First Attempts: Finding Target Thickness with Acoustics

[jfmckenna]

First of all, hello down there in Oz and NZ and of course to the rest of you. I hear it's awfully hot down there this time of year. My first post here while I imagine most of you down under are still asleep! Hello from Virginia USA. I spend most of my time on the OLF but I found out that this place is probably more apt to post on this style of building then anywhere else, so here I am.

Anyway I am building 3 guitars at the moment. This will be my 65-8th guitar and I have decided to look into using this method. Or at least in conjunction with what I have developed over the years. So far I am pleased to see that it comes close to what I already do. In short, many years ago I learned a trick for thicknessing tops. That was to proceed to thin the top such that when you hold the top on each side and shake it back and forth it sounds like that whoop whoop whoop of a piece of sheet metal. Then at some point I got into deflection testing and found that deflections that I targeted for certain model instruments jibed well with the sheet metal trick. Now I am using this 3rd method and am finding that it jibes with my deflection. So far so good.

But this post is in particular about this one top, out of the 3, that is a bit difficult to analyse. It's for an OM or 000 Sized guitar. With a torrified Red Spruce top.

The recordings were done using a Rhodes NT3 condenser mic into a PC with a Delta 10/10 sound card. The 9v battery powered mic is plugged directly into the card with an XLR cable. For tapping I made a small dowel hammer with cork on the tips

The three plots are below Fl, Fc, and Flc respectively:

The Fl plot:


The Fc Plot:


The Flc Plot:


The Calculations:


I'm happy with the final numbers as the top is indeed pretty stiff feeling by hand though I have not done my deflection testing yet to compare. But it's the Fc plot with those double peaks that is difficult to interpret. The peaks are actually close enough to where they don't change the final number by a whole lot but still, which peak should be used? Should an average of the two peaks be used?

In general I find the plots difficult to analyses. The book gives ranges of normal values which helps. I guess you are supposed to just ignore some of the noise in the first part of the plot. Like for example in the Fc plot that first peak around 50Hz, is that just a 'shadow' of the Flc peak coming through? And then again on the Flc plot you see a 'shadow' of the twin peak Fc plot?

I found that for the 3 tops I've done the Fl plot was the most obvious and easy to read.

Couple other questions I have is:

1) Do you thickness right to the target thickness or do you leave some room for finish sanding?
2) The book talks about thinning the edges of the lower bout after the box is closed, something I've done in classical guitar building. But what does doing this suggest about going through all this work only to make pretty significant changes to the resulting target values?
3) A bit off topic but, when measuring the response of a complete guitar do you do the tapping with the strings on tuned to pitch?

Regards.

[Trevor Gore]

In general I find the plots difficult to analyses. The book gives ranges of normal values which helps. I guess you are supposed to just ignore some of the noise in the first part of the plot. Like for example in the Fc plot that first peak around 50Hz, is that just a 'shadow' of the Flc peak coming through? And then again on the Flc plot you see a 'shadow' of the twin peak Fc plot?

Generally, Fl and Flc are very easy to find using the techniques shown in the book (hold on a node, tap on an antinode, hold on a node of the mode you want with it also being an antinode of the modes you don't want. Place the mic at an antinode you want to see). The peaks you see will always be the lowest frequency, high amplitude peak in the target range. High amplitude peaks at 50Hz or 60Hz should always be suspected of being due to mains induced hum. Other noise in your environment can be analysed by doing the tap test procedure without tapping the wood. You will get a spectral analysis of the background noise and therefore will know what to ignore. The old G-tune app has a noise filter which samples the background and gets rid of it for you, which is one of the reasons I use it.

For Flc, hold on the node as shown in the book (long edge, centreline, top), have the mic behind the wood at one lower corner and tap the front of the wood on that same corner very lightly. Works for me every time.

Fc can certainly be the hardest to get (but is not always difficult). The problem tends to arise due to the situation discussed on P 4-26 and 4-27. Changing the aspect ratio of the wood can help, if you have that luxury. Again you are looking for the lowest frequency peak that is not due to another mode. (See above about tapping).

In extreme cases where you can't get decent tap results, there is always static testing, Section 4.4.4.

1) Do you thickness right to the target thickness or do you leave some room for finish sanding?

If you work cleanly enough to avoid dings (pretty much essential on woods like WRC, redwood and Engelmann) you can thickness straight to target, which is what I do.

2) The book talks about thinning the edges of the lower bout after the box is closed, something I've done in classical guitar building. But what does doing this suggest about going through all this work only to make pretty significant changes to the resulting target values?

I recall you asking a similar question before. The answer hasn't changed and is here.

3) A bit off topic but, when measuring the response of a complete guitar do you do the tapping with the strings on tuned to pitch?

I haven't found the tuning to make any difference, but anything that adds mass to the top (like saddles, bridge pins and even the strings themselves) will make a difference. If the strings are slack and the saddle is loose (for example), they can act as if they are not attached to the top. We are interested in the response of the guitar as played, so you need to accurately mimic that.

[jfmckenna]

Thank you Trevor. I really appreciate the response and your time.

Regards.