Official Luthiers Forum!

I use the usual 5 lb @ 18" span in my setup. My weight is an empty scotch tin filled with pennies. There is a substantial difference in readings having it stand on it's end in the center of the top, and on it's side. Upright flexes the top much more. Which do you think more appropriate, upright or on side?

I don't think it matters as much as long as you are consistent in your method of before and after . I measure before and after bracing .

_________________
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

WHY would there be a difference?

Because upright negates much of the cross grain stiffnes because the weight doesn't reach the edges of the top. The piece I'm working on now deflects .140 with the can sideways and .165 upright. The difference in the end result to get to my target deflection would be .103in thick in the first case and .114 in the second, quite a discrepancy in overall strength.
Alan, would you be willing to share your deflection setup?

I use a cut scrap piece of pipe or a long narrow weight standing upright rather then a can. Try to keep as small amount of contact to the wood as possible. Also flip the piece your testing over and measure/ average the two. Flex memory plays in and you just need an average.

_________________
Todd Lunneborg
http://www.tlguitars.com/
http://www.fretboardjournal.com/

If you have the weight in a large container as the weight deflects the top , the weight is concentrated at 2 points .This them acts and changes the delfection by weight as the weight is not on the center . When on edge you have the weight concentrated one a line in relation to the top center and the container . In a sense the flat bottom will act like 2 weights on the top not one . With the weight off center and you still have the same weight but it will be closer to the supports .

_________________
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

To get good results that can be correlated with other people's data, it is important
to get the weight to be a line, as stated. That line should also contact the plate
across the entire width of the top, not be concentrated across only part of the
width. You get very different deflection profiles when you are not getting this
full-width, uniform line contact. I'd encourage you to switch to a piece of pipe
or solid cylindrical rod rather than using your tin.

_________________
Gene

Politicians and diapers must be changed often, and for the same reason- Mark Twain

That makes sense. However the tin is round, so on it's side it only contacts on a point. It is a little short though. Let's see, I think Laphroaig has a longer tin. If not, Smokehead does, though it's a little green for my taste, however sometime you just have to sacrifice...

I'll second that. It also makes it easy to calculate the modulus of elasticity doing it this way. Not that that's really necessary.

Here's my set up, sorry if you seen it before.


And a mock up of it in use. I can rotate the "pressure foot" and the rails/supports to test the opposite axis, thats why it's so long, I also set a weight up on top of the platform when in actual use.

_________________
Jim Watts
http://jameswattsguitars.com

Jim, pretty slick. Do you use it just to find out a pieces relative stiffness or do you aim for specific numbers? I ask because it seems you can't do a whole top in your setup.

I'm just looking for relative stiffness and density to help me grade my wood for stiffness/weight. I've gotten pretty good at just doing by hand and ear too, but I can assign numbers to my wood doing this way

_________________
Jim Watts
http://jameswattsguitars.com

What deflection parameters are you looking for after bracing....especially for a SS?
Kent

_________________
Wood Creations by Kent A. Bailey
EXCELLENCE IN SCULPTURE, CARVING, LUTHIER, ARCHITECTURAL MILLWORK AND DESIGN

http://www.kabart.com

Kent,
who was that question addressed to? If me, I aim for .200 under 5lb at 18"" span, but only because I read that is what others aim for. I cannot comment yet on it's usefulness as a target number. I need to build for a while with it to see what I think. Wish I'd started doing it on the first 50 guitars I made, I'd probably be a lot farther along...
....oops, sorry Kent that's before bracing. After, I have no idea. Yet.

I generally use a vibration test, rather than deflection, so my setup might not do you much good.

The point of my question was exactly that there should not be a difference in the readings if what you're measuring is just the lengthwise stiffness: if there's a difference there must be some other factor causing it. Certainly crosswise bending would be high on the list of possibilites, so you'd need to get that out of your setup by both supporting and loading the piece uniformly all the way across.

Wood 'cold creeps', so it's likely that delays in taking the deflection measurement will give a different reading. Zeroing on the loaded piece, and then reading the deflection immediately after taking the load off minimizes the time involved.

Dave Hurd notes that he often gets different readings by fliping the wood over. That doesn't surprise me.

Alan, thanks for the reply. I can see how your setup would not cross over well. I've watched your vid several times over, and I found it instructive. However, much of it was over my head so I'm looking for a good chladni 101 book. I'll be watching it many times over to let as much through my thick skull as I can. And then I'll start posting questions about it... Deflection is just a starting point for now.
Thanks everyone so far...

Assuming line contact as stated above, the biggest source of error I've found is the flatness of the panel. I recently did some experiments on the repeatability of various longitudinal stiffness testing methods for spruce tops. I tried to answer these questions:
1. If a top-half or glued-up panel has slight warpage after stickering (typically twist, but also cup), how is the stiffness measurement affected? Can it be corrected?
2. Are the left and right halves of a bookmatched panel different in stiffness?
3. Is it better to do stiffness testing in both directions ("top" surface facing up and down, with downward loading on both), or is testing in one direction sufficient?

Here's what I found (I'm glad to share the raw data):

1. Warpage causes substantial errors in deflection measurements, and appears to be the largest source of error for this type of testing. Twist warp causes uneven contact (gaps) between the spruce and the supports. As load is applied to a warped panel, the resulting deflections are artificially high since the some of the load goes into closing these gaps. Errors can be up to 40%. Once the gaps are closed, the stiffness is linear and repeatable.
--> Solution: apply a moderate tare load (1-2 lbs) that is sufficient to close the gaps. Then zero the deflection indicator, apply the test load, and record deflection.

2. Assuming warp errors are controlled, I found no substantial difference between the left and right halves of a bookmatched pair (8 L/R pairs, average difference of 2% (absolute values)). If warp errors are not controlled, the L/R apparent "difference" can be up to 30%.

3. I first tested narrow (3" wide x 18" long) spruce strips, and found no substantial difference for the direction of loading ("top" surface facing up v. down, average difference = 2% for 12 samples,). Thus, the wood itself is not different on one side vs. the other, as expected for just ~1/8" of thickness. But, when testing full width glued-up tops, the consistency in deflection was worse when flipping sides (11% average difference for 4 tops). Since the wood itself isn't different, and the tare load closes the warpage gaps, I think this poorer repeatability is due to residual warpage in the panel (slight domes, waves, etc) which may behave differently in tension v. compression. So, for glued-up panels, it's probably best to test in both directions and take the average of the deflections. (But, if gaps are not closed by a tare load, an average of the deflections would be artificially high--in that case the lower deflection reading (stiffer) is probably more accurate.)

Main conclusions:
The flatter the better.
If a panel seems different after flipping, it's because it isn't flat enough.
Tare loads help a lot.

_________________
David Malicky

GOOD STUFF David!

The tare load is a fantastic idea!

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Yeah I've been doing it with an initial load too, only way to go IMHO.