Category: Tool Making

Some time ago, I mastered controlling grain size and manipulating hardness of 26c3 and O1 steel. If this is your first post, you can find what those are with a google search.

The cycle that I published and used works for both of these well. I eventually sent samples to a metallurgist as I posed on a woodworking forum that I doubt that these ever vary by more than a point. This caused some people to call BS, but the hardness part is definitely true. Small variations in hardness around 60-64 cause a great difference in how a plain steel performs on certain stones. These stones are chosen because I know that one of them will begin to slow down on 62 hardness plain steel, a lot, and another will become slow on plain steel that’s more than about 65.

The metallurgist that I sent the samples to, I have a feeling, doesn’t care for my methods. I get that. I don’t normalize steel. I think for plain steel, it’s not necessary, but one has to be willing to snap samples and do some actual testing of results before determining that it’s worth sending something for testing.

When I sent my samples, I didn’t care that much about toughness, but the market for knife makers is much bigger than independent tool makers, so a lot of the testing has a big emphasis on toughness. Toughness has a lot of influence on whether or not someone will break a knife, and in the world of knives, a broken knife is a huge problem.

I sent the samples, asked about hardness, and then got the hardness results first. I was happy with them. The person doing the testing had concerns about the level of finish I did with the test samples (figure a fraction of a stick of gum size) because I had to precisely size and finish these little samples freehand. The tolerances are pretty tight for freehand. However, toughness results were also fine. Compared to hardness, both were as good as the metallurgist’s published results – that is, the toughness and hardness balance was good.

Here is the chart I received – I markered in the 1095 result. That will make sense in a second.

These results are no fluke, but I haven’t worked up and snapped samples for anything else, so I couldn’t guess anything other than hardness, which I can tell on a stone.

Here’s Where I Went Wrong

Out of curiosity, I then sent sample coupons of 1084 and 1095 later, and I heated XHP and sent coupons of that. When I first attempted to match XHP to V11 steel, I heated it very hot very quickly, but later learned that one of the reasons for high heat is to normalize. I’m not normalizing steel, so I didn’t heat the coupons to the same high temperature – in my mind figuring that since I wasn’t dissolving chromium but just hardening a sample, the high temperature wasn’t necessary.

I was wrong. The sample should’ve tested about 60/61 hardness at my tempering level. I’d made two knives with lower heat and they also seemed a little soft – you could just barely file the knives before tempering. They probably are – the test coupon came back 57.3 hardness or some figure within a couple of tenths of that.

I didn’t do my due diligence and attempted to reason into a change without testing it. I also never snapped 1084 or 1095, but just assumed that they would work just as well with the method that I use for 26c3 and O1. Well, they don’t. Off of the top of my head, I believe the 1095 averaged 63.1 hardness and 4.1 or 4.3 ft lbs of toughness. This was with a 400F temper, which is off the charts compared to published results without liquid nitrogen – something I don’t have. I did make a couple of samples of 1095 irons and they are OK, but will nick a little too easily and they were always harder than expected.

Why? I assumed that my prior heat treat cycle would just work and I didn’t so much as snap a single sample before sending anything off for testing.

Bad idea. 1084 was even worse. 61.5 hardness or so on average and toughness even worse than 1095, which is the wrong direction. I was unhappy with the results, but I don’t use any of these steels for anything, and for my own consumption, if I could bump an XHP sample to 58 hardness and just buff the edge to make up for the slight lack of hardness, no real problem. Toughness was good enough that breaking wouldn’t be a problem.

I was so pleased with the 26c3 results and that I’d made a claim of being able to make samples separately without varying hardness that I reasoned that there shouldn’t be an issue, especially with 1084, not testing anything. It’s always suggested as being easier to heat treat than O1 or 26c3. I’ll put aside XHP for now until I have a chance to run another sample to high heat and see if it completely thwarts a good file on a sharp corner. I failed to do something that only takes ten minutes at the most. Create a small coupon for myself and observe grain size.

Yesterday, I finally addressed the 1084 issue by making a couple of samples. I was shocked to find out both with grain snapping and testing with a magnet that what works for 26c3 in a forge is overheating 1084 – a lot. My eye is trained to see the point where 26c3 transitions to nonmagnetic. 1084 transitions (unexpectedly to me, but probably not to others) at a cooler appearing color. And then the second part of the mystery.

I intentionally overheat 26c3 during a quench. It takes about 10 extra seconds to do this and the result is imperceptible change in grain, but higher hardness. This is a nice thing when making chisels, and maybe not worth as much excitement making knives.

You can compare my results to a simple schedule of expected hardness and toughness here – I tempered steel at 390F for 26c3, two tempers. To be conservative, I think it’s also useful to check toughness a point or two softer than my averages as one end of the coupons is a point or two (it varied) less hard because it’s held by tongs and it’s not changing from hot to cold as fast in the quench. As in, my samples average about 63.8, but you could compare chart toughness at 62 and I think that’s fair.

Link to Chart of Expected Results

Link to more about 26c3

No matter how you look at it, the forge results are unexpectedly – at least to me – good. I never sent any of these coupons for analysis until I’d made 26c3 chisels that hold up in side beside testing better than anything else I have that’s not Japanese. But I still figured the samples would have some kind of shortcoming. They didn’t. I think I could do 100 of these on 30 different days and they’d very little.

So, What’s the Problem

The 10 seconds of chasing the heat higher to prep for a fast quench with 26c3, and the same color of thermal cycles that I’d trained myself to see by eye – both grow grain a little in 1095 and a whole lot in 1084.

If I’d done 10 minutes of due diligence and maybe 1 hour of various trials and a little bit of analysis, I’d have known not to send either sample, and in not doing that, I wasted my time and the time of the person doing the testing. But I think in advising against heat treatment, they may have at least gotten some enjoyment out of the fail. I didn’t get a chart for the second set, so I have no chart to display. The results, though, match what I felt with 1095 irons – hard and a little bit under tough for the hardness, and half of the toughness that they would’ve been done properly in a furnace a point or two softer.

I finally started this with an offcut yesterday. It took less than 10 minutes to make these samples. These are 1/10th of an inch thick bar stock. Having little for analysis, what I do to refine cycles (and did for 26c3) is snap samples along with some kind of performance test. The first sample is just whatever the steel is heated once without any intentional overshot and quench.

1084 – low temperature quick quench

Then, I will intentionally overheat a sample, and see what it looks like. This is with about 10 seconds of chasing the quench temperature higher than needed to get better hardness. Exactly what I do with 26c3, which shows no visual change in grain size.

This is a shock, and the difference in coarseness is drastic. I’m sure my tested samples were at least as large as this, but they may have grown during thermal cycling. I never broke one to look, but i have looked at 1095 in the past. I didn’t break one of these because I didn’t have extra steel after cutting samples, and if I did break 1095 at or near the same time, it wouldn’t have looked quite this bad. based on the test results.

The steel that is supposedly an easy starting point is no good for what I usually do. so, I made a third sample, increased the grain size slightly and then used lower temperature thermal cycles than what I do for 26c3, and heated it quickly a little hotter than the first sample.

1084 is my worst steel, so it’s the one I’m trying to conquer first.

I can take the unbroken part of the middle sample now and confirm that the grain will get close to #1 above, then I’m well on my way.

And sitting around and guessing at the various problems – a complete time waster (“was it the steel? Maybe it wasn’t rolled and treated well? Maybe it was mixed up with something else?”).

I don’t think any of those happened. I know now that I made bad samples because of an assumption that the cycle used for 26c3 should be fine for any plain steel.

If you are going to do heat treatment in a forge and develop something that’s relatively easy – The 26c3 cycle is easy for me – reflexive at this point – you absolutely have to snap samples and confirm that you can heat treat without increasing grain size before moving on to anything else.

I’ll post tomorrow or in the next few days about the very simple quick method of doing this.

When you begin to make your own tools and branch out from O-1 steel, there’s an interesting array of alloys that can be done well in a forge. 1095, 26c3 (my favorite for chisel making, better than anything commercially available outside of Japan), 1084. Of course there are plenty of mellow steels in terms of heat treating complexity like A-2, and V11 (probably CTS-XHP).

O-1 is the first steel I ever heat treated, and it works well. In my heat treat article I advised if you’re not looking to learn much, just buy spheroidized O-1 steel, or really any good brand of O-1 steel, heat it quickly a full color past the point where it becomes nonmagnetic, quench it and temper it.

But I branched out, 26c3 was something I’d never heard of, and my results with 1084 have been mediocre as it’s not something I’m interested in using and it probably needs a simpler cycle than the one I’ve published on here.

I’ve rehardened various tools and blades over time and found as documented on here that even among three stanley irons and a chisel later, none are the same alloy. Where did all of these steels go? The answer is, I think, two parts: 1) some may not have been used by the general public in he first place whether or not they are still available, and others are out of production.

This leads us to 1095, which I expected to be a little softer and with good toughness and uniformity because the micrographs are very fine. In fact, it looks twice as fine as 26c3 when I snap samples. In my samples, it’s 1/3rd to 1/2 as tough and quenches and tempers harder than one might expect. In my case, 400ºF tempered samples came back at 63.1 c-scale hardness and relatively low toughness somewhat in line with what industrial charts would predict at that hardness. I could and may continue to experiment with it to see if I can improve the results slightly, but realistically, optimizing something that’s not ideal just leads to a little better version of the same thing. With 1095, I have a few irons that are high hardness, very crisp and easy sharpening, but they develop tiny nicks more often than I like. Put a different way, if I was running a business and had only 1095 plane irons and chisels, it would operate fine, but I’m in toolmaker mode.

1095 Knives and Saws – Knives aren’t Really 1095, Probably for the Better

Woodworkers think of 1095 as spring steel. Charts for knife companies show that “1095 isn’t a very hard steel, but it’s tough, and it’s 55 hardness” or something of that sort. The latter is just idiots collecting specs that they’ve found somewhere and making a chart, leading to more unintentional idiots presenting them on woodworking forums and making assertions. Often with the bend to them that what’s really missing in history is superior modern steels. This becomes a full lagoon of mixed animal turds once the discussion starts. I’m past that, but you can catch me once in a while going off the rails and blasting someone who claims to be an expert responding to specific questions with confident answers that make no sense and are sometimes perfectly incorrect – as in, exactly the opposite of reality.

If you have a question about steels, you’re free to ask me. If I don’t know the answer, I’ll point you to someone who does.

I’m sure this isn’t confined to woodworking.

So, back to 1095. 1095 saws are probably just that, 1095 spring steel. What is it? More or less 1% carbon, some manganese for hardenability and not a whole lot else. Spring just refers to the temper, or how hard it is. At spring temper, you can bend steel readily and if you don’t intentionally find its limit, it will return to its original shape. 1095 knives, however…..I doubt any are 1095. But many are or were marketed as that. What they really appear to be is steels that are relatively low alloy, but that have additives to improve toughness vs. 1095, and in some cases, probably to make the alloy more amenable to less costly industrial heat treatment processes.

I long believed that if I could start making things from steel, it would just be a matter of finding really plain steel (iron and carbon) to get vintage steel-like results. Like really vintage – 200 years old instead of 100, and then I could find one with higher carbon and get biting hardness and uniform wear – that is, a strong chisel edge and no nicking when try planing and finish planing. I was off base. Small amounts of alloying elements used correctly can make things easier for industrial heat treat, but they also can yield better results. We are generally talking fractions of a percent, and not like the chromium additions in huge amounts to A2, D2 or V11 (which is probably CTS-XHP) steels. If you like those steels, by all means, carry on. I don’t like them as a woodworker so I’m not inclined to use them as a toolmaker. They don’t offer anything for an experienced woodworker working with hand tools as the best they can offer is a fair exchange for sharpening and grinding efforts. If you’re a beginning, this isn’t going to make sense, but if you’re not, I’m referencing the fact that at some point, you will become so fast and accurate with sharpening that everything is proportional to time until wear resistant parts of steels don’t protect them from nicking. As soon as that’s factored in, things tilt back toward the plain steels. There are few users that I’m aware of that can take advantage of this. I know some, and I’m one of them. Most experienced woodworkers who use a lot of hand tools end up in this group – looking for fineness, edge stability and a steel that agrees well with stones.

Once you find yourself in this situation, you can ignore the folks who want to talk about bessemer converters or powder metals, or whatever else. Those are all interesting topics, but they seem to lead to the idea that something different is always better, or comments about grain size. I encountered one last week made by a self appointed expert stating that V11 (XHP) is as fine- grained as O1 steel – confidently stated. Except it’s false. V11 has a large volume of chromium carbides – exactly where it gets its wear resistant. For the volume, they’re well distributed and round. However, the micrograph shows that the carbides coarsen the overall particle size by a lot – perhaps 5 times more coarse than O1 – or more. V11 is more like taking A2, adding more carbon and a bunch of chromium and then distributing the carbides evenly. The largest carbides in both are about the same length.

Pointing out things like this (at length, it’s my style -not my chosen style, just my style) is usually not appreciated and those offended by being corrected when they’re providing not just bad opinions, but constant true factual falsehoods – those folks generally don’t like being corrected and probably don’t care if they’re correct in the first place.

Sifting through this – to the knives – phew. What’s wrong with a biting sharp knife out of 1095 that’s a little brittle? I don’t think I mentioned it above – real 1095 at high hardness would work fine as long as you don’t bend it or twist it in a pocket knife. Guess what people do. I think a huge part of the market of knives, nearly all of it, is made with the benders and icebreakers in mind. Why? Imagine you don’t know anything about knives. You open a metal container or split a piece of wood by prying and the tip breaks off. And maybe it’s part of a set. What do you do? Even if it’s years later, apparently the answer is that you complain to the manufacturer and demand a replacement.

What if you got a knife that was boring, soft and had a fat dull feeling profile. 90% of the market or greater doesn’t care. So, we all end up with knives that are less easy to break even if they aren’t as good at slicing and cutting.

What appears to be the case with Kabar and others who are known for making easy to sharpen, but a little soft, 1095 knives is that the name is used as a branding term for plain carbon steel, but the actual steel is a 1% steel that has some additives – especially chromium or chromium and vanadium.

Cure horror movie music. That’s a term – chrome vanadium – that woodworkers hate. It’s also often described as a soft steel that’s not very good because many tools made with it are proudly stamped chrome vanadium and they contain a smaller amount of carbon than we’d like because limiting carbon prevents dealing with reduced toughness that can occur if excess carbon dissolves into a steel lattice. But, there is a whole class of chrome vanadium steels where the chromium and vanadium are small amounts and the resulting steel is much more plain than even O1. Carbon can go well above 1% and these steels can be used in straight razors. However, the CrV variants that are available inexpensively in europe don’t seem to be marketed here and importing them thanks to shipping costs doesn’t make any sense.

What Does this Have to do with this Blog?

The answer is simple. Whatever alloy stanley used for the block plane iron, I’d like to find it. It looks like plain steel with a little bit of tungsten in it based on less than ideal carbide dispersion.

Whatever alloy Kabar used, I’d like to find it. I can see older spec sheets. 1% and 1.1% steels with small additions of chromium in some cases and “B” versions with a small amount of vanadium added. Very small, but enough to keep grain size fine in less than perfect heat treatment. Not enough to make a bunch of hard to sharpen vanadium carbides that are more suited to turning tools.

These steels, in this case several made by Sharon Steel, I was finally able to find. But only one alloy in one size. The rest, I’ll have to keep looking. The bottom line with these things – the 1095s in knives that aren’t 1095, is that many are gone and there isn’t a hot market for retail sale of steels that are water hardening and lower wear resistance.

So, I’m all the way up to finding one. 50-100 sharon steel – “1095” with 0.6% chromium added and maybe some other trace changes, but the chromium addition is the key. Will it be the 1095 that makes a great plane iron? It’ll have to be, because I can find it only in 3″ wide bars, 0.145″ thick and 3 feet long. it’s out of production, and the retailer selling it (USA knife maker) is just selling found old stock. Too thin for chisels, too thin for tapered plane irons, too thick for stanley irons. I can grind it into knives or make infill plane irons with it.

Unfortunately, other than one or two finds here and there, getting the feel of the steels in place 125 years ago, or pushing the hardness up on something that was well loved (like the Kabar knives), the opportunity is mostly lost to history.

This has gone long, but luckily as of this post, I have already made one infill iron from Sharon Steel 50-100. I’ll post about it separately as you may need a break. If you got to this without any pictures, ghee-whiz nonsense, or misdirection to get you to buy something that benefits me or generates a commission – congratulations. You’re probably in the minority.

Something I’ve forgotten about for quite some time is ever attempting to make a tapered bench plane iron freehand, or even in a jig. This may sound like something that should be reserved for machine tools, but I’m convinced that the golden age of laminated plane irons (mid to late 1800s in England) existed not based on precise jigging, but rather on a job grinder who was able to grind taper into irons on a wet wheel quickly and also grind some concavity into the backs of irons.

I realize also that most folks may have never thought about this concavity, but visualize the old plane irons as follows: The top of the iron that faces the cap iron will be flat. This is the side opposite of the bevel.

The back or bevel side of the iron beds against the plane bed. If you can help it, you want the iron to bed as close to the mouth as possible and again at the top of the bed with not much significant contact between the two. At least not on a double iron plane.

A well sprung cap iron will create some of that concavity on its own by bending the iron itself, but but this bias exists on older irons, anyway. I think the reason is because without it, the point of contact still could be a little bit further up the iron’s length vs. being right at the top of the bevel on the back. Getting these little details in your favor and not high centering or having uneven bed pressure makes a plane lock down tight, work well though minor seasonal movement over time, and adjust appropriately. As a maker, these biases are important – not only to ensure success, but to ensure excellent success.

But most tapered plane irons that I’ve seen new – other than japanese irons – are just two flat planes that aren’t parallel with each other – a sign that they’re either created in a jig (older irons) or just machined and ground (newer).

A Couple of the Doses of the Cant’s

About a decade ago, I wanted to try to harden a plane iron, but I knew as much as most people know. Almost nothing. A professional plane maker advised me not to bother as I’d be able to make a tolerable iron but never a good one. I’m past that.

Another professional plane maker gave me a different “can’t do it” around the same time when I asked if there was a way to taper plane irons for a hobbyist. The advice at that time was that it would take a minimum of $3,000 (more now) just to be able to begin to think about tapering.

I wasn’t thinking the same way – the precise machine way, and came up with a way to taper moulding plane irons well enough, but it was crude. Crude but it worked and was almost free.

My point isn’t to celebrate that I got past the can’ts, but more to mention that doing it – whatever *it* is – is often solving a problem and maybe you can’t solve it right away. I could taper moulding plane irons pretty easily. I tapered a thick plane for an infill shooter years later with a cheap ryobi belt sander idler, but it was a slow process – 2 hours of dusty filthy burnt hands with less control than I’d really like. These things have been incremental, satisfying curiosity, the involvement of handing the process – by hand – is bliss, rather than the thrill of solving a problem, and the the rigid boredom of repetition thereafter.

For reasons I don’t have a clue about, I bought something completely unrelated and got the wild idea last week that now that I have an 8″ contact wheel belt grinder, I could probably make taper plane irons. So I’ve made two. The first one was a bit slow, but by iron #2, the grinding adds about half an hour of time to making a larger plane iron. In this case, the entire process from start to finish is about an hour before a quenched iron goes into the freezer.

What was the unrelated item? 1095 Cro-Van. 1095 itself is a good spring steel. As a knife steel or chisel steel, or even a plane iron steel, it’s not that great. The steel itself triggered this idea – what to make – because I could only find it in a thickness that would probably be suitable for an infill plane. And somehow, that got me thinking about tapering it (too thin) and over to the rack for O-1 and off to the races.

No Great Reason – Sometimes Just for Making

I don’t really have a need to make these irons. My second double iron plane was a cocobolo smoother that I like, but I really like a stanley 4. That smoother has an I&H sorby iron that is a touch soft – unusual for that brand. Irons that are a bit soft and made of water hardening steel don’t last very long smoothing wood, but set the shaving a step or two thicker and they work wonderfully. About like a try plane – for people who measure shavings, somewhere from half a hundredth to a hundredth thick shaving in cherry and a little less if the wood gets harder. If you’re working entirely by hand, especially on something like sticking for mouldings where you also want a pretty good finish, that’s very useful.

It occurred to me that I could make a copy of the iron, but a smarter move would’ve just been to run the original iron through a quick set of thermal cycling and then reharden it. I made the irons instead. In the back of my mind is also that I’ve always made wooden planes with irons that someone else made. There’s no compromise to the design of a good iron and cap iron that I’m willing to go to, so making something like a single iron plane with a flat iron in a wooden plane isn’t going to happen – why bother when you’re working at my level – the level of an experienced dolt, not just an inexperienced dolt.

To apply the taper, I used a wooden block, a pair of vise grips, a belt sander with a platen and the contact wheel on a belt grinder. It worked great. That doesn’t explain the process well, but it’s literally what I used.

The first attempt of anything for me is always something I’m wiling to fail at and learn from, but the first iron turned out fine. It’s not perfect, it’s fine. This line of thinking – knowing what you want to do but then going and doing it and counting problems as guiding you to the right solution is important – it dominates sitting around and pondering hypothetical problems and successes. Knowing exactly what you want to make is important, but knowing exactly how you’re going to make it isn’t. If you’re creative, you figure that out. The second iron took half as long to make and I made it more accurately because of what I learned from the first.

I don’t really have much of a point other than that I’ve crossed something off of my list. The idea of working freehand often and being creative with a belt sander platen, the idler and the contact wheel on a belt grinder is why this was successful, and I’d put the idea of making a large tapered iron on the back burner long ago – if it takes 3 hours to make when you can find good Mathieson and other irons online, there’s no real point.

It’s also a reminder that if you don’t allow yourself to get boxed into things other people have had success with, the little things you learn to do by feel will open the door to do some things you may not have intended. Once you get to the point that you know exactly what you want to make, start making it. If some part is hard or fiddly, figure out another way to do it. I can almost guarantee that you’ll be able to make something twice before you’d ever get through with one paralysis analysis version, and it’s pleasant if you don’t take yourself so seriously that you feel like you’re too good to fail or that you can’t throw away something you make.

Lastly, understanding all of these little aspects of what makes a good cap iron or a good plane iron, including this curvature on the back of an iron, has resulted in being able to stuff both of the new plane irons into their subject planes with no adjustment to the wedges – they fit and work ideally right away. That wasn’t by accident. I liked the proportions of the original irons and have the ability to match the taper. The biases that an intelligent maker came up with in the past carried the rest of the load.

I don’t have a way to video this process, and also don’t have a wiling camera to help. Being realistic, I don’t think too many people would want to try it, so other the bits and pieces mentioned here, I think I’ll defer and not go into detail about the method – it’s less important than celebrating something that would’ve seemed too difficult 10 years ago but now is fairly low effort. And celebrating not falling into the trap of following the can’ts or “won’t be any good if you make it by hand”.

A Few Pictures

I’ve got a fair number of wooden planes, but not an enormous number by any means. Maybe a dozen bench planes and smoothers. I do have infill planes that will take the same size irons, so some of this is done with those – both used in this case are norris A13s – one beater of a smoother and one long panel plane that I forgot to photo.

There may be some disagreement, but the reality is that the spring in a double iron plane makes really small adjustments in a smoother less easy. It also makes the plane a real treat with a shaving a little thicker – like a truck with heavy suspension. I use the infills to take thinner shavings and test the surface quality the iron makes, and after getting the edge fully prepared, ensure that a the iron can plane a hundred or a couple of hundred feet with a fine edge and not develop any nicking. So, these pictures are generally at random and the ones that show no planes at all are the edges of the boards being planed – and the reflected image on them.

I haven’t yet made a cap iron or a cap iron screw, but I think I’ll live long enough to do that.

Two items of note before going into detail about tempering: 1) you can temper anything pretty easily if the tempering temperature is below 450F. That means that if you have an overhard Japanese tool or anything else that’s just junk to you, it’s likely that it could be improved tempering. 2) I am not a knifemaker, and i don’t use high speed or very high carbide content steels that have a very high temper or two tempering ranges.

If you’re going to temper complex steels, you’ll need to get something designed to temper high speed steels.

Otherwise, the focus here is tempering recently quenched tools, but for most things used in woodworking hand tools, a good quality oven or toaster oven will be all that you need.

What is tempering?

Quenching relatively simple steels is a matter of converting nonmagnetic microstructures of steel – austenite – to martensite, and doing so as completely as possible. You can read knifesteelnerds.com if you want to know more about those states, but for our purposes, we assume a quenched tool starting with brittle martensite and peak hardness. For almost all uses other than a blunt scraper, you’ll find untempered steel doesn’t hold up to anything – not even sharpening.

Tempering is a process of heating the untempered microstructure to the same microstructure with somewhat less hardness and more toughness. For hand tools, I’ve found with O1, XHP (similar to or same as V11), 1095, 26c3, etc. to prefer 375F to 425F. As a rule of thumb, you’ll trade about a point of hardness for each 25 degrees. that means big temperature variations are unacceptable and we’d really like to achieve something more like a 10 degree deviation at most.

How much tempering, when can it be done?

A single temper with really simple steels will take care of 90% of the result and leave you with something usable to very good. Some retained austenite can be converted to martensite during the first temper, so official schedules generally recommend tempering twice.

Your job if tempering steel is to get steel through its thickness to be an even tempering temperature. Whether you do that in hot oil, in a kitchen oven, whatever it may be doesn’t matter. However, other than very simple small applications, I’d recommend staying away from tempering over a flame.

Most schedules suggest tempering right away. If you arrived here after reading from the quenching page that it’s nice to park steel in a freezer to really chase as high of initial hardness as possible – at least on a budget where liquid nitrogen is out of the question – then that really means it doesn’t matter that much. Tempering sooner will reduce warpage, but good technique quenching should limit that and we’re chasing hardness above perfection in warp avoidance. Long story short, temper sometime before you use a tool

I’d recommend tempering twice. And if you make something you really like, have a stable setup.

Tight Budget Equipment

The range that we’re discussing here – 375F-425F is well covered by a toaster oven. Toaster ovens probably vary a lot in quality, but I bought the cheapest toaster oven I could find on eBay – something like $24 shipped. the toaster oven cycles on and off and surface temperatures below the elements deviate wildly, so an easy way to mitigate that is to use the trays that come with the oven and block direct heat from whatever you’re tempering, and put the tool you’re tempering in a sandwich of metal to increase mass, have something else take the direct heat from the heating elements and have the overall tool temperature variation be smaller.

400F tempering generally results in a color change on the surface of bright steel to about light straw color. if you see purples and blues, the edges of your tools are getting too hot, and the remedy is you start over and re-quench. However, a little blue on the bevel of an iron may be misleading if the entire subject item isn’t blue – try the tool first.

You’ll also want to invest in one of the cheap hanging thermometers (analog) that are designed for ovens. If you use a kitchen oven instead of a toaster oven, you’ll find in some cases that your oven is not close to what it says the temperature is, and that while the oven cycles, the temperature varies like a sine wave pattern. In all cases in my experience, even convection ovens have drastically different temperatures in different spots. So, use the same spot in the same oven with the same setup. Consistency is key. Right next to the glass on a toaster oven is also, of course, not going to be as hot as the center of the toaster oven.

If you have a non-contact thermometer, you can also put something non-reflective in the oven with the piece you’re tempering and get a reading. More sources of information are better. In my case, as long as there is some shielding to the steel being tempered, I’ve found the indicated temperature on the toaster oven to be within about 10 degrees of what it says. It’s easier to get the same temper than it is with a larger oven.

Exactly what I use and Do

I use the aforementioned inexpensive toaster oven. This rarely heats lunch – it’s in my shop and quench oil makes it smell like a smoky candle. If you can only temper in a regular oven, O1 steel and vegetable oil is a good idea or your house will stink like petroleum or smoking candles.

You can see in this picture that there is a lower tray. This toaster oven came with a second flat tray that can go under or above, and I will sometimes put the steel between both trays. It takes a very long time for the “metal sandwich” to get to temperature doing that, though, so use of the top tray to shield away from heat is usually a only if the metal sandwich outsides are thinner.

In the picture are two heavy aluminum plates, and thinner may mean unwanted older plane irons, or 1/4″ mild steel bar stock.

I temper twice, an hour per cycle. The steel needs to cool between tempering cycles, but not long, so I pull it from the sandwich, and put it on the floor or the bench for a minute or two and then cool in water and then start the second cycle. There’s no need for a big rest between cycles. That term, an hour, means an hour after the thermometer shown reaches a stable temperature.

For ease, if quenching isn’t done right before bed, I will throw the aluminum plates in the toaster oven as soon as I drop the steel off after quench in the freezer, and then after an hour or so, start the tempering process.

For simple steels, that’s it. needling away at ensuring temperature stability and accurate measuring is important, and the toaster oven, plus the thermometer, plus a third check if available is nice.

There are other alternatives for temperature stability, too. If you have a small metal can and can put chisels, knives or irons within the can and not touching the sides, that’ll work well. You can also use a container with sand in it – I use the metal sandwich because it’s available and works. Grade yourself on outcome, not style. If you quench and temper accurately, even with a cheap setup, you’ll see almost no hardness variation with oil hardening steels and variation of less than 1 at the extremes with water hardening steel.

One last point – the Japanese tools

I find 375-425F to be a great range for everything listed above, and I’m partial to the sweetness/hardness combination of 400F. However, Japanese tools that specify 65/66 hardness are at the lower end of hitachi’s (white/blue steel) range. You’ll see comments that this or that blacksmith is tempering at 100 Celsius, or 212F. There’s nothing to gain tempering tiny amounts, but it does provide bragging rights for people with no nerve endings.

If you find yourself with an older japanese tool that’s extremely extremely hard, it may be 25 or 50 degrees of temper away from being a real treat. You’ll need to remove any wooden parts, abide by the accuracy discussion mentioned above and start tempering at 325F and then increase by 10-25 degrees depending on your patience until you find something you like.

White steel and steels like 26c3 will be 2 points or more harder than O1 or A2 at the same temper, so sweetness in the result can come with high hardness. 26c3 can hit 64 hardness after a hand quench and temper at 400F. The samples that I sent for testing averaged 63.8, and O1 averaged closer to 61.5 on the c scale. 26c3 is similar to japanese white 1B. Had I tempered it around 325F, it would’ve been less tough than I like, but would’ve hit 66 hardness or so.

Unless you’re working pallet loads of paulownia, you’ll probably find 325-350F will get you a nice japanese chisel from one that started overhard, and you may still find yourself with a very sweet 64F chisel at a 375F temper.

Higher and Higher Temper Can Yield Subpar Results

Larrin Thomas (knifesteelnerds.com) writes about tempering embrittlement or something similarly named. Apologies for poor handling of proper nouns if I mangle names or proper noun, but what this means is more important than exactly what it’s called.

The useful fact in this case is that simple steels often reach a point where adding temper leaves them both softer and less tough. Or, put differently, the result is worse no matter how you look at it as the tradeoff to get better toughness by reducing hardness is lost. We don’t want to delve into this. From what I gather, this range for simple steels can vary, but is above 425F. If you want a softer steel than O1 at 425F (about 60/61 hardness), you’re better off switching to something with less carbon and tempering it in its sweet spot than you are trying to temper a harder steel softer and softer.

The basic hardening process with a furnace involves three steps. I do not do heavy forging, so if you do, I think you will have difficulty getting the same results as these as it’s no easy task to heat steel to forging heat and then refine the grain afterward.

The Three Steps – All Before Tempering

1. Shape, heat, anneal steel. There’s no reason to get complicated here. I shape steel during this. Heating is hot enough to hammer a general taper to the steel to reduce the amount of grinding that I need to do. I do this three times. Hotter than critical temperature by a full shade and let the steel cool while hammering to black. Overheating here may cause headaches later
   • If you’re dealing with good stock, nothing is gained here if you’re not shaping.
   • Too, you could do a whole bunch of things here, like a single heat and anneal in vermiculite. I don’t do this, but it’s popular and a google search will help you find out more about it.
2. Thermal cycle steel prior to quenching. In this case, heat steel until just prior to critical. Realistically, it is probably changing phase but not completely, and then partial quench in oil until the steel is magnetic again. This is quick and there is no holding at temperature – the quenches should be done before the steel becomes completely magnetic. Let the steel cool to air cool or close to it after the third iteration.
3. Quickly heat steel to nonmagnetic evenly and as fast as you can heat it. Allow the steel to get to a full color past nonmagnetic and then quench in appropriate oil until fully cool. Or…
   • Finish the top part of the quench in quench oil, and then once the item being hardened is mostly cool, switch to ice water
   • Same as the prior bullet, but just room temperature water and then toss whatever is being hardened in the freezer

A couple of rules:

1. The steel is never kept at high temperature for long. As little as possible. Heat changes the structure of the steel. If steel needs to be at a high temperature for duration to normalize, this process probably will not work well.
2. Quench oil needs to be able to transition the first part of the cooling fast enough. The faster transition occurs, the harder and the deeper the hardness of the steel will be. Steels that are made for cooling in oil are more tolerant of slower oils – like O1 steel. If you’re using 1095 or 26c3 or some other water hardening steel, the results will be subpar in a slower oil. The quench oil I use for everything is Parks 50. It’s not cheap and the smoke point is low, but the hardness results are excellent and consistent.
3. The first part of the quench needs to be fast. The second part doesn’t need to be as fast, but the lower the terminal temperature, the better the result. If you are finishing steel in ice water or the freezer, it needs to get there very quickly or you’ll get no gain. Don’t put hot steel in the freezer – the hot part of whatever you’re working with will be working against your effort to cool the steel in the freezer. Continuous immediate cooling down to the lowest temperature possible yields higher hardness.
4. Chasing higher hardness is also correlated with more warping. If you’re using water hardening steels, you’ll be dealing with warp. You can introduce plates or chilled plates to the middle of this process to reduce warping, and mind bevels on things like plane irons. It’s fine to pre-cut cutting bevels, but don’t go overboard or they will help induce far more warping than you want.

The next part of this is a burden on you. When you are learning to heat treat steel, test it. How? You will end up with small cutoffs or scrap, and when you do, those bits of steel are ideal for testing process. Take something well known, like a quality file, snap it, and take a picture of the grain. Files have excess carbide, so they make a good starting target. The carbides make the grain appear a little bit coarse. If you’re using O1 or 1095 or something with less carbon, you should be able to snap your own samples after they’re cool – and before tempering – and look at the grain with a hand held microscope and see that you can create fineness superior to the file steel.

Take digital magnified pictures of your snapped samples and keep them with process notes. Snapping the steel means just that. Before you temper steel, you can hit it with a hammer or put it in a vise and break it. Files will also break easily. The only steel I’ve found to have toughness right out of the quench that can be unnerving is 52100, but 52100 has surplus chromium carbides and I don’t think it’s as good for woodworking as O1, so I don’t use it. It benefits greatly from a furnace normalizing cycle and I’m trying to avoid a furnace. It’s slower and with 1095 and O1, I can match published schedules for hardness and toughness in combination. With 26c3, I can better published data, and only with 1084 so far in this type of steel have I created a dud in tested samples. I never bothered to follow up because I don’t think 1084 has any advantage over O1.

What you’re looking to do with your making of samples is train your eye at the same time. Training yourself to see what red looks like before steel becomes nonmagnetic. Training yourself to see the color at nonmagnetic transition so that you can heat steel a shade further before quench. Getting repetition.

You need your eyes for this. Little checks and tricks and tempilstiks or whatever are too slow, this is all quick.

If someone tells you this kind of experimenting is a waste of time, then you don’t really need to hear from them. To cycle small samples takes only minutes – it’s an ideal thing to keep your brain working and thinking and analyzing when you may not have the time to do something more substantive in the shop.

If all of this seems like too much, just buy good O1 steel (starrett, precision, whatever it may be), heat it quickly to a shade past nonmagnetic and quench it. You’ll never do anything in woodworking where the results just from that will be lacking.

Tempering is next.