It is currently Mon Mar 18, 2019 9:54 pm

All times are UTC - 6 hours




Post new topic Reply to topic  [ 60 posts ]  Go to page Previous  1, 2, 3, 4
Author Message
 
PostPosted: Sat Jan 12, 2019 8:12 pm 
Offline
User avatar

Joined: Sun Dec 12, 2004 4:12 pm
Posts: 1981
Location: Malua Bay, on the NSW Nature Coast
Tunborough wrote:
The resulting flute won't play in tune for three reasons:
- The headjoint insert isn't in yet.
- I haven't got the calibration factors right.
- The WIDesigner model needs work.

But with luck, the tuning you measure from this geometry will get us closer on all three of these.

Sounds like a good enough reason to proceed.

OK, done all that and had quick play with it. Regard this as preliminary only. It plays very differently to my normal playing flute, so I need to adjust to it. Ignore the spurii (the notes not in the D or G scales).

Code:
            Note         Samples       Delta in Cents
              D4      790      8
             D#4      31      -18
              E4      513      21
              F4      83      4
             F#4      773      12
              G4      515      10
             G#4      98      4
              A4      1631      22
             A#4      191      -16
              B4      1772      14
              C5      553      15
             C#5      492      -7
              D5      1549      11
             D#5      73      -7
              E5      830      7
              F5      32      0
             F#5      827      1
              G5      260      1
             G#5      74      15
              A5      1028          -12
             A#5      56      0
              B5      1036          -20
              C6      996      1
             C#6      680      -2
              D6      588      8


I think that's pretty stunning for our first attempt. It's got the classic top-end droop of a non-compensated cylindrical bore, but of course we expect that. Bottom end notes are a bit sharp, but I imagine the introduction of the head insert will reduce that. But that could be me too, so don't jump to conclusions.

We will have to agree on fingerings. All the usuals are of course not negotiable. But, for the more negotiable notes, these work well:
Code:
C6      oxo xxo
C#6     oxx xoo
D6      oxx xox


Well done! Is this an historic moment? Or not quite yet?


Top
 Profile  
Reply with quote  
 
PostPosted: Sat Jan 12, 2019 9:37 pm 
Offline
User avatar

Joined: Sun Dec 12, 2004 4:12 pm
Posts: 1981
Location: Malua Bay, on the NSW Nature Coast
I should add, I did a really dopey thing. I lined up the embouchure hole with the finger holes. But I never play like that, I have the head rotated back towards me, so that my left elbow is down near my chest. Now I'm playing the Orange Flute with my left elbow stuck out straight in front of me. Oh well something to fix in iteration II. If we don't have reason to invent a tuning slide in the meantime....

Just had a play, boring out a scrap of the pipe and turning down another to act as a tuning slide. That could work if we need it....

I should also mention that, at this point I feel the flute is pretty weak. Could be a number of things involved:
- embouchure hole a little undersized and not undercut
- the plastic is light and flexible compared to our woods and polys
- the wall thickness is thin
- the harmonics are not well lined up (compare the same note in each octave).

That last one hopefully will be improved by the insert, but not the others. But that's OK. We're not trying to make a world-record shattering flute at the moment, just an in-tune one. We can deal with the other issues later.


Top
 Profile  
Reply with quote  
 
PostPosted: Sun Jan 13, 2019 9:34 am 
Offline

Joined: Sun Dec 05, 2010 2:59 pm
Posts: 896
Location: Southwestern Ontario
Not bad, not bad at all. With an adjustment to the calibration, WIDesigner gets within 6 cents of all of your measured values (using your fingerings), except for:

D5: -9
E5: -10
B5: +12
C6: +21
C#6: +11

In some respects, changing the calibration is cheating. You should have to calibrate only once for a given flute. However, the new calibration factors still give good agreement with your measurements before drilling the toneholes. We just didn't have enough data points to get better numbers. Now, we've got lots of data points, and really, if the predictions are within +/- 10 cents, I'm delighted.

Now, the moment of truth. I'm assuming that the cylindrical headjoint insert will reduce the bore to 18.8 mm. In that case. the insert should run from the stopper to 65 mm past the centre of the embouchure hole. If the I.D. is more like 18.7 mm, use 63 mm, and for I.D. 18.6 mm use 59 mm.

The tonehole layout assumed the following fingerings. If you want to use your fingerings, we'd have to try again with a slightly different layout.

C6: OXO XXX
C#6: OOO OOO
D6: OXX XXX


Top
 Profile  
Reply with quote  
 
PostPosted: Sun Jan 13, 2019 5:17 pm 
Offline
User avatar

Joined: Sun Dec 12, 2004 4:12 pm
Posts: 1981
Location: Malua Bay, on the NSW Nature Coast
OK, very interesting.

Very roughly achieved and poorly played, but very encouraging. (I've removed the spurii this time)

Code:
             Note        Samples   Delta in Cents
              D4      1102      2
              E4      1314      6
             F#4      1098      0
              G4      1682      -1
              A4      1918      5
              B4      1510      5
              C5      616      2
             C#5      734      -10
              D5      2083      13
              E5      744      8
             F#5      442      4
              G5      1491      7
              A5      1646      -2
              B5      1953      -6
              C6      1216      -12
             C#6      756      -35
              D6      789      13



A couple of observations:

C6 a bit flat. Responds well to the traditional tuning oxo xxo

C#6 is still struggling. I've never felt good about ooo ooo in the 2nd octave. The traditional oxx xoo works very well. (Really a 3rd partial of F#4, just as oxx ooo is a 3rd partial of G4)

Note that D5 is a little sharp of D4. I wondered how we'd go with this. I'm assuming (perhaps wrongly) that it's because of the naturally better venting D notes get. In our conical flutes, we use some backreamed flare at the bottom to combat this issue. Or go for the extra 2 holes in the foot region to artificially reduce D's venting, which sometimes then introduces other issues!

Now, I reckon I'm about ready to go set up camp outside our local electrical trades outlet and busk some money off the sparkies....

It's still only half as good as my own playing flute, but if I were on a desert island, I'd be very happy to have it!

Where to now, Tunborough?


Top
 Profile  
Reply with quote  
 
PostPosted: Sun Jan 13, 2019 8:42 pm 
Offline

Joined: Sun Dec 05, 2010 2:59 pm
Posts: 896
Location: Southwestern Ontario
Here is what WIDesigner predicted for the flute with the headjoint insert.
Code:
Note    Target      Predicted     Cents           Min f           Max f
D4      293.66        292.79       -5.1           279.28          293.82
E4      329.63        331.11        7.7           314.93          333.10
F#4     369.99        369.79       -0.9           350.71          372.94
G4      392.00        391.06       -4.2           370.26          394.83
A4      440.00        441.81        7.1           416.68          446.93
B4      493.88        496.36        8.7           466.25          503.12
C5      523.25        527.53       14.1           495.72          535.02
C#5     554.37        550.93      -10.8           515.66          559.44
D5      587.33        586.93       -1.2           570.95          596.40
E5      659.26        657.92       -3.5           639.95          668.35
F#5     739.99        739.63       -0.8           718.13          752.50
G5      783.99        784.90        2.0           761.08          798.91
A5      880.00        880.61        1.2           853.91          897.41
B5      987.77        991.28        6.1           960.82         1012.10
C6     1046.50       1042.73       -6.2          1020.63         1065.04
C#6    1108.70       1096.81      -18.7          1063.99         1121.61
D6     1174.70       1182.84       11.9          1172.21         1211.36

With your fingerings, the last three notes become:
C6     1046.50       1069.64       37.9          1043.41         1090.38
C#6    1108.70       1123.24       22.5          1100.56         1146.05
D6     1174.70       1184.90       15.0          1173.38         1211.44

Comparing that to your measured values, all of the predictions are within 10 cents except these.

C5: +12
D5: -14
E5: -12
B5: +12
C#6: +17

The worst prediction from the last batch, 21 cents out for C6, is only 6 cents out this time around. At least some of the difference may be playing style. Are you accustomed to giving D5 and E5 a little extra push because they are in the second octave, when they don't need that on this flute?

WIDesigner did predict that C#6 would be flat, although not nearly as flat as you found. It predicts that it would be possible (just) to blow C#6 in tune if you leaned into it more than you possibly could in normal playing. Is this accurate? (It predicts the air would have to be moving about twice as fast to get C#6 in tune as for C#5.) For C#6, your preferred fingering would probably be more in tune. C6 will probably end up too sharp, and D6 will be about the same.


Top
 Profile  
Reply with quote  
 
PostPosted: Sun Jan 13, 2019 9:07 pm 
Offline

Joined: Sun Dec 05, 2010 2:59 pm
Posts: 896
Location: Southwestern Ontario
Terry McGee wrote:
It's still only half as good as my own playing flute, but if I were on a desert island, I'd be very happy to have it!

Where to now, Tunborough?
Well, I'd say we had two theses to test:

1) A stepped cylindrical headjoint can produce a cylindrical flute with good intonation between the octaves.

2) WIDesigner can predict the playing frequencies of a transverse flute within +/- 10 cents.

I'd say we have successfully demonstrated our first thesis. The intonation, as you measured it, is remarkably even through the whole range. C#6 is flat, but I knew it was going to be at least somewhat flat. If we tried again, using your preferred fingering and perhaps sacrificing some of the tuning of C5/C#5, I'm sure we could do better. You said the flute was pretty weak. Again, we could probably do better with a second try, and the weakness isn't the result of stepped cylindrical insert, since it was present before the insert went in.

Anyone interested in making simple cylindrical flutes might want to consider using a stepped cylindrical headjoint instead of a wedge or other tapering.

I think we have a qualified success on the second thesis. With calibration, WIDesigner was able to predict most, but not all, of the playing frequencies. WIDesigner was right about the stepped cylindrical headjoint. It was able to predict the effect of a change to the flute geometry. I do need to spend more time with data you and paddler have collected, to see if it warrants changes to the underlying flute mouthpiece model. And I'd be happy to receive any more data that you can provide.

Thanks so much for all your work to make this exercise a success.


Top
 Profile  
Reply with quote  
 
PostPosted: Sun Jan 13, 2019 10:45 pm 
Offline
User avatar

Joined: Sat Nov 03, 2007 7:19 pm
Posts: 425
Location: Hood River, Oregon, USA
I would consider this a success. The degree of accuracy in tuning predictions is impressive. Plus or minus 10 cents or so
is really nothing. You get more variability than that as your embouchure changes day to day, and it is very easy for a player
to lip notes up or down by this amount. Its totally unreasonable and unrealistic to expect "perfect" predictions because
the player has such a huge influence on the tuning. So, in my mind this is an impressive result for WIDesigner. Although,
it must be said that the geometry of this particular flute is very simple, being basically cylindrical with one step.

Whether the flute produced is a good flute or not is a whole different question though. Being reasonably well in tune is a
necessary condition for a flute to be considered good, but it is far from being a sufficient one. In my opinion it is a very tall
order indeed to expect a computer model to be able to design a good flute. And I do not expect that to happen any time soon,
if ever. However, to me, it is becoming quite clear how such a model can be used by a flute maker to streamline existing
exploratory processes for making a good flute.


Top
 Profile  
Reply with quote  
 
PostPosted: Mon Jan 14, 2019 1:32 am 
Offline
User avatar

Joined: Sun Dec 12, 2004 4:12 pm
Posts: 1981
Location: Malua Bay, on the NSW Nature Coast
Tunborough wrote:
Thanks so much for all your work to make this exercise a success.

Heh heh, thanks for your work getting us this far! And well done!

I agree that the relative weakness of the flute is unlikely to be a WID issue. We'd really need to build a stepped bore flute in better materials than this cheap plastic stuff. I'm aware that there are pressure pipes and poly irrigation risers that probably have a bit more weight in them, but I haven't looked into it. I will if nobody else does!

The flute definitely feels and sounds better now that the partials are better aligned. There's something a bit doleful about the sound and feel of a plain cylinder flute, compared to the sweetness of a well-tuned instrument. Which is interesting, as it's not as simple as the harmonics being out of tune. They aren't when playing the low octave, as the jet switching resynchronises them at every cycle. Yet we hear and feel something. Interesting.

So, it would appear we have a powerful new tool to assist us to bring in the flutes of the future. And to understand better the flutes of the past.


Top
 Profile  
Reply with quote  
 
PostPosted: Mon Jan 14, 2019 2:17 am 
Offline
User avatar

Joined: Sun Dec 12, 2004 4:12 pm
Posts: 1981
Location: Malua Bay, on the NSW Nature Coast
Oops, and I should add. We're naturally quick to blame the software for any variations between predicted and measured. But I'm at least equally culpable. We have no way of knowing how typical a blower I am. Am I hearing errors and attempting to ameliorate them? Or do I have a systematic sharpness or flatness going up scale? Or is my unfamiliarity with this style of flute mean I'm blowing harder or softer than WID imagines? We really need some other mouths and ears involved.


Top
 Profile  
Reply with quote  
 
PostPosted: Mon Jan 14, 2019 11:09 am 
Offline

Joined: Sun Dec 05, 2010 2:59 pm
Posts: 896
Location: Southwestern Ontario
paddler wrote:
Whether the flute produced is a good flute or not is a whole different question though. Being reasonably well in tune is a necessary condition for a flute to be considered good, but it is far from being a sufficient one. In my opinion it is a very tall order indeed to expect a computer model to be able to design a good flute. And I do not expect that to happen any time soon, if ever. However, to me, it is becoming quite clear how such a model can be used by a flute maker to streamline existing exploratory processes for making a good flute.
Agreed on all counts. I had a very specific question for WIDesigner: For a flute with a cylindrical body and a headjoint that contracts toward the stopper, what geometry gives the best intonation? The answer, a stepped cylinder, was quite surprising, but it it looks like it does the job. Is it the best geometry for intonation? We can't be sure, but it looks pretty good. Does it make for a good flute in other ways? The only way to know for sure is to build them. ... And refine them based on the flute maker's own experience.

paddler wrote:
Although, it must be said that the geometry of this particular flute is very simple, being basically cylindrical with one step.
The bore profile is quite simple, but the bore profile algorithms are arguably the most time-tested, trustworthy part of WIDesigner (as long as we don't get into extreme flaring bells). The tonehole model is pretty sound. The parts that I have least confidence in are the flute mouthpiece model and the headspace model, and both acquitted themselves very well in this exercise.


Top
 Profile  
Reply with quote  
 
PostPosted: Sun Feb 17, 2019 3:07 am 
Offline
User avatar

Joined: Fri Nov 01, 2002 6:00 pm
Posts: 2217
Location: The North of Scotland
From a discussion about octave tuning on whistles (http://forums.chiffandfipple.com/viewtopic.php?f=1&t=108141) Tunborough kindly pointed me to this topic, which I read read only yesterday with great interest. I can now add results of my own experiments just done:

Occasionally I've build some PVC flutes, using 3/4" Schedule 40 US waterpipe. This is a great material for flutes, as the wall is about 3mm thick (26.8mm OD, 19.8mm ID). I have used a Fajardo wedge, cut out of another PVC pipe, for octave tuning in the past, but it is a bit tricky to cut and to glue in place. So this thread and the approach I took with my whistle designs made me think of using a cylinder for a bore restriction. I've got 20mm PVC conduit tubing of about 1.1mm wall thickness, which fits snugly into the Schedule 40 pipe. First I made an experiment with an existing PVC flute in Eb, which was non tunable and had no Fajardo wedge fitted, and displayed a flat second octave, getting really flat the higher the notes are, as usual. I fitted pieces of 20mm conduit, trying a variety of length, and also positions. What worked best was a length of 50mm (40mm and 60mm did not do so well), and fitting it just below the embouchure hole, so the hole is just free of this restriction. Now I tried the same on a D flute, again non-tunable, and a 50mm tube insert below the embouchure hole had similar good effect. On both flutes the tuning went about 20 cents flat, adjusting my electronic tuner from A 440Hz to A 435 to compensate showed good intonation in both octaves. My ear tells me the same, when I overblow each note into the next register and hear the two notes are in good harmony an octave apart.

So I am very happy with this result and looking forward to make a flute with that 20 cents adjustment, to get a fully in-tune flute at A 440Hz, and transfer that to a tunable PVC flute.

Now I wonder what that all means in terms of the modelling?

PS: applied the same with success to a C PVC flute and a D Gypsy PVC flute, and my only tunable D PVC flute here. For calculating holes for a non-tunable flute I guess I can set the design A frequency to 445Hz, so by adding the bore restriction cylinder it will result in a flute based on A 440Hz, about 20 cents lower. Or I change the calculator and add that factor to the calculations.

_________________
Hans Bracker - Handcrafted Whistles from Scotland - http://www.music.bracker.uk
Custom built aluminium whistles in any key and other speciality whistles and flutes


Top
 Profile  
Reply with quote  
 
PostPosted: Mon Feb 18, 2019 5:32 am 
Offline
User avatar

Joined: Fri Nov 01, 2002 6:00 pm
Posts: 2217
Location: The North of Scotland
Since then I changed the length of the insert to 40mm, which seems to work better than the 50mm today! 30mm was not giving enough second octave lifting. So something around 40 to 50mm seems okay. I report back if i get more refined results.

_________________
Hans Bracker - Handcrafted Whistles from Scotland - http://www.music.bracker.uk
Custom built aluminium whistles in any key and other speciality whistles and flutes


Top
 Profile  
Reply with quote  
 
PostPosted: Wed Feb 20, 2019 7:19 am 
Offline

Joined: Sun Dec 05, 2010 2:59 pm
Posts: 896
Location: Southwestern Ontario
hans wrote:
Now I wonder what that all means in terms of the modelling?

PS: applied the same with success to a C PVC flute and a D Gypsy PVC flute, and my only tunable D PVC flute here. For calculating holes for a non-tunable flute I guess I can set the design A frequency to 445Hz, so by adding the bore restriction cylinder it will result in a flute based on A 440Hz, about 20 cents lower. Or I change the calculator and add that factor to the calculations.
The position and size of the insert that WIDesigner recommends will depend on the flute. For some designs, it wants the insert to start higher up than the bottom edge of the embouchure hole. With the right design, you can get a slight improvement by pulling the stopper out a bit, and extending the insert all the way to the stopper. However, that's only a slight improvement, and for all practical purposes, it does look like it makes sense to start the insert near the bottom edge of the embouchure hole.

I modelled slightly thicker and thinner walls on the insert, and it looks like a wall size around 0.5 mm works well.

For designing a flute that's in tune with an insert, well ... The current release of WIDesigner doesn't include the new features that I'm using, so you'd have to wait for the next release if you wanted to use WIDesigner. I took a design that was optimized with an insert (of 48 mm), and modelled taking out the insert; if I pushed the stopper in 3 mm, I got a flute that was in perfect tune in the second octave, and 10 to 20 cents sharp from E4 through C5. Does that help?


Top
 Profile  
Reply with quote  
 
PostPosted: Wed Feb 20, 2019 9:31 am 
Offline
User avatar

Joined: Fri Nov 01, 2002 6:00 pm
Posts: 2217
Location: The North of Scotland
Tunborough wrote:
I modelled slightly thicker and thinner walls on the insert, and it looks like a wall size around 0.5 mm works well.

For designing a flute that's in tune with an insert, well ... The current release of WIDesigner doesn't include the new features that I'm using, so you'd have to wait for the next release if you wanted to use WIDesigner. I took a design that was optimized with an insert (of 48 mm), and modelled taking out the insert; if I pushed the stopper in 3 mm, I got a flute that was in perfect tune in the second octave, and 10 to 20 cents sharp from E4 through C5. Does that help?

Just done a quick test with an insert 0.5mm wall thickness (so the bore gets reduced from 20mm to 19 mm over 50mm length below the embouchure hole). The resulting sharpening of higher second octave notes is not sufficient. My previous ca 1mm wall thickness insert works a lot better. I also find this works good enough with my existing flutes, designed with A=440Hz. I do not need to change the design to 435Hz I think. I still have to account for some top-side undercutting I used on some holes on some flutes, which also has the effect of sharpening the second octave relative to the first. I'll continue my experimental research...

_________________
Hans Bracker - Handcrafted Whistles from Scotland - http://www.music.bracker.uk
Custom built aluminium whistles in any key and other speciality whistles and flutes


Top
 Profile  
Reply with quote  
 
PostPosted: Wed Feb 20, 2019 7:42 pm 
Offline

Joined: Sun Dec 05, 2010 2:59 pm
Posts: 896
Location: Southwestern Ontario
Sorry, I confused your Killarney whistle insert with the insert you were using on PVC flutes. Yes, the modelling suggests a 1 mm insert can do a good job, too. Not surprisingly, it doesn't have to be as long. Also, it works best a bit farther down the tube; for one example, WIDesigner suggested a 35 mm insert that started 11 mm below the embouchure hole.

Again, when I took out the insert, and pushed the stopper 5 mm closer to the embouchure hole, I got a flute that was in tune in the second octave, and 10-20 cents flat from E4 through C5. With the insert removed and the stopper pulled out instead, the lower octave was a touch sharp, and most of the upper octave was 20-30 cents flat. ... for what that's worth.


Top
 Profile  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 60 posts ]  Go to page Previous  1, 2, 3, 4

All times are UTC - 6 hours


Who is online

Users browsing this forum: No registered users and 7 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group
[ Time : 0.099s | 11 Queries | GZIP : On ]
(dh)