Off topic, I know, but Tim quoted Trevor as saying:
"The Red (especially) and the Sitka are much denser (>30%) than the Engelmann/Lutz, so to hit what is essentially a modal frequency target (which is what my formulae do) they have to be made stiffer, therefore thicker, therefore heavier, therefore thicker...."
I thickness to a target stiffness, more or less, and let the frequency float (more or less). Denser softwood does have higher Young's modulus in general, so I usually make those tops thinner. They do end up heavier, of course, but not all that much heavier unless they're outliers, with lower than expected E values along the grain relative to the density. I don't see enormous deviations in frequencies.
I do my stiffness and damping tests the old way. Plates are set on foam pads and a signal generator is used to find the fundamental bending modes along and across the grain. The motion of a small piece of iron stuck to the top, usually in the center of the end, is read with a magnet and coil driving a DVM (MV-AC) to find the peak frequencies and 3dB down points. These are then used to calculate the E and Q values in the usual way.
This takes a bit more time than simply tapping and running an FFT on the computer. One thing that repays it is that you can see when the bending mode shapes are distorted, due to another mode close in pitch, or changing runout in the piece, for example. These things don't always show up in a tap spectrum. Basically, when the node lines are curved you're looking at a combination of properties of two stiffness along and across the grain, so the numbers are not exactly what you think. Looking at the modes won't disentangle them, of course, but at least you know when things are not what they might seem.
Also, it's wise to keep in mind that the equations we use to calculate the E values are limited in accuracy. For one thing, of course, you can't get results more accurate than your least accurate measurement. I use a metric dial caliper for thickness, and can interpolate to maybe .5mm, or 10% of a 5mm thick piece. On top of that, the equations themselves are simplified to make them solvable. According to one researcher they actually only give you a result that it within about 10% of the 'real' value, even if your measurements are accurate. There are other issues as well.
The point is that those numbers you get are better than nothing, by a long shot, but not etched in stone. The more care you put in, the better and more useful the numbers are likely to be, but they will never be definitive.
In light of that one thing I was told when I started out in the measurement game takes on more meaning. What's going to hurt you in taking these measurements will be really egregious narrow-band deviations in the apparatus. That is; if the mic has a very strong peak or dip at a particular frequency, and the resonance you're looking at is in that range, you're going to get bad data. However, any mic that does that is probably going to sound terrible when you use it to record normal speech or music. So; just record something and play it back. If it sounds OK, then the mic is probably good enough.
|
). ![[:Y:]](./images/smilies/smiley20.gif "Thumbs Up")
