Plane iron clearance angle study

[Patrick Chase]

Hi James,

My son, who is a data scientist, would tell me that I do not have enough data!

No amount of data can answer an unanswerable question. There is no ideal clearance angle, merely subjective tradeoffs.

Many people here argue about the results of this study. May I suggest that you, or anybody interested, sharpen a blade at 28° and then test it. Then resharpen at 35° - or any other angle that you would be interested - and test it again. Doesn't take long - maybe faster than writing here.

Derek has told you that he tried that and prefers 35 in at least some circumstances.
David has told you that he tried that and prefers 35 (same caveat).
I've told you that I've tried that and prefer 35 (same caveat).

Does that mean that 35 is "the ideal"? Absolutely not. But it should be a huge clue that individual working style and preference are factors here, and that there consequently is no ideal, period.

For that matter I told you in the other thread that I'd switched my 5 y/o son's plane to 30 deg because in his case a pound or two (probably less) of cutting force is a big enough deal to be worth sacrificing other attributes.

I just found this website http://homepages.sover.net/~nichael/...ters/caop.html where the clearance angle role is very well explained

"The clearance angle is the angle between the bevel on the back of the blade and the wood surface. It is the angle by which the blade clears the freshly cut surface (Fig.A-l:2). This clearance is required, for the following reason. The thrust as the blade moves forward distorts the wood a small amount, until the wood yields to the cutting action. Part of this distortion is a downward compression. As the blade moves on, the freshly cut wood springs back and would tend to lift the plane blade if it had no clearance. Greater downward pressure would be required for a smooth cut, and friction would increase."

Sure, but nobody here is proposing a blade that "[has] no clearance" as posited by that quotation. All it says is that you need *some* amount of clearance.

Even if lower clearance does slightly increase cutting forces (as I've acknowledged repeatedly in this thread), why do you think that should trump all other concerns (edge life, surface quality, etc)? Like everything in life it's a tradeoff, and obsessively latching onto a single metric is problematic.

EDIT: Softened last para

[James Waldron]

Hi James,

My son, who is a data scientist, would tell me that I do not have enough data!

Many people here argue about the results of this study. May I suggest that you, or anybody interested, sharpen a blade at 28° and then test it. Then resharpen at 35° - or any other angle that you would be interested - and test it again. Doesn't take long - maybe faster than writing here.

I just found this website http://homepages.sover.net/~nichael/...ters/caop.html where the clearance angle role is very well explained

"The clearance angle is the angle between the bevel on the back of the blade and the wood surface. It is the angle by which the blade clears the freshly cut surface (Fig.A-l:2). This clearance is required, for the following reason. The thrust as the blade moves forward distorts the wood a small amount, until the wood yields to the cutting action. Part of this distortion is a downward compression. As the blade moves on, the freshly cut wood springs back and would tend to lift the plane blade if it had no clearance. Greater downward pressure would be required for a smooth cut, and friction would increase."

I fully agree with your son. That is indeed one central point of my prior post.

I'm not proposing to do any testing. I've far too much else to demand my time to launch such a Promethean task as my prior post indicates. I wrote then and again now to suggest something far different: you work is not in any way false or even questioned, by any one here and certainly not by me; the effort is worthwhile, as it adds data we have not had before. What I have written to say, apparently unclearly and inadequately, is that your data are not sufficient to draw conclusions about optima, by you or by anyone else, because the feature you have explored is one of many variables in a complex system; you need more data and while you may choose to hold your conclusions dear, while the logic of your thought may seem compelling, and your defenses of your rationale may be eloquent, so too are those of voices raised in opposition to yours. None can be certainly correct, as all are drawing inferences that are inadequately supported by the limited data.

It is always good to remember that a thing may be logical and yet not be true.

But be of good cheer. Such a debate is fun, it may spark further experimentation and additional data for our contemplation, and no furry animals are harmed in the engagement. Unless someone takes it all too far. Taking it too far is the point of my prior post: a courteous and collegial debate is what is needed here and not an exchange of charges and counter-charges. Indeed, my prior post was address, in my intent, more to the somewhat adversarial tenor of some others on the thread, and I did not have you specifically in mind.

As to the language you quoted above, it falls short of validating your hypothesis of the optimum angle. It speaks only to your assertion of a specific physical basis for - in several steps - producing some supposed support for your inferences, although with no supporting data and a citation of an authority with unknown and uncertain provenance. Remember, you found it on the internet! It is, at best and even if totally true and correct, a "red herring."

In short, you and everyone else should take a breath. You're all beginning to "violate my safe space."

[Patrick Chase]

OK, I'm backing off, regardless of who replies.

I agree with James that Normand's work is valuable and that it established something that's useful to know (how to optimize clearance for longest cutting life at some unknown fixed downforce).

My objection was to the conclusion that it had demonstrated that some clearance angle "is the best" (exact words from #1).

[James Waldron]

OK, I'm backing off, regardless of who replies.

I agree with James that Normand's work is valuable and that it established something that's useful to know (how to optimize clearance for longest cutting life at some unknown fixed downforce).

My objection was to the conclusion that it had demonstrated that some clearance angle "is the best" (exact words from #1).

Whoa! Don't be backing off because of me! This latest post has a very nice tone and puts your point clearly and without question, is a reasonable and reasoned position. I don't want to stop the debate, just get it back to a less contentious level.

For example, it would interesting to consider your point made one more post back that: "No amount of data can answer an unanswerable question. There is no ideal clearance angle, merely subjective tradeoffs."

In one sense, you're right because even when the system is completely quantified with all the necessary data, there will be up-sides and down-sides to each peak in performance, and those are likely to differ. The choice of which properties to optimize and which to tolerate at less than best will ultimately introduce a subjective (or at least seemingly so) aspect to out several choices among several peaks of performance.

In another sense, you're wrong, in that any physical system that is fully explored and quantified for all the variables can be defined by the calculus to locate the peaks of performance (only rarely would there be a single peak of performance for all purposes). Then, in one way of looking at the possible choices, the seeming subjective choice of a particular peak would most often be dictated by the result we require of the system and is only subjective in that we thought we had a choice to make, but considering what we wish to achieve, there was a particular peak that best served our goals and dictated our choice after all.

On the other hand, planing wood with a hand plane is far from such quantification; there are so many variables and so few opportunities to test appropriately that we may never get 'er done. So for now and for the foreseeable future, we have broad guidelines we can follow based on empirical experience. And we can build fine furniture with that, so it may not matter a great deal to the craft that there is more to learn here. And we will learn more as the curious look again into the variables and generate additional, new results to debate and digest. Good on us all.

[Patrick Chase]

In one sense, you're right because even when the system is completely quantified with all the necessary data, there will be up-sides and down-sides to each peak in performance, and those are likely to differ. The choice of which properties to optimize and which to tolerate at less than best will ultimately introduce a subjective (or at least seemingly so) aspect to out several choices among several peaks of performance.

In another sense, you're wrong, in that any physical system that is fully explored and quantified for all the variables can be defined by the calculus to locate the peaks of performance (only rarely would there be a single peak of performance for all purposes).

There won't even be a single peak for the same purpose for different users, the reason being that there is no single definition of "performance".

Normand seems to care a lot about cutting force. I don't care at all as long as it's within, say, 20% of the minimum for the cutting angle. We will consequently never agree on a single "optimal" angle, even for the exact same application (wood, grain orientation, cut depth, plane, etc etc etc). As noted in a previous post my 5 y/o son's current optimum is closer to Normand's than to mine though.

Then, in one way of looking at the possible choices, the seeming subjective choice of a particular peak would most often be dictated by the result we require of the system and is only subjective in that we thought we had a choice to make, but considering what we wish to achieve, there was a particular peak that best served our goals and dictated our choice after all.

True, in theory with enough data you could quantify the tradeoffs to the point where each person could derive their individual optimum from a model. As you point out, planing is so complex as to make that potentially impracticable. In addition, that assumes that everybody accurately understands their own preferences, and in my experience that's seldom the case. I think that most people (me included) iterate their way to a preferred configuration without fully understanding why they got there.

[Matthew Springer]

One thing I've really enjoyed about this thread was thinking about the clearance angle at all. I was sharpening my BD bench planes at 40 deg (moar is better and 45 was to steep to fit in the jig). I also beleived myself brilliant for having "discovered" this.

Because I naively assumed that would be best for edge retention. Makes intuitive sense since that part of the blade isn't in contact with the wood. Right?

Except it's totally wrong. Yet another human endeavor where the simple, intuitive answer turns out to be too simple. It turns our reasoning by personal incredulity has issues; who knew?

[Patrick Chase]

One thing I've really enjoyed about this thread was thinking about the clearance angle at all. I was sharpening my BD bench planes at 40 deg and 45 was to steep to fit in the jig.

Because I naively assumed that would be best for retention. Makes intuitive sense since that part of the blade isn't in contact with the wood. Right?

Except it's totally wrong. Yet another human endeavor where the simple, intuitive answer turns out to be too simple.

Out of curiosity were you seeing serious cutting issues at 5 deg?

I'm asking because there seems to be wide diversity of opinion about where Really Bad Things (tm) start to happen. I've seen severe cutting issues at 5 deg on softer woods but not harder ones, while Sellars claims that a blade with 3 deg relief cuts "as well as" one with 15 (which I find hard to believe).

[Matthew Springer]

Out of curiosity were you seeing serious cutting issues at 5 deg?

Yes but I assumed it was me not getting things sharp enough or a host of other factors. I probalby didnt sharpen steeper than 38 in practice with micro bevels. The single biggest issue is that the plane is just really hard to push. (side note:Which also wouldn't make any sense unless the wood actually was in contact with the bevel side). It feels like you're taking much too thick a shaving even though they're pretty thin and they get really curly, so you back off and start to get more "dusty" shavings almost like a card scraper but backwards.

The problems were most manifest in the jointer plane (7) and jack (5-1/2) when I skewed the plane. I have a york pitch 4-1/2 so that was fine. I'd get really really bad tearout when I skewed, especially with the jointer. I was also experimenting with the chip breaker trick on some curly maple for a passthrough arch in the house so I thought that also might have something to do with it.

[Pat Barry]

A very interesting and stimulating debate indeed.

On the other hand, it is worthy of note that the debate is about the conclusions to be drawn from the data and not about the data itself. I think that distinction is of some import. Viewed as a set of data, measuring what is (at least close to) a single variable test, it's meaning is contradictory to other testing that other folks have done (at least as I understand the history here and elsewhere).

As with any testing regime with multiple variables, only a relatively large number of tests varying multiple permutations of variables can provide an adequate dataset for the application of multivariate calculus to derive any truly solidly grounded results. Multivariable systems often produce unexpected conclusions when proper mathematical analysis is applied. In advance of such an undertaking, it is not possible to know with any fundamental certainty what hypotheses about the exiting data are correct or even reliable across a useful spectrum of usage.

In that context, "follow historic precedent" is one decent stop gap choice, "it works for me" is another, and so on. Interesting as these debates may be, no one should "bet the farm" on a single proposition hypothesized from these data or in light of others. The most we can realistically say is that exemplars that demonstrate failure may help us discern the boundaries of effectiveness. They do not lead us to optima.

I suggest that we need not "... rage, rage against the light ...." in debating this isolated dataset.

This is too deep for me. Because of that I don't know if you have a point or are just stringing a line of nonsense.

[george wilson]

Is any of this kind of stuff going to make anyone a more SKILLED woodworker????? NO,NO,NAH.

[James Waldron]

This is too deep for me. Because of that I don't know if you have a point or are just stringing a line of nonsense.

Multivarient calculus is difficult for anyone and it's awfully hard for anyone to foresee what results it may produce. And since we don't have the data, it doesn't much matter. I was just trying to get everyone to realize that there were far too many angels dancing on the head of this pin and everyone needed to take a deep breath and stop arguing past one another. There were instances of folks not taking the trouble to assess and understand what others were saying.

I personally think this is a rich field for learning if we can swap our factual experiences (data) and not get too hung up on extrapolations and inferences from the data (opinions) which we can neither prove nor disprove beyond "it works for me" or "I tried it an I couldn't get it to work."

Maybe you could PM Patrick to ask his view privately. It seems he gets it. It seems he doesn't fully agree with my position about picking optimum setup, but he isn't too far from where I am. I'm sorry if you find it hard to deal with. Not everybody needs this. If it's a bit over the top, I apologize.

[Stewie Simpson]

Is any of this kind of stuff going to make anyone a more SKILLED woodworker????? NO,NO,NAH.

George; have a read of #53..

Stewie;

[Patrick Chase]

Multivarient calculus is difficult for anyone and it's awfully hard for anyone to foresee what results it may produce. And since we don't have the data, it doesn't much matter. I was just trying to get everyone to realize that there were far too many angels dancing on the head of this pin and everyone needed to take a deep breath and stop arguing past one another. There were instances of folks not taking the trouble to assess and understand what others were saying.

I suspect Pat did the full science/engineering calc sequence in his undergrad.

If we're invoking multivariate calc (or worse still graduate-level mathematical physics - now THERE'S a good time) for something like this then it's a lost cause IMO. I understand where you're coming from, but I wouldn't attempt to apply that to myself much less advocate it for the "general population".

Stewie's basically right - the main value in a discussion like this is in learning what to avoid (5 deg clearance for example, which is ironically exactly where this all started in the previous thread) so that you're not banging your head into a wall like Matthew. Optimizing is of far less importance.

[Pat Barry]

Multivarient calculus is difficult for anyone and it's awfully hard for anyone to foresee what results it may produce. And since we don't have the data, it doesn't much matter. I was just trying to get everyone to realize that there were far too many angels dancing on the head of this pin and everyone needed to take a deep breath and stop arguing past one another. There were instances of folks not taking the trouble to assess and understand what others were saying.

I personally think this is a rich field for learning if we can swap our factual experiences (data) and not get too hung up on extrapolations and inferences from the data (opinions) which we can neither prove nor disprove beyond "it works for me" or "I tried it an I couldn't get it to work."

Maybe you could PM Patrick to ask his view privately. It seems he gets it. It seems he doesn't fully agree with my position about picking optimum setup, but he isn't too far from where I am. I'm sorry if you find it hard to deal with. Not everybody needs this. If it's a bit over the top, I apologize.

I had enough multiVARIABLE calculus in college back in the day to know that there is no such thing as multvarient calculus. Beyond that though, seeing as how you have a good handle on the calculus, please humor us with the variable and equation set we can all use to analyze this problem. Please don't talk down to us though. Show us your data please. I suspect all we will get is a lot of words strung together but no real substance.

[Patrick Chase]

I had enough multiVARIABLE calculus in college back in the day to know that there is no such thing as multvarient calculus.

It's also called multivariate calculus depending on professor and institution. Never multivariant though AFAIK :-).