Accuracy

[John Clifford]

Warning: rant ahead. I just watched another video by a well-known luthier who claims that every component of his guitars is machined to 0.001” accuracy. And this is a guy who doesn’t use CNC machines. With all due respect (and I do have great respect for this guy’s skills), this is just nonsense. It is simply impossible to achieve that level of consistent accuracy with wood, yet I see these claims being made frequently in the world of luthiery. For one thing, how would you even know whether you had achieved this feat? You can’t see it. Go look at your ruler with 0.01” marks. If you’re my age you probably have trouble seeing those. If each of those marks was divided by 10, do you really think you could tell the difference between 0.003” and 0.004”? OK, I know, you have a digital caliper that goes to 0.001”. I have an amp that goes to 11. These are just numbers displayed by a device. When I measure an ebony fretboard with my digital caliper, the display jumps around by at least 0.002” just depending on how much fingertip pressure I put on the caliper and how far I insert the wood. Move it to a different spot on the same freshly thickness-sanded fretboard, and there is even more variance. Second, even if you could measure it, you can’t machine it that accurately. The finest plane shaving I can get measures around 0.002” (or 0.001” or 0.003”, depending on above factors). Some of you may be able to make thinner ones. But the fact that you can make a shaving that might (or might not) be 0.001” thick doesn’t mean you can shape an entire fretboard - let alone an entire guitar - to that level of accuracy. I know some people will say yes, but I can FEEL the difference. Well, there’s no arguing with people’s feelings, but I would love to see this level of sensitivity demonstrated in a scientific double-blind test. Finally, as we should all know, a given piece of wood will change shape and thickness by more than 0.001” just about every day, all by itself. So what even would be the point of trying to achieve that level of machining accuracy?

OK, I asked for it. Now let me have it.

Humbly yours,
John Clifford

[Steve Sawyer]

No argument from me on this. Absolutely ridiculous. And this us from a guy who would be lost without my digital calipers, 1-2-3 blocks, setuo bars, feeler gauges and digital readouts on most of my power equipment.

Not only is it not achievable, it yields no benefit. Save that degree of accuracy for where it matters in machine setup and guitar setup. Striving for a high level of accuracy is beneficial. Insistence on achieving it is a fool's errand.

[Brian Evans]

I learned the difference between accuracy and precision in electronics school. On exams we were told the equipment that we were using in our virtual lab. and lost points if the answers were more precise than the instruments were capable of. If we where told we were using a slide rule (yep, that long ago) we had no more than three significant digits allowed. Anyway, most of my tools read out to .001", then I go ahead and mark with a pencil - .010 at best - scribe with a marking knife - .020" - and cut with a saw - .030". If I manage to middle out all of that, then I'm back to where I started, .001", but not by design, by luck.

[Andy Bounsall]

It is a completely ridiculous claim. Like a lot of marketing hype, however, it doesn’t necessarily have to be proven, just stated. To the uninformed it suggests that a high level of skill and precision is involved. To the informed, it just sounds foolish.

[Michael McBroom]

This is a claim I certainly wouldn't make, even though I use a milling machine with tolerances of 0.001" in my builds. There is one specific operation where I take 0.020" off the front edge of the fingerboard to help with intonation. I am glad to have my milling machine for this one step because very small increments can affect intonation.

When I first started building guitars, I cut fret slots by hand, measuring out their locations using a machinist's rule with1/100 graduations. The results were hit or miss. A couple guitars I obviously missed getting the frets placed right since it's impossible for them to be tuned accurately. Later I began to use templates, which removed inaccuracies. How accurate are the templates? I dunno, I've never tried measuring the resulting fret slots. But the guitars I've built using templates play in tune, at least. So, what is the accuracy of these templates? 0.001"? Probably not. But better than 0.010".

So I guess the way I feel about it is, 0.001" accuracy is something to strive for, but one shouldn't get too upset when it isn't realized. Wood, by its very nature, doesn't hold tight tolerances that well.

[John Clifford]

Michael, yes fretting is the one area where some plausible claim can be made that you should strive for better than .01" accuracy. I actually did the math on this, using a theoretical 650mm scale length and disregarding real-world string compensation issues. It turns out that to get a 1 cent difference in tuning on the high E string at the 12th fret, you would need to move that fret .188mm, or .0074 inches. In other words, if your fret placement is accurate to the nearest .01", your tuning should be accurate to roughly the nearest cent. Of course, this varies with different scales and fret locations.

I guess my whole reason for starting this thread is that it is plenty hard enough to make a really good guitar, and if we want to help each other be able to do that, we should focus on the things that really do matter and be honest about the things that really don't.

[Clay Schaeffer]

Extreme accuracy of fret spacing is not absolutely necessary for a number of reasons:
The average person can't hear differences of "1 cent"
String stiffness affects the pitch and overtones of the note - changing string gauges can change the pitch of notes as you go up the fretboard.
"Action" also affects pitch. Different string heights above the fretboard require different compensation
The equal tempered tuning system is inherently inaccurate and throws things off more than a half millimeter misplaced fret will.

The last number in the table shows the number of cents "off" from "just intonation" the equal tempered system is.
In the following table the sizes of various just intervals are compared against their equal-tempered counterparts, given as a ratio as well as cents.

Name Exact value in 12-TET Decimal value in 12-TET Cents Just intonation interval Cents in just intonation Difference
Unison (C) 2​0⁄12 = 1 1 0 ​1⁄1 = 1 0 0
Minor second (C♯/D♭) 2​1⁄12 = 12√2 1.059463 100 ​16⁄15 = 1.06666… 111.73 −11.73
Major second (D) 2​2⁄12 = 6√2 1.122462 200 ​9⁄8 = 1.125 203.91 −3.91
Minor third (D♯/E♭) 2​3⁄12 = 4√2 1.189207 300 ​6⁄5 = 1.2 315.64 −15.64
Major third (E) 2​4⁄12 = 3√2 1.259921 400 ​5⁄4 = 1.25 386.31 +13.69
Perfect fourth (F) 2​5⁄12 = 12√32 1.334840 500 ​4⁄3 = 1.33333… 498.04 +1.96
Tritone (F♯/G♭) 2​6⁄12 = √2 1.414214 600 ​7⁄5 = 1.4
​10⁄7 = 1.42857... 582.51
617.49 +17.49
−17.49
Perfect fifth (G) 2​7⁄12 = 12√128 1.498307 700 ​3⁄2 = 1.5 701.96 −1.96
Minor sixth (G♯/A♭) 2​8⁄12 = 3√4 1.587401 800 ​8⁄5 = 1.6 813.69 −13.69
Major sixth (A) 2​9⁄12 = 4√8 1.681793 900 ​5⁄3 = 1.66666… 884.36 +15.64
Minor seventh (A♯/B♭) 2​10⁄12 = 6√32 1.781797 1000 ​16⁄9 = 1.77777… 996.09 +3.91
Major seventh (B) 2​11⁄12 = 12√2048 1.887749 1100 ​15⁄8 = 1.875 1088.27 +11.73
Octave (C) 2​12⁄12 = 2 2 1200 ​2⁄1 = 2 1200.00 0

It is good practice to get your fret within a 1/4 millimeter of the calculated distance, but in many cases not absolutely necessary.

[John Clifford]

Yes, and an optimally setup guitar with perfect fret placement and saddle-only compensation will have up to 5 cents error from the equally tempered scale, according to Trevor Gore. Of course it's still important to get the fret placement as accurate as you can, to avoid compounding errors.

[David King]

I make all kinds of small wooden parts and I hold a tolerance of .001" all day long. I've never touched a CNC machine. I don't see a problem with this claim but I don't use woodworking machines either having a Bridgeport and a Hardinge at my disposal.
Granted there's not much point to holding those tolerances in larger wood parts since woods like maple and ebony can change dimensionally by several thousandths over night. I would hope that everyone here can keep their fret heights to within a thousandth or two at the most even if they aren't measuring it otherwise their instruments wouldn't play.

[John Clifford]

In fret leveling, you can realistically achieve that kind of tolerance (at least temporarily). But that's metalworking.

[David King]

I'd also point out that "wood" is a very broad generic term that covers everything from balsa and pawlonia to African blackwood and snakewood. I've heard that in remoter parts of Africa there are craftsman who will carve replacement engine parts, valve rockers and the like from African blackwood using a knife and a dial caliper. Apparently these parts can resist a hot oil bath and keep an engine running indefinitely or until the metal replacements can be procured.