Testing existing top and back

Talk about musical instrument construction, setup and repair.

Moderators: kiwigeo, Jeremy D

Post Reply
artus
Gidgee
Posts: 8
Joined: Fri Aug 03, 2018 7:23 am

Testing existing top and back

Post by artus » Sun Feb 10, 2019 4:56 am

Folks,

Many years ago (late '70s) I built my first guitar- I built it from Brazilian RW and Spruce. My abilities did not do justice to the wood. My abilities have progressed a bit and now have disassembled the guitar for rebuilding. I have tried gathering the spectrographic data to investigate thicknesses and bracing, but the spectrograms are not what I expect and I believe I may not be selecting the best node and antinode spots. I get confusing mutiple peaks. Or perhaps I can't approach the redesign this way. I also have been looking at the static deflection from test bed loads -- again, I am not confident in the results -- selecting the effective width is problematic and I may eventually figure that out. Any thoughts on how I might better approach the problem. Particularly, I want to do the best job possible for the rosewood. I can redo the top with new wood, but did like the rosette I have in the old one.

Thanks,
Jack

blackalex1952
Blackwood
Posts: 676
Joined: Tue Aug 27, 2013 6:36 pm
Location: North East Victoria

Re: Testing existing top and back

Post by blackalex1952 » Mon Feb 11, 2019 1:54 pm

If you post some of your data, Jack,it may help with comments. I use Visual Analyser for my tests and have found that the mic level is important for clear results along with an external soundcard or box. It's also important to check that the external soundcard is selected each time as the version of VA that I have defaults to the internal soundcard. So I have to select the external card each time I load VA on the computer that I have. It is pointless to use the internal microphone in the computer because noise is the issue, eg computer fan, inherent noise in the lousy internal soundcard, distortion etc....
Did you do any Chladni tests? These can confirm the VA readings. I get consistently repeatable results with VA by tapping at the bridge.
If you trawl this forum you will find some great deflection test rigs have been posted. I have found that the best deflection rig I have come up with has a perspex plate sitting on top of the dial indicator shaft, which will hold the mass for the deflection tests. This method minimises any friction, limiting it to the dial indicator shaft only.
Cheers! -Ross
"Everything I say on the topic is based solely upon inexperience and assumption!"

jeffhigh
Blackwood
Posts: 1512
Joined: Thu Feb 07, 2008 5:50 am
Location: Caves Beach, NSW
Contact:

Re: Testing existing top and back

Post by jeffhigh » Wed Feb 13, 2019 8:20 am

If you are testing "existing top and back" disassembled from the body as your title implies, all the regular material testing methods will not work since they rely on using rectangular uniform pieces.

artus
Gidgee
Posts: 8
Joined: Fri Aug 03, 2018 7:23 am

Re: Testing existing top and back

Post by artus » Wed Feb 13, 2019 1:59 pm

The top and back are free and the bracing has been removed. I am attaching spreadsheets for both the spectrogram and static tests; I trust none of the results. My best approach may be to do a new top with known properties and use the back with the current thickness (2.4 mm.)
Once I have a reassembled top and sides, I can tweak the back bracing by clamping the back on.
Thanks for responses -- Jack
Attachments
Screen Shot 2019-02-12 at 5.29.16 PM.png
Spectrographic
Screen Shot 2019-02-12 at 5.30.26 PM.png
Static

artus
Gidgee
Posts: 8
Joined: Fri Aug 03, 2018 7:23 am

Re: Testing existing top and back

Post by artus » Wed Feb 13, 2019 2:01 pm

Sorry duplicated post somehow

blackalex1952
Blackwood
Posts: 676
Joined: Tue Aug 27, 2013 6:36 pm
Location: North East Victoria

Re: Testing existing top and back

Post by blackalex1952 » Wed Feb 13, 2019 2:57 pm

My best approach may be to do a new top with known properties and use the back with the current thickness (2.4 mm.)
This sounds like the most simple and direct response to the issue...keep it simple and BUILD!!! Jeff's point re testing the disassembled plates is valid. The soundboard does most of the sound production, so that is the place to begin. Your back seems like a good thickness. Prior to disassembly, you could have played with brace heights using a finger plane and sanding blocks to get an idea of what was actually happening. (Assuming the guitar has a soundhole big enough to get your hand into.) But now, of course, you have a guitar kit without a good soundboard.
-cheers! Ross
"Everything I say on the topic is based solely upon inexperience and assumption!"

Online
Woodsy23
Blackwood
Posts: 112
Joined: Sat Mar 16, 2013 10:23 pm
Location: Sydney

Re: Testing existing top and back

Post by Woodsy23 » Thu Feb 14, 2019 1:27 pm

Jack,
From your original post, I understand that you have tried to obtain the long and cross grain E values (and Glc) for the wood of the top and back plates and have used both the tap testing (spectrographs) and deflection tests. Both methods are described in "the books" but with the main focus on the tap testing method. As mentioned by Jeff (and in the books) the tap testing method for determining El, Ec and Glc only work for rectangular pieces of wood. Also, the tap tests are usually carried out with thicker than final plates. If the rectangular plates are too close to the final value (which is not known at that stage anyway), the tap frequencies can be too low to measure reliably.

However, the defection tests should give reliable results for the El and Ec. All you have to do is support the shaped guitar top of back plate along two parallel lines and apply a line load half way between these. The parallel supports don't need to be at the extremities of the shaped plate. First measure the deflection at the load point with the supports and load perpendicular to the grain (for El) and then parallel to the grain (for Ec). The thickness of the "test beam" will be constant and measurable. For the width of the "test beam" use the average width of the shaped plate between the two supports.

Determining Glc by deflection testing is a bit more difficult. However this material property has the least effect on the target thickness, according to the formulas in the book. If you use a Glc value from published data for the particular species, that should be close enough and any difference to the actual value of Glc would probably only make a difference of 0.05mm to the target plate thickness, which is within achievable tolerance.

Once you have the El, Ec and Glc, your can determine the target thickness (and hope your plates are not already too thin).

I hope I have understood your situation correctly.
Richard

artus
Gidgee
Posts: 8
Joined: Fri Aug 03, 2018 7:23 am

Re: Testing existing top and back

Post by artus » Thu Feb 14, 2019 1:50 pm

Thanks Richard. The static test results are from a test jig using parallel rods and a linear load at center of span. I will look into published Glc values and recalculate target thicknesses. I'll rerun the static tests a few more times to get a sense of the reproducibility. Thanks for suggestions.

Jack

artus
Gidgee
Posts: 8
Joined: Fri Aug 03, 2018 7:23 am

Re: Testing existing top and back

Post by artus » Sun Feb 17, 2019 5:09 am

Folks,

I reran static testing multiple times and consistently get targets for both top and back of 2.1 mm, slightly less than current thicknesses of 2.3 mm . These calcs used a Glc of 0.72 GPa for Sitka and 2.0 GPa for the Rosewood. Note, I found a mistake in the Static spreadsheet I attached earlier and that explained the weird results. Thanks again,

Jack

jeffhigh
Blackwood
Posts: 1512
Joined: Thu Feb 07, 2008 5:50 am
Location: Caves Beach, NSW
Contact:

Re: Testing existing top and back

Post by jeffhigh » Wed Feb 20, 2019 7:21 pm

I'd suggest you just use the existing top and back at current thicknesses, brace and assemble and then test.
You can always thin the plates in situ.
The results from static testing of a shaped back and top are going to be suspect and can't just be rectified my averaging the width as suggested.
For one thing the span to width ratio will be too low to allow the assumption of a simply supported span.

Post Reply

Who is online

Users browsing this forum: Woodsy23 and 14 guests