Trombone Air Flow Indicator

[ccktek]

Trombone instructors ask their students to strive to play with constant air flow through the instrument.  Some instructors advise practicing by lip-buzzing into a mouthpiece connected to a disposable medical spirometer.  The acoustic impedance of such a kludge is vastly different from that of a trombone, so the feel of playing a trombone is lost.  I describe here a device that monitors air flow into a trombone while it is being played.

Every trombone mouthpiece consists of a large cup followed by a focal narrowing (throat) followed by a  bore (backbore) of larger diameter than the throat.  This construction lends itself to use in a Venturi flowmeter, in which pressure-sampling orifices are placed in the cup and throat.  The pressure differential between the two orifices is an indicator of flow.  I drilled into the cup and backbore close to the throat of an old sacrificial mouthpiece (and later a new one) and cemented stainless steel tubes into the drilled holes.  A cemented baffle made from a bisected plumbing washer provides a barrier between the sampling orifice in the cup and the direct air jet entering the cup.*  Flexible plastic tubing leads away from the mouthpiece to a display unit.  I used tubing from medical IV sets.

Initially I measured pressure differences with a differential transducer followed by a D.C. amplifier.  This approach failed because a large-amplitude alternating pressure component superimposed on a much smaller average level evidently swamped the differential transducer.  The transducer output, even with lowpass filtering, did not reliably reflect air flow.  I achieved improvement with acoustic filtering of the mouthpiece signals by inserting a tiny pinhole within each sampling tube at the mouthpiece.  These resistive series elements in conjunction with the air volume in the plastic tubing formed lowpass filtering that effectively removed the alternating pressure component proximal to the transducer.  Then all was well until the tiny orifices became obstructed with fluid, which they did frequently, requiring clearing with compressed air.  However, the concept was proved.

The key to a successful design was the use of two gauge-type transducers, one for each sampling orifice, each followed by electronic lowpass filtering, the two signals then compared in a differential amplifier.  The tiny resistive orifices were eliminated, so the air pathways from mouthpiece to transducers were wide open.

The 100K resistors at the transducer outputs followed by 1uf capacitors to ground provide the lowpass filtering.  The signals then go to a differential amplifier which drives a variable gain amplifier to set sensitivity; this stage drives the bargraph. The transducers are NXP model MPXV7007, specified to measure -1 to 1 psi, corresponding to 0.5 to 4.5V output.   The LED display is a Barmeter Electronics model AE151S29Z 05CA7021H.  The odd power supply potentials are due to tailoring to parts on hand.

In the oscilloscope photo Ch1 & Ch2 are filtered and unfiltered backbore signals; Ch3 & Ch4 are filtered and unfiltered cup signals.  The center line is zero volts.  Note the difference in sensitivities between the filtered and unfiltered traces -- the p-p unfiltered signal voltages are more than an order of magnitude greater than the filtered averages.  Instantaneous pressures go well below atmospheric.  The note played was a middle Bb (about 233Hz) at moderate volume.

*Purists may point out that the baffle affects acoustic impedance and therefore the playing characteristics of the mouthpiece and trombone.  This is true, but the effect is not great -- about the same as experienced between unmodified standard mouthpieces having different specifications.

[coppercone2]

the lowest fluid impedance flow sensor that can be used in this situation is a carefully positioned hot wire anemometer in my opinion. Like a ultra fine gauge of platinum stranded through the trumbone through two pinholes which are sealed.

Not sure however, I do wonder if a thin wire can be presieved.

However for precise flow measurement I think air temperature heat transfer effects due to the trombone valving will be difficult to measure in some high level of precision

[djacobow]

What an interesting project. I don't have much to add, but as an EE and trombonist, I approve.

The hot wire method has been used in MIDI wind controllers for ages, going back to the venerable WX-7 and probably before. I wonder how they manage to deal with variation from the ambient temperature, the varying temperature of breath, and the changing temperature of the instrument itself. A trombone warms up quite a bit as it is played. (I've played in marching bands in subzero weather where we had to keep blowing through the instrument to keep the slide from seizing.)

Among all the horns I've played (some with F key, some without) as well as the many different mouthpieces, there has been substantial variation in the "resistance" I felt to air flow, and even more variation in what feels like the "inertia" of airflow. I suspect that the introduction of a wire, or some other obstructive measuring device (small turbine, etc) would be small relative to this inter-instrument variation.

Let's see a video of this thing in action!

[coppercone2]

you can maybe use another wire (as a RTD) to measure temperature. I want to say you can make a 'specialty' RTD very small so long you have beefy amplifiers to deal with it. I.e. keep in mind such a RTD would need to operate like +-10 degrees of room temperature (maybe more for hot breath) but you don't need to design for mass market with +-200C in mind of dynamic range.

I want to say I saw a single strand RTD used for chemistry use embedded in glassware in a specialty setup.

[MarkF]

I play the keyboard myself.

But I'm curious why you didn't make a male/female bushing (i.e. extension) that would fit between the mouth piece and the horn. That way the students could use their own mouth piece.  Maybe a slightly longer horn is just as bad?

Mouth pieces come in different sizes to fit the musician don't they?

[ccktek]

MarkF, I thought about doing it your way, but the Venturi flowmeter requires a focal  constriction.  If an additional assembly with a second constriction were added between the mouthpiece and the lead pipe I suspect there would be a fairly drastic change in the playing characteristics of the horn.  And of course the added length to the lead pipe would put the horn farther away from the player, would flatten intonation, and would alter relative slide positions, at least somewhat.  However, it's a pity that each mouthpiece used with the device must be essentially ruined.  Might still be worth a try.

[djacobow]

A partial solution would be switching to a screw-rim mouthpiece. Essentially, the lip portion unscrews from the cup. I think I've even seen variations where the shank can be changed, too. Then you're just hacking up one part of the mouthpiece.

http://mouthpiecefinder.conn-selmer.com/index.php/en-us/learn-about-mouthpieces/screw-rim-mouthpieces/


All that said, I think something that could be shoved down the horn of the instrument might be better, like one of those electronic practice mutes, but with air measurement capabilities.

[DaJMasta]

What about something like a hot wire anemometer integrated into the side of the cup/bore?  You're probably better off with a plastic mouthpiece, then, but you could probably drill out a metal one and pot the sensor in epoxy or something - there would be some protrusion in the bore, and whatever potting compound you use would probably reflect the sound differently from the normal mouthpiece wall (though probably less than the baffle version), but if you can keep the size down, you can probably make good measurements (of course, you could run into humidity and ambient temperature change issues, but these may just come out in the sensor calibration).

I think the risk of using a pressure sensor is the nature of the sound waves themselves.  With a 1ms response time for the sensor, you're theoretical top frequency is only 500Hz, which a trombone's fundamental can pass in the highest range, but the at times significant upper harmonic components could lead to aliasing issues.  It's an interesting approach to the problem, though.  I wonder if you can pull out some other interesting information about the playing from analyzing the output.