Blowing machine

Whistle
1

So, Terry, we’re just shy of 11 years on this. How’s the blowing machine going?

More seriously, a blowing machine would make a huge difference in our ability to model whistles. Ideally, this blowing machine would deliver a controlled volume flow (ml per second, say) to the whistle. I’m guessing that we’d need a range around 100 ml/sec through 1000 ml/sec. A university lab in France has a blowing machine that delivers a controlled air pressure, but for whistles the volume flow has a more direct impact on what the whistle is doing. (What really matters is the air speed across the window, but volume flow is close enough.)

Does anyone know how I can buy or build such a device on a limited budget?

2

If I wanted to construct a device like you describe it would be fairly easy. You need an appropriately sized blower with some form of stable variable speed control (VFD, ECM motor, DC motor with speed control…), duct transitions from the blower outlet down to whistle mouthpiece connection. A length of straight duct at a reasonable constant diameter to setup a flowmeter (I would make one from a pitot static probe and inclined tube manometer, but you could likely also use an orifice meter with a similar manometer).

3

Heh heh, what Latticino said. Although I must say my first instinct was more organic. Burns night is coming up. Hang around the car park outside your nearest Burns club until they have the Piping in of the Haggis. When the Piper staggers out later, kidnap them.

I’m wondering if you need a full blowing machine, or could get away with considerably less? You for example are a blowing machine. Would it be enough to be able to measure your airflow into the whistle?

A consideration is what will you be doing while the machine is blowing? If it requires recording or measurement of sound or sound volume, you’re going to need a silent blower. Or storage. That could be as simple as a inner tube with a board and a few bricks on it to set the pressure.

If you could outline how the experiments might work, we can possibly come up with some low cost ideas. I do have some resources here that I may be able to test out approaches or at least needs before you go any further.

The figure 1000mL/Sec is scary - that would be the equivalent of running out of breath after about 6 seconds!

4

An ordinary hairdryer should deliver a pretty stable airflow, and can blow unheated air. Perhaps in combination with an air velocity meter and some kind of tubing?

5

But too noisy for acoustic purposes, I should think.

6

I’ve thought about doing something similar: making a setup to allow sound recordings at known pressures + flow rates, allowing quantitative comparisons among brands.

Well, . . .

1)The mouthpiece of a whistle is essentially a “fixed orifice”. There will be a constant relation between pressure + flow (with, possibly, some variation due to temp, density, humidity).

  1. One “off-the-shelf” way to get steady, quiet flow would be to use charged-up air compressor and a pressure regulator.

  2. There are off-the-shelf flow sensors, mostly aimed at respiratory-therapy. ~$100 for the sensor itself.

  3. One question I have is what role the human anatomy (mouth cavity, throat, lung volume) have on the “total acoustic field”.

Entirely doable. Some expense required. Would take some of the mystery out of “back-pressure”.

7

OK, did a bit of playing around “after work”. I have a compressor in the workshop feeding a line that runs around the place, with convenient outlets along the way. It’s the usual arrangement of compressor followed by pressure regulator, which also contains a chamber for dealing with condensed air humidity. I took a line from one of the outlets to another similar pressure regulator (set lower again), then to a flow regulator (think like a water tap but with a finer thread to make fine adjustment easier). So now I have regulated pressure and flow available to me, and a convenient lever valve to turn flow on and off. I connected that through a flow meter of the “floating ball” variety. As you increase the flow, the ball rises further in its tapered tube.

I used some typical lab poly tube to connect that to the top of the whistle, but needed a bridging scrap of thin-walled irrigation pipe to go over both the whistle beak and the poly tube. I used electrician’s insulating tape at both ends of that piece to keep it all airtight. I could make better adaptors if pressed.

Having set the flow regulator to minimum, I fired up the compressor, and opened the valve gingerly. A slight hiss of air. As I increased flow, I got first the low octave, then the high octave to speak. Promising. I used a strip of insulating tape to cover the fingerholes, as by now I was running out of hands. (Some years back, a friend gave me a little bronze statuette of the Indian god Shiva. She felt I “needed a hand” in the workshop.) So now I could explore the flow needed to obtain the various notes on this old small bore whistle. My flow meter only goes up to 20L/Min so that’s why I stop at " >20" (greater than 20) in the list below. I do have a second one of these and could probably hook it up in parallel and add the results together if this would be any help.

D 8-10 L/Min
E 9-12
F# 9-13
G 10-14
A 10-15
B 12-16
c# 12-21
d 13-20 (oxx xxx); 14-20 (xxx xxx)
e 16 - >20
f# 17 - >20
g 17 - >20
a 18 - >20
b 20 - >20
c# >20

There is of course a lot of subjectivity in what’s acceptable on each note! But taking the average would probably be pretty safe if the aim is “where does the note play well”. I could also be guided by a tuner as the compressor doesn’t need to kick in for a long time. Interesting to watch the tuner while tweaking the air-flow regulator knob - it really gives very fine control.

I realise I’m not sure what type of whistle this is. It’s brass tube with a green plastic head. The top of the mouth end is flat, and it has a strongly formed ridge across it after the ramp. But I could repeat the test above with any of the whistles I have if needed.

Getting back to setting up to do this at home, Tunborough, the big ticket item is obviously the compressor, but depending on what you’re trying to achieve, you could probably get away with bottled compressed air. Or even a hand pump and an old car tube?

8

And curiosity getting the better of me, I went back and hooked up the second 0-20L/Min flow gauge in parallel with the first. They bob up and down in very good balance (which is promising), and I can now get up to 40L/min which for example on an A fingering gets me into the ear-shrieking third harmonic of xxo ooo, E7.

9

If the air pressure inside your mouth is an acceptable first approximation of whatever the blowing machine is intended to help quantify, and you can hold a piece of thin lab tubing in the corner of your mouth while playing, an inexpensive digital manometer might prove useful. Rigging up a water-column manometer would be even cheaper and more precise but the digital alternative is decidedly more convenient. It also has the significant advantage of indicating the temperature of air being blown into the instrument. Just be sure that it can measure the pressure range of interest.

Pipe organ tuners use pressure gauges as essential adjuncts to electronic tuners, measuring the pressure at the entrance to the pipe. When I was actively involved in the tuning and voicing of recorders, this was also the way we settled quality-control arguments.

When I started following the various discussions here about the comparison of the performance attributes of whistles of different designs and manufacture, I picked up a digital manometer to see if air-pressure measurements might also be worth noting. It was easy enough to confirm the anticipated correlation between qualitative reports of “easier blowing” and “high B too screechy” with hard numbers. There was no apparent need for full comparison but if it’s consistent with the topic of the present thread, I can easily enough provide a sample specification.

What is the alternative measurement of volumetric flow expected to reveal that cannot be adequately quantified by air pressure? If, as Tunborough says, “what really matters is the air speed across the window” how does the one parameter correlate more closely to it than the other?

10

Very cool, very neat, Terry! And so much nicer than what I tried many years back: blowing a whistle and having at the same time a tube from a home-made fluid manometer (a pvc U tube with water in it) in my mouth. The pressure readings I got were similar to your flow readings, namely the second octave notes needed about double the pressure/air flow than the note in the first octave. But what was most exciting to see is differences between whistles of different design/make, and of different key, comparing low to high whistles. If I remember correctly, it was also interesting to investigate the respective range of flow/pressure, for which a note would speak.

I was always aiming to optimise my design, so that the whistle would speak easily, with little pressure, and also have a relative great range in which a note would speak, before jumping very cleanly into the next octave.

With your setup all this could be very well demonstrated and quantified.

11

Ah, stringbed and hans, I was going to raise the matter of manometers. It would be easy to take a T junction off the tube feeding the whistle to take it to a manometer. I wondered if it might make us aware of any ‘nonlinearities’ in the whistles’ pressure vs flow characteristics.

I have both a simple handheld digital manometer and the pvc tube water type. It’s a while since I used either. I found the battery in the digital one had died, but a new battery seems to have kicked it back into life.

And the water had evaporated from the u-tube manometer. I tried to introduce some fresh water but it stopped at the bottom. Small brown blobs seemed to be preventing it going around the curve. A closer look showed that the small brown blobs have legs. I suspect we’re seeing mudwasps that never made it. I may have to lash out on a new length of poly tube!

12

Very interesting. The first few of responses highlight some additional requirements… The blowing machine must be quiet enough that it doesn’t interfere with pitch measurements, it must have continuously variable output over a wide range (which I do, but a hairdryer does not), and its output must be stable for long enough to take stable pitch measurements (which rules me out).

What I want to do is measure the output frequency of the whistle as the flow is varied. What’s the frequency at the lowest flow at which a note just speaks? What’s the highest frequency just before a note jumps to the next register? What’s the lowest frequency in the second register before the note drops down to the first register? How does the frequency vary with flow between these extremes? With this information, we can design a whistle that stays in tune with a regular increase in breath flow as we go up the scale.

Volume flow is easier to work with than pressure: the air speed across the window is directly proportional to the volume flow; all we need to know is the area of the windway exit. The relationship with pressure is more complicated, and I’m not certain there aren’t non-linearities involved, especially around a register shift.

That we don’t know yet. There is something happening, because skilled recorder players have more control over frequencies around the register shift than novices.

13

Somewhere, almost certainly on this forum, I have seen a reference to a thesis (USA university) that included research involving a pressure sensing capillary tube in the mouth of a flute or recorder player. I think it could also be connected to a microphone.

I can’t find it now, but I did find this: https://hal.science/hal-00964988 which seems to be research in the general area your folks are discussing. There is a section headed “Pressure controlled artificial mouth”.

Regarding the experiments with it (other thread), to ease the setting up of permutations of variables, would it be possible to make whistle heads in two halves, split across at the front of the block, so that by clamping them together different windway and blade geometries could be combined?

14

When I was investigating Grey Larsen’s preference for small bore, small hole flutes back in 2003, Professor Neville Fletcher and I submitted him to the capillary tube pressure measurement approach. It was either that, I explained to Grey, or we’d need to drill a small hole in the back of his neck. We found that Grey’s “East Galway” playing style uses significantly higher mouth pressures than the usual classical flute players Neville had been measuring. Plus a tighter embouchure and jet offset to push more of the energy into the partials.

I wouldn’t expect to find that lung, mouth cavities etc had too much of a bearing on whistles, as they are decoupled from the instrument by the duct, but it wouldn’t be hard to interpose some suitable sized cavities in the blowing machine line to see what we find. Lungs are big - the typical six litres of lungs is about 1.5 US gallons. By comparison, mouths are very small, typically 70mL for men and 55mL for women. But then there’s a lot of stuff between lungs and fipple.

15

The hairdryer could be made variable by spilling some of its output, but it would still be very noisy.

What I want to do is measure the output frequency of the whistle as the flow is varied. What’s the frequency at the lowest flow at which a note just speaks? What’s the highest frequency just before a note jumps to the next register? What’s the lowest frequency in the second register before the note drops down to the first register? How does the frequency vary with flow between these extremes? With this information, we can design a whistle that stays in tune with a regular increase in breath flow as we go up the scale.

OK, I did a little playing with this in the current setup. Tuning the whistle (by pulling out the head) so that A (xxo ooo) was at 440 in the middle of its comfortable playing range, I then looked at xxx xxx. I had previously reported that as playing OK between 8 and 10L/min, but adding the tuner meant I could now see this is quite flat, starting at about -60 cents around 10L/min, needing 12L/min for 0 cents, and breaking at +10 cents at 13 L/min. It then jumped to the 2nd partial -55 cents at 14L/min, 0 cents at around 22L/min, and breaking around +10 cents at a slightly higher pressure.

I had gone back to the single 20L/min flow meter, to maximise precision, but then found I needed to go over that. This system would benefit from easily switchable flow meters!

Oh, and I did note a high level of hysteresis - once you had performed a regime change say into the upper octave (2nd partial), you could drop the flow rate quite a long way lower before it flopped back to the fundamental regime. I think we knew that instinctively, but it was nice to see it happen “in the lab”. I’d want to sort out my flow rate monitor switching before investigating that further. (Heh heh, I’d have to make sure that the flow rate monitor switch didn’t drop me into the lower mode by momentarily interupting flow. I can probably leave one monitor permanently connected, and just drop the second one in parallel when I need more range.)

[quote=“trill, post:6, topic:105572”]

  1. One question I have is what role the human anatomy (mouth cavity, throat, lung volume) have on the “total acoustic field”.
    [/quote]

That we don’t know yet. There is something happening, because skilled recorder players have more control over frequencies around the register shift than novices.

And I imagine one of those skills is learning to exploit the hysteresis I mentioned above. A chirp of pressure will produce the regime change, you can then back off a bit to prevent going too sharp or too shrill without fear of dropping back down.

16

This is extremely worthwhile reading for the description of the measuring instrumentation alone. The conclusions based on the experimental data — all expressed in terms of air pressure — will otherwise be no surprise to anyone here. Digital manometers normally permit pressure measurement at two points. On a whistle, the oral cavity is the obvious and accessible one, and the point of comparison would be somewhere inside the bore. The article indicates where that might usefully be, but getting there to measure it is invasive.

The tuners and voicers at Moeck had access to a blowing machine that was essentially what we’ve been talking about designing here. The air feed was set volumetrically and the air pressure was measured at the entrance to the windway. I never saw or heard of anyone using it the entire time I was there. We also all had water-column manometers at our workbenches and used them for such things as calibrating differences between our individual blowing pressures and logging what we did when custom tuning for VIP clients.



I studied the recorder with Bernard Krainis for several years. One of the techniques that he taught — labeling it as counterintuitive but effective — was how to maximize and vary the size and shape of the oral cavity. He said he based it on watching the way Frans Brüggen played and confirmed in discussion with him, that Frans was consciously harnessing the effect. A quick check a few mins ago showed that the pressure in the cavity drops as it is enlarged. The effect on tone quality is perceptible without any shift in pitch, which I assume means an instinctive increase in air flow. There was no immediate indication of any of this making a diff when shifting modes of vibration but I’ll spend more time with that.

An automated blowing device may be useful for measuring the steady-state behavior of a whistle. However, it would also have to emulate the initial transients of notes bumped into different registers as done by a reasonably skilled actual player. Repeating what I still feel to be a pivotal question — given that so much of the available data is quantified in terms of pressure, and that any one of us can measure that aspect of their own playing, what purpose is there in shifting to volumetric data? Terry’s chronicle of his on-going efforts is informative and fascinating, and it is easy enough to understand the L/min measurements in relative terms. But it would sure be nice to be able to compare them in a more absolute sense, for example, to the workshop literature about voicing organ pipes.

17

I agree. Terry, perhaps you could add pressure measurement data to your flow data, so we can see some relationships, and be able to replicate experiments with simple pressure measurements?

18

I’ll have a look at it, hans. The setup is a bit messy, but let’s see if it can handle getting messier!

Now, here’s today’s Frightening Fact. If I calculate correctly, the volume of a 12mm bore tin whistle is just over 30mL. But in doing these tests, I’ve been pumping an average of about 15L/min through it. That’s 15,000mL, or about 500 air changes per minute. That’s 30,000 air changes per hour. Convince me I’m wrong!

19

The air flow is split at the edge. The illustrations I’ve seen of the process show the larger part diverted to the outside of the instrument.

20

A quick look at the maths and physics of gas flow through orifices suggests you need to measure both pressure and flow for the various windways.

What is the effect of ‘forming an embouchure’ with the lips on a whistle? Injecting turbulent air into windway? Does it really effect tone? Does tonguing briefly change the nature of the airflow arriving at the windway?