Blowing machine

[trill]

... Hooked up the usual hapless Feadog Mk 1, and covered all 6 finger holes with tape. . . Feeding it about 3.5L/Min, I get about 500Hz of wafty tone.

Sir, is there any chance you could repeat the test with the tone-hole-tube removed ? Flowing air into just the "naked head" ?

Inquiring minds . . .

[Tunborough]

What do we think?

And is it possible that an indicator of the bottom of the low regime is the level of flow where changing the fingering DOES change the note?

I would say we can distinguish edge tones from "real" notes (known in the literature as the first hydrodynamic mode) by what happens when the flow increases (at the same fingering). If the pitch drops suddenly, you were hearing edge tones; if the pitch rises inexorably (possibly with sudden leaps to a new partial), you were hearing a real note.

[Terry McGee]

. . . What do we think? . . .

My current belief is the our beloved whistles are simply "oscillator+resonator".

Once upon a time, I removed the head of whistle + blew. I found that the pitch varied smoothly. More pressure=higher-pitch (+more volume). No "jumps". A simple variable-frequency-oscillator.

The tube (with tone-holes) is the primary (free-free) resonator. (Player anatomy, including highly-damped, variable-volume lungs would be an additional fixed-free resonant cavity).

Seems to fit the above data.

And is it possible that an indicator of the bottom of the low regime is the level of flow where changing the fingering DOES change the note?

That's what I'm thinking.

Keep in mind it is more complex than just an oscillator and a resonator. You can simulate that set-up by putting the window under your lip, and saying "shhhhhhhhhhhhhhhhh" (a source of whitish noise), while fingering the octave. You'll hear the resonance in the tube accentuating the frequencies it likes. But there's no gain.

When we play the whistle by blowing down it, we form a broad jet that strikes the edge, shooting a blast of air down the whistle. It reflects back and switches the jet out of the whistle, sending a negative pressure wave down the whistle, etc etc. It's that jet-switching action that provides the gain in the system. So it's not a free-lance oscillator, it's one controlled by the resonator. And that's where the magic comes in.

[Terry McGee]

. . . so I've just run this experiment...

Sir, is there any chance you could make sound recordings of your tests ? Either WAV or MP3 ? I would be very interested.

It's a fair question. But it makes me wonder if I'm going that far, should I video the test? (with audio of course!) I think I could squeeze the camera in down there, on a tripod, peering over my right shoulder. It would give viewers a better idea of how the system works and what I'm doing, and I could probably mutter things like 10.5 Litres per minute, 7mm of pressure, n Hz, G5-27cents, 93% volume, after pausing to let viewers hear the regime we've just entered. Or perhaps even point the camera to each measurement in turn, although I might run out of hands doing that! Is an experiment worth a shot?

If so, what would be the best place to lodge the video? I have used YouTube some years back to lodge evidence of some carillon problems so that I could discuss them with the owner and overseas manufacturers. It proved very successful! But maybe there are better options than YouTube these days?

Would this mean I'd need to brush my hair?

[Terry McGee]

... Hooked up the usual hapless Feadog Mk 1, and covered all 6 finger holes with tape. . . Feeding it about 3.5L/Min, I get about 500Hz of wafty tone.

Sir, is there any chance you could repeat the test with the tone-hole-tube removed ? Flowing air into just the "naked head" ?

Inquiring minds . . .

I think I did try that, but I'll do it again and get back to you. Just making a few changes down there to make things easier.

[Terry McGee]

What do we think?

And is it possible that an indicator of the bottom of the low regime is the level of flow where changing the fingering DOES change the note?

I would say we can distinguish edge tones from "real" notes (known in the literature as the first hydrodynamic mode) by what happens when the flow increases (at the same fingering). If the pitch drops suddenly, you were hearing edge tones; if the pitch rises inexorably (possibly with sudden leaps to a new partial), you were hearing a real note.

OK, I'll repeat what I was doing and look for any of that evidence.

[Terry McGee]

What do we think?

And is it possible that an indicator of the bottom of the low regime is the level of flow where changing the fingering DOES change the note?

I would say we can distinguish edge tones from "real" notes (known in the literature as the first hydrodynamic mode) by what happens when the flow increases (at the same fingering). If the pitch drops suddenly, you were hearing edge tones; if the pitch rises inexorably (possibly with sudden leaps to a new partial), you were hearing a real note.

OK, revisited the very low flow rates in the G fingering.

Using the 5L/Min gauge at first for better readability:

3L/Min wafty G5 +5cents. Covering the 4th hole alters the pitch by a semitone
3.5L/Min, a secondary lower tone joins very quietly in the background

3.8L/Min, A4+30 established, replacing the wafty G5. Pitch now not affected by closing 4th hole
4.2 B4
4.4 C5
4.7 C#5
4.9 D5 None of the above group affected by closing next hole.

Reached max on 5L gauge, so shifted to 20L/Min gauge:

6L/Min D#5 Next hole drops pitch 15c
6.3 E5 Next hole drops pitch 25c
6.8 F5 Next hole drops pitch 50c
8 F#5 Next hole drops pitch 90c
12.8 G5 Next hole breaks to instability

So, it seems the ultra-low (circa 3L/Min) wafty note is affected appropriately when you close hole 4, but the very low notes (A4 up to D5) are not affected in pitch. As we increase the flow, the notes from 6L (D5) to 8L (F#5) are increasingly affected by closing the 4th hole. Finally the 12.8L/Min flow to produce a G5 on the edge of breaking is too much for the F# fingering and breaks to instability.

Does that tell us anything?

[Tunborough]

Well, it tells us that the sounds from 3.8 L/min through 12.8 L/min are part of the "real" first register (not aeolian sounds). It also tells us that your two flowmeters don't agree: the 20 L/min gauge definitely reads higher than the 5 L/min gauge, perhaps close to 1 L/min. How do you calibrate a flowmeter, or even check its calibration?

Why does XXXXOO give the same pitches as XXXOOO at low air speeds? That I don't know. Very curious. I guess we need to try this on other fingerings.

[Terry McGee]

Well, it tells us that the sounds from 3.8 L/min through 12.8 L/min are part of the "real" first register (not aeolian sounds).

You don't think it tells us that the very low stuff is real, but then gets taken over by something that is not real (where changing the fingering does nothing), but then slowly morphs back into real? Weird, I know....

It also tells us that your two flowmeters don't agree: the 20 L/min gauge definitely reads higher than the 5 L/min gauge, perhaps close to 1 L/min. How do you calibrate a flowmeter, or even check its calibration?

I put the 5L meter in series with the 20L meter. With the 5L reading full scale, the 20L meter read 6L. I swapped the two 20L meters, the second one also read 6L. And of course I have no way of knowing which one is right! It perhaps tells us that I should just stick with the 20L as at least it's not throwing in sudden steps.

You can't calibrate these - they depend for operation on the size of the floating bead vs the taper inside the body. If we wan't better accuracy, readability and precision, we'd need to go to some other kind of meter. Or perhaps at least a more upmarket and taller version of these.

Hmmm. I do have a Magnahelic Differential Pressure meter. If we came up with a "calibrated" or at least "adjustable" resistance to put across it so that FSD (sorry, Full Scale Deflection") occurred at 40L, we could at least read down to the flows we need on a single scale. Have to multiply all the readings by 4. Or is that where the squared formula kicks in making it a bit more complex than that? Still, we have spreadsheets, don't we?



There might be better approaches out there.

Why does XXXXOO give the same pitches as XXXOOO at low air speeds? That I don't know. Very curious. I guess we need to try this on other fingerings.

And other whistles! Just in case....

[Terry McGee]

... Hooked up the usual hapless Feadog Mk 1, and covered all 6 finger holes with tape. . . Feeding it about 3.5L/Min, I get about 500Hz of wafty tone.

Sir, is there any chance you could repeat the test with the tone-hole-tube removed ? Flowing air into just the "naked head" ?

Inquiring minds . . .

Just did that.

Below 8.3L/Min, nothing.
Then sizzly notes that the tuner had trouble trying to recognise, giving away to a fairly stable rise in pitch according to flow rate:
F7 @ 10.5L/Min
F#7 @ 11.6
G7 @ 13.7
G#7 @ 15.8
and getting a bit weird after that.

I guess it's acting like a very short whistle?

[trill]

Terry,

My goodness . . . you've been very busy !

Thank you for running that case + posting the results. I really appreciate it.

I must say, my curiosity has been running wild.

I remembered I had access to a "regulated" supply of gas: some medical oxygen + the flow-regulator that goes with it. Here it is:




I now have an idea of what "wafty" means !

Honestly, the amount of information you have posted is tough for me to process all at once.

It may take me a while. But, I'll get to it. This is fascinating !

I'd really like to understand the note from a few days back:

. . . Feeding it about 3.5L/Min, I get about 500Hz of wafty tone. B + 20 cents, where we expect D. So well below the bottom of the low regime . . .

Cheers,
trill

[Terry McGee]

Honestly, the amount of information you have posted is tough for me to process all at once.

It may take me a while. But, I'll get to it. This is fascinating !

Don't beat yourself up over it - this is a lot of info, and it's something I've never seen before either. These sorts of tests have probably been run on transverse flutes from time to time, but I'm not aware of anyone doing it on whistles. And if they did, I haven't seen any published outcomes. So we're in unchartered waters. And I'm very aware that I'm just dipping a toe in here and there to try to understand what to expect, and to determine if our experimental rig is up to the task. So we're still very much in the exploratory phase, not yet at the bread & butter data collection phase.

I'd really like to understand the note from a few days back:

. . . Feeding it about 3.5L/Min, I get about 500Hz of wafty tone. B + 20 cents, where we expect D. So well below the bottom of the low regime . . .

Sorry, I'm being too cryptic. Let's see if I can phrase it better, and do feel free to come back if I'm not making sense. There's not much point in me posting this stuff if nobody can understand it! I meant:

I have the whistle's 6 finger holes covered with tape, so we are expecting to hear the low D, D5, about 585Hz. Feeding it about 3.5L/Min, I get a tone quite a lot lower in pitch. It's about 500Hz, registering B4+20 cents. And, interestingly (I just went down to check!) partially shading the end of the whistle with my thumb doesn't flatten the note. (Fully covering the end of the whistle does kill the note though.)

As I crank the flow up, we smoothly progress through the notes until we get to D5. By that time, shading the open end of the whistle does flatten the note. So seemingly consistent to what I reported earlier today on the G fingering - until we get a reasonable amount of flow up, it's as if it's not behaving like a woodwind should.

And by the time we have some good flow, the note has power, whereas it had been weak and wafty. (Tenuous, lacking conviction, the sound of a fairy at the bottom of your garden blowing lightly on a daffodil, trying not to attract attention.)

[Terry McGee]

I said above:

As I crank the flow up, we smoothly progress through the notes until we get to D5. By that time, shading the open end of the whistle does flatten the note. So seemingly consistent to what I reported earlier today on the G fingering - until we get a reasonable amount of flow up, it's as if it's not behaving like a woodwind should.

But is that exactly the point? When we starve a whistle of flow, it plays extraordinarily flat, and the jet-switching action with it's remarkable gain doesn't kick in. And the whistle doesn't really react if we cover the next hole down. It's only as we approach decent flow rates (the kind of rates that humans use instinctively) that we see pitches, levels and pitch sensitivity approach what we expect. Note how the pitch and the flattening effect of covering the next hole is changing dramatically as we increase flow:

6L/Min D#5 Next hole drops pitch 15c
6.3 E5 Next hole drops pitch 25c
6.8 F5 Next hole drops pitch 50c
8 F#5 Next hole drops pitch 90c
12.8 G5 Next hole breaks to instability

[Tunborough]

These sorts of tests have probably been run on transverse flutes from time to time, but I'm not aware of anyone doing it on whistles. And if they did, I haven't seen any published outcomes.

As I understand it, they have been doing similar things with organ pipes for generations. In France, they've been doing measurements on whistle-like organ pipes and recorders for a while. Here's an interesting one published in 2000:

Segoufin, C., Fabre, B., Verge, M. P., Hirschberg, A., & Wijnands, A. P. J. (2000). Experimental study of the influence of the mouth geometry on sound production in a recorder-like instrument : windway length and chamfers. Acustica United with Acta Acustica, 86(4), 649-661.


Most of the interest is on the transitions between partials at normal flows, rather than the behaviour at low flows. The Auvray paper I quoted above does measure down to the levels we're looking at, but doesn't say much when their model predicts you shouldn't be able to get down to that level. As far as I know, no one has looked at the behaviour that Terry has found where all fingerings sound the same at low flows.

[Tunborough]

Hmmm. I do have a Magnahelic Differential Pressure meter. If we came up with a "calibrated" or at least "adjustable" resistance to put across it so that FSD (sorry, Full Scale Deflection") occurred at 40L, we could at least read down to the flows we need on a single scale. Have to multiply all the readings by 4. Or is that where the squared formula kicks in making it a bit more complex than that? Still, we have spreadsheets, don't we?

That could work with an orifice plate for measuring the fluid flow. But yes, the squared formula gets involved.

Beyond the orifice plate, for the general relationship between flow and pressure, we have to decide which formula...

Bernoulli works on short channels (like the orifice plate) with laminar (not turbulent) flow: V = sqrt(2 p / density)

Hagen-Poisseuille works on long channels with laminar flow (your long drinking straw):

When the flow gets turbulent, we have Darcy-Weisbach:

This is getting out of my depth.