Official Luthiers Forum!

Iʻm at the point of working on the intonation of my newly finished OM-14 style guitar. Iʻm mostly playing all frets on all strings and seeing how far theyʻre off - and finding strings and frets where the pitch is unstable and moves around a lot - generally discovering how much of a seat-of-the-pants task this is turning out to be <g>...

This particular guitar started as a StewMac OM kit, with a fingerboard pre-slotted for the frets. As I do my testing and charting, I see a general pattern: almost all frets on almost all strings are sharp, and the sharpness is all up and down the fingerboard. So hereʻs my first theoretical question:

Suppose I measure a guitar and every single fret on every string is +10 cents. Am I correct to assume that the frets are placed correctly and thereʻs simply a bit too much fingerboard between the nut and the first fret? Assume that my first fret is placed 35 mm from the nut - can I assume that since +10 cents is 1/10 of a semitone, I can simply cut off 3.5 mm from the nut end of the fingerboard and move the nut up? Will that drop the pitch on each fret and each string by 10 cents?

This is of course a theoretical question - Iʻm not planning to take the Sawzall to the fingerboard yet. But I want to know if that part of my thought process is correct; that shortening the fingerboard at the nut end will lower the pitch at each fret by the same amount.

The second "theoretical" question is more of an observation and request for your thoughts. Iʻm playing a lot of Hawaiian slack key music, and in particular, a lot of tunes that take the bottom string down to C. FYI, the guitar is strung with AʻAddario phosphor bronze light/medium strings - the medium bottom 2 strings feel a bit less like wet spaghetti on those low tunings, while the light tops are nice for smooth melodies with ease of sustain. But the bottom string on the new guitar goes very sharp - itʻs somewhat sharp by the 5th fret, and way up over +10 by the 7th fret - then itʻs sharp the rest of the way up. Weʻre talking mucho sharp - say 15-20 cents on all frets from 7th to the top, while frets 1-6 are only a few cents sharp.

Iʻm not sure what needs to happen in order to come closest to good intonation. And I observe that on my Larivee L-10, a guitar thatʻs generally extremely well in tune, I see the same thing on the bottom string - unless I tune the open string flat, my 5th fret and up are very noticeably sharp when I tune in those wide C tunings.

So what is to do? Iʻm guessing that Iʻll have to make a saddle with a wide top that goes back toward the tuning pins in order to have enough width to start correctig the bottom string. Any thoughts?

When you glued on the bridge did you add about .10" to the scale length at the saddle slot? Your observation of sharpness all the way up the neck would indicate not. When preparing to glue on the bridge I use Mottola's compensation chart for the numbers, https://www.liutaiomottola.com/formulae ... sation.htm.

Yes, I added some compensation length when I glued the bridge. I canʻt figure out how to post charts here, but the top 4 strings donʻt go wildly sharp from 1st fret to the upper frets. Hereʻs a very quick summary of what Iʻm seeing - Iʻm just measuring with the Peterson tuner on my iPad, and the guitar isnʻt at all set up yet - nut and saddle a bit high, neck close to flat but I havenʻt checked it carefully yet. Iʻm measuring with a slack key tuning thatʻs pretty similar to several I play in: DGdgbeʻ. Hereʻs what I see:

fret 1 fret 3 fret 5 fret 7 fret 10 fret 12
1st str. +6 +7 +7 +6 +7 +10
2nd str +4 ? +5 +5 +8 +8
3rd str +6 +2 +2 +2 +2 +2
4th str -3 -9 +20 +20 +8 +5
5th str +11 +10 +8 +8 ? 0
6th str +3 +3 +2 +2 +18 +18

** Sorry, I see that when I post this, the chart formatting disappears. I wish I knew how to just post an image of the chart, but I donʻt....

Some of these look pretty strange, and some of the measurements may be odd because I put on slightly used strings for setup and some of them arenʻt very true. But all strings except 4 are sharp at the first fret. 1 and 2 only gain 4 cents from 1st to 12th fret; the 3rd string actually drops. So it looks to me like Iʻll want to shorten the FB at the nut, mostly on the treble side. The 4th string isnʻt sharp at the 1st fret and gets sharper going up. The 6th string is a bit sharp at the nut and goes wildly sharp at the top frets (my Larivee L-10 also does this on the bottom string) - I assume that this is going to take moving the saddle back toward the tuning pins a lot, just for the 6th string.

Do understand - iʻm really just taking quick measurements and thinking things over - not making many changes yet. The nut is just a bit high now. Yesterday I back beveled the saddle along its whole length since all strings were sharp at the 12th fret - it made intonation generally better and the numbers above are after I did that.

You can wrap text with [ code ] [ /code ] tags (remove the spaces) to make it display with fixed-width font and extra spaces retained. But the editor window has a variable-width font, so it still won't look exactly the same. Best to write in Notepad or other text editor with a fixed-width font and copy-paste into the forum editor.


As for your question of shortening the nut-to-1st-fret distance... wars have been fought over what effect it will have, because it depends on your perspective. I think it will give the result you want, but it makes more sense to me if I ignore the open note and tune the string so all the fretted notes are correct. Then the nut position can be moved to adjust the open pitch without changing the string tension which affects all the others.

You should probably use a new set of strings when finetuning the compensation. 29 cents difference in error between frets 3 and 5 sounds more like a string problem than compensation problem. Or a fret slot cut in the wrong place, except that would affect the other strings too.

Dennis - please explain the end of that first paragraph in a bit more detail - Iʻm still trying to get my head around all of this. If I ignore the 4th string numbers that are so totally strange, my general thought was that if numbers are all positive, youʻd want to move the nut closer to the 1st fret so as to move all the numbers closer to zero - then if you have a string with numbers that climb more positive toward the upper frets, youʻd move the saddle back toward the bridge pins for that string, but if you have a string with numbers that drop, youʻd move the saddle forward.

If the saddle were in the perfect place but a string had the same plus number on all frets, youʻd certainly move the nut towards the first fret - but itʻs never that simple! For example, I assume that for both the 1st and 2nd strings the nut is too far from the 1st fret and the saddle is a bit too near the nut, since both strings go 4 cents more plus at the 12th fret. But the 3rd string is different - it looks also like the nut is too far from the 1st fret, but changing the nut in the same way you would for strings 1 and 2 would mean that the saddle would have come towards the nut a bit for the 3rd string.

This is all me playing mind games to try to understand whatʻs happening with the intonation - I do understand that correction is a bit of an inexact game that aims to get you so that almost everything is within a few cents or so. Iʻm sure that if I measure my Larivee, which is quite well in tune except for the 6th string, Iʻll bet that Iʻll find that the numbers run + or - 2 or 3. What Iʻm really doing is trying to figure out how to make the numbers get better and how close to zero is good enough <g>...

That is correct. The confusion comes from the unspoken assumption that you will also be reducing the string tension as you move the nut toward the 1st fret. If the tension is held constant, the open note goes up in pitch while the fretted notes remain unchanged. It's really the same thing, since they will all be in harmony once the open note is as sharp as the rest. And on a guitar you'll have to retune the strings anyway to take the nut in and out. But if you build a test rig with movable nut and saddle, you can move them without changing tension.

Of course you need to drop the tension a bit when you shorten the string length by moving the nut forward. You also have to raise the tension a bit when you shift the saddle back. Nobody questions that. That's a matter of physics, not 'perspective'. It sometimes happens that by the time you get done compensating both the nut and the saddle you end up with the same string length you started with: both have been moved 'south' by the same distance. The intonation would not be right if you hadn't moved them, though.

Nut compensation is fairly new, and it does take some getting used to, but the physics is really pretty basic, and it usually works out well.

One thing that can throw it out is motion of the top at the bridge location caused by a strong resonance near a played pitch. If, for example, the 'main top' ('loudspeaker') resonance comes in just below the pitch of the open G string it will shift that string pitch upward when the tension is 'correct'. You don't tune using a tension chart, of course; you stop tightening the string when the meter reads it as in tune. At that point the string is actually a little slacker than it 'should' be, and that shows up as you try the next few notes up from the open string: they're flat. This effect is more pronounced on lower notes, where the top moves more, but it can throw out single notes almost anywhere. I suspect we're seeing some of that on Peter's guitar.

In a case like Peter's I'd start by using something like pieces of toothpicks wedged in under the strings to shift the take-off points of the nut forward without, it's hoped, raising the action at the nut end appreciably. This should shift all the pitches on that string down (after re-tuning) by about the same number of cents. Saddle compensation mostly shifts the high positions, but those are not so far out here.

Moving the nut and moving the saddle have different effects. In the following, the error graph is a plot of frequency error (vertical) vs. fret position (horizontal).

Moving the nut offsets the frequency of each fretted note by a constant amount relative to the open string. In other words, moving the nut will raise the error graph up and down, no change in tilt.

Moving the saddle offsets the frequency of each fretted note by a different amount relative to the open string. The change in frequency increases as you go up the fretboard. In other words, moving the saddle will tilt the error graph, no net change vertically.

This comes from Trevor Gore's book. You can use my spreadsheets to calculate nut and saddle compensation to minimize this error.

Greg

Thanks guys - it sounds like itʻs time to make myself a wooden zero fret and see what happens - itʻs a cheapʻnʻeasy way to test things out, and I hadnʻt thought of it. Youʻll be the first to know, once I try it.

Interesting experiment! I took a piece of #14 Romex wire, flattened the bottom a bit to get it down to a bit higher than the nut, and slid it into place. I havenʻt measured much yet, just spot checks, but itʻs clear that all strings are better in tune. If I measure from the nut to the middle of the wire, Iʻve only moved the nut about 1 mm closer to the frets - the formula says that should be less than a -3 cent change but it seems much more, both with the tuner and when I play and listen. More later after I do some measuring, and thanks for the idea!