will you only compare the results over the same ranges of C, with C(start) and C(end) being the same for all? Then if one steel is capable of achieving a higher Ci than other steels, this advantage will be disregarded.

If you want optimum, then you will specify (and vary for the different steels in your test);
blade width; thickness; grind angle; edge angle(s); grit finish; sharpening procedure using specified sharpening abrasives; and ?

Will there be many different combinations of the blade/edge parameters above where the Ci(o)'s for any given steel will be equal? How will you determine which combination of these is optimal for each steel? And will any practical considerations be given in selecting the optimal parameters? Otherwise you may have some blades that look a lot like a serrated razor blades or maybe a hacksaw blade.

My opinion of anyone being able to specify these optimal parameters for each steel has already been stated, and I am not holding my breath to see any science capable of quantifying the optimum parameters above for any steel (not to mention that the optimum parameters would change depending on what is being cut).

So which is better, to start the test using the best hints available to set the baseline, and then use the results to make direct comparisons between blades of varying geometries, edge finishes, and angles?

Or to keep the blade parameters above constant for each test run, and make your comparisons from there. An added benefit is that further tests varying the baseline blade/edge parameters would develop into an understanding of which geometries and edge finishes are optimal for each steel.

 

[gator68]

For any who are stuck on the fact that different steels perform differently at varying geometries, finishes, and sharpening, and think that the baseline for the testing used, re. Equal geometry, sharpening, and measured initial sharpness - is wrong.

Let's say we proceed with a methodology comparing the steels at their "optimum" blade and edge parameters for slicing cutting ability, Ci(o).

First a preliminary question - Will the different steels have different Ci(o)'s with optimal geometry and sharpening? This is something that needs to be considered. If yes, should the steel get credit for a high Ci(o) in the test? If no, then you are contending that all steel's Ci(o) are equal *(or you may not care one way or the other).

Will you start the test & make comparisons from each steels Ci(o)? Then you will start each blade from its optimum until a set stop point C "ending" and compare total material cut?
...

This is all a very big IF. There is no evidence that the Diamondblade people have done anything like this. They presented results where for some reason their blade had a much higher initial Ci, then cut until a fixed C was reached. Despite having the same initial sharpening treatment, they did not seem interested in understanding or explaining why the FFD2 was so much better initially. This different starting point has a big effect on the results. You might as well start with a blunt SV90 blade and then say how your FFD2 blade cut so much more rope.

 

[gator68]

I don't have a problem with what it does at the edge, it's the trying to sell the soft spine that gets left behind that bugs me.

I agree. If the blade is not hardened through, you are essentially cutting the lifetime of the blade down to a third. Anyone seen pictures of old skinning knives that were used so much they now look like fillet knives?

There is a huge difference between katanas and knives, and I see no functional reason for a knife to have a hardened edge and unhardened blade.

 

[PatriotDan]

Hardheart, Gator68, do you consider differential heat treat beneficial for larger camp knives in the range of say 7 - 10 inches? How about knives meant for soldiers?

Thanks.

 

[db]

Quote:
Originally Posted by cds4byu
d) sharpen blades with 600 grit diamond belt, using fixture to maintain edge geometry. Same number of passes, same pressure.
e) Remove wire edge (burr) with cardboard wheel and CrO compound

Ah, there it is, same pressure and number of passes. Again, if there is a significant difference in sharpening response, different sharpenings would have to be evaluated to see where FFD2 really stood out, or perhaps even fell behind. Minimizing burr formation would be something I would personally try to do when checking sharpening response, and see how well the FF steel does. Verhoeven has already shown benefit, and Ben Dale of Edge Pro espouses a burr free sharpening method as the best for edge longevity.

If they did use a Arkansaw stone, or even a AO or SIC belt I'd agree there could be a significant difference in forming the edge. However I've found sharpening with diamonds there isn't much of a difference between different steels, at least none that I could notice. I haven't sharpened everything but I have sharpened S90V, full hard M2 and 1095, D2, A2, and a few softer AUS 8 blades with diamonds.

 

[hardheart]

No, I think using a more suitable steel and stock size is a better idea. But, I personally don't pry with my knives, so it isn't a big deal either way. That's why I say it isn't necessarily a plus or minus (unless you modify the profile as gator mentions) If I do ever have to pry with a knife, I don't want it to bend.

edit-yeah, I used arkansas to make it an easier and more graphic demonstration of how steel/abrasive mixes could definitely effect sharpening results. Diamond will cut either easily, which is why I wonder more about how much burr each one ended up with, or how much they fatigued the metal on the cardboard wheel. They may have oversharpened any of the blades, and could be looking at poorer results. heck, the FFD2 may cut even longer