Interesting difference on hole 4. I wonder if it is more than coincidental with the problems I had with that hole.
I have some pipe cut to length so I will try your spacings tomorrow.
I will ignore the thicker-wall route.
As for inaccuracies in the calculations in TWJCalc, well I take responsibility though I just ported Daniels algorithm. If you have a different algorithm you are willing to share I could build that into the program as well. I have to admit that I am mostly ignorant when it comes to the physics of whistles - but can turn out a reasonable user interface. I will check TWJCalc against TWCalc tomorrow, and see if I have dropped a stitch somewhere with regard to pipe length…
And I just started with a mock-up to see if I could manage a bass-A.
It should be interesting to see what you come up with, DrPhill. I have not had a chance to work through any of my designs with TWJCalc yet but I have tried TWCalc 3.3 against my current working designs. This is in no way a hit on Daniel. He’s doing a great service by making his code available and I greatly appreciate his contributions to whistle making. What I have found is that his current calculations predict a longer tube than what I calculate. I have not gone into analyzing why that is yet. (Things are a bit busy here lately.) I suspect it has to do with an end-correction factor, but that’s just a guess. Also one maker’s calculator may not work accurately for another maker’s designs. That’s why many of us tend to modify the calculations to fit how we make things.
I use a modified version of Pete Kosels Flutomat.
Look at the HTML source of the page, and you see the javascript. FindLocations2() is the main function used for the calculations.
the comment to this function:
This is a non-iterative procedure equivalent to the above procedure. It involves use of quadratic solutions of the Benade equations obtained by “simple but tedious algebraic manipulation”.
The function FindLocations() is not used at all. It is still there as a relict, I guess abandoned in favour of FindLocations2().
My own changes to bring the calculations more in line with what I was observing building flutes and whistles are
a change to the bottom end correction in function C_end().
I use
return 0.5*Bore;
but sometimes I adjust the factor a bit.
a change to the embouchure correction (window correction for a whistle), to get a more accurate total length. For whistle design I use in function C_emb():
return 2.5*(Bore/2)(Bore/2)/Math.sqrt(Demb1Demb2);
Demb1 and Demb2 are window length and width. This is not too bad, but window wall height plays also an important role, so overall the total length from foot end to window block is not totally accurate, but better.
Sorry for having gone into these math details. We can discuss this further privately, if that is more appropriate!
Thanks Hans, that helps a lot. I dont mind the maths - I can follow algorithms and translate them to Java pretty easily. I will need to do a small refactor in my code, then, after implementing this algorithm, I will be able to give the user the option of which calculation to use.
As far as TWJCalc length - I am in error. TWCalc gives a Cork-to-en length of 748.22mm. Still longer than your calculation, but closer. I probably forgot to update the initial (uncorrected) value on the screen at the end of calculation. Will fix that as my next programming task.
I looked at the relative cut-off values for the example you gave - hole four is quite far out of line with the others. I know diddley-squat about whistle design - I was trying to get all the relative cut-off values about the same and larger than two, while keeping the hole sizes reasonable. Was my approach flawed? Are the relative cut-off values a good measure?
I would not bother too much with relative cut-off values. As long as they are bigger than 2 you are fine. For a good third octave you may wish for values bigger than 4, but then you are going into serious big holes, and you need keys to cover them!
For a big bore I would try to use big holes. But they need to be reachable, so I’d need to shrink some quite a bit. Shrink them too much, and the note gets too weak. It’s all about compromise between tone and ergonomics.
Ok, thanks, that makes sense. For my bass A whistle I have resorted to thumb-holes for T1 and B1 which removes some of the ‘hand stretch’ constraints, and a bend in the neck to make up for short arms… I have been deliberately making use of the extra freedom this gives me by choosing large holes further down the pipe.
I had in my mind that higher cut-off would give ‘better tone’. I am not sure where I got that idea.
But TWJCalc shows 774mm length, my own shows 714mm, with a window 8 by 16mm.
I put an id of 32mm, window 8x16, wall 2mm, and all holes 12mm (B1 should be smaller, but whatever) into my hole position calculator spreadsheet and for length it gave me:
712mm labium to bottom, so 716mm from the center of the window,
and 776mm for an effective length.
So maybe the length calculated by TWJCalc is effective length.
The 8x16 window is effectively a 60mm length of 32mm id tube.
I have corrected TWJCalc, but the fixed version is not up on SourceForge yet. By fixed, I mean that it now agrees with TWCalc. I guess I never noticed the error, as I have always cut pipe to length using the tuner as a guide, rather than a ruler. I started doing that because I could not be sure how a 45 degree bend was going to affect the length. Maybe I should buy a straight coupler and determine the real length, just for calibration. I wonder how much effect the bend has? I could imagine the effective distance to a hole on the front being longer than the effective distance to the same hole on the back…
Hans: I tried the measurements that you gave, and the holes were a lot closer, especially hole 4. A couple were a little flat (I dont mind that at all), a couple were spot on, and a couple were a little sharp. The discrepancies could be due to my inaccurate drilling/measuring techniques. But it would seem as if flutomat style calculations are closer for my purposes. I will have a look at reproducing them. The resultant whistle was noticeably quieter than with the larger holes, and the top of the second octave was difficult. Bigger holes are probably needed.
Highwood: thanks for the corroboration. Did your hole 4 location work out closer to Hans (flutomat based) or TWJCalc values? Are you using a flutomat based calculation, or one you have derived from scratch?
And for you both: What is the best way to express window distance to the labium, to the fipple block, or to the centre?
Phill, I never meant you use 12mm holes! I just set it as a value to compare the calculators! Bigger holes are better, as long as you can cover them!
I don’t know, I have not settled on that one, as the calculated value is too inaccurate as yet, and the tuning slide compensates.
For a flute I use the centre of the embouchure hole, so for a whistle I tended to think the centre of the window may be the point of reference. Practically the top edge of the windway exit may be better. This is not the same as the block, I tend to think the block intrudes into the pipe at the window, but the window is formed by labium edge, sides, and top of windway exit.
But knowing the two calculations are different does not help nearly as much as knowing which is closer to reality (at least in this case). The answer is the flutomat-based one (in this case). It only cost a small amount of pipe and a few minutes drilling.
I don’t know, I have not settled on that one, as the calculated value is too inaccurate as yet, and the tuning slide compensates.
For a flute I use the centre of the embouchure hole, so for a whistle I tended to think the centre of the window may be the point of reference. Practically the top edge of the windway exit may be better. This is not the same as the block, I tend to think the block intrudes into the pipe at the window, but the window is formed by labium edge, sides, and top of windway exit.[/quote]
To answer your original questions…“Can anyone help? how do (voicing) dimensions effect tone.”
The voicing, including the parts…the windway, the emboucure hole(window),the labium ramp and the “ears”(sides) are the “sound generation device” of the duct flute. Every aspect of the voicing can change the tone or sound generation. Here are some “basics”.
The Windway: Directs the airstream across the window to strike the labium ramp edge.(75% above the edge,25%below). The size of the windway opening controls the airstream velocity which controls pitch(frequency of oscillations). The “lips” of the windway control the shape of the airstream>(Example:rounded lips = fan shape or sharp lips = bar shape)
Labium Ramp Angle: Controls pitch bending ability and register shift respose. (Example:A shallow angle allows greater pitch bending with small changes in breath pressure and faster register shifts {Tabor Pipe}. A steep angle gives stable pitch with small changes in breath pressure and slower register shifts {Recorder}.
Labium Ramp edge: A sharp edge requires a higher velocity airstream to oscillate, while a rounded edge requires a lower velocity airstream. The process is called “Generating a Von Karman Vortex Street”.
I am using calculations that I derived from scratch - that is I use the equations that Pete Kosels flutomat is derived from (which have been kicking around for a long time - 1930s I think), but do not make some of the trig approximations before solving. The errors from the approximation give holes that are tuned flat - and this is a good thing, at least better than being sharp, since one can easily drill out a hole but it is harder to make it smaller.
For what is worth my spreadsheet gives
394.17
345.80
291.06
219.77
195.62
121.46
with all 12mm holes, measured from the bottom - note that most of my holes are about 4 mm further up the instrument, and that my 4th hole is quite a bit further up.
Note:
1/ I would not make a whistle with the fourth hole as big as all the rest.
2/ I think these holes are a touch to small and would guess this instrument would have trouble playing the octave 000 000 (or G#)
And for you both: What is the best way to express window distance to the labium, to the fipple block, or to the centre?
I’m not quite sure what this question is asking - but answering anyway!
The window calculation is problematic since it is not a round hole and the wall thickness varies, so even if you assume that a hole calculation is accurate (which it is not several assumptions/approximations are made which of course is par for the course in modeling something) it is hard to calculate this accurately - I even have a fudge factor in my spreadsheet to change as needed for this.
Anyway the measurement as for other holes is to the center of the hole, not to the fipple block (which may not be even with the windway exit - such as in a Generation Whistle) or to the labium. Of course it is simple math to calculate these distances and I find it easier to have a measurement to the labium for construction purposes.
Of course this does not matter much as long as you are close and making a tunable whistle - if I’m not sure about the lengths I just leave a little extra length on the top of the body (the holes are drilled measured from the bell) and then trim off the top of the body or bottom of the head as needed to get the whistle in tune (the details of this will depend on your tuning slide design).
One thought to share:
The first two whistles I made were pretty good. But then:
The more whistles I made the more I realized I knew less, until it started to make sense and I thought I knew more, about which time a whistle came along that did not work or worked and should not have and I knew I knew less until I made some sense of it and I knew a little more, maybe, and at some point I knew that I could make a whistle head that worked well but I also know that making a small change might completely break the balance. And now I’m working on a almost completely different design!
I can empathise with your last point, if, as I think, you are saying ‘the more I learn, the less I know’. That is how life seems to me. Like walking up a hill - the brow is right in front of you. Just a couple of steps. Except when you get there you realise that what you saw was not the summit. Or if it was, then there is a bigger summit after. (Now, why do I always visualise life as an uphill walk?)
Making my own whistle has added some to my understanding and appreciation of my bought whistles. Its easy for me to take such for granted - ‘oh the cross fingered Cnat is off’, ‘the bell note breaks too easily’, ‘the high a and b are too shrill’. Then I play my home made one, and realise just what fine whistles the others actually are… And then remember that these are largely hand made, reliably, repeatedly, and affordabley. The makers have my admiration.
I have just had the time to fully contemplate your helpful points Thomas. Thank you.
What is the ‘normal’ range of a whistle Labium Ramp angle? I like the idea of more stable pitch, and I have been tending to very shallow angles. So steepening the angle of the labium ramp seems the way to go for my next whistle head, but I do not want to make a recorder…
Does the width of the window/windway have effects other than 'back pressure? I am tempted to increase back pressure by reducing the windway from 15 to 10 mm. The embouchure hole width will be similarly reduced.
All other things being equal (which of course won’t be true!) a narrower windway will take less air, and be quieter, and the pitch will be flatter. Tone - I would be guessing so I won’t, I have not played with window width very much. Just my 2 cents.
Not for my sake please! I suggest you try and find the formulas which best model whistles. And that may be a tall order! But what is the point in offering all kinds of algorithms for the same purpose? Surely one will be closer to reality than another. Unless they vary for the circumstances. I see a long road of research ahead!
[quote=“hans” I suggest you try and find the formulas which best model whistles. And that may be a tall order! But what is the point in offering all kinds of algorithms for the same purpose? Surely one will be closer to reality than another. Unless they vary for the circumstances. I see a long road of research ahead![/quote]
I do not have the skills or the tools to make whistles accurately enough to test for fine differences. I do not have the tools or the ear training to measure the differences between the whistles. Nor do I understand the physics of whistles well enough to raise hypotheses to test. But apart from that…
Some of these shortcomings could be rectified, but, seriously speaking, the task you suggest is probably beyond me.
My target was to make a playable bass A, and with the help of many here I am getting close. I admit this makes me a consumer of knowledge that others have gleaned, but I dont think I can pay back in the same coin.
TWJCalc was a spin-off from my learning process, and if it can help somebody else that would be very pleasing. For those whistle researchers who can write in Java it could provide a framework for developing their own algorithms, maybe even for sharing and comparing their algorithms with others should they wish. I am happy to develop the program in response to the users needs, and provide support and help for those using it, but I don’t think I am going to be able to do meaningful whistle research.
