Steel testing underway...

No idea on dulling medium.
But for sharpness testing - how about "cutting weight"?
You build a balance weight with uneven arms. On longer arm you place a specific weight. On the other arm - you mount or place a piece of string/line/other cut material.
When force needed to cut the string is larger then counterweight, cutting action will lift counter weight. Ergo - at this moment blade is dull.

Clever idea, I like the fact that it doesn't require a scale to be used...

My only thought is that the tension on the string would have to be very regulated to ensure consistency, and that might just introduce another error.

I'll see if I can lash something up at the shop and play with the idea.
 
Please pardon another dumb question: You did clean the edges (strop on your pants or other cloth to remove build-up of debris) before each slice of notebook paper, yes? And you are cutting the paper at approximately the same angle each time? I know that these are simple/obvious points, but numbers like that point to a key oversight...

Again, just following along. Thank you.

Both good suggestions. The paper was always being cut at the same angle yes.

I wasn't cleaning the blade edge though, it doesn't seem like it's something that the thicker paper would be especially sensitive to, but i'll try it out.
 
So I just went to my local art store. Walked out with a ream of 75gram copier paper and 9 2x3" sheets of corrugated cardboard (54 sq ft).

I was using cardboard from boxes previously, I was cleaning the tape and such off them, but perhaps there was some residue left over that was skewing the results.

I'll try the whole deal again tonight with all this 'clean' material. If I'm not getting consistent results this time then I'm going to just move on...

Wish me luck!
 
Bad news guys.

I got a friend who is more versed in statistics to have a look over the numbers from the edge-retention testing in round one. He has said that the numbers are likely not statistically significant, ie: they're not much better than noise and are likely garbage. Here is the post in question: http://www.bladeforums.com/forums/s...teel-testing-underway?p=12602474#post12602474

EDIT: He did point out that blade #6 (440C) did noticeably worse than the others, that's the only real conclusion to be drawn.

I've edited the post so that people don't find it on Google and take it as gospel.

I'm not the type to try to cover up a mistake like that, I'd much rather you all knew so that you didn't make decisions based on bad data!

Sorry for the hassle. I think the toughness testing is still valid.

From what I've seen so far I think edge-retention testing is very, very, hard to do right and requires many runs with the same blades to produce a useful dataset. I'm considering how I could use my forthcoming CNC mill as an automation platform to do some useful edge-retention testing, but that's a while off yet.

I haven't decided yet whether I will actually do the last round of edge retention testing tonight. At best it's likely going to be a subjective test...
 
Last edited:
I got a friend who is more versed in statistics to have a look over the numbers from the edge-retention testing in round one. He has said that the numbers are likely not statistically significant, ie: they're not much better than noise and are likely garbage.
...
EDIT: He did point out that blade #6 (440C) did noticeably worse than the others, that's the only real conclusion to be drawn.
...
From what I've seen so far I think edge-retention testing is very, very, hard to do right and requires many runs with the same blades to produce a useful dataset. I'm considering how I could use my forthcoming CNC mill as an automation platform to do some useful edge-retention testing...

Of course you haven't established 'statistical significance' with such differences and so few replicates. Were you trying to? That is what CATRA is for - high volume automated testing of slicing edge retention. I thought that you were simply assessing general properties, i.e. trends, not looking for statistical significance. Is it necessary to do so?

For the purposes of knife-steel, properties like edge-retention and impact toughness allow general categorization of the steels - high/mid/low wear, high/mid/low toughness. For impact toughness, most high-chromium steels at knife-blade hardness (i.e. 58-60 Rc) can endure 20-30 J/cm2 up to ~40 with PM or very fine grains; whereas tools steels like O1 & A2 tend to reach from 40 to 70; and high-toughness steels breach 100 J/cm2. For wear-resistance, even CATRA tends to categorize knives/steels rather than rate individual performance because measurement variability is so high. "Amateur" Ankerson here on BF gives ~10 categories of wear-resistance for his testing.

When companies market their steels, they prefer comparative bar-charts of relative values to actual measured values because they know that the measurements have little 'real-world' meaning. The specific value is less important than the general range of endurance. Statistical significance isn't necessary for generalizations, and generalizations are all that is necessary to demonstrate capacity. In summation, don't be too hard on yourself! This is great work and demonstrates what you are capable of with these steels as well as showing where you can improve.

Now, the degree of variance in your most recent test is still a matter of concern because it doesn't even show a trend. Given your method, a variation +/-20 is quite consistent, +/-40 somewhat less so, >40 is pushing it. Remember that your method tests how intact the very edge of the knife is, something that drops away very quickly in the stabilization process. Others have performed tests like this on literally thousands of knives or the same knife dozens if not hundreds of times in order to accrue a trend-line of edge-loss that is 95% confident. You don't want to waste time doing that. Keep up the good work, and stabilize that second set a little :thumbup:
 
Of course you haven't established 'statistical significance' with such differences and so few replicates. Were you trying to? That is what CATRA is for - high volume automated testing of slicing edge retention. I thought that you were simply assessing general properties, i.e. trends, not looking for statistical significance. Is it necessary to do so?

For the purposes of knife-steel, properties like edge-retention and impact toughness allow general categorization of the steels - high/mid/low wear, high/mid/low toughness. For impact toughness, most high-chromium steels at knife-blade hardness (i.e. 58-60 Rc) can endure 20-30 J/cm2 up to ~40 with PM or very fine grains; whereas tools steels like O1 & A2 tend to reach from 40 to 70; and high-toughness steels breach 100 J/cm2. For wear-resistance, even CATRA tends to categorize knives/steels rather than rate individual performance because measurement variability is so high. "Amateur" Ankerson here on BF gives ~10 categories or wear-resistance for his testing.

When companies market their steels, they prefer comparative bar-charts of relative values to actual measured values because they know that the measurements have little 'real-world' meaning. The specific value is less important than the general range of endurance. Statistical significance isn't necessary for generalizations, and generalizations are all that is necessary to demonstrate capacity. In summation, don't be too hard on yourself! This is great work and demonstrates what you are capable of with these steels as well as showing where you can improve.

Now, the degree of variance in your most recent test is still a matter of concern because it doesn't even show a trend. Given our method, a variation +/-20 is quite consistent, +/-40 somewhat less so, >40 is pushing it. Remember that your method tests how intact the very edge of the knife is, something that drops away very quickly in the stabilization process. Others have performed tests like this on literally thousands of knives or the same knife dozens if not hundreds of times in order to accrue a trend-line of edge-loss that is 95% confident. You don't want to waste time doing that. Keep up the good work, and stabilize that second set a little :thumbup:

Thanks for the encouragement mate, it's appreciated. It's worth noting that you've just given a bunch of useful information in your post regarding edge testing that I personally haven't seen anywhere else!

I'll admit I was thinking 'statistically significant data' when I first started testing, but that was just out of naivety. Now I'd be more than happy with general trends, and eliminating outliers.

I just prepared about 31,000" of cardboard stock ready for cutting up. I'm pumped to go establish some general trends :D
 
Ok! So I just finished running the test blades through all the cardboard I had... In total I cut 19,440" of cardboard (1620 feet). I was planning on doing several runs, but the first blade was still cutting at the end of the allocated cardboard, so I kept going until I had used up 1/6 of the total amount on that one blade, then I did the same with all the others.

Each blade cut 3,240" of cardboard.

And I am completely unable to tell them apart by sharpness.

Each blade:
1) No longer shaves at all
2) Happily cuts copier paper
3) Cleanly cuts phonebook paper 45º across the grain (from the side)
4) Won't cleanly cut phonebook paper with the grain (from the top)
5) Still cleanly cuts corrugated cardboard

If you guys have any suggestions for other simple tests that might help discriminate between them I'd like to hear them! On the face of it though it seems that they all held their edge quite acceptably.

EDIT: Here are some photos of the test stock and the resulting carnage:
OqCozvxl.jpg

H1ki8RHl.jpg
 
Last edited:
Ok! So I have moved onto the impact testing, tip strength, and break testing.

First up, impact testing:

Impact was administered with the help of a smallish crowbar. Each blade was hit both in line with the blade and then at about 45º off axis. The inline blows did very little damage to any of the blades, certainly nothing that couldn't have been sharpened out quite easily. To be honest this likely means that any of the blades would handle most low-level abuse quite nicely.

However the off-axis blows really showed up some differences between the blades:

Code:
[FONT=Courier New]
Worst
#12 - large 7/16" wide chip taken out of the blade
#9 - smallish chip at first (about 1/32" across), but the blade spontaneously cracked several minutes after the blow. The cracked extended 3/8" up the bevel
#11 - decent sized chip 3/16" across
#8 - small bent/torn section of edge 1/16" long
#10 - small chipped/rolled section of edge 1/16" across
#7 - small section of edge rolled, no chipping
Best
[/FONT]

#7
MYbOIpcl.jpg


#8
sEgD92Cl.jpg


#9
NMyRgkHl.jpg


#10
eSuiwECl.jpg


#11
P8owO2Jl.jpg


#12
OkBPOsYl.jpg


Tip-strength testing:

Each blade had its tip hammered into a piece of very hard maple using 3 moderate hammer blows on the butt. The handle was then used to lever the tip out sideways. If the tip did not fail on the first try, then the test was repeated until the tip failed, the tip was levered out in opposite directions each time.

Code:
[FONT=Courier New]
Worst
Failed on first try:
#9 (the tip was a little thinner than the others)
#7

Failed on 2nd try:
#12

Failed on 4th try:
#8

Broke on 6th try:
#11

Survived all 6 attempts:
#10
Best
[/FONT]

Winners: #8, #10, #11, #12

The tip on #10 bent quite a bit each time, so I get the feeling that perhaps the tip was softer than the others? Regardless I consider any of the blades that survived at least one attempt to be viable, this was a pretty brutal test given that the wood was extremely hard and well seasoned, and that each tip was embedded in about 3/8".

Break test:

Overall I was very impressed with the break testing this time! With the exception of blade #9 I consider all of the results to be acceptable. Each blade was very noticeably bent before failure, much more so than during the first round of tests. The only reason I consider this test important is because if a user of one of my knives decides to use it as a prybar then I want the knife to give them a strong visual indication that they're doing something bad well before it breaks.

The angles were measured by the angle of the tang of the knife.

Code:
[FONT=Courier New]
#7 - 42º
#8 - 42º
#9 - 10º (This blade failed early due to the crack that formed during the impact test earlier)
#10 - 90º (Wow!!!)
#11 - 35º
#12 - 42º
[/FONT]


#7 - 42º
sO7TuA6l.png


#8 - 42º
AD9qGMZl.png


#9 - 10º (This blade failed early due to the crack that formed during the impact test earlier)
EovZMKKl.png


#10 - 90º (!!!)
r7jyQREl.png


#11 - 35º
kZ0NPCxl.png


#12 - 42º
VXkT4Rnl.png


Coming up: my picks from the bunch, then we get to find out which blade is which!
 
So, given that from what I saw it was pretty hard to tell the different blades apart by edge retention, I'm going to basically ignore that part of the testing. Even taking micrographs of the edges at 150x magnification did not show up any differences between them after the edge-retention testing.

Basically it comes down to the choice being made on the basis of toughness and abuse resistance. I don't want my customers to abuse my knives, but if they have to during an emergency then I sure don't want the blades to fail on them without a seriously good reason!

That means my picks overall are:

Worst:
#9
#12
#11
#7
#8
#10
Best:

EDIT: I should say that I would be happy with either #8 or #10. The extra toughness that #10 showed during the bend test is great, but not really necessary when it comes down to it!

I'm going to go unwrap the blades now and find out which one is which! I'm excited!
 
Results time!

First things first, another quick disclaimer:

These results are specific to my blade geometry, heat-treatments, testing methods and so on. You'll likely get different results. Everything here should be seen as 'just another data point' rather than the last word on any of the steels involved.

Worst:
Blade #9 - A2 @ 62.5HRC (heat-treat ver2)
Blade #12 - 440C @ 59.5HRC
Blade #11 - A2 @ 60.5HRC (heat-treat ver2)
Blade #7 - CPM154 @ 62HRC
Blade #8 - O1 @ 59.7HRC
Blade #10 - CPM3V @ 60.5HRC
Best:

EDIT: Anyone seeing the results here for A2 should note that an experimental heat-treat was used for this test, not a standard A2 heat-treat.

I have to say I'm surprised that the A2 dropped back in terms of performance with the revised heat-treat. I think that the heat-treat I tried was likely optimized for edge-retention rather than extreme toughness. Putting it up against CPM3V is a pretty tall yardstick after all!

It's also worth noting that the differences in the other blades because of the revised heat-treats was pretty awesome! All of the other blades benefited from the updated heat-treats (except the toughness of A2, I bet edge retention on A2 has gone up if I could measure it). CPM3V in particular saw huge improvements, bending to 90º while fully hardened is simply amazing!
 
Last edited:
Here's what all the blades look like after being broken:

pu5wN0gl.jpg


And here is blade #10, after it's infamous 90º bend!

RwemlKJl.jpg
 
Compiled overall test results:

Given that the edge retention testing was not terribly significant I'm focusing on the toughness aspects of the testing. Below I have compiled a list of the results from both sets of tests.

I compiled the results using the blade numbers so I wouldn't remember which blade was which, then replaced the blade numbers with proper descriptions.

Break test:

Code:
[FONT=Courier New]
Worst:
Blade 9 - A2 Ver2 (62.5HRC) - 10º
Blade 1 - CPM154 Ver1 (59.6 HRC) - 15º
Blade 6 - 440C Ver1 (59.1 HRC) - 25º
Blade 3 - O1 forge  (55-60HRC) - 85º (bent strangely, plastic failure)
Blade 4 - O1 Ver1 (60.6 HRC) - 30º
Blade 11 - A2 Ver2 (60.5HRC) - 35º
Blade 7 - CPM154 Ver2 (62HRC) - 42º
Blade 8 - O1 Ver2 (59.7HRC) - 42º
Blade 12 - 440C Ver2 (59.5HRC) - 42º
Blade 2 - A2 Ver1 (59.5 HRC) - 55º
Blade 5 - CPM3V Ver1 (61.1 HRC) - 60º
Blade 10 - CPM3V Ver2 (60.5HRC) - 90º (Wow!!!)
Best:
[/FONT]

Tip strength test:

Code:
[FONT=Courier New]
Fail:
Blade 9 - A2 Ver2 (62.5HRC)
Blade 7 - CPM154 Ver2 (62HRC)
Blade 3 - O1 forge  (55-60HRC)
Blade 4 - O1 Ver1 (60.6 HRC)
Blade 6 - 440C Ver1 (59.1 HRC)

Pass:
Blade 2 - A2 Ver1 (59.5 HRC)
Blade 1 - CPM154 Ver1 (59.6 HRC)
Blade 8 - O1 Ver2 (59.7HRC)
Blade 10 - CPM3V Ver2 (60.5HRC)
Blade 11 - A2 Ver2 (60.5HRC)
Blade 12 - 440C Ver2 (59.5HRC)
Blade 5 - CPM3V Ver1 (61.1 HRC)
[/FONT]


Impact testing:

Code:
[FONT=Courier New]
Worst:
Blade 12 - 440C Ver2 (59.5HRC) - large 7/16" wide chip taken out of the blade
Blade 4 - O1 Ver1 (60.6 HRC) - large section of edge chipped out
Blade 9 - A2 Ver2 (62.5HRC) - smallish chip at first (about 1/32" across), but the blade spontaneously cracked several minutes after the blow. The crack extended 3/8" up the bevel
Blade 6 - 440C Ver1 (59.1 HRC) - decent chips and some bent sections
Blade 11 - A2 Ver2 (60.5HRC) - decent sized chip 3/16" across
Blade 5 - CPM3V Ver1 (61.1 HRC) - section of edge bent and deformed
Blade 8 - O1 Ver2 (59.7HRC) - small bent/torn section of edge 1/16" long
Blade 3 - O1 forge  (55-60HRC) - some small amounts of chipping and rolling
Blade 1 - CPM154 Ver1 (59.6 HRC) - very little damage, some slight chipping
Blade 10 - CPM3V Ver2 (60.5HRC) - small chipped/rolled section of edge 1/16" across
Blade 2 - A2 Ver1 (59.5 HRC) - almost no damage, slight rolled section of edge
Blade 7 - CPM154 Ver2 (62HRC) - small section of edge rolled, no chipping
Best:
[/FONT]

The only real failure in edge-retention was 440C Ver1, which held an edge noticeably worse that the others. Every other blade is likely pretty close overall in retaining an edge.

Overall ranking:

Point were assigned to each blade according to how it placed overall in each test. 1st was 12 points. 12th is 1 point. Not breaking in the tip test is 6 points.

Code:
[FONT=Courier New]
Points:
Blade 1 : 2 + 6 + 9 = 17
Blade 2 : 10 + 11 + 6 = 27
Blade 3 : 4 + 0 + 8 = 12
Blade 4 : 5 + 0 + 2 = 7
Blade 5 : 11 + 6 + 6 = 23
Blade 6 : 3 + 0 + 4 = 7
Blade 7 : 7 + 0 + 12 = 19
Blade 8 : 8 + 6 + 7 = 21
Blade 9 : 1 + 0 + 3 = 4
Blade 10 : 12 + 6 + 10 = 28
Blade 11 : 6 + 6 + 5 = 17
Blade 12 : 9 + 6 + 1 = 16
[/FONT]


Best:
Blade 10 - CPM3V Ver2 (60.5HRC)
Blade 2 - A2 Ver1 (59.5 HRC)
Blade 5 - CPM3V Ver1 (61.1 HRC)
Blade 8 - O1 Ver2 (59.7HRC)

Blade 7 - CPM154 Ver2 (62HRC)
Blade 1 - CPM154 Ver1 (59.6 HRC)
Blade 11 - A2 Ver2 (60.5HRC)
Blade 12 - 440C Ver2 (59.5HRC)
Blade 3 - O1 forge (55-60HRC)
Blade 4 - O1 Ver1 (60.6 HRC)
Blade 6 - 440C Ver1 (59.1 HRC)
Blade 9 - A2 Ver2 (62.5HRC)
Worst:

Overall I would consider the top 4 blades to be good choices for a hard-use knife. Blade #7 (CPM154 Ver2) would also probably be fantastic as long as care was taken to design in a strong tip.

Again, it's worth noting that your results will vary, especially with different edge geometries. Different heat-treat will obviously make a huge difference as well.

It's been a lot of fun doing all this, I appreciate you all following along and offering advice and encouragement!
 
And here is the compiled heat-treatments that were done for both rounds of testing.

Round 1:

CPM3V Ver1 (blade #5):
Preheat to 1500ºF, Equalize
Ramp to 2050ºF, hold for 20 minutes
Quench in air to below 125ºF
Temper at 1000ºF, 3 times, 2 hours each -> 61.1HRC

440C Ver1 (blade #6):
Preheat to 1425ºF, equalize
Ramp to 1900ºF, hold for 20 minutes
Quench in air to room temperature
Temper at 300ºF, 2 times, 2 hours each -> 59.1HRC

CPM154 Ver1 (blade #1):
Preheat to 1400ºF, equalize
Ramp to 1900ºF, hold for 60 minutes
Quench in oil to below 125ºF
Cryo-quench into dry-ice & isopropanol for 20 minutes
Temper at 400ºF, 2 times, 2 hours each -> 59.6HRC

O1 Ver1 (blade #4):
Preheat to 1250ºF, equalize
Ramp to 1500ºF, hold for 15 minutes
Quench in oil to 150ºF
Temper at 400ºF, 2 hours
Cryo-quench into dry-ice & isopropanol for 20 minutes
Temper at 400ºF, 2 hours -> 60.6HRC

A2 Ver1 (blade #2):
Preheat to 1100ºF, equalize
Ramp to 1775ºF, hold for 35 minutes
Quench in air to below 150ºF
Temper at 400ºF, 2 hours
Cryo-quench into dry-ice & isopropanol for 20 minutes
Temper at 400ºF, 2 hours -> 59.5HRC

O1 Forge (blade #3):
Heat in forge until dull cherry-red color
Quench in oil immediately
Temper at 400ºF, 2 hours, twice -> 55-60HRC

Round 2:

A2 Ver2 (blade #9 & #11):
Double coat in anti-scale (ATP-641)
Stress relieve: Ramp to 1200ºF, hold 2 hours, furnace cool to 900ºF, cool in still air to room temperature
Ramp to 1740ºF @ 400ºF/hr, hold 20 minutes
Oil quench until no longer glowing, then cool in still air to room temperature
Sub-zero treatment in dry ice and isopropanol for 45 minutes
1 blade: Temper @ 4250ºF, twice, 2hrs each time -> 62.5HRC
1 blade: Temper @ 600ºF, twice, 2hrs each time -> 60.5HRC

CPM154 Ver2 (blade #7):
Wrap in stainless foil envelope
Stress relieve: Ramp to 1200ºF, hold 2 hours, furnace cool to 900ºF, cool in still air to room temperature
Preheat to 1400ºF, hold 10 minutes
Ramp to 1900ºF, hold 60 minutes
Plate quench to ambient
Sub-zero treatment in dry ice and isopropanol for 45 minutes
Temper @ 400ºF, twice, 2hrs each time -> 62HRC

O1 Ver2 (blade #8 ):
Double coat in anti-scale (ATP-641)
Stress relieve: Ramp to 1200ºF, hold 2 hours, furnace cool to 900ºF, cool in still air to room temperature
Preheat to 1200ºF, hold 10 minutes
Ramp to 1470ºF, hold 30 minutes
Quench into oil to ambient
Sub-zero treatment in dry ice and isopropanol for 45 minutes
Temper @ 400ºF, twice, 2hrs each time -> 59.7HRC

CPM3V Ver2 (blade #10):
Enclose in stainless foil envelope
Stress relieve: Ramp to 1200ºF, hold 2 hours, furnace cool to 900ºF, cool in still air to room temperature
Preheat to 1500ºF, hold 10 minutes
Ramp to 1975ºF, hold 30 minutes
Plate quench to ambient
Sub-zero treatment in dry ice and isopropanol for 45 minutes
Temper @ 975ºF, three times, 2 hrs each time -> 60.5HRC

440C Ver2 (blade #12):
Enclose in stainless foil envelope
Stress relieve: Ramp to 1200ºF, hold 2 hours, furnace cool to 900ºF, cool in still air to room temperature
Preheat to 1500ºF, hold 10 minutes
Ramp to 1875ºF, hold 30 minutes
Plate quench to ambient
Sub-zero treatment in dry ice and isopropanol for 45 minutes
Temper @ 400ºF, 2 times, 2 hrs each time -> 59.5HRC
 
I found some more info about A2, subzero quench and resulting hardness.
a2.GIF


This might answer question why A2 got so hard.
 
Great find Idaho!
I had to look up the temperature conversion, so here they are if anyone else needs them...

950ºC -> 1742ºF
1000ºC -> 1832ºF
1050ºC -> 1922ºF
1100ºC -> 2012ºF

The first graph is the closest to the heat-treatment used during the second round of testing (it's bang-on temperature wise actually). It's interesting that it shows a much lower hardness at the same tempering temperatures that I used.

I think there's a possibility that something went wrong with the heat-treatment for the A2 in the second round. For instance, perhaps I kept the blades in the quench oil too long? Perhaps the soak at temperature was too long or too short (the graphs don't mention soak times)?

Interesting to see too that the graphs show the results of dry-ice (-80ºC) and Liquid nitrogen (-180ºC) as so similar, at least in terms of hardness.

Thanks for digging that up Idaho! This is why I love sharing results, as other people like yourself will always add meaningful info!

-A
 
It's interesting that it shows a much lower hardness at the same tempering temperatures that I used.

I think there's a possibility that something went wrong with the heat-treatment for the A2 in the second round. For instance, perhaps I kept the blades in the quench oil too long? Perhaps the soak at temperature was too long or too short (the graphs don't mention soak times)?
-A

Something to definitely look into, Aaron.
 
Back
Top