Official Luthiers Forum!

I've done some deflection testing of tops and top-halves and sometimes find errors due to plates that aren't perfectly flat. I sticker them, but often there is still a slight warp, and this creates a small gap somewhere between the plate and the round dowel support(s). The gap usually closes when I place the center load, but then part of the dial indicator measurement is simply due to that air gap. The problem seems more apparent on glued-up tops than top-halves, since there's less warp on the narrower plates.

I've noticed that when I flip one of these slightly warped plates over and test again, the deflection measurement is quite a bit different (e.g., 0.200" to 0.230"), whereas for the really flat plates, the 2-sided deflections are nearly identical. Also, if I carefully close the gap with finger pressure at the dowel support, the resulting dial indicator change is very close to the difference in those prior 2-sided deflection measurements (e.g., 0.230 - 0.200 = 0.030").

I'm thinking of a few things to manage this...
- Use the smaller of the 2-sided deflection measurements since it's likely to be more accurate -- but a warp will probably exaggerate both.
- Place shims to fill gaps prior to testing -- time consuming.
- Test the unglued top halves since they are flatter, find E, then calculate what final thickness is needed for the target deflection.
- Place a small tare load to close the gaps, zero the dial indicator, then place the test weight.
- Put the warped plate back in the sticker stack and hope.

I'm wondering if anyone else has run into this issue, and what you do to manage it?

Thanks, David

_________________
David Malicky

I'd pick this one (might use full width iron bars just outside the supports) and do it EXACTLY THE SAME on all tests.

_________________
Dave
Milton, ON

For a simply supported beam. the deflection is proportional to the load.
You could just use an initial load at the centre to preload the test piece and zero the dial at that position. Then add the test load additional to that and read the dial.

However,testing the unglued halfs and calculating E may be a better approach for giving you longterm comparable information.

Are you using full width loads or just a concentrated centre loading? This makes a difference too especally whan using different plate widths.

David,

I don't have a scientific reason for doing it but i take both measurements (both sides) and average them.

_________________
Burton
http://www.legeytinstruments.com
Brookline, MA.

I use a sample of the top , I like .125 square 6inch long apply a standard weight to the jig I have and measure the deflection. I think that as long as you base your data on the same methodology ,you should have a base for consistency .

_________________
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

I know it wasn't the main jist of your question, but what's the point of testing only one half at a time if the point is to target consistency in final thicknessing? Determining E may be a useful piece of data but I get trying to back-calculate your way to a final thickness by it. Unless someone can enlighten me.

_________________
Wes McMillian
Santo, TX
http://www.wesmcmillian.com

David Hurd noted the tendancy for getting different delfection readings on two sides of a top in his 'Left Brain Lutherie', and recommended averaging them. The tare method sounds good, too.

This is one reason I like to find the Young's modulus by frequency. Another is that you get the Q value.

In an article on wood properties in the old Catgut 'Newsletter', James Woodhouse pointed out that there is no simple test for a plate that will give you the exact values for the Young's moduli. There are always other things, like the shear moduli and Poisson's ratios, that are 'mixed in'. He showed how a number of vibration tests of different modes of the plates, and some mathematics, would improve the readings, at the cost of some complication. I have to say that the math was beyond me.

There's the added point that wood is not metal: wood varies in properties from point to point. In theory you could chop up that top into, say, 2" squares, measure each one carefully, and then plot out the variation before gluing them all back together, but buyers have enough trouble accepting 3-piece tops, let alone 20 piece ones with glue lines across the grain ("Hey is that the 'silking' I've head so much about?" "Oh yeah; neat isn't it!").

The bottom line is that I don't think we're realistically going to get any closer to the 'real' values than, say, about +/- 10%. Given that the stiffness of the top will vary as the cube of the thickness, a 3% difference in thickness will give a 10% change in stiffness. Are you so sure you can control your top thickness that closely? Do you really know how stiff the top should be to that level of precision? If your answer to either of those is 'yes', then you're a better luthier than I am.

Thanks, all, for the helpful replies. For the 4th option (tare load), I was thinking of a center load (like taring a container on a scale), as Jeff suggested. But Dave's idea of weights above the dowels may work, too (it could change the 'pinned' end condition to a semi-fixed end condition, but yes, if done carefully and consistently it could work for my own use). Jeff, yes I do full width loading.

For those that just average the readings, I'm curious how much different your two readings are. Maybe your plates are flatter than mine... from the tests I've done so far, it looks like the main factor in my tests producing different numbers for the two sides is warp in the plates.

And there's another way that warp could give different readings when testing both sides, even if there are no gaps at the dowels--any arch, dome, or waviness in the plate could respond differently based on load direction. So if gaps exist, the lower deflection # is probably more accurate since the higher # is inflated from the air gap; but if arch/dome and no gaps, averaging is probably better.

Wes, yes I agree deflection tests on the final top would be most accurate, so long as they are flat. The only advantage of testing half-plates is their higher flatness.

Alan, I see your point on the advantages of vibration tests, and on the level of accuracy needed. That Woodhouse article sounds interesting. Getting Q would be very helpful, too... I will look in the archives for more info on vib testing. Nope, I'm definitely not a better luthier than you, or you all! Plus/minus 10% would be plenty accurate for me, but some of the deflection errors I'm getting appear beyond that, and I like to minimize method errors before I start a database.

Thanks again, David

_________________
David Malicky

I only have 3 samples so far. E is calculated for each with 3 different weights 2.25lb to 5lb. Weights are 1' long steel bars. I measure each side then average.

Average Plate MOE (E), psi. Data are listed as side 1, side 2

1. Euro 1.75E+06, 1.6E+06
2. Sitka 1.6E+06, 1.3E+06
3. Adi 1.7E+06, 1.7E+06

_________________
Steve Smith
"Music is what feelings sound like"