Carving site recommendations for plate tuning

[Don Noon]

6 hours ago, Peter K-G said:

We must be all in great need to repeat ourselves

 

For the record I'll repeat:

If you want them (M5/B1) to correlate it's possible up to R^2 1.0.

(Anders will say that there is something wrong if R^2 1.0 and I will say  My data and Wikipedia says that it's possible)

To which I'll repeat:  So what?

Available data shows tha particulart B mode frequencies are not critical for goodness, a particular M5 frequency is not critical for goodness, and a particular relation between plate modes and body modes is not critical for goodness.

I can believe that with careful control of humidity, arching, and diddled graduations and bass bar you can get a high correlation between M5 and a B mode... I just don't see why anyone would think it is worth the effort.

[Peter K-G]

38 minutes ago, Don Noon said:

To which I'll repeat:  So what?

Available data shows tha particulart B mode frequencies are not critical for goodness, a particular M5 frequency is not critical for goodness, and a particular relation between plate modes and body modes is not critical for goodness.

I can believe that with careful control of humidity, arching, and diddled graduations and bass bar you can get a high correlation between M5 and a B mode... I just don't see why anyone would think it is worth the effort.

While on repeating  

I would never diddle with graduation.

The mission is tonal copy and I think that if you set all the parameters the same you can come close: M5 and later B1 modes are worthless without the same weight and graduation scheme. Thickness is adjusted for the wood's density. If density is the same thickness is the same.

 

 

[Anders Buen]

22 hours ago, Peter K-G said:

For the record I'll repeat:

If you want them (M5/B1) to correlate it's possible up to R^2 1.0.

(Anders will say that there is something wrong if R^2 1.0 and I will say  My data and Wikipedia says that it's possible)

Statistically R^2 = 1 would mean that all variation in the frequency of B1- (?) could be explained by the variation in the M5 top plate frequency. We know that e.g. adding a chinrest in general will move the B1 frequencies a bit, sometimes quite a lot. Now, the variation in the top plate can´t explain that and 100% prediction from the top plate M5 frequency is not possible. We can dig up other examples too. From my large dataset about 30% of the variation maybe is reasonable. One need to do multiple regression or maybe some other type of statistics mining technique to figure out this. The input data need to be of high quality. 

Patrick Kreits book claim very high prediction power like that, but by looking at the data proivided there, and having experience with violin acoustic datasets, one can tell that they are likely to be model predictions and are not real measured data. I think I have explained where I think the fault is in that model too, somewhere here on MN.

The problem with Dons and Bissingers datasets is that they are not large enough. And sets with data from before and after regraduation are probably better. Using models from statistics, like this, probably also works better for predicting the perturbations, the expected changes, from regraduation than for a new build of a given model.

[Anders Buen]

On 9/21/2020 at 2:00 AM, Don Noon said:

I have also noticed that higher arching tends to suppress the "B" frequency range, giving a more "old Italian" tone but less overall power.  I haven't gotten to the point where I can assign the effect to the top or back plate, though.

I agree that good violins don't necessarily have to have a low B frequency band, although I primarily like the strong overtones and punch it gives to the G and D string, and don't really like the offensive power and unevenness that usually shows up on the E string.  I know Curtin thinks that the "old Italian sound" character is not necessarily preferred these days  The real problem is excessive peaks in the B range, which are not that easy to control.

Anyway, thanks, Anders, for all this incredibly detailed real-world work, which I find far more useful that just theory.

Thanks back Don, I follow your updates and appreciate your comments and input very much.