Edge Retention

CATRA Tests of M390 Knives

Thanks to Yuval Peleg, Patrck Roehrman, Hugo Costa, and Stefan Stromberg for becoming Knife Steel Nerds Patreon supporters!

CATRA Testing of M390 Knives

Recently two sets of CATRA studies of M390 knives have been reported. One was conducted by Mike Latham of CollectorKnives, and the other by a group of people including Clint of the Alchemy_1 Youtube channel. These reports are the product of a minor controversy over the Rockwell hardness of production knives and the relation between Rockwell hardness and edge retention. Here are links to both reports:

CollectorKnives: Verifying Cutting Analytics

Alchemy_1: CATRA Report

Background on CATRA Testing and Hardness

I have written several articles about CATRA testing and the effect of Rockwell hardness on properties. Here are links to a few relevant articles:

Rockwell Hardness is the Megapixels of Knife Steel Specs

Maximizing Edge Retention – What CATRA Reveals about the Optimal Edge

Which Steel Has the Best Edge Retention? Part 1

Which Steel Has the Best Edge Retention? Part 2

Those articles introduce what Rockwell hardness means, what properties it controls, and introduces CATRA testing and what it measures. I will not be re-introducing it here, other than to say that hardness is a measure of strength, and also improves resistance to wear. An improvement in strength and wear resistance helps with slicing edge retention. CATRA is a slicing edge retention test where the knife cuts through cardstock with a 5% silica content which abrades the edge. The amount of cardstock cut with each slice is recorded over a 60 back-and-forth stroke test. CATRA edge retention is controlled by edge geometry (edge angle and thickness behind the edge), steel hardness, and the volume and type of carbides.

M390 and S30V Edge Retention

Most of these reported tests were performed on M390 steel which has 2.5% vanadium carbide and 17.5% chromium carbide. One knife in S30V was tested which has 4% vanadium carbide and 10.5% chromium carbide. Since both of those steels have 4% vanadium, the chromium carbide hardness is increased somewhat because the chromium carbide becomes somewhat enriched in vanadium. You can see where S30V and M390 sit in terms of edge retention in this chart of experimentally measured CATRA edge retention, which presents values relative to 58-59 Rc 440C (100%):

Predicting Edge Retention

The greater overall carbide content of M390 provides it an edge retention advantage over S30V. In my earlier CATRA articles I used statistical analysis of a large dataset to provide a prediction of the CATRA test based on hardness, edge angle, and amount of each carbide type:

TCC (mm) = -157 + 15.8*Hardness (Rc) – 17.8*EdgeAngle(°) + 5.0*Fe3C(%) + 11.2*CrC(%) + 14.6*CrVC(%) + 26.2*MC(%) + 9.5*M6C(%) + 20.9*MN(%) + 19.4*CrN(%)

Vanadium carbide is calculated using the MC coefficient, where M can refer to either V, Nb, or W. You can read more about the different carbide types and why we use “M” in this article on carbides. The vanadium-enriched chromium carbides are calculated with what I have listed as CrVC. So if we had a knife with 16 degrees per side (32° inclusive) at 60 Rc in M390 we would estimate 542 mm of cardstock cut and 481 mm for S30V. The equation above predicts 15.8 mm for ever extra point on the Rc scale though that varies some based on steel, edge angle, and experimental scatter. Here are results from a few different conditions to show what I mean, where each circle is an individual test:

Independent M390 CATRA Tests

To see how the equation is doing with predictions of production knives I compared the estimated edge retention with the results of these new M390 tests:

There are good examples of experimental scatter here. The three lionSteel knives performed very similarly despite having some small differences in hardness and measured edge angle. The Kershaw and Spyderco M390 knives performed a bit better, perhaps because of a bit higher hardness when at the same edge angle. The Spyderco M390 knife oddly outperformed all of the others at 785 mm, which I attribute to experimental scatter. However, it could be that thickness behind the edge is the differentiating factor between lionSteel, Kershaw, and Spyderco knives. Thickness behind the edge measurements are not available in the reports.

Comparing the two Spyderco knives, there is a significant jump between S30V and M390, though that is also a comparison with the outlier M390 result higher than the rest. But if the comparison is made with the Kershaw M390 knife instead it still appears that M390 has better edge retention than S30V, which is expected.

Hardness comparisons are a bit more challenging since the comparisons are primarily between different knives. We must not confuse comparisons between different knives with different edge profiles, thicknesses etc as being tests of the steel or hardness only. There are extra confounding variables. I don’t know if we can necessarily compare the hardness trend between the lionSteel, Kershaw, and Spyderco M390 knives due to the geometry differences between them. There is still a trend with hardness but it’s hard to tell if the geometry differences or experimental scatter are drowning it out:

Prediction vs Measurements

Across the board the predicted CATRA measurement is lower than what was measured, averaging 142 mm lower, or 122 mm if the Spyderco M390 knife is ignored. This is somewhat annoying because I was not sure what the difference could be attributed to. Possibilities include:

  1. Thickness behind the edge – The edge thickness was not measured either in the original dataset I used or in these new tests so comparisons could not be made. It could be that these tested knives are thinner than the knives used in the dataset, which were primarily simple rectangular test knives, not consumer pocket knives.
  2. Sharpening – The new tests were performed with knives sharpened by CATRA with their special sharpener which they claim leads to superior edge retention. Edge finish did not show much effect in the previous tests on 154CM knives, but I speculated in that article that the diamond sharpening plates used is the reason behind the flat trend, because others have reported a strong effect of edge finish on slicing edge retention. Diamond plates have been reported to lead to a coarse edge regardless of grit size. Typically coarser grits lead to better slicing edge retention.
  3. Experimental scatter – As pointed out in the above analysis, values can vary somewhat. If the knives were re-tested they would have a slightly different result. However, the knives here have a consistent delta with the prediction and it is unlikely that all of them would randomly be low.
  4. Stroke length of test – I was not particularly satisfied by the three possibilities listed above, and then knifemaker Shawn Houston pointed out an important detail to me – CATRA used a 25 mm stroke length because these were short pocket knives. The more typical stroke length is 40 mm which is what was used in the entire dataset that became my predictive equation. In the CATRA report it says, “The normal test is for the blade to be assessed over a 40 mm cutting stroke; however, due to the size of the blade this was reduced to 25 mm and the results factored to give an equivalent performance result. Factoring is based on a test data obtained from a single straight blade where we are able to obtain data for 25 a 40 mm cycles on the same edge.” Obviously a single test knife is not sufficient for developing a “factor” to convert from 25 to 40 mm, as that factor would change based on steel tested and especially edge angle. In an email to Shawn, a representative of CATRA actually said something similar: “I must strongly advise you not to compare 40 mm test results with 25 mm test ones even by using a multiplier. Tests carried out by CATRA show that due to the dynamic movements of the 3 slides operating during the cutting process this ratio changes depending on the actual cut level.”

Therefore my opinion is that the difference between the predictions and experimental values are due to the difference in stroke length of the test. The predictive equation is not perfect though I have confidence in it as a relative comparison (higher hardness steel with more carbide at a lower edge angle will cut better). However, the source of the consistent delta between the prediction and these experiments was curious enough to warrant a little investigation.

Miscellaneous

CATRA measured both Vickers and Rockwell hardness on the knives, apparently because they felt that Vickers would be more accurate. They pointed out that the Rc values were about 1 point lower than the values obtained from converting Vickers to Rc. However, conversions between different hardness tests is not an exact science. In an article on high alloy tool steels, for example, it was found that the ASTM conversion scale can be off somewhat [1]. They proposed a new equation of Vickers = 111*exp(0.0316*HRC). Below is a comparison between the hardness as CATRA converted it and the proposed improved conversion in the cited article. Using the new conversion there is very little difference between Vickers and Rockwell C. Either test appears to be relatively consistent.

CATRA reports both TCC (total cardstock cut) and ICP (Initial cutting performance). The TCC is the cardstock cut over the entire 60 cycle test. The ICP is the cardstock cut over the first three cycles. CATRA calls the ICP a measure of “cutting ability (sharpness) of the blade” at the start of the test. However, it is certainly not a test of sharpness but more accurately cutting ability so ignore the parenthetical sharpness. Sharpness and cutting ability are different as described in this article: Sharpness vs Cutting Ability. Sharpness can only be deduced from a cutting ability test if the edge geometry is identical. However, the CATRA test cannot be used as a measure of pure cutting ability either, because even cycle number 1 is affected by edge retention and wear resistance of the knife because each cycle has both a forward and backward stroke. The first half-cycle wears the edge, then the edge is drawn over the cardstock again, resulting in a lower value for steels/knives with lower edge retention. So ICP should not be used as a representation of cutting ability or sharpness. You can read more here.

Summary

These CATRA edge retention tests were a fun comparison between different knives and steels. I don’t think there was much that was too surprising here in terms of comparisons between different knives. M390 had better edge retention than S30V, as expected, and increased hardness led to increased edge retention. Experimental scatter and differences between knives can make hardness or steel comparisons difficult, however. Identical knives are always best, of course, to reduce the number of variables being tested. There was a difference between predicted and measured edge retention, which appears to be attributable to the difference in stroke length which was compensated with a simple multiplying “factor.” The multiplying factor does not adequately convert between the 25 and 40 mm stroke length tests.


[1] Samal, P., and J. Newkirk. “Properties of Powder Metallurgy Tool Steels.” (2015).

8 thoughts on “CATRA Tests of M390 Knives”

  1. you said:
    both of these steels have 4% vanadium… but only s30v has 4% vanadium, you said s390 has 2.5 % vanadium?

  2. Any objection to me taking re-purposing your Catra edge retention data to publish with the Cedric & Ada cut tests? I think it will be interesting to see how these align to his testing.

    I assume there’s a specified sharpening process or sharpener used for these before they’re tested? What’s the edge angle, if so?

    Also, is there any data on initial sharpness? Can we assume they’re all at the same sharpness level?

    1. Maybe I forgot to say in the article that all were sharpened by CATRA on their special sharpener. The measured edge angles are shown in the reports and repeated in my table. No measurement was taken of sharpness. They have a special sharpness tester but I assume they charge extra to actually use it. They should have a similar level of sharpness assuming their sharpener has good repeatability.

      It’s not my CATRA data so I can’t give permission to use it.

    2. Hello Larrin and David,

      First, the article was a good read, Larrin. I want to add a bit more context to the group’s perspective and speak to using the data from Project Action.

      Yes, there has been talk about hardness directly relating to edge retention. However, our perspective is not solely that. We take into account chemical composition, protocol, geometry, and hardness.

      A good example of this is the Kershaw outlier you mentioned. It is the hardest knife in the test batch, but it is not the highest performing, as far as “toughness”/edge retention.

      As far as using the data, it is the community’s data. The test was funded, to a large extent, by the community. Therefore, all documents have been provided to the community for the use of the community.

      I am not here to plug my channel. However, I have linked to the video where I discuss the results of our test. In the description of this video, there are links to both reports.

      Thank you, Larrin, for always providing great content and thoughtful perspective.

      Clint “Alchemy1”

  3. Thanks for the Info. It’s clear to me that some manufacturers know how to heat treat m390 properly while others fail miserably such as lionsteel and viper.

  4. Not really sure what algorithm CATRA is using for the Hv/Hrc conversions, but I noticed it was high a few months ago. Although I do feel the Vickers is a more accurate reading, that is lost if you err the conversion. I have been using Taylors table at http://www.taylorspecialsteels.co.uk/pages/main/conchart.htm and it seems to line up very close with your algorithm. I sent a blade (the Dom you reference above) that Peter’s had tested at 59.5Hrc and the Vickers came back with 722Hv; which via Taylors is @59.4Hrc and the algorithm above rates 59.26Hrc.

Leave a Reply

Your email address will not be published. Required fields are marked *