440C vs N690C on Carpet

[Cliff Stamp]

I had a little carpet and was curious about a comparison of 440C and N690Co in two similar Spyderco blades :



-Chicago : 440C
-Pingo : N690Co

The knives had the original edge angles maintained :

-edge bevels at slightly more than 15 dps, less than 20 dps
-apex bevel set with a 600 grit DMT stone

I made cuts until the knives were very dull :

-could not slice 1/8" light jute under a 500 gram load

At this point the apex would reflect light, struggle to slice newsprint and only do short slices on photocopy paper. However they could still keep cutting carpet, it takes a LONG time for the edge to stop being able to cut at all if you are willing to just keep pressing harder.

The edges that Spyderco does apply angle wise are well suited to this work. While I generally run lower edge angles, once you go over 15 dps the edge becomes durable enough so that they won't even take damage even on high carbide steels in this work. Here is the worst spot of damage on the Chicago after one of the runs :



It is barely visible under 50X magnification. The Pingo was similar. The apex would reflect light and in fact the entire apex/micro-bevel would be worn away, but there was little damage extending into the edge itself.

I only did two runs with each blade, so these results are very tentative, but they make sense, I would not expect a significant difference between the two :

-Chicago : 100 (25) slices
-Pingo : 125 (10) slices

Interestingly enough, the Pingo was a much nicer knife to work with for the cutting :

-more comfortable in hand
-less belly kept the blade from slipping out of cuts

[jabba359]

-Chicago : 100 (25) slices
-Pingo : 125 (10) slices

I've wondered before, but what do the numbers inside the parentheses mean?

[Cliff Stamp]

Jabba,

They are ranges. 100 (25) means that the average results are 100 +/- 25. The 25 is basically the standard deviation. It shows the certainty or precision of the result.

[jabba359]

Got it. Thanks.

To pick your brain just a bit more, how do you determine the deviation value?

[Surfingringo]

i saw the title and was kind of hoping that this was about the serrata. I was interested to see how that cast 440c performed on a task like this. Seems like some of the qualities that steel is purported to have might work well on material like carpet. I dunno.

[Cliff Stamp]

To pick your brain just a bit more, how do you determine the deviation value?

There are a number of ways to estimate the deviation, the most common method is to use the standard deviation which is is a root sum of deviations squared. That just means is it basically an average deviation if you look at how each result compares to the average result. I typically use that as it works well in most situations. Any spread sheet will just calculate it for you, as will almost all modern calculators with even basic statistic packages.

[jabba359]

Thanks for explaining.

From a design perspective, I would have expected the Pingo to be the better knife (all around, not edge retention). My Chicago rarely gets use because the blade is so short. If I'm going to clip a knife to my pocket, then I typically want a bit more cutting edge out of the blade. I like my Pingo more than my Chicago for that one reason alone.

[Cliff Stamp]

I was interested to see how that cast 440c performed on a task like this. Seems like some of the qualities that steel is purported to have might work well on material like carpet.

It is an interesting material to cut because it is so abrasive. Practically I would have sharpened it after 50 cuts as it was already suffering from lack of performance. Initially when the blade is properly sharpened it just does long zip cuts. As work is done and the blade blunts then you have to switch to shorter cuts, multiple slices. But the time I hit ~100 cuts with each blade it was taking almost 10 times as many cuts to make one section of carpet. If I wasn't doing some kind of comparison I would have sharpened it before then.

The very coarse carbide structure of cast 440C might create some kind of uneven wear pattern. It might be that the tungsten carbidized blades might fare well either. I have one of them and it might be interesting to try. I also have some serrated blades. I am going to run a few more of the Chicago and Pingo first though. The nice thing about it is that since it takes blades down so fast then it is very quick to do comparisons. Trying to look at something on wood, hemp and even cardboard takes way longer.

[arty]

How do you get a standard deviation and a mean when the N = 1? The SD tells you where most of the scores in a distribution will fall and assumes a normal distribution.

[Cliff Stamp]

How do you get a standard deviation and a mean when the N = 1?

I didn't do one run, I did two with both knives and averaged the results. However even if you do one run you can still quote a standard deviation if the population statistics are known, or can be inferred. For example if you make a measurement with a normal weight scale you can quote a reading such as 1.6 (1) grams where the +/- 1 comes from the population statistics and it represents the precision of the scale from the inherent properties. Given the carpet cutting I have done in the past, i would infer a 50% +/- on one run for carpet cutting when the number of cuts is large and the carpet is random sampled. This is likely a bit larger than the actual deviation however I prefer to err on the side of caution and produce type II errors (lack of a claim of significance) vs type I errors (false claims of significance).

The SD tells you where most of the scores in a distribution will fall and assumes a normal distribution.

A standard deviation doesn't assume a normal distribution, the standard deviation is distribution specific, i.e., it can be calculated for any given distribution. A standard deviation is just the root of the variance, and the variance is just the expected value of the squared deviation of a given distribution. The assumption of normality comes in when you do something like say "95% of measurements fall within +/- 2 standard deviations", that is only true for the normal distribution. Many measurements are not-normal, a lot tend to be binomial as they are based on discrete events which are not continuous. For example when you measure something with a ruler and it is say 10.1 cm then if you were to measure it repeatedly it would follow a binomial, but if you drew a normal curve it would well approximate it. The reason we use normal statistic inference often anyway is that they are simply much easier to calculate.

In some cases I use median based statistics because there are issues with outliers which make the distribution really non-normal. In that case I usually use the approximation to the standard deviation which is to use the common root mean formula and multiply it by 1.25. It can be directly calculated (the median deviations) but the calculation is more involved and quite frankly I again am always going to err on the side of caution so I am looking in general for large approximation to spreads, not the smallest ones.

[KevinOubre]

Cliff is that the same stock of carpet from the Normark/Wilson/K2 video? Seems almost like to tore donw an old apartment. You have a huge amount of that stuff

[Cliff Stamp]

Cliff is that the same stock of carpet from the Normark/Wilson/K2 video?

Yes and no, in the same way if you replace the handle of a hammer is it still the same hammer? It contains some of that carpet but I have removed a lot of it from cutting and added to it. The main source of it is renovations. I help friends/family out with these projects as long as I can salvage it. I do this for two reasons. First, I actually do want/need the things I take which have no value to almost anyone else. However as important this "payment" keeps them from feeling they owe me anything as that can lead to a lopsided relationship. This is why I always have massive amount of scrap wood, ropes, cardboard, carpet, tires, etc. . I repurpose a lot of it, some directly and some indirectly.

[arty]

Perhaps it is just me, but I don't compute standard deviations when I don't have enough observations, and it doesn't make sense to me to compute it if you don't have a normal distribution of scores. While you can certainly compute a mean of 2 scores, other measures of central tendancy, like medians, are more appropriate if you don't have independent observations.
Just me perhaps, but I have a different view of statistics.

[Cliff Stamp]

...and it doesn't make sense to me to compute it if you don't have a normal distribution of scores.

Where did you get the idea that a standard deviation is only used for normal distribution? All distributions have standard deviations as it is just the root of the variance. When you are introduced to statistics the normal distribution is the one which is first shown because it has some very nice properties and in general a lot of physical behaviors are normal-like. However it isn't the case that only normal distributions have statistics, they all have means, standard deviations, etc. .

[JD Spydo]

I have found that carpet is truly rough on knife blades ( especially plain edged blades)>> but I've also found burlap sacks to be almost as much of an edge destroyer as carpet is on knife blades. I don't know if it is the backing of the carpet or if it is something added to the polymer strands themselves but it's about as bad as cardboard in some cases.

I've had several knives with 440C and it is a respectable blade steel in my opinon>> but I like VG-10 better for everyday uses. And for cutting materials like carpet I usually use one of my Spyderedged blades for that job.

I guess my question is this>> what are the materials in the carpet that makes it so rough on knives or edged tools in general? Also did you do any tests at all with any of your Spyderedged blades>> the reason I'm asking is because I would love to try my fully Spyderedged Temperance 1 fixed blade model with VG-10 steel>> also I do find Spyderedges or most modern serration patterns tend to cut carpet and other fibrous materials better than most plain edged blades do from my own personal experiences.

I hope I'm not too far off your subject matter but this seems to be a job that serrated edges might excel at. I've had great success cutting fibrous materials with my fixed blade SE Temperance 1 model.

[Cliff Stamp]

JD,

I am going to do some work with serrated patterns shortly. I am still not really confident in a way to measure the sharpness of them due to the tremendous influence the geometry has on how they cut. The sharpness isn't thus a good indicator in general of how a serrated blade can cut compared to a plain edge. In any case, I will be doing some work in a week or do depending on what else comes up. It is a very interesting question. One of the things I might look at first is just a very rough comparison of how long would I cut before I would sharpen which is a very rough/low precision comparison but an interesting one.

I cut mainly used carpet which is just full of dirt. Aside from that, many modern carpets contain some pretty harsh fibers, the backings contain glue and often contain fibreglass or similar in the weave.

[sal]

Thanx Cliff,

That concurs with our CATRA findings, and about what I would expect.

Looking forward to your opinions tests for serrations. Might be an interesting standard to have a plain edge tested along side the serrated model with same steels. :cool:

sal

[ChrisinHove]

I find the Chicago a useful size, plus it gives me lots of sharpening practice!

[Skywalker]

My statistics classes would've been so much more interesting if we were using knife performance examples...

Thanks for the informative post, Cliff. N690Co is supposed to perform similarly to VG-10, correct? The compositions certainly look similar.

[Stuart Ackerman]

N690, N690Co and VG10 are very similar chemical make-ups, and perform as if there was no difference, as long as you compare same Rockwell hardnesses and similar grind profiles....