Blowing machine

[Terry McGee]

Now Tunborough and stringbed, a little while back I goaded trill and david_h into revealing their interests in what we're doing here. And I made assumptions that everybody was familiar with the two of you. It was possibly overreach, perhaps it would be good to put something on the record? And DrPhill - you've outed yourself - where are you coming from in this? And reaching out to any lurkers - feel free to step out of the shadows into the harsh light of day. What do you hope we can get out of this?

And to all, what's our next step?

[Tunborough]

perhaps it would be good to put something on the record?

I've posted a number of times about WIDesigner, some software for wind instrument modelling and optimization. WIDesigner includes an empirical model of the whistle mouthpiece, based on data I had available at the time from blowing a lot of notes on a lot of whistles. To improve the model, I need more dependable data, of the sort I can get only with a mechanical blowing device such as Terry has built.

And to all, what's our next step?

Terry, you fortuitously thought to include in your apparatus both a flow meter and a pressure meter. Most investigators in this area include only one or the other, usually a pressure meter, and assume that Bernoulli had the entire answer for converting between them. Your numbers demonstrate that Bernoulli isn't the whole answer. I'm satisfied that every whistle mouthpiece has an associated constant that allows us to convert between them with a reasonable degree of accuracy. I"m getting close to being able to estimate that constant from the mouthpiece geometry. This would mean others, like me, who wish to reproduce your blowing machine don't need to measure both flow and pressure, we can measure whichever is most convenient.

To this end, I'd appreciate if you could re-measure pressure vs. flow on a number of samples with the new apparatus - the digital manometer and the new pressure tap. I'd be most interested in the 30 mm x 4 mm calibrator, the 30 m x 2 mm calibrator, the three tapered calibrators, the Feadog, the Killarney, and the old Generation, in the flow range 5 to 30 L/min. That's a lot of measuring, so I understand it could take you a while to get to it. No rush. You've got flutes to make.

The numbers I'd like to collect for myself, whenever I get around to building my own blowing machine, cover two areas. First, for a given note, how does the frequency vary with the flow rate. WIDesigner includes a model for this to help it get a note in tune at a flow rate that the player is comfortable with in the context of a tune. Second, for each note on a whistle, what is the highest frequency before it jumps to the next register, and what is the lowest frequency before it drops down a register. WIDesigner uses this to ensure that notes can be played in tune at some flow rate.

There's another experiment for the blowing machine that would be interesting ... Take the Feadog and play through the range like you wish it could be played, with whatever blowing pressure feels most comfortable at that point in the range, with no attempt to keep it in tune. Measure the actual frequencies that the whistle produces when you play it like that. Then use the blowing machine to measure the flow rates that it takes to reproduce those frequencies through the range. What does that set of flow rates say about how you like to play the whistle? (I specified the Feadog, because I have one here that I've measured the geometry of.)

[trill]

. . . And to all, what's our next step? . . .

Personally, my next step is to re-do my Bernoulli model to include the pressure+velocity upstream of the whistle head. My first model assumed zero-velocity upstream: a large plenum. I was trying to figure out why my discharge coefficients were off. That was why I asked for the tube length + diameter upstream.

*Thank you* for explaining the diameters and showing your bench setup ! Honestly, the datapoints you just gave for the two diameter tubes [pre+post T] might be able to resolve the Oxy-air uncertainty of your flowmeter. It's not an orifice, but it *is* a known change in diameter, governed by Bernoulli.) [personally, my money is that its designed for oxygen]

More when I get to it.

PS: glad to know Dwyer has easy-tech (low cost!) *air* flowmeters in our range of interest. My only concern is that the scale is 10cm in length. How much resolution can be achieved "eye-balling" ? But hey, its only a hobby . . .

[edited a few times . . .]

[trill]

. . . .every whistle mouthpiece has an associated constant that . . .

Does the term "discharge coefficient" ring a bell ? It's a first-look at how geometry (area specifically) influences" resistance.

**edited to add**: The "constant" may vary with flow. That is kind of my expectation. After all, viscous losses rise with velocity. (Just a hunch). But, that's why having measurements over the whole flow range is useful. We could see if + how-much it changes with flow rate.

. . . estimate that constant . . .

I propose we send Terry a "care package" which contains a bundle of whistles: different brands, different sounds, ... Measure 'em all !

(in his spare time, in between flutes, of course, after the flow meter readings are blessed)

[DrPhill]

And DrPhill - you've outed yourself - where are you coming from in this?

Oh dear, really I am really just an interested bystander. I am a problem solver at hear, and treating this as a spectator sport. I did write some software for these calculation (TWJCalc) but did/do not really understand the equations. I relied heavily on Daniel Bingamons code (with his permission) .

[Terry McGee]

Terry, you fortuitously thought to include in your apparatus both a flow meter and a pressure meter. Most investigators in this area include only one or the other, usually a pressure meter, and assume that Bernoulli had the entire answer for converting between them. Your numbers demonstrate that Bernoulli isn't the whole answer. I'm satisfied that every whistle mouthpiece has an associated constant that allows us to convert between them with a reasonable degree of accuracy. I"m getting close to being able to estimate that constant from the mouthpiece geometry. This would mean others, like me, who wish to reproduce your blowing machine don't need to measure both flow and pressure, we can measure whichever is most convenient.

To this end, I'd appreciate if you could re-measure pressure vs. flow on a number of samples with the new apparatus - the digital manometer and the new pressure tap. I'd be most interested in the 30 mm x 4 mm calibrator, the 30 m x 2 mm calibrator, the three tapered calibrators, the Feadog, the Killarney, and the old Generation, in the flow range 5 to 30 L/min. That's a lot of measuring, so I understand it could take you a while to get to it. No rush. You've got flutes to make.

Indeed I have, and a lot of other issues unfolding here which are proving time consuming. And we've learned that there may still be surprises in store, so all of that suggests we go at it in a measured way (so to speak!). I'll do a run on the calibrators, then wait to hear that that is making sense. Then probably hit the whistles one by one, with breaks for feedback.

The numbers I'd like to collect for myself, whenever I get around to building my own blowing machine, cover two areas. First, for a given note, how does the frequency vary with the flow rate. WIDesigner includes a model for this to help it get a note in tune at a flow rate that the player is comfortable with in the context of a tune. Second, for each note on a whistle, what is the highest frequency before it jumps to the next register, and what is the lowest frequency before it drops down a register. WIDesigner uses this to ensure that notes can be played in tune at some flow rate.

Now, we can't combine that with the whistle tests mentioned above?

There's another experiment for the blowing machine that would be interesting ... Take the Feadog and play through the range like you wish it could be played, with whatever blowing pressure feels most comfortable at that point in the range, with no attempt to keep it in tune. Measure the actual frequencies that the whistle produces when you play it like that. Then use the blowing machine to measure the flow rates that it takes to reproduce those frequencies through the range. What does that set of flow rates say about how you like to play the whistle? (I specified the Feadog, because I have one here that I've measured the geometry of.)

Yes, this is the sort of testing that I would envisage a serious whistle making company might employ before churning out a million or so whistles*. Arguably a bit beyond the boutique makers, but maybe we can help fill the gap there.

*As we saw with my old Mellow D tweaking, they didn't even bother to get the tuning near right before rushing them out the door. I don't know if the modern Mellow D whistles have been tweaked - anyone got a recent one to try out?

[Terry McGee]

OK, given it looks like I've got quite a bit of measuring to do, I've rerouted the air line so I can leave it safely in place. Settling in for the long haul....

And here's first go at the two calibrators mentioned. For the record, using the new Whistle Connector and the Digital Manometer.

Code: Select all

30 x 4mm Calibrator			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	
5	4.5	2.1	0.42
10	19	4.4	0.44
15	43	6.6	0.44
20	80	8.9	0.45
25	111	10.5	0.42
30	165	12.8	0.43
			
			
30 x 2mm Calibrator			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	
5	80	8.9	1.79
10	334	18.3	1.83
15	768	27.7	1.85

The 30 by 4mm one went without anything noteworthy, but the 30 by 2mm one was different:
- I couldn't get beyond 15L/Min without the Digital Manometer complaining. It spits the dummy at 1400mm, which is far better than my analog U-tube, which would by then have repainted the ceiling blue again.
- The readings were "noisier" than the 30 by 4mm or whistle measurements. Then again, the pressures are much higher, and that might explain it. Or we might be seeing more turbulent flow?

Let me know if that's satisfactory and enough before I progress to a whistle....

[Tunborough]

The numbers I'd like to collect for myself, whenever I get around to building my own blowing machine, cover two areas. First, for a given note, how does the frequency vary with the flow rate. WIDesigner includes a model for this to help it get a note in tune at a flow rate that the player is comfortable with in the context of a tune. Second, for each note on a whistle, what is the highest frequency before it jumps to the next register, and what is the lowest frequency before it drops down a register. WIDesigner uses this to ensure that notes can be played in tune at some flow rate.

Now, we can't combine that with the whistle tests mentioned above?

For these other studies, I'm looking for a lot more numbers on more notes on more whistles, with a lot of geometry measurements to accompany them. For now, I'll be happy with some flow and pressure measurements. If you're still game after that, we can consider frequency measurements on your Feadog, since I have one here for reference.

[Tunborough]

Let me know if that's satisfactory and enough before I progress to a whistle....

Satisfactory so far.

The 2 mm calibrator measurements are a lot different than the last round using the water manometer with no dish soap, but thanks to the extra data point you wrested out of the digital manometer it still lines up nicely with our square relationship. I would like to see results from some of the tapered calibrators before we progress to whistles. The two calibrators alone aren't enough of a test of the direction I'm going in.

[Terry McGee]

OK, here's three more.....

Code: Select all

5.18 x 10, 4 x 20 Calibrator			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0.00
5	4.5	2.1	0.42
10	17	4.1	0.41
15	39	6.2	0.42
20	73	8.5	0.43
25	99	9.9	0.40
30	151.5	12.3	0.41
			
			
4.97 x 20, 4 x 10 Calibrator			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0.00
5	4	2.0	0.40
10	17	4.1	0.41
15	38	6.2	0.41
20	70.5	8.4	0.42
25	97.5	9.9	0.39
30	148	12.2	0.41
			
			
5 x 10, 4 x 10, 2.9 x 10 Calibrator			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0.00
5	13	3.6	0.72
10	56	7.5	0.75
15	127	11.3	0.75
20	240	15.5	0.77
25	341	18.5	0.74
30	526	22.9	0.76

Note, I've checked the end diameters and may have come up with slightly different values. Calipers don't like measuring small inside diameters much!

Let me know how these look, and whether it's time to move on to a whistle. Any preference as to which whistle?

[david_h]

(I specified the Feadog, because I have one here that I've measured the geometry of.)

It may be worth comparing photos of your Feadogs (including the top see-through sticker). There may be two different versions of the Mk1. See photos in these posts.

viewtopic.php?p=1242193#p1242193
viewtopic.php?p=1242199#p1242199

Maybe even a third because from memory there was a photo (possible the missing Mr Gumby one in one of the those discussions) of a Feadog with a Mk1-style head and sticker that had a smaller F# hole (more like a Generation).

[Terry McGee]

Heh heh, I was thinking the same thing, david_h.

Mine is exactly like the one on the left in this image:



But it's many years since it had a sticker!

You make a good point about making sure these are the same whistles. We should trade some dimensions.

It would be nice to be able to put some date ranges on these!

I also have 2 later Feadogs which I'd identify as the third from the left, the Mk II.5. The elder one of these is OK, but the newer one is ghastly. I'm keeping it only to learn from it what not to do!

[Tunborough]

My Feadog definitely looks like a Mk 1, too. Bore is 12 mm I.D., 265.5 mm from the sounding blade to the end of the tube. Window is 5.25 mm long and 8.2 mm wide. Noting the caution from david_h about hole sizes, here is what I've measured. All positions measured from the sounding blade.

Code: Select all

Posn	Space	Dia	Depth
117.5		5.0	0.35
138.0	20.5	6.6	0.35
161.0	23.0	5.6	0.35
183.5	22.5	5.7	0.35
201.0	17.5	7.7	0.35
226.5	25.5	6.9	0.35

[david_h]

Mine is exactly like the one on the left in this image:

How exactly? This image has two similarly shaped heads:



Both of those have been referred to as Mk1 on this forum. From this post: viewtopic.php?p=1242199#p1242199 and see the two below.

Mine is the same as the one on the right with the raised oval around the windway entrance, the "by McCullough Piggot Mfg Ltd" label and so far as I can compare the smoother edges to the tone holes. The Teo label is similar apart from the 'made by' name ( there is a photo one of the threads)

[Tunborough]

Let me know how these look, and whether it's time to move on to a whistle. Any preference as to which whistle?

Those look remarkably consistent with the first two. We're ready to try whistles, and I'd be most interested in the old Generation and the Feadog. For the whistles, if it's not too much trouble, I'd like to see more data points, say every 3 or 4 L/min rather than 5. Thanks for all the numbers.