Melodeon Reed Transients – An Indicative Study

Over the last year, I have been looking at the acoustics of the Charango, a small stringed instrument from the Andean region of South America, using a Polytec Laser Vibrometer. Now that this is all done (got a distinction in it by the way) I got permission to use the equipment for around 5 hours in one day to try and take some measurements of melodeon reeds. Unfortunately I did have only 5 hours and didn’t have much equipment other than the vibrometer, so I was limited in what I could do. In addition, I don’t have MATLAB on this computer, so I am limited in the analysis as well. I hope however that the following is of interest to those intrigued by how the box works. I haven’t concluded anything game-changing, however I hope that this study shows that within 24 hours, interesting experiments can be carried out. If there is anyone out there attached to a research centre who is interested in studying these instruments then please get in touch with me – I would love to hear from you. This is not a formal report, this is not a formal study and nothing should be taken to be conclusive. I think however that it can be considered to be indicative.


A series of experiments were carried out in order to investigate the behaviour of melodeon reed transients. The movement of individually excited reeds was examined, together with the degree to which they influenced other reeds around them. The effect of isolating the block from the table was examined. It was concluded that the time scales are such that melodeon reed transients could affect reed response. In addition, it was shown that partner reeds are coupled and that the palletboard may have some effect on both the sound and the response of reeds. Finally it has been concluded that although more work is needed to investigate these instruments, interesting data may be gathered in a short space of time.

Motivation and Introduction

Accordion and melodeon reeds have not been well studied in the past, so aspects of their behaviour are not well understood. It is believed that studying the instrument would enable better instruments to be built, which is the principal motivation behind doing so. The main motivation behind this study however is to demonstrate that interesting experiments can be done in a very short space of time with the right equipment.

There are two main mechanisms by which a melodeon reed may move. The first is self-excitation, where the reed vibrates due to a constant airflow imposed by the bellows. In order to investigate this, a compressed air supply would need to be sourced and this was unfortunately impractical in the time allotted. The second mechanism is vibration without airflow – where a displacement is imparted to the tip of the reed and the reed left to vibrate freely. This is relevant to a real instrument because once the pallet is closed and the airflow interrupted, the reed will have a displacement. It is this transient movement of the reed which is to be examined. If the transient is long, i.e. if the reed has little damping, then the reed may still have some residual energy when it is next sounded. This would mean that the response could be improved, as it will take less time to excite the reed fully when airflow is next passed through the reed. In addition, if one reed can impart vibration to another, then the response of a whole box may be improved.

Equipment and Method

A Polytec Laser Vibrometer was used to measure directly the velocity of a particular reed from the author’s D/G Saltarelle Irish Bouebe. Reflective tape was attached to the part of the reed focussed upon by the laser, such that a good signal was returned to the device. The reed was excited in a consistent fashion by using a pencil to displace the reed by a set amount (henceforth ‘pinging’). This pencil was shaped such that when removed, it allowed the reed to vibrate freely. The reedblock was clamped to a rigid table, supported by foam (which has a high damping ratio) both under the block and under the clamp.

A number of experiments were carried out. Most did not give any meaningful data and there were some problems with repeatability. Below is a small subset of experiments from which some conclusions may be drawn.

A single reed – the low F# reed on the D row – was pinged and measured using the laser vibrometer. A MATLAB program written by Professor Jim Woodhouse was then used to gain the maximum amplitude, the ‘half life’ (time taken for the vibrations to die down to half of their original amplitude) and the ‘10% value’ (time taken for the vibrations to die down to 10% of their original amplitude). This experiment was repeated three times.

The movement of the same reed was then measured when the other F# reed was pinged. The Bouebe is a two voice instrument, which means that there are two reeds for every note, tuned slightly apart so as to create ‘beating’ in the overall sound. This other F# reed is situated on the other side of the reedblock from the original reed. This was repeated three times.

These two experiments were then repeated with the foam removed. The object of this was to try to simulate the existence of a palletboard (the board which the pallets rest on and to which the reedblocks are clamped), as opposed to the reedblock being tested in isolation.

Results and Analysis

Fig 1: F# reed pinged and measured whilst supported by foam:

The half life of Figure 1 is 0.108s and the 10% value is 0.417s. If we assume that we are playing a reel at say 105bpm (i.e. 1.75 bps) then one quaver beat takes approximately 0.14 seconds. So for repeated, staccato notes, the reed is probably still vibrating noticeably, meaning that it could conceivably have an impact on the response. Of course, merely saying that the time scales are adequate does not mean that this effect is noticeable – some rather more complex experiments would need to be performed to ascertain whether this is so.

Fig 2: F# reed measured due to pinging the opposite F# reed whilst supported by foam:

Figure 2 shows that pinging one F# reed, as expected, excites the other one. Indeed, the effect of ‘beating’ can be seen in the movement of the reed, clear evidence that the two reeds are coupled. This is something which has been noted by tuners – opposite reeds can affect one another. The maximum amplitude of Figure 2 is very roughly 6% of the amplitude of Figure 1.

Fig 3: F# reed pinged and measured whilst not supported by foam:

Figure 3 shows little difference in shape when the foam is removed, but perhaps a slight difference in amplitude and time. It is not clear from those two graphs however whether this is within the variability of the experiment.

Fig 4: F# reed measured due to pinging the opposite F# reed whilst not supported by foam:

Figure 4 however is very different from Figure 2. Again, the effect of beating can vaguely be seen, but the amplitude is much lower – to the extent that the effect of the noise floor becomes noticeable. The maximum amplitude is roughly 1.2% of the amplitude of Figure 3.

Fig 5: Graph showing the maximum value, half life and 10% value of pinging and measuring the F# reed with and without foam:

Figure 5, finally, shows three sets of data with foam contrasted to two sets of data without. There is a slight difference between the data with foam and the data without, however it is not clear whether any possible effect of this is noticeable.


The evidence shows that there is some residual kinetic energy in the reeds for time scales found in playing. This opens up the possibility of residual vibrations playing a part in the responsiveness of the box. This is something which is definitely noticeable on the bass end, where the reeds are much longer and heavier, however this research shows that it could also be noticeable on the treble end. Experiments not presented here due to lack of space show that this effect decreases significantly as reeds get smaller. However, responsiveness is less of an issue at those length scales, as there is less mass to move.

In addition, the evidence shows that the two reeds associated with a particular note are coupled in their vibration. It was intended to examine whether the reed pinging could significantly influence other reeds around it in addition to its partner, however the degree to which this occurred was small. This mechanism might be more noticeable when the reeds are self-excited, but this analysis does not show that it is significant.

When the foam was removed, the degree to which partner reeds influence each other was reduced. It is thought that this is due to vibrations passing through the block into the table, where they dispersed due to the large volume. This shows that vibration from reeds is passed to the palletboard and as such, it could theoretically have an influence on the movement of the reeds. Whether this is noticeable is another matter. An interesting experiment to carry out, which I was hoping to do but ran out of time, is to look at whether removing the foam allows reeds on an adjacent block to vibrate. My suspicion is that even when the reed is moving in its self-excited phase – i.e. when air is passing through it – and even when the rather thick, high impedance table is replaced with a relatively thin piece of plywood, as in a real instrument, this effect is negligible or non-existent. Furthermore, too much energy being taken away from a reed is not necessarily desirable and too much vibration of the palletboard can lead to some odd effects.

What this analysis does show then is that the existence of the palletboard does have a definite impact on the movement of the reeds. If I were to hypothesise, I would say that it is more important to retain as much energy in the reed as possible, rather than allow some of it to pass to other reeds, and that the negative effects of soundboard vibration would quickly outweigh any positives. So I would probably advocate a stiff palletboard, i.e. one made of aluminium. This has the added benefit that you can make it thinner, meaning less diversion of the air through to the blocks.

This analysis also shows that there could be a change in sound between clamped reedblocks as opposed to glued. Clamped reedblocks have some gasket material under them, which could prevent vibrations from moving across the palletboard. However, I suspect that the real difference in sound is due to the reedblocks being held more rigidly, meaning that they are prevented from vibrating appreciably. I will talk more about this in a later post.


  1. It is possible that residual kinetic energy in reeds once the pallet has been closed could noticeably affect the response.
  2. Partner reeds are coupled with each other.
  3. The existence of a palletboard can change the sound, decreasing the maximum amplitude and increasing the time taken to die down. In addition, it can decrease the degree of coupling between partner reeds.
  4. It is possible to get interesting, if not conclusive, results in under 5 hours of experimentation.


I would like to thank Mr Ioannis Mitsos for allowing me use of the equipment whilst he was in the final stages of doing his PhD testing. It was very generous of him to allow me to use it and I am very grateful.

I would also like to thank Professor Jim Woodhouse, for granting me permission to use the equipment in the first place and for teaching me how to use it back in October 2011.


One thought on “Melodeon Reed Transients – An Indicative Study

  1. Pingback: When thinking about instruments… | Music and Melodeons

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s