Project performed in
2017, P2I6 mécatronique at INSA Lyon , France
12 students
Xylophone demonstration with the robotic orchestra
Manufacturing and Multi-Modal Tuning of Xylophone Bars
We use 40x10 mm² aluminium to manufacture the xylophone bars. The
material is AU4G1 (Aluminum 2024 for USA).
Prismatic bars vibration is anharmonic: f2/f1=2.76 and f3/f1=5.40
We need to remove material at the center of the bar to have harmonic
bars.
We want to have: f2/f1=4 and f3/f1=9.8
This is the most common tuning for Marimba. The tuning for xylophone is
different.
Step 1:
Xylophone bars shape calculation with Matlab and Zhao inspired code
Graphical result for 10 mm thick bars. 19 bars calculated from C5. Here
is the modified code Here
is the original Zhao code
The code is useful for Step 1 and step 6.
The code is very efficient with aluminium because the physical
properties of aluminium is very well controled.
The code exports the shape as csv
The code is a bit messy . Have a look to
Rich code on Supermediocree; Mayb it's easier to
understand....
Step 2:
csv import to the CAD software Solidedge
Step 3:
Shape manufacturing with CNC. The CNC program as generated with NXcam
from the Solidedge CAD software.
Step 4:
Drill hole with salt method
Step: 5 Experimental frequencies analysis
We use the following applications:
Da tuner:(Android)
Very good for fundemental but not harmonic
It is possible to hit aluminium bar near the node to have the 1st
harmonic intensity higher than the fundamental.
Here is a demonstration:
FFT Spectrum:(Android)
Good
for quick harmonic vizualisation but the sample size are too small. The
harmonics frequency precision is not good enought. It can't be used for
fine tuning.
Here can be seen the fondamental (522 Hz) and first harmonic (2085 Hz)
Audacity:(Windows/Mac/Linux)
It is possible to specify a 32768 sample size and the harmonics can be
seen with a very prcecise frequency.
A
quicker method is to use Datuner and hit the bar near the hole to have
the first harmonic frequency instead of the fondamental.
Step 6:
Fine tuning position calculation with matlab. White columns are the frequencies after milling and before tuning. Blues columns is the position (pos) and the depth (depth) where material must be removed.
Theorical frequency
Measured
Offset
%
File
Note
f0
f1
f2
f0
f1
f2
f0
f1
f2
f0
f1
f2
pos
depth
C5
523,3
2093,2
5128,34
525
2112
5147
1,7
18,8
18,66
0,32
0,9
0,36
76,47
0,79
554,42
2217,67
5433,29
558
2240
5426
3,58
22,33
-7,29
0,65
1,01
-0,13
82,54
0,42
D5
587,38
2349,54
5756,37
592
2394
5729
4,62
44,46
-27,37
0,79
1,89
-0,48
74,18
0,94
622,31
2489,25
6098,66
628
2513
6055
5,69
23,75
-43,66
0,91
0,95
-0,72
81,16
0,31
E5
659,32
2637,27
6461,3
664
2661
6407
4,68
23,73
-54,3
0,71
0,9
-0,84
78,22
0,34
F5
698,52
2794,09
6845,51
721
2862
6808
22,48
67,91
-37,51
3,22
2,43
-0,55
77,83
0,45
740,06
2960,23
7252,57
771
3048
7210
30,94
87,77
-42,57
4,18
2,96
-0,59
76,21
0,5
G5
784,06
3136,26
7683,83
818
3231
7638
33,94
94,74
-45,83
4,33
3,02
-0,6
74,04
0,51
830,69
3322,75
8140,73
869
3431
8333
38,31
108,25
192,27
4,61
3,26
2,36
71,93
0,52
A5
880,08
3520,33
8624,81
905
3588
8625
24,92
67,67
0,19
2,83
1,92
0
70,43
0,36
932,41
3729,66
9137,66
963
3812
8945
30,59
82,34
-192,66
3,28
2,21
-2,11
68,42
0,39
B5
987,86
3951,44
9681,02
1014
4016
9427
26,14
64,56
-254,02
2,65
1,63
-2,62
66,48
0,31
C6
1046,6
4186,4
10256,68
1078
4252
9628
31,4
65,6
-628,68
3
1,57
-6,13
65,59
0,32
Step 7
The metal is removed at File "pos" acording to the matlab calculated
table with a half round file. The depth is controled with Datuner on f0
because too difficult du measure with a micrometer.
When f0 is tune on the theorical frequency, f1 and f2 are tuned too!
Half round file
Resonators
Cut
view of resonnators: Resonnator are made with a Ø40 mm PVC pipe.
Stoppers in PVC(blue) are CNC machined. A O ring (not shown
on
the figure) allows an easy adjustment.
We
play a sine frequency at 1 cm from the resonnator. We adjust the
stoppers position to the theorical frequency of the bar (the bars were
not ready at this step of manufacturing).