Official Luthiers Forum!

I saw a failure like that once on a Lowden. In that case it was a cedar top, and the bridge just popped off one afternoon with the guitar hanging on the wall of the shop. Close inspection showed a sharp incised line all the way around the edge of the bridge in the top: the glue didn't fail, but the wood peeled up due to the stress riser at the edge. It doesn't take much. It's hard to tell from the photo whether there was any such thing on the bridge you just saw. It's hard to account for the edge being so clean all the way around, with no tear out into the finish otherwise, though.

Interesting that Collings does a rabbet on the bridge. I thought that was just a custom builder thing. Does anyone know if other larger shops do it too?

I don't know how much of a stress riser that would be as it's only the thickness of a piece of paper (although I'm sure it's some), but it seems like every square mm of glueing area you could save would be a better investment.

Just curious, I remember that you use a vacuum bridge clamp. Have you always used that type?

If the glue joint between the bridge and top are stressed, you want them to separate cleanly to relieve the stress, instead of warping the top.

_________________
"Act your age, not your shoe size" - Prince

Maybe adding some info, below is an FE analysis focusing on bridge-top joint stresses (done last year, posted on mimf). There are a lot of caveats with an analysis like this, particularly on the actual stress values near sharp corners like the tail of the bridge. This FEA models the braces, bridgeplate, crude body, saddle, string loads (at sadde, ramp, and bridgeplate), and orthotropic wood properties, but I haven't done the glue or its properties yet (it's a wood-wood bond right now). There is no rabbet; just a flat-bottom bridge on a flat top. And it's linear elastic; no microyielding, plasticity, etc. Models like this are best for comparative purposes. That said, this pic shows the stress perpendicular to the top surface, 0.001" below the top surface:



Here is a zoom-in of the tail edge, at the transition to the wings (where stresses are highest):


It's easy to see how peeling starts at that tail edge. Once a tiny bit peels, the stress concentration just moves forward and bridge unzips. The ballpark peak stresses at that rear edge (0.001" into the spruce) are around 700 psi tensile. 0.005" into the wood, they are down to ~350 psi.

Page 5-13 of the Wood Handbook shows the tensile strength perpendicular-to-grain for spruce is around 370 psi. http://www.fpl.fs.fed.us/documnts/fplgt ... ter_05.pdf

A few thoughts:
- Since stresses decrease rapidly below the surface at that rear edge, glue penetration there is very helpful. Anyone know what depth is typical for softwoods?
- Since the FEA peak stresses are greater than the actual mean strength, there is probably microyielding occurring in the wood and/or glue, and it's not surprising we see bridges peel off occasionally. Of course, some spruce is a lot weaker than the published mean strength.
- Anything that would weaken the wood near that rear corner is a bad thing.
- Dowel pins in the middle are too far away to help the peeling at the back edge. The pins would help only after the joint had peeled up to them.
- I've done a few variations on the bridge plate... overlap and a thicker plate reduce the peel stresses pretty substantially.
Someday I'll do systematic runs and write it up.

_________________
David Malicky

To me, it doesn't look like the finish has been routed away. Instead it looks to have been masked off prior to finish. There is a ghost around the perimeter of the bridge location. That's why it looks as though the bridge has been rabbited. My point is that I doubt that a blade was used to score around the bridge for routing, cutting the fibers of the guitar top.

Just what my eye sees...

All speculation, but...

I'm going for subsurface damage caused either by the thicknessing process or by the finish removal process at the bridge location, most likely the latter (because I don't see "shadows" in the rest of the finish). If the tool used to remove the finish in the bridge area is not really sharp you can remove the finish but damage the wood fibres beneath to the extent that they might look OK but in fact are barely attached. Glue down a bridge on top of that and it is then just a matter of time. The bridge detaches with the delamination immediately subsurface and parallel to the surface, as we see.

_________________
Trevor Gore, Luthier. Australian hand made acoustic guitars, classical guitars; custom guitar design and build; guitar design instruction.

http://www.goreguitars.com.au

David M, thanks for posting the graphic from the FEA. A picture is definitely worth 1K words in this case, and not just in the instance of the peak stress portion at the rear of the bridge. Very informative.... thanks!

This has morphed into a really great thread.

_________________
It's not what you don't know that hurts you, it's what you do know that's wrong.

Filippo, interesting, your laminated plate sounds like a stiff design. Jay, yes, understanding how ball-end loads propogate through the joint was one of the original motivations. Good to hear the FEA is interesting.

For some completeness, here are the corresponding shear stresses at that same 0.001" depth into the spruce (same color scale):


So, the shear stresses are quieter than the normal (perpendicular) stresses. The tail edge is much quieter than the front -- this is probably because the forward load path (directly from bridge to neck block) is much stiffer than the aft load path (bridge to tail of body, then through back and sides, then to neck block). Loads and stress follow the path of greatest stiffness. Wood is much stronger in shear-parallel-to-grain than tension-perpendicular-to-grain, so this shear plot is probably not so important for the wood failure in this thread. But maybe interesting.

_________________
David Malicky

I have seen this a few times. Often a heat or temp shock if it pops right off. Rare but it does happen.

_________________
John Hall
blues creek guitars
Authorized CF Martin Repair
Co President of ASIA
You Don't know what you don't know until you know it

In my line of work (semiconductors) we see this type of failure when we glue two materials together that contract at different rates when cooled (different CTE). The stresses are very localized to the glued area and are uni-directional as the shrinkage occurs.
Is it possible in this case that the bridge, or more importantly the glue, was somehow caused to shrink much faster or more than the spruce (cold temp is what does it for us)? If the glue shrank much more than the spruce it would essentially pull the wood apart at the junction of glue intrusion into the wood.
Just a thought.
Dave

My Yamaha failed like that after a decade and a half of ownership. I bought it in Japan, then took it to San Diego, and then to Illinois. I contributed it to bringing it three complete different environments with any real effort on my part to control the humidity / temperature it was stored in.

A joint will fail at it's weakest point. In this case that was the wood itself. You can clean up and reglue, but I wouldn't be surprised if it took more of the soundboard the after it's brought up to concert pitch.

I don't know if all Yamahas are created equal but one Yamaha I fixed had one of those improperly glued bridge that looked as though it was pasted onto the wet finish with the dry finish still under the bridge, so the bridge didn't get fully bonded to the soundboard.

_________________
Cat-gut strings are made from kitten guts, stretched out to near breaking point and then hardened with grue saliva. As a result these give a feeling of Pain and anguish whenever played, and often end up playing themselves backwards as part of satanic rituals.

Typhoon Guitars
http://www.typhoon-guitars.com

Sure it does. Even jointed and sanded wood has open pores. Wood glue grips into those pores.

_________________
Another day, another dollar.

What Todd said.

AS I understand it, the strongest glue joints have between .002" and .008" of dried glue in them. 'Keying' just makes deeper scratches that end up being thicker glue lines than the optimum. Smooth, well fitted surfaces are what you need to keep the bond thickness in the right range.

I was told that the Forest Product Lab, in a study of wood aircraft structures back in WW II, found that surfaces that had been worked within 15 minutes of gluing ended up bonding better. That's the 'high surface energy' Todd mentioned: working the wood breaks chemical bonds, which take a while to re-form, and if you put glue on those open bond sites, you get a better joint.

The test for surface energy is to spray it with a mist of water. If it spreads out into a film the surface energy is high: the polar water molecules are attracted to the open bond sites. If the water forms droplets the energy is low. I tried that on a fossil mammoth ivory bridge. The stuff is notably hard to glue, and I suspected it might have naturally low surface energy, since most of the chemistry will have already happened. Sure enough, a surface even a few hours old caused the water to bead up, while it would spread out on a fresh surface. I gave the gluing surface a quick scrape just before spreading the HHG on it, and so ( about two years) far it's held.