Martin neck reset

One of the most difficult and complex jobs a guitar tech can be faced with is a neck reset. Acoustic guitars typically need to have their necks reset every several decades or so, as the 200 some-odd pounds of tension pulling on the neck at all times slowly pulls the neck forward, changing the angle at which it joins the body. It’s pretty easy for anybody to determine whether their instrument needs a neck reset: if the action is high, the neck itself is straight, and the bridge saddle is low in the bridge, it’s a clear indication that the geometry of the guitar has shifted. The proper way to fix this is to take the neck off, reshape the neck/body joint, and put it all back together. I recently completed a neck reset on a 1974 Martin D-28: a fairly complex job made even more difficult by the fact that this Martin didn’t have a truss rod. This Martin neck reset was going to be a doozy!

When this guitar showed up in the shop, I instantly noticed that it needed a neck reset: the action was at almost 4mm at the 12th fret, and the saddle was completely buried in the bridge. Sighting down the neck (ALWAYS from the headstock), it was easy to see how far underset the neck was. It’s hard to photograph a bad neck angle, but I tried anyway – hopefully you’ll get the idea:


Here’s a shot of the saddle. It’s so low that the strings don’t exert much downforce on the saddle, which can cause a weak tone by the fact that vibrations aren’t transmitted efficiently into the soundboard:


Laying a straightedge on the fingerboard reveals exactly how underset this neck was. On a perfect neck set, the straight edge should line up EXACTLY with the top of the bridge. On this guitar, it was underset by almost 3mm (yes, that’s a lot!)!


Amazingly, Martin guitars didn’t introduce a truss rod into their guitars until 1985! Previously, the used a hollow square tube to support the neck – but over the decades the string tension usually wins out, and this guitar was no exception. It’s neck was curved into an unacceptable amount of relief (.3mm), which would make an accurate neck reset very difficult. Typically I’ll straighten the neck, take measurements, and then calculate how much wood I need to remove from the neck heel to angle the neck back accurately. Since there was no way to adjust the neck with a truss rod, I opted to heat press the neck to straighten it. How I do this is clamp the neck into what is essentially metal sandwich, with a heating element in between two thick pieces of aluminum, and then cook the neck for about 45 minutes. What this does is soften the glue between the fingerboard and neck, and when the glue cools, it resets the two pieces of wood in the new straight position. While there’s no way to guarantee the wood is going to straighten the way I’d like, I’ve learned over the years that it’s a fairly safe bet, and this time was no exception.


Now, with the neck in a much more acceptable state of straightness, I was able to take some good measurements and calculate how much material I needed to shave off the heel. With measurements in hand, it was time to remove the neck. The first order of business was to drill a couple of holes in the fingerboard down into the dovetail joint, to create access for the steaming needle I’d insert later. I removed the first fret in from the start of the body, which is typically where the end of the dovetail is, and drilled two holes, avoiding the metal hollow tube running down the center of the neck:


Next I heated the fingerboard extension, and carefully cut through the glue with a knife, and separated the fretboard from the guitar’s top:

IMG_2769 IMG_2770

Then I warmed up my modified steam machine, loaded the guitar into my neck removal jig, and began steaming the neck joint. Typically melting the glue and easing the neck out of it’s joint takes about 10 minutes, but this guitar’s neck came off in less than 5. Right on!




It was clear why the neck came off so easily: it had been removed before, and wasn’t reset correctly. The dovetail wasn’t tight, and there was only a half-hearted effort to create a solid joint, with the addition of a couple small pieces of maple veneer at the bottom of the dovetail. Not pro! I cleaned up the two sides of the dovetail, removing the excess wood, glue, and making the sides flat.


Next I clamped the neck to one of the legs of my workbench, and started carving away at the heel. Since the sides of the guitar are curved, carving the heel in this way allows the edges of the heel to “cup” the sides, and makes for a solid foundation against the the walls of the guitar. IMG_2778IMG_2780 IMG_2781

I had previously calculated how much wood I needed to take off the bottom of the heel for an accurate neck angle: .99mm. I marked this on the heel cap, made a makeshift sanding block, and sanded the heel down to the line:

IMG_2785IMG_2786 IMG_2787

Of course, I checked and double checked my work – even a single stroke of the sanding block could ruin the neck set. This is one of the most tedious aspects of completing a neck reset: I’d make a few passes with the sanding block, the reinstall the neck, clamp it in place, restring the guitar, tune it up to pitch, check the neck angle, twist, and pitch… and repeat. And repeat. And repeat. I must have done this about 10 times before I was satisfied with how the neck set was looking. Yeah, it’s a pain… fortunately this only needs to be done once ever 30 years or so!

To finesse the neck into shape, and to make sure the neck and the sides fit seamlessly, I dragged a piece of sandpaper through against the neck while it held the two pieces in place. Only a couple of passes were necessary, and even though I hadn’t removed but a whisper of wood from the neck, I still string the whole thing up and double checked, just to make sure the neck set hadn’t wandered off.


The next step was to rebuild the dovetail joint, to make for a strong mechanical joint. I smoothed both surfaces of each side of the dovetail, and added some mahogany veneer to the neck, using the world’s cutest tiny clamps:

IMG_2791 IMG_2793

Yet another tedious stage to face: test fitting, sanding the veneer down, test fitting, sanding some more, etc., etc.,  etc. While time consuming (and to be honest, slightly irritating), it was absolutely worth it: the dovetail joint was so tight now that it would hold under string tension without the use of glue! Sweet!

After once again double checking the neck for proper angle, pitch, and twist, I glued it up, clamped it all together, and let it sit for 48 hours.


Alright! Neck reset done! The next few steps were critical: I needed to refret the guitar, which not only required planing the fingerboard, but actually carving in the proper amount of into the fingerboard (since there’s no truss rod) – all while not ruining the neck reset I had just completed. I won’t go into tons of detail on refretting, as I’ve covered it extensively elsewhere on this blog. What I did differently with this refret was this: I strung up the guitar with a very light set of strings, set the guitar into my neck jig, and planed the board flat under simulated string tension. Once the guitar was strung up with the owner’s preferred heavy gauge strings, the neck *should* pull into a proper amount of relief. Admittedly, this requires a bit of guesswork, but at least it’s educated guesswork.



Finally, after refretting the guitar, making a new bone nut and saddle, the moment of truth arrived. I strung it up with a set of 13s and inspected my handiwork – and I have to say, I’m pretty proud of this. It came out absolutely PERFECT: the neck was dead centered (the outer two strings exactly 3mm from the fingerboard’s edge), the angle was absolutely perfect (a straight edge lined up right on top of the bridge), and the action from string to string had come down by over half it’s previous height (and could come even lower had the owner desired it). Huzzah!

This guitar showed up almost completely unplayable – and now it plays like butter. Best of all, it even sounds better! Win! Since the neck joint is so solid, I wouldn’t expect this to need another neck rest for at least another 30-40 years, maybe longer. I imagine I won’t be alive to do the job next time – I hope the next tech takes as much care as I did!

Permanent link to this article: