Blowing machine

[Tunborough]

I came to understand that there were two blowing styles, and that affected the tuning: the tuning that players could perceive, and the tuning that I could quantify with a Strobotuner.

That's very interesting, Walt. Can you describe the differences between the two different blowing styles?

[Terry McGee]

Could you run 3 data points ?

What I have in mind is:
1) hook the digi-manometer up to *both* the T and the spigot
2) read the pressure in "differential" mode
3) run 3 cases: 36, 38, and 40 lpm
4) Old Gen or Killarney on the end.

Yes, I'm trying to run down whether your flowmeters are calibrated for oxygen or air.

Here we go:

Code: Select all

Manometer in Differential mode, +ve I/P to spigot, -ve to T-junction, Old Generation whistle	
L/Min	MMH2O
36	4.5
38	5
40	6
	
22C in room, 21C in whistle window	
All readings “a bit noisy”	

Note the connections. The T-junction comes after the flow meters, the spigot is further down the chain, just before the whistle. I have the normal input to the manometer connected to the spigot, and the differential input connected to the T-junction. I checked the reverse setting, indeed it gives negative pressure readings.

So what do we learn?

[trill]

. . . Here we go . . .

Terry,

Thank you for running the cases and including your comments on temps+noise with the data.

First observation: the measurements are at the very low end of the instrument's scale.

More observations when I do the supporting calcs.

A couple questions:

1) About the last digit in the display (to the right of the decimal point): does it flip between 0.0 and 0.5 or does it give increments of 0.1 ?

2) How would you describe the noise ? e.g. flipped entire digits left of the decimal ? many times a second ?

3) How was the "whistle window" temp measured ?

Thanks again.

More when I get it done.

[Terry McGee]

First observation: the measurements are at the very low end of the instrument's scale.

They sure are, but I guess that's not surprising, given that the spigot and the T-joiner are not far apart, and the bores are big compared to the whistle's windway.

1) About the last digit in the display (to the right of the decimal point): does it flip between 0.0 and 0.5 or does it give increments of 0.1 ?

The unit gives increments of 0.1CM, but keep in mind I'm reporting it in MM. So the least significant digit is 0.1CM or 1MM. But if it say vacillates fairly equally between readings of 0.4 and 0.5CM, I am reporting that as 4.5MM. Posh people would call that "interpolating", but "coping mechanism" works too.

2) How would you describe the noise ? e.g. flipped entire digits left of the decimal ? many times a second ?

The 6mm quoted for 40L/Min might say flicker between 0.5, 0.6 and 0.7, occasionally 0.4CM. At the sampling rate, whatever that is. But spent more time on 0.6 than the others. It's subjective....

3) How was the "whistle window" temp measured ?

Just holding the thermocouple gauge so that the tip was in the window with the air rushing out on to it. When pulled away from the window, it went back to room temperature 22C. "Slightly scientific". The meter only has a least significant digit of 1º. I can switch it to ºF which might give an illusion of more resolution! At the moment it reads a stable 21C, or vacillates between 70 and 71F. If we called that 70.5F, it translates back to 21.4C.

This one is a naked thermocouple, so it has a very fast response time. I also have an enclosed thermocouple, but it's as slow as a wet week. Posh people would call them "integrating"...

[Terry McGee]

I came to understand that there were two blowing styles, and that affected the tuning: the tuning that players could perceive, and the tuning that I could quantify with a Strobotuner.

That's very interesting, Walt. Can you describe the differences between the two different blowing styles?

I can imagine that there are two broad types of players, or perhaps more likely, a spectrum of players bounded by two extremes.

At one end, the player cursed with perfect pitch, doomed to constantly modulate their pressures to best match the wildly varying pitches of those around them, including Flailing Fred the Failing Fiddle Player and Gregarious Gary the Gross Guitarist who never got around to sorting out the intonation issues of his cheap and poorly set-up instrument.

And at the other end, the players who make some attempt to get into tune at the start of the session ("can anyone give me a C natural?") but are then happy to play along at the pressure that feels good and sounds loud enough.

Would that be in the same ballpark as your experiences, Walt?

The uniting thing of course is that both would benefit from whistles that are better tuned at predictable pressures. Less work for group 1, and more success for group 2.

[david_h]

I now try to be in group 1. I don't have perfect pitch but have learned that there is an unpleasant sound when whistles and flutes are not in tune on a note. I automatically try to adjust to make things better (even if I am not the problem). I am reasonably successful on flute. Part of my interest in this discussion is that I would like to get better at it on whistle and expect a great deal of experiment with a tuner to be needed to identify which of my assemblage of whistles suites me best.

I was in group 2. It may have changed now but when I started on whistle all of the tutor books said what the fingering was for each note but none said one had to blow just right to get the correct pitch. My ear was not attuned to the proper intervals and I was use to varied dynamics from singing at school. So I was happy to "play along at the pressure that feels good and sounds loud enough" and pleased that it sounded like a tune.

I think whistlers (and their teachers) could learn from recorder players (and their teachers) here. They know there are no dynamics and learn to play in harmony with others. In several pub sessions whistle players who also play recorder have become my 'go to' pitch reference.

A surprising number of experienced workshop leaders who don't play whistle or flute don't realise that for an initial tune-up a cold instrument wont be in tune but that the player can fake it, then sort it out after a few minutes of playing. Just once I was told 'OK - how about the second octave?'.

[waltsweet]

About the blowing style. I've met players of the Boehm flute, competenet musicians, who struggle with tuning issues on the "Irish" flute. Mostly, it's about the idea of "octave stretch" (or "droop" as I call it, related to Benade's observations). The "droop" is totally manageable, but it takes some awareness and adjustment, especially when comparing brands of flutes.

Now to whistles. The Copeland, for example, has some stretch to the octaves. Now if a player comes from playing the ITM flute, there's no problem. Further, to boost the input on the lower octave, and to be more delicate on the upper, this approach will balance the loudness and presence in the ensemble. In another camp, the Generation, as a reference, does not stretch much (nor does the Boehmflute). I find that this type, very popular, is played in a rather even-toned manner, without trying for the complement of expression heard on a flute. Plenty of musicianship there, but with particular focus (commpare the flourishes characteristic of harpsichord music?). Years back, we made the "PRO" whistle with a tapered bore. It was disconcerting to hear some complaints about flatness, especially D2; what I figured-out is that a few customers played with a VERY light pressure, requiring the slide to be pushed-in all the way. Of course, an extreme move like this distorts the musical scale, bringing C# too close to D2, hence the complaint. Dad obliged by hacking-off the bottom 5/16" (8mm) with the table saw.

My friend, the recorder-maker, does have a blowing machine; the Magnehelic ensures even pressure. Players will adjust their pressure to arrive at pitch that supports the musical consonance, so the pressure-to-pitch relationship is most important. Now there's a can of worms I don't need to open!

In the drumcorps world, I recently made a new model fife on which I could point to the improvement of a couple critical notes. Other notes had improved pitch, but doing so created a conflict with the traditional expectaions and ensemble playing. My friend said, "But nobody asked you to [change those notes]". So, even if I can justify my intent and apply all the science I can muster, I will be judged by other criteria, sometimes arbitrary. "We want the instrument to be better, but not different." And now for my next trick, I will change straw into gold . . .

Terry, I've certainly experienced the types of musicians you describe. But here, I'm describing how whistle players manage the usual compass on their instrument, how blowing habits affect the results as a set, and how I'm always trying to get a handle on what people say vs. what the maker should do about it. There's been a huge growth in the number of whistle-makers. Not so many are like Copeland, so it seems to me the other type will dominate. Regarding the issue with my "PRO" whistle, I suppose I could have changed the design to conform, but others were already doing so.

[Terry McGee]

Hi all

Sorry for going AWOL yesterday, there were matters in the Nation's Capital that demanded my attention....

Tunborough, herewith the Old Generation results.

Code: Select all

Old Generation			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0
4	5.5	2.3	0.59
8	15	3.9	0.48
12	36	6.0	0.50
16	64	8.0	0.50
20	103	10.1	0.51
24	132	11.5	0.48
28	192	13.9	0.49
32	244	15.6	0.49
			
Average Resistance	0.50 

Again note the dodgy looking 4L reading. If I set up the 4L on the 5L/Min gauge, the pressure drops to 5mm, giving a "resistance" of 0.56. Still on the high side.
And note that the 20L/Min reading has been associated with the highest "resistance" calculation for other whistles and calibrators, so maybe that flow is reading a little low? But again, hey, are we really expecting 2 decimal points out of these low-resolution meters?

[Terry McGee]

And here's the Killarney:

Code: Select all

Killarney			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0
4	11	3.3	0.83
8	27	5.2	0.65
12	62	7.9	0.66
16	110	10.5	0.66
20	178	13.3	0.67
24	220	14.8	0.62
28	307.5	17.5	0.63
32	397.5	19.9	0.62
			
Average Resistance	0.67 

Note the 4L/Min Resistance dramatically higher. Average Resistance without 4L/Min is 0.64

[Terry McGee]

And I thought to add this one as an example of a larger bored, free-flowing whistle, but interestingly, the small-bored Feadog Mk1 exhibits slightly lower "resistance". The Mellow D has a longer windway but also ends smaller.

Code: Select all

Mellow D Tweaked			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0
4	4	2.0	0.50
8	12	3.5	0.43
12	27	5.2	0.43
16	49.5	7.0	0.44
20	82	9.1	0.45
24	102	10.1	0.42
28	151.5	12.3	0.44
32	192	13.9	0.43
			
Average Resistance	0.44
Average Resistance without 4L/Min 0.44

[david_h]

Some observations on the data from the last four whistles.

I understand that Resistance is being calculated using the square root of the pressure because that is the relationship we would expect from Bernoulli's Equation.
Fitting a power relationship curve to those observation gives a power of greater than 0.5 (i.e. square root) for all of them. In the order Feadog Mk1, Old Generation, Killarney, Mellow D they are: 0.55, 0.53, 0.56, 0.53, Having all four of them on the same side of the expected value raises suspicions that either there is systematic bias in the data or the model is wrong. The flow meter being calibrated for another gas shouldn't throw things of (just the Resistance) but the higher flow rates being measured at higher pressure might. Will do some sums.

If it is a square root relationship we can fit a straight line to Flow vs sqrt(P) and see how the points lie in relationship to that. Looking at such plots hints that the individual points tend lie off the line in the same direction by roughly the same amount for each whistle.

The distances of the points from the lines, the residuals, in units of flow are below. It needs a plot of Flow vs residual to be really obvious but these are not random differences. Something systematic is happening and if it's not to do with the flow regime my guess is it's scale errors on the flowmeter. The 4 L/min value is not particularly 'worse' than others - it shows up in Terry's ratios because it's nearest the origin.

Code: Select all

Flow	MellDtw	Kill	OldGen	FdgMk1
0	0.242	0.785	0.321	0.603
4	-0.391	-0.610	-0.418	-0.470
8	0.134	0.284	0.286	0.232
12	0.196	-0.130	-0.040	-0.081
16	0.027	-0.381	-0.161	-0.393
20	-0.838	-0.958	-0.487	-0.706
24	0.846	0.590	0.629	0.757
28	-0.250	0.176	-0.316	-0.316
32	0.044	0.252	0.182	0.386

[david_h]

I am scratching my head about this correction equation now

From that if the 'new' gas was denser than that for which the meter was calibrated the meter would be under-reading - the float would not be lifting as much. Is that right?

Either way the difference is not enough to explain the relationship of flow to pressure not being exactly a square root.

However, missing out the 4 l/min value (the zero is already out for a power function) results in powers of 0.52, 0.50, 0.52 and 0.50 so maybe Bernoulli rules and it's just a data imprecision issue - with the one Tunborough and Terry suspect being a culprit.

My feeling is that the data give no reason to doubt an overall square root relationship between flow and pressure but that the departure of measurements from that follow a pattern. The pattern could be to do with air flow in windways or it could be be to do with the flow measurement. If we assumed the latter we could correct for it using the average across the whistles so may get very slightly better resistance measures. There is not much gain from modelling noise but If we were wrong and the pattern is significant we would be throwing away information about the air flow that Tunborough's modelling might explain.

[Terry McGee]

I wondered if there might be something odd about the first of the 20L/Min flowmeters, so swapped them over. But saw pretty much the same results. That doesn't rule out the possibility they are both wrong of course - they are of the same brand and model, bought at the same time from memory.

So then I took more data points at the bottom, and it turns out that the "resistance" increases even more as you go down lower!

Code: Select all

Mellow D Tweaked, Flow Meters swapped			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0
2	3	1.7	0.87
3	3.5	1.9	0.62
4	4	2.0	0.50
5	5.5	2.3	0.47
6	6.5	2.5	0.42
8	12	3.5	0.43
10	19	4.4	0.44
12	28	5.3	0.44
16	53	7.3	0.46
20	85	9.2	0.46
24	104	10.2	0.42
28	149	12.2	0.44
32	189	13.7	0.43
			
Average Resistance	0.49
Average Resistance from 6L/Min up 0.44

[Terry McGee]

Just explored that same area using a calibrator, to make sure it just isn't a whistle thing. But the rise in resistance down low is still visible....

Code: Select all

30 x 3.94mm Calibrator			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0.00
2	2	1.4	0.71
3	3	1.7	0.58
4	4	2.0	0.50
5	4.5	2.1	0.42
6	6	2.4	0.41
7	8	2.8	0.40
8	10	3.2	0.40
9	14	3.7	0.42
10	17	4.1	0.41
20	78	8.8	0.44
			
Average Resistance	0.47
Average Resistance from 5L/Min up 0.41

[Terry McGee]

And now, still with the calibrator, but using the 5L/Min meter up to 5L, reduces the effect. So I'm guessing it is a flow meter issue. And that if we want low readings, we have to use the 5L/Min meter for them?

Code: Select all

30 x 3.94mm Calibrator, using 5L/Min meter			
Flow	MM(H20)	√A/P	Resistance
0	0	0.0	0.00
1	0	0.0	0.00
2	1	1.0	0.50
3	2	1.4	0.47
4	3	1.7	0.43
5	5	2.2	0.45
6	6	2.4	0.41
7	8	2.8	0.40
8	10	3.2	0.40
9	14	3.7	0.42
10	17	4.1	0.41
20	78	8.8	0.44
			
Average Resistance	0.43
Average Resistance from 6L/Min up 0.41