Official Luthiers Forum!

I'm interested in how smooth the inside of the top should be prepared before gluing braces. This article has some interesting info on planing vs sanding:
http://saliguitars.com/casj_influence_cutting.pdf
Especially, see Page 4, Table 2, the "Delta-j" column, and compare case 6 (60 grit sanding) to cases 4 or 5 (planing).
"Delta-j" is the log decrement, so higher Delta is more damping. Sanding gets a 0.211, while planing averages 0.152.
Note their final sanding pass is extremely aggressive: 60 grit at 1mm (0.040").

If using sanding, I know a tight sequence (80, 100, 120, 150, 180, 220) is best, although that's a lot of work. Somewhere I read that 220 is roughly equivalent to planing (bad pun).

Then I ran across this article, showing how abrasive planing (drum or wide belt sanding) crushes cells far below the surface (See Figure 4 on p 113):
http://www.fpl.fs.fed.us/documnts/pdf1986/murma86a.pdf
That is Doug fir, which I imagine is less crushable than spruce? (Hate to think of those times I pushed cedar through my drum sander.) But, they used flatsawn boards, whereas quartersawn wouldn't have the hard latewood layer on top of soft earlywood. Does anyone know how much crushing occurs with the side-by-side layers of QS? They found no real solution to it, except to suggest light sanding (not drum or wide belt) or maybe flooding the surface with CA. The last 3 paragraphs on page 116 are good reading.

How deep does shellac or HHG penetrate the surface? Or, would that at least help to glue together the torn surface fibers? (Applied after the braces are glued.)

Has anyone compared tops thicknessed by planing vs. those by sanding and/or abrasive planing?

_________________
David Malicky

Yes, but a long time ago! I came to the conclusion (from a small number of tests) that the surface damage caused by abrasive thicknessing extended subsurface a distance approximately the same as the diameter of the average grit size used, as this was commensurate with the approximate reduction in Young's modulus I was measuring. I've never bothered pursuing this conclusion further because a) I'd convinced myself and b) I've never owned an abrasive thicknesser to do exhaustive tests with. So take that as anecdote if you like.

Now here's some more anecdote, but based on a lot of experience. My guitars almost always come out stiffer than other peoples using basically the same design methods and materials but different wood working techniques. I use hand planes for almost everything they can be used for, whereas most other people seem to use abrasives if they can get away with it. Which brings me to speculate in few areas:

1) (Sub)surface damage caused by abrasives results in effective loss of stiffness (reduced Young's modulus) as explained above, and I, at least, am fairly comfortable with that conclusion

2) Gluing onto a mushy surface, or gluing two mushy surfaces together could cause spectacular variations in apparent stiffness in the resulting structure compared to the same structure with hand planed surfaces, much like building on a sand foundation compared to a rock foundation

Hypotheses 1) and 2) either together or separately go a long way to explaining the extra stiffness I see in my guitars, which allows me to use less material.

The Sali paper only discusses rotary planing. I would expect hand planing to produce an even better result in terms of effective stiffness and maybe even damping. I haven't been able to detect any differences that the change in log dec might produce in the finished guitar, but that is a very hard comparison to make, given other variations between builders. It would require me to build a bunch of guitars primarily using abrasives and I'm not going to do that any time soon!

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

http://www.goreguitars.com.au

Interesting....

I thickness sand backs and sides, and the surface of the top (less risky than planing rosettes for me if I'm using spalted wood or similar that needs to be reduced down significantly. But I think I'll go back to the hand plane to bring the top to final thickness from the inside. I actually find it very enjoyable to hear and feel good steel slicing off thin, translucent shavings of spruce. And it's pretty quick.

I tend to scrape backs to get a gluing surface.

Thanks, all, that's very helpful! Sounds like yet another way the hand-builders have an edge, at least to some degree. The correlations of damage to grit size and ~half-grit size make sense.

Trevor, thanks for those findings and anecdotes -- great info. No, gluing to a mushy surface doesn't sound good. I also wonder about the outside face since it's furthest from the neutral axis, though it at least gets finish. I see how it would be very difficult to relate surface prep to log decrement results from whole guitars.

Todd, it's great to hear that a ROS can remove the scratches down to 0.001" or less. Thanks also for the mass info. I admit I'm in the abrasive and power tool camp and would like to stay there, at least for the volume of my student guitar project.

I still wonder about subsurface damage from the thicknesser that the finer ROS grits can't get to. That Fig 4 from the second study was a 0.003" pass with 120 grit, and it extends the full frame of the pic -- about 0.030" deep. But, QS should behave better, as the latewood columns should protect the earlywood -- unless they simply buckle. I'll see if I can find any studies that have looked at damage in QS softwoods... if not I think I'll be learning how to make microscope slides.

_________________
David Malicky

Just before gluing the braces I plane the surface with a well sharpened high angle plane, lightest possible cut. This only removes a very little material with no effect on the tuning. At least with hide glue it is important for a good bond. I made a lot of tests and samples used older spruce usually come apart more easily.

_________________
Build log

This whole discussion strikes me as the luthier's version of the "How many angels can dance on the head of a pin" debate of centuries ago.
However one thicknesses a top plate, the results are whatever one decides are right (perhaps measureably thicker if abraded or thinner if planed). The determination is not made on a measurement alone and <that> is where we individual makers have the edge.
Indeed most of us who use a thickness sander (I've used planes too) use incremental hand sanding or scraping to taper and graduate tops from center to edge (about 2.3mm to 1.8mm on average for me) so any heavily scored (I use 80 grit for tops) surface is gone by the time braces are glued.
Also the degree to which the glue (I use fish) impregnates a slightly abraded surface makes this effect negligable as well I think.

Remember too that the nominal thickness of the top off the sander is far from the finished thickness taking into account top finsih sanding not to mention leveling and dressing bindings and purflings. Any experienced builder will leave some headroom in their thicknessing for this as well regardless of their initial technique.

David, Thanks for your input. I certainly agree each maker determines their top thickness so it gives the sound they want using their building style. And I certainly wish the discussion were pointless. But I don't see how it is, given the kind of results Trevor mentioned, and the depth of subsurface damage in the 2nd study. Damage such as in Fig 4 (reproduced below) can not be removed by further sanding, unless you're prepared to lightly sand the entire top another 0.030" or so.

Todd, yes, I agree those factors would be predictors of damage, and that industrial surfacers usually hog off much more material than we do, but that Fig 4 was from repeated 0.003" passes using 120 grit.

_________________
David Malicky

Awe man!!! Ok my drum sander is up for sale now anyone?

In the first article if I am reading it correctly, table one says that for sanding they are using 60grit and taking off 1mm per run. That seems very aggressive to me and nothing I would ever do to a guitar top. I wonder if it's the force of the drum, or belt in this case, that actually crushes cells rather than the plowing and tearing of the grit.

And any way I always finish my sound boards off with a hand plane and scrapers, nothing pops out the finish better than a blade.

i've been wondering about this as well. in building furniture if i can get away with it i prefer to finish with a scraper. it seems to compress and show off the grain in a way that even the finest grit sandpaper can't. i've wondered if this compressing of the grain would be better for the inside of the whole box, sides and back included.

another question i've had and this maybe for another thread is has anybody ever applied a finish to the inside of the box? i imagine it might make it more trebley a tinny, but i have no idea. in my carpentry days i've seen trim that wasn't primed on the surface against the wall cause the material to cup. could finishing the inside of a guitar help or is there an effect on the tone that would make it not worth it?

_________________
http://www.wideskyguitars.com
https://www.instagram.com/wide_sky_guitars/

Hi Patch - most luthiers do not finish the inside but a few do. Tim McKnight comes to mind. Check out this video about French polishing the inside of a guitar. http://www.youtube.com/watch?v=308bh3DHSzY

Todd - I'm not sure I follow. What unbalanced construction are you referring to here? Thanks.

Ah, gotcha. Still, I don't see why sealing the inside like we do the outside wouldn't yield some beneficial protection against changes in humidity. I do realize there are downsides against which any benefits must be weighed. If there is truly no significant upside then it's a no brainer I guess.

So how do you guys prepare the braces to be glued? Most radius dishes have 60 grit, so Im sure it's a bad idea to glue braces after being shaped on just 60 grit. I would think properly preparing braces to be glued is just as important as then plates. (I just thought of this after reading through this thread again, this forum is great!)

Thanks Todd. Sounds like a worthwhile exercise. If I get a chance to do it I'll post my findings.

Those kind of experiments are pretty easy to do, although it's also easy to have confounding variables mixed in, leading to spurious results. Some suggestions...
- Make many 'matched pairs' of samples, either bookmatched or sequential cuts from the same plate. One of each pair gets treated; its mate is the control. The treated samples get either 1 coat or 2. Or, give them all 1 coat, run a test, then all or half get a second coat, retest.
- Before treating them, check that the matched pairs respond the ~same to humidity changes, or that you can't tell a difference.
- Coat the end-grain with wax, as moisture moves out of it the fastest by far.
- Waxed shellac should be a much better moisture barrier than dewaxed.
- Use a dial or digital caliper to check expansion accurately. I make my samples 5 3/4" 'wide' (cross-grain dimension), so they fit a 6" caliper. That also makes bookmatching easy, and top scraps can be converted to samples.
- Alternately or additionally, use a digital scale with 0.01g readout... moisture gain and expansion track each other closely.

_________________
David Malicky

Though I don't normally build guitars, I do seal the inside of my bandura tops specifically for this purpose. I've built both ways and the ones without shellac on the inside definitely move more and more unpredictably than the ones I've sealed with shellac. Banduras have a much bigger top than a guitar though so that may be why it seems to help so much. Shellac by the by, is one of the best vapor sealing finishes we have.

That said, on the one full size guitar I've made I didn't seal the inside of the top and it seems to be holding up just fine.

p.s. I do sometimes chuckle in my sleep when I dream about a future repair man cursing my name for shellacking the inside of my instruments

_________________
http://www.birkonium.com CNC Products for Luthiers
http://banduramaker.blogspot.com

Hahaha, nice.