Bi-Directional Reed -- Experimental Results

[rlgph]

Some months ago i posted a proposal for a bi-directional concertina reed. (To be clear, i am using "concertina" here generically; it could actually be an accordion reed.) I have copied below a paragraph from my original post to describe the proposed reed, along with a couple of clarifying drawings:

 

Take a usual (though it may need to be thicker or thinner in practice) reed frame and file part of the inside end down at an angle, as shown in the cross section of Fig 1 in the picture below. Repeat with a second frame in such a way that it is a mirror image of the first. Now sandwich a flat reed tongue between the two frames as shown in Fig 2. The left end of the reed is clamped between the two frames, and the other end is free to oscillate. Note that when the reed is in its equilibrium position (as shown in Fig 2) there is an air passage between the top and bottom at the free end of the reed tongue. The setting of the free end of the reed is such that it comes very close to the top and bottom frames at the end of the sideways V so that it will effectively cut any air flow occurring from top to bottom or vice versa. (In actuality, the V might need to be shaped like and arc.)

I later bought a "for-parts" Chinese concertina for experimenting with the proposal. I propped one of the buttons of the right side open, and turned the end around so that the reed box was easily accessible. The two photos show the apparatus. The nearest reed in the first photo is the only one whose pad is open. That's where the prototype bi-directional reed (shown more clearly in the second photo) was located.

 

 

 

 

The prototype shown at the right in the second photo consists of one of the accordion reed units with a second frame cut from another reed unit stuck to it with masking tape. Both frames were filed slightly to produce the V shape seen in Fig. 2 above. Prior to adding the additional frame, i bent the tongue back so that it was as flat as i could get it. (I.e., not bent upward at the end.) Insuring that the reed freely moved through both frames was the most difficult part of the experiment, and until i had the idea of turning the end around to expose the reed box, i had no success in getting everything aligned properly.

 

I'm going to put the actual sound files in the post just below this one.

[rlgph]

Two mp3 files are given below. The first one is the reed without the second frame added, but with the notch filed in the end and the tongue flattened. It produces a sound only on the bellows draw. The second mp3 is the recording of the prototype which adds an additional frame to yield the bi-directional reed, which produces a sound for both directions of the bellows motion.

 

Unidirectional reed: http://rlgreene.net/BiDiReed/UniDir.mp3

 

Bidirectional reed: http://rlgreene.net/BiDiReed/BiDir.mp3

 

To visualize the difference in the two cases, i have made a screen shot of the two recordings shown side-by-side as displayed by Audacity. See below:

 

 

The one on the left is the unidirectional reed; the one on the right is the bi-directional reed.

 

I have a few preliminary comments about the results:

 

1. The volume of the bidirectional reed is less, presumably because the rate of air flow is less. I suspect that fine tuning of the V notch or other filing of the frames could largely correct this difference.

 

2. There is an audible and visible difference in quality of sound on the compression part of the cycle of the bidirectional reed compared to the draw. That is the part of the cycle where the reed initially moves into the added frame. Perhaps this is not surprising because the reed tongue is physically attached to one frame (which presumably was made or adapted to the specific tongue), but not attached to the added frame, nor was that frame made especially for the tongue. Moreover, i took no special care to file both frames equally. In brief, my prototype is not truly symmetrical.

 

3. Looking at the bi-directional signal on the right, you can see there is a substantial initial transient of the compression part of the cycle that eventually settles down to a steady state signal shortly before the direction of the bellows reverses. I don't know why this occurs, but it is probably the major cause of the qualitative sound difference between the draw and compression parts of the cycle.

 

4. To my ear, the uni-directional and the bi-directional reeds sound pretty much at the same pitch. I, in contrast to Tom, had predicted an octave difference. My thinking was that if the the air flow is cut off twice per cycle (by each frame) rather than once per cycle, the sound frequency should double. I have to think about this some more, because i'm still not convinced by his explanation. I wonder if one of the frames of the prototype is not cutting the airflow effectively during the oscillation of the tongue.

 

I look forward to your thoughts.

Edited January 28, 2016 by rlgph

[Wolf Molkentin]

To my ear, the uni-directional and the bi-directional reeds sound pretty much at the same pitch. I, in contrast to Tom, had predicted an octave difference. My thinking was that if the the air flow is cut off twice per cycle (by each frame) rather than once per cycle, the sound frequency should double. I have to think about this some more, because i'm still not convinced by his explanation. I wonder if one of the frames of the prototype is not cutting the airflow effectively during the oscillation of the tongue.

Good to have some results and graphs now!

 

As to the details, albeit admittedly not fully understanding the tone generation of a free reed I'd rather doubt there could be any "doubling" the frequency by providing more cuts. Otherwise any free reed would already produce a tone one octave above its own resonant frequency (as to be heard without any frames), wouldn't it? and in fact it doesn't, does it?

 

Best wishes - Wolf

[rlgph]

Good to have some results and graphs now!As to the details, albeit admittedly not fully understanding the tone generation of a free reed I'd rather doubt there could be any "doubling" the frequency by providing more cuts. Otherwise any free reed would already produce a tone one octave above its own resonant frequency (as to be heard without any frames), wouldn't it? and in fact it doesn't, does it?Best wishes - Wolf

My understanding is that the sound from the reed itself is almost inaudible. Rather, the air that moves through the reed assembly is essentially cut off during the time the reed is within the frame, and allowed to flow at other times. This results in a series of "puffs" of air, which are audible. The time span during which the flow is cut off occurs once every complete oscillation of the tongue in a conventional reed, so that the dominant frequency that we hear is the same as the frequency at which tongue oscillates.

[Chris Ghent]

How is the reset between one direction and the other? Is there a lag or does the note resume instantly?

[Wolf Molkentin]

My understanding is that the sound from the reed itself is almost inaudible. Rather, the air that moves through the reed assembly is essentially cut off during the time the reed is within the frame, and allowed to flow at other times. This results in a series of "puffs" of air, which are audible...

Albeit we're in fact not hearing a sort of amplified sound from the reed itself the audible sound appears nevertheless to be basically of the same pitch (with external factors and influences causing only minor shifts).

[rlgph]

How is the reset between one direction and the other? Is there a lag or does the note resume instantly?

The lag is what you can hear in the BiDir.mp3 recording or see in the Audacity output; i haven't massaged the sound files except to cut clutter associated with starting and stopping the recorder at the beginning and end of the recording.

[rlgph]

Albeit we're in fact not hearing a sort of amplified sound from the reed itself the audible sound appears nevertheless to be basically of the same pitch (with external factors and influences causing only minor shifts).

It's the same pitch because the air flow is cut significantly once per cycle of the tongue, and thus has the same frequency as the tongue's oscillation. If we could contrive to cut the air flow twice per cycle (each cut of the same duration and effectiveness), the frequency of the "puffs" would be twice as high. I thought the extra frame would do that, but apparently the frame on the higher pressure side of the reed assembly is not cutting the air as effectively or with the same duration as the frame on the lower pressure side. (There is an asymmetry there as i discussed somewhere in the long string of previous posts related to this issue. The experimental results suggest to me that the asymmetry is larger than i expected.)

Edited January 29, 2016 by rlgph

[JimLucas]

My thinking was that if the the air flow is cut off twice per cycle (by each frame) rather than once per cycle, the sound frequency should double.

 

I see that you changed your analysis while I was writing up my own interpretation, which follows. So it seems we're now in closer agreement than when I quoted the above.

• Firstly, a full cycle of an oscillation -- even of a simple sine wave -- has two zeros, not one. This corresponds to the motion of a standard reed, where the airflow is "cut off" as the reed passes through the frame from its maximum distance on the one side to its maximum distance on the other, then again as it returns to its "starting" maximum.
• Note that if there were no bellows-induced airflow, the direction of reed-induced flow would be opposite on the two different sides of the reed, corresponding to the positive and negative peaks of a simple oscillation. It is this oscillation which determines the basic frequency (and even the harmonics?) of the particular reed.
• The otherwise-steady bellows-induced airflow is modulated by this small-amplitude oscillation, thereby amplifying the small oscillation. This is the same principle as electronic amplification by triode vacuum tubes (where the reed corresponds to the grid) and transistors (where the reed corresponds to the gate).
• This also explains why your bidirectional reed doesn't double in frequency. The oscillation of the reed isn't really any different... only its ability to start (and continue?) oscillating under both directions of bellows pressure. The air flow is still "cut off" only once each time the reed passes through the ("double") frame, i.e., twice per cycle. It isn't "cut off" once for each half of the frame, because the "V" section is so narrow and the reed is moving so fast (at maximum velocity) there that it doesn't spend enough time there to "cut on" in any meaningful sense. (It certainly doesn't change direction in that interval.)

[rlgph]

 

My thinking was that if the the air flow is cut off twice per cycle (by each frame) rather than once per cycle, the sound frequency should double.

I see that you changed your analysis while I was writing up my own interpretation, which follows. So it seems we're now in closer agreement than when I quoted the above.

• Firstly, a full cycle of an oscillation -- even of a simple sine wave -- has two zeros, not one. This corresponds to the motion of a standard reed, where the airflow is "cut off" as the reed passes through the frame from its maximum distance on the one side to its maximum distance on the other, then again as it returns to its "starting" maximum.

Not necessarily. A sine wave with a vertical offset is still a sine wave. I'm not just being cute here because the air flow through the reed assembly provides a vertical shift that vastly dominates the variation due to the tongue without the air flow.

• Note that if there were no bellows-induced airflow, the direction of reed-induced flow would be opposite on the two different sides of the reed, corresponding to the positive and negative peaks of a simple oscillation. It is this oscillation which determines the basic frequency (and even the harmonics?) of the particular reed.

The oscillation of the tongue does determine the basic frequency, but not in the sense that i interpret your statement to mean. The tongue controls it because it alternately allows the air flow and cuts it. The frequency of the sound is determined by how many times the air flow is stopped/started per cycle of the tongue. In a standard reed this occurs once per cycle.

• The otherwise-steady bellows-induced airflow is modulated by this small-amplitude oscillation, thereby amplifying the small oscillation. This is the same principle as electronic amplification by triode vacuum tubes (where the reed corresponds to the grid) and transistors (where the reed corresponds to the gate).

I'm changing my original comment here. I don't think the triode analogy applies here. Here the air flow is modulated by the motion of the mechanical oscillating tongue, whose energy is comparable to that contained in the moving air in that region.

• This also explains why your bidirectional reed doesn't double in frequency. The oscillation of the reed isn't really any different... only its ability to start (and continue?) oscillating under both directions of bellows pressure. The air flow is still "cut off" only once each time the reed passes through the ("double") frame, i.e., twice per cycle. It isn't "cut off" once for each half of the frame, because the "V" section is so narrow and the reed is moving so fast (at maximum velocity) there that it doesn't spend enough time there to "cut on" in any meaningful sense. (It certainly doesn't change direction in that interval.)

If the tongue were oscillating symmetrically so that it moved as far into the frame on the high pressure side as it does into the frame on the low pressure side, it would stop the air flow twice per cycle, resulting in the air flow pulsing with twice the frequency of the tongue. However, the oscillation is not symmetric because of the dominating air flow due to the pressure difference. I did say in that previous thread that it was possible that the tongue would not move far enough into the frame on the high pressure side to provide an effective cutoff in that part of the cycle. I did not expect that would be the case, but the experimental results indicate that my expectation was wrong.

 

In any event, i think the important point is that the double frame allows the tongue to start oscillating, regardless of which direction the air flow comes. Furthermore, from a practical point of view, it's a good thing the frequency of the pulsing air isn't doubled so that the reed tongue doesn't have to be made larger to compensate.

Edited January 29, 2016 by rlgph

[ttonon]

Rlgph, I congratulate you on your success so far, with a novel way to construct a free reed. I think you’ve demonstrated that it's possible to make an “asymmetric” free reed into a symmetric free reed, with the ability to voice upon either air flow direction.

 

I think most of us here realize that, although this is a good start, there’s need for improvement, if this design is to be widely utilized. Before getting into too much detail, can you explain more about how you fed air to these reeds? Was the concertina disassembled and was the apparatus fixed at one end, as pictured, or did the sounds come from an assembled instrument, with you holding the bellows with both hands? Was the microphone mounted securely so that its position relative to the reed did not change during voicing? Where was that position?

 

In the Audacity traces of the Bi-reed, could you identify which set of traces correspond to push and pull of the bellows? You indicate that the second, fourth, sixth and eighth traces with the larger amplitude are compressions, but could you be clearer about that? For the Uni-reed, was the voicing done with positive (push) or negative (pull) pressure in the bellows?

For the Uni-reed, Audacity gives the frequency of the fundamental partial as 292.0 Hz, and that trace indicates that the bellows pressure was not very constant during voicing. I point out that Adaucity can give small variations in fundamental frequency, depending upon the choice of the number of sample points. The more points, the more accuracy, if the trace is strictly periodic. And if I may suggest, that in further experiments, you use some kind of mechanized air flow apparatus.

 

For the Bi–reed, the odd number traces seem to be louder, despite the visual indication of lower amplitude. However, I notice that traces 2, 4, and 8 contain an audible click, which I assume is noise, and trace 6, which doesn’t have this click, does not show such a large “transient” amplitude. What about that click? I don’t know if this is really a transient, because the air delivery may not be constant, but if it is a transient, at about 500 msec, it’s quite long as far as musical instruments go. Here, I think Chris’s question is pertinent. It might be helpful if you performed a “bellows shake,” trying to determine the minimum time required to establish the musical tone.

 

With the odd number traces, I also notice a slight drop in pitch towards the end. From Audacity, I get for the third trace, a pitch of 296.1 Hz for the steady part and 293.4 Hz for the end part, a drop of almost 3 Hz. Small but noticeable, and it’s puzzling how that could occur. Usually the pitch of a free reed will drop as bellows pressure is increased, and there is shown no increase in volume. Do you have any idea what would cause this drop in pitch?

 

To me, the most interesting aspect of this experiment is the hollow, feathery sound that accompanies the odd numbered traces, and this may provide a hint at how to improve the musical tone of the free reed. For years, I’ve wondered how to improve the sound of the western (asymmetric) free reed (WFR). To me, the Asian (symmetric) free reeds have much more interesting sound, simply because they incorporate a resonating air column for their voicing, and the WFR does not. For instance, you can find the sound of the Bawu on the internet, and because of its mounting in a tube, it has the hollow tone of the clarinet.

 

But it’s not just the tone itself. I think it would be an improvement to the tone of the WFR if each reed can be made to have its own signature sound, such that it could stand out better among other musical sounds, and in particular, among its own musical tones (as in playing chords).

 

Looking at it from theory, the chief characteristics of the sound spectrum of the WFR are the lack of a distinctive start transient and the presence of a very large number of harmonics (partials whose frequencies exhibit very nearly exact whole number ratios). I believe modification of either one of these features would improve the musical tone of WFR instruments.

 

The WFR produces sustained (steady) oscillations in the musical tone. Other musical instruments that also produce sustained tones, such as the beating reed instruments, the bowed string instruments, and the Asian free reed instruments, do not generally have such large numbers of overtones. I think their tones have more distinguishing character than those of the WFR, which have such a large number of partials with coincident frequencies that perhaps our hearing is overwhelmed by them.

 

The individual notes of musical instruments incorporating strong start transients, such as stringed, and other, instruments that are hammered or plucked, can easily maintain their own identity among other, similar tones. In fact, the Hammond B3 electronic organ, I think, became such a jazz favorite largely because of its distinctive start transient.

 

The instruments that produce their entire tones by a transient, such as the piano, guitar, xylophone, etc., contain inharmonicity in their partials, because of nonlinearities in their sound sources. Some of these even contain relatively large numbers of partials, but these partials can be discerned among themselves, perhaps because the partials do not exactly coincide in frequency. So again, maybe our hearing system doesn’t get confused.

 

Getting back to the tone in traces 1, 3, 5, and 7, perhaps improvements to the tone of the WFR can be made by introducing a small amount of noise into the tone, such as what I detect here, which would cause inharmonicity - perhaps a good thing. Concerning the other chief feature of the WFR sound spectrum, perhaps a strong start transient can be introduced by a mechanized finger that plucks the tongue the instant the key is pressed. Of course, complex construction is a bad thing, so any attempted improvement must be very simple.

 

Regards,

Tom

www.bluesbox.biz

[rlgph]

Tom,

 

The recordings i made were with the apparatus as pictured in the first posting of this thread. As you can see from the empty chamber next to the bidirectional reed, the hidden part of the reed assembly is connected by an air passage directly to the inside of the bellows; the outer frame that you see is at the ambient pressure. The recordings were initiated by pulling on the bellows, so the inner portion of the reed assembly is the one that sounds first. That is the part that the tongue is attached to, and so it's not surprising that the 1st, 3rd, etc. signals look more similar to those from the unidirectional recording.

 

My guess is that the large transients for the push part of the cycle are related to the fact that the tongue is not directly attached to the outer frame, and the two frames are attached to each other only with masking tape on the sides. There's probably some vibration occurring between them that may be producing some extraneous sound. Also, i made no attempt to insure that the two frames were equally filed at the free end of the tongue. So, while your speculations could well be correct, one should make recordings of a truly symmetric bidirectional reed to see if the transients persist.

 

I freely admit that i am a sloppy experimentalist. I have neither the equipment nor the inclination to take this further. As a (retired) theoretical physicist, i found the bidirectional reed to be an interesting theoretical speculation. What i did experimentally was simply to convince myself that such a reed would work in principle, since it did not appear to me that anyone else with greater expertise and appropriate equipment had any interest in testing the idea.

 

Although i hope someone else will investigate further and answer your questions and others, i will be surprised if that is the case since (in my understanding) the large majority of concertinas are anglos, for which a bidirectional reed is of no interest.

 

Best wishes,

 

ron

[JimLucas]

Something significant, though I'm afraid not technically useful, has appeared on eBay, as reported in this topic.

[rlgph]

Something significant, though I'm afraid not technically useful, has appeared on eBay, as reported in this topic.

Thanks for providing the link. i subsequently followed the link provided by Stephen Chambers; it was interesting reading about the variety of approaches to a bi-directional reed.

Edited February 20, 2016 by rlgph