Fluid Dynamics—Building a better MIDI Breath Controller

[Freixas]

Background

A MIDI breath controller converts breathing to MIDI values, which are then used to control a MIDI instrument and produce sound. The ones I'm familiar with work through the use of a pressure chip.

For example, the TEC Breath and Bite Controller 2 uses the MPCV5010GP temperature-compensated pressure sensor. Here's a photo:

A naive design would have a tube connected to the pressure sensor. This would detect pressure, but does not map well to breathing. If we stop blowing into the tube, the pressure should drop to ambient. Breath controllers have a T-intersection near the pressure sensor. One end goes to the sensor, the other leads to an air exit. You can see the exit hole in the left side of the photo.

If the exit hole is too big, one has to blow very hard to register any pressure at all. If it is too small, the pressure won't release.

I've adjusted mine as well as I can. I can generally feel some back pressure and I run out of air sooner than with a comparable wind instrument.

The Question

Would anyone have ideas for alternate ways of measuring air flow through a tube? The ideal behavior would be:

• There is little or no back-pressure.
• The mechanism has to respond to changes in the mass flow, air speed, or pressure (any of these will work) within a few milliseconds.
• The design has to work no matter what angle it is positioned at.

Some ideas I've had are:

• Some sort of sealed pinwheel. The problem with this, of course, is getting it to stop when the air stops flowing.
• A small flap on a spring: when the air is flowing, the flap opens and the movement is detected. When the air flow diminishes, the spring restores the flap to its resting position.

The pressure sensors are probably a lot more convenient and are off-the-shelf parts. This may not be a practical DIY project, but I would still be interested in any ideas.

Bonus Points

The typical breath controller produces a lot of noise on the exhaust port. This is usually ignored when one is using the controller for studio work, but it's obnoxious for live performance. The exhaust port has to be relatively short to perform properly. Too long an exhaust port makes the pressure sensor work as though there were no exhaust port. I've create a muffler consisting of a short tube filled with aquarium floss. This creates back-pressure, though.

If someone came up with a practical design for measuring air flow through a tube without interfering with the flow, we still have to deal with the air exiting the tube, so I'm looking for a "muffler" for the exit that blocks or eliminates the sound without creating back pressure. That's sort of what car mufflers do, but I think they have to be "tuned" for the frequencies one wants to quiet. Copying a car muffler design might not prove useful.

 

[Borek]

My first thought: differential pressure measurement (perhaps realized with two sensors?), something similar to Pitot's tube, should be (at least in principle) able to measure the flow speed through the tube.

 

[Freixas]

My first thought: differential pressure measurement (perhaps realized with two sensors?), something similar to Pitot's tube, should be (at least in principle) able to measure the flow speed through the tube.

It's not really important if we measure air speed, pressure or mass flow. Any of these can be converted to the other. What is important is not impeding the air flow (or rather, impeding it as little as possible) as we measure whatever we're measuring.

I'm having a hard time determining whether a Pitot tube would help. We already have one pressure sensor; having two wouldn't seem to help. The Pitot tube itself measures relative pressures with a single sensor, but it doesn't seem much different than the pressure sensor I described--you still have a dead end for pressure sensing and an open end to allow for quick pressure changes.

 

[Freixas]

I attempted searches based on breath-controlled wheelchairs for quadriplegics.

https://scienceinfo.net/control-your-wheelchair-by-breathing.html
http://www.midisolutions.com/prodbth.htm

I'm confused.

The first link is a pointer to an article about a breath controller for wheelchair users. It provides no details of the implementation mechanism or even a link to the source of the information. A Google search lead me to an explanation that says it uses "sniffing" to control a wheelchair. Breath control works through the mouth, so this particular invention is likely not relevant--it could be, but I couldn't find any details of the implementation.

The second link is just a box for interfacing with a Yamaha BC2 breath controller. The Yamaha BC2 is not a relevant answer; if you know how it works and could describe it, then it might prove relevant.

The latter MIDI controller addresses the open tube question.

How?

I'm really looking for a method of measuring mass flow, air flow, or pressure (with the criteria I gave above), not an end product.

 

[jrmichler]

Would anyone have ideas for alternate ways of measuring air flow through a tube? The ideal behavior would be:

• There is little or no back-pressure.
• The mechanism has to respond to changes in the mass flow, air speed, or pressure (any of these will work) within a few milliseconds.
• The design has to work no matter what angle it is positioned at.

Some ideas:

1) Pitot tube (search the term) with the two connections to a differential pressure sensor. Search term differential pressure sensor will find good general information. Then go to an electronic distributor such as Digikey, Newark, Mouser or others to find a specific sensor with datasheet. Depending on how hard you blow through what size tube, you may want a sensor with full scale output somewhere between 1 inch of water and 10 inches of water. A pitot tube made from 1/16" ID plastic or brass tubing will fit inside a blowing tube 1/4" or larger diameter.

2) A mass air flow sensor with fast reaction time can be made from a thin resistance wire. Find a small incandescent light bulb, smash off the glass, mount it with the filament in a tube, and connect to a Wheatstone bridge with amplifier. I suggest starting with a flashlight bulb if you can find one, or an automotive dome light bulb. I know that this could be made to work because I once tested the idea when trying to build a fast response temperature sensor. It was much better at sensing air movement than temperature variation.

Either of the above two ideas should meet all three of the above requirements, including response time.

 

[berkeman]

A MIDI breath controller converts breathing to MIDI values, which are then used to control a MIDI instrument and produce sound.

Sorry, what is the transfer function? What parameters of "breathing" (pressure? rate? depth? etc.?) are used to produce what aspects of "sound" (volume, pitch, vibrato, what?)? [1]

I've adjusted mine as well as I can. I can generally feel some back pressure and I run out of air sooner than with a comparable wind instrument.

So it's something that you blow through like you're blowing through a trumpet or some musical instrument? I was under the impression that a MIDI breath controller was a fine control device of some kind...?

I'm confused.

Join the club.

What exactly are you trying to accomplish? What "MIDI parameters" do you want to control real time with your mouth? Are you limited to using breath pressure (or an inferior parameter air flow, since we all have limited lung capacity), or can you use things like lip and tongue pressure as part ot the mouth sensor?

--------------------------------------------------
[1] https://en.wikipedia.org/wiki/MIDI

 

[Freixas]

Some ideas:

1) Pitot tube (search the term) with the two connections to a differential pressure sensor. Search term differential pressure sensor will find good general information. Then go to an electronic distributor such as Digikey, Newark, Mouser or others to find a specific sensor with datasheet. Depending on how hard you blow through what size tube, you may want a sensor with full scale output somewhere between 1 inch of water and 10 inches of water. A pitot tube made from 1/16" ID plastic or brass tubing will fit inside a blowing tube 1/4" or larger diameter.

2) A mass air flow sensor with fast reaction time can be made from a thin resistance wire. Find a small incandescent light bulb, smash off the glass, mount it with the filament in a tube, and connect to a Wheatstone bridge with amplifier. I suggest starting with a flashlight bulb if you can find one, or an automotive dome light bulb. I know that this could be made to work because I once tested the idea when trying to build a fast response temperature sensor. It was much better at sensing air movement than temperature variation.

Either of the above two ideas should meet all three of the above requirements, including response time.

Thanks. You're the second person to mention a pitot tube. I had looked it up and can't say I quite get it, but I appreciate the detail you provided.

The mass flow sensor you described sounds really intriguing!

 

[Freixas]

Sorry, what is the transfer function? What parameters of "breathing" (pressure? rate? depth? etc.?) are used to produce what aspects of "sound" (volume, pitch, vibrato, what?)?

The breathing parameters are mass flow, air flow (speed), or pressure. Anyone of those would work. I believe that if you know the diameter of a tube, you can convert anyone of these to the other.

So it's something that you blow through like you're blowing through a trumpet or some musical instrument? I was under the impression that this was a fine control device of some kind...?

It is something you blow through, yes. And it is sort of like blowing into some musical instruments, although generally breath controllers don't support much of what you can do with, say, a trumpet or a reed instrument. Think of it like blowing into a straw and generating values based on how hard you blow.

What exactly are you trying to accomplish? What "MIDI parameters" do you want to control real time with your mouth? Are you limited to sending pressure (or an inferior parameter air flow, since we all have limited lung capacity), or can you use things like lip and tongue pressure as part ot the mouth sensor?

I'm trying to create a breath controller that doesn't require as much effort to use (effort in the sense of breathing effort).

With regards to this question, I'm only concerned with whatever one can generate using lung power. There are breath controllers that support other parameters such as bite pressure. The way these are handled are as separate controllers.

Old-style MIDI supports integer values from 0 to 127 from one controller. Information is sent when a value changes. The information consists of a controller value (breath control is generally 2) and the value. Most breath controllers I know measure pressure. If you were to add a bite sensor, it would be coded using some other controller value; the bite strength would still be from 0 to 127. But, as I said, let's focus just on the breath control.

Let me add that you don't need to worry about what I'll do with this input. MIDI controllers can be mapped to anything. If I have two virtual MIDI instruments, I might map the breath controller to different characteristics. When one designs a controller, one doesn't worry about what it will be used to control.

 

[Dullard]

Hot-wire anemometer.

 

[Freixas]

Hot-wire anemometer.

This sounds similar to @jrmichler suggested. Great idea, thanks!

 

[berkeman]

This sounds similar to @jrmichler suggested. Great idea, thanks!

I only skimmed the MIDI Wikipedia article, but it looks like you need your device to be pretty low power unless you get power from somewhere else, no?

 

[Freixas]

I only skimmed the MIDI Wikipedia article, but it looks like you need your device to be pretty low power unless you get power from somewhere else, no?

View attachment 299829

Are commenting on the hot-wire anemometer idea? Does it need a lot of power?

Most MIDI devices these days have USB connections or have MIDI and USB connections. But since I'm talking about creating a new type of breath controller, I could get power from anywhere. Ideally, one would want it to be capable of working either using USB power or household batteries.

 

[berkeman]

Are commenting on the hot-wire anemometer idea? Does it need a lot of power?

Yes I was, but I don't know enough about them to say how much power they take. The "hot wire" part is what made me think that it may take at least medium power, to heat up the wire... I'm off to Google...

 

[hmmm27]

Tube, with a variably-constrictive exit (so you can set it up the way you like), holding a photoelectric cell.

Light the flame ; oxygen flow (ie: breath) determines how bright it glows, or how close to the photocell the flame gets ; sensor output through an Arduino to translate and convert to MIDI signals.

Pipe angle determines a working range offset ; flame flicker adds an organic element.
Blow hard enough and flames shoot out the end.
Refill lamp oil reservoir as necessary.
Don't suck.


seriously though : if the return to zero is too slow, maybe your pressure sensor is busted, calibration is off or software sucks. Is there a mfr's user forum ?

I've used a BC1 (never really got the hang of it, though) ; I don't recall any problems with slew rate.

 

[Dullard]

This sounds similar to @jrmichler suggested. Great idea, thanks!

jrmilcher's bridge is one way to implement a hot-wire anemometer. I provided the general term to make your searches more productive. I'm not sure how fast you need this thing to be - I once did a similar job (mass flow meter) by alternately heating and measuring the resistance of a thermistor - it's easily accomplished with most modern microcontrollers. 'Breath' temperature is pretty well-defined - that fact could let you do all kinds of cute algorithmic shortcuts.

 

[Freixas]

seriously though : if the return to zero is too slow, maybe your pressure sensor is busted, calibration is off or software sucks. Is there a mfr's user forum ?

I never said the return too zero was too slow. I said that if you constructed a breath controller hooked directly to a pressure sensor such as the MPCV5010GP, it would not work well. This is not how the TEC BC2 is designed—it, like all other breath controllers I know, use a T (Y?) intersection with one end leading the the pressure sensor and the other function as an exit to ambient.

I've had a lot of discussions with the manufacturers. You get two options: a big exhaust opening (the opening size is adjustable) where blowing through offers little resistance but also is not good for measuring high pressures or a small exhaust opening where it's easy to get high pressure, but which feels like a lot of work.

For comparison, I play the melodica, which is a keyboard instrument with brass reeds controlled by breath. It's a lot less effort to play.

 

[Freixas]

Yes I was, but I don't know enough about them to say how much power they take. The "hot wire" part is what made me think that it may take at least medium power, to heat up the wire... I'm off to Google...

Assuming that this is the same idea suggested by @jrmichler, it doesn't seem like it should take a lot of power. He suggested using a wire from a flashlight bulb and flashlights generally operate on batteries. And it looks like most hot-wire anemometers you can buy operator on batteries. On the other hand, the are expensive and, I'm told, fragile. It is also sensitive to humid environments, and blowing into a tube creates a very humid environment.

 

[hmmm27]

I never said the return too zero was too slow.

Oh, okay, then what did you actually mean by...

If we stop blowing into the tube, the pressure should drop to ambient.

I've adjusted mine as well as I can. I can generally feel some back pressure and I run out of air sooner than with a comparable wind instrument.

At the risk of misconstruing your intent, again...
It sounds like you need a mapping table (or algorithm) for the conversion of pressure-sensor-to-CC values ; one that isn't linear (or at least is different from the default one in your current device).

You could mention that to one or more of the many manufacturers you've talked to : the phrase "pressure mapping table or algorithm" might jar something loose.

The product you mentioned earlier has one (more than one, actually, since it does things other than pressure-sensing as well) : the graph is halfway down the product page.

The typical breath controller produces a lot of noise on the exhaust port.

If you're building your own, I'd suggest using a wider tube body, and cloth or a length of foam-rubber as the "mute". That should reduce the hiss to effectively nothing. If you're modifying an existing device, stick a cork in it ; one with multiple holes.

 

[Freixas]

Oh, okay, then what did you actually mean by...

What I meant by "If we stop blowing into the tube, the pressure should drop to ambient." was that any alternate design also needs to meet this criteria. What I didn't say was that the TEC BC2 didn't meet this criteria. My question wasn't about building another breath controller similar to the TEC BC2, it was about other technologies that might accomplish the same thing without affecting the flow if air, which the TEC BC2 does.

At the risk of misconstruing your intent, again...
It sounds like you need a mapping table (or algorithm) for the conversion of pressure-sensor-to-CC values ; one that isn't linear (or at least is different from the default one in your current device).

You could mention that to one or more of the many manufacturers you've talked to : the phrase "pressure mapping table or algorithm" might jar something loose.

Yes, the TEC BC2 has this. And I've played with various curves. It doesn't solve the problem.

The product you mentioned earlier has one (more than one, actually, since it does things other than pressure-sensing as well) : the graph is halfway down the product page.

Yes, I know.

If you're building your own, I'd suggest using a wider tube body, and cloth or a length of foam-rubber as the "mute". That should reduce the hiss to effectively nothing. If you're modifying an existing device, stick a cork in it ; one with multiple holes.

I mentioned the muffling system I had designed--a tube with aquarium floss. Yes, this silences the noise, but it produces back-pressure. This is all in the OP.

Car mufflers are designed to reduce noise without creating any back-pressure, but I don't know anything about how they are designed. The ideal answer to the muffler question (and it might be asking too much) would be a set of directions for how to create a mini-car-like muffler or some equivalent alternate solution.

Adding foam or a cork with holes are solutions similar to what I already have.

 

[Tom.G]

As suggested by @jrmichler, a small flashlight bulb such as the #222 would have a small filament. The 222 is rated at 2.25V @0.27A. However it is so tiny with a thick glass shell it would be challenging to remove the glass.

With much thinner glass but a bit larger is the #PR2, 2.38V @ 0.5A.

Of course you would not be running either of them anywhere near their rated current. Sounds like time to experiment!

(above information from: https://www.bulbtown.com/)

And on to Mufflers:
A muffler is basically where the inlet gas goes thru some constriction feeding into an enclosed volume (an expansion chamber). Automotive mufflers often have three such sections; in either series or parallel depending on the particular design.

Here is an image of the series design:
https://en.wikipedia.org/wiki/File:MufflerInterior.jpg

And the first reasonable video of the parallel design that turned up in a search:


Cheers,
Tom

 

[Freixas]

And on to Mufflers:
A muffler is basically where the inlet gas goes thru some constriction feeding into an enclosed volume (an expansion chamber). Automotive mufflers often have three such sections; in either series or parallel depending on the particular design.

Thanks, Tom. The video was interesting in that it gives some design parameters. I'd have to check the frequency of the output sound from my little tube. He gave the formula for calculating the distance between the first two walls, but not the next. The whole thing is quite complex, isn't it?

An alternate approach would be to think about what causes the sound in the first place. For instance, would the exit be quieter if it were flared? Another thought is maybe using a longer hose and directing the air into a box with sound-proof walls. Sound-proofing foam, unfortunately is not sized for small boxes--I'd want something just a few inches in any dimension.

In the end, I suspect I would have to do a lot of trial-and-error work and the things I would have to try wouldn't necessarily be easy to build.

 

[Borek]

I only skimmed the MIDI Wikipedia article, but it looks like you need your device to be pretty low power unless you get power from somewhere else, no?

MIDI is mostly a transmission protocol, all devices typically have their own power.

 

[hmmm27]

Okay, I've been thinking of a relaxed exhale as the zero ; the OP I imagine uses a more controlled exhalation. Sorry about that, dude.

 

[Freixas]

Okay, I've been thinking of a relaxed exhale as the zero ; the OP I imagine uses a more controlled exhalation. Sorry about that, dude.

I'm not sure what you mean by "relaxed exhale". If you mean that air flows through the tube at a relatively low velocity, then that is indeed not zero. Zero means that the average velocity of the air molecules is 0.

If we have an open system, then air flow = 0 implies the relative pressure = 0. In an open system, if we know the tube size, measuring the mass flow, air flow (average velocity), and pressure should all work because we can translate anyone of these to any other. If we have a closed system, velocity could be 0 while the relative pressure could be > 0. This would be what happens if you connect a tube directly to a pressure sensor.

For a breath controller, you need an exhaust port close to the pressure sensor. If one stops blowing, the air in the tube quickly returns to ambient pressure, which should register as 0. People sometimes try to build a DIY breath controller without the exhaust port and quickly learn that it doesn't work.

 

[Freixas]

I'd like to summary the responses so far.

There have been two interesting suggestions: Pitot tubes and hot-wire anemometers. The latter sounds great, but I suspect it wouldn't survive the moisture-laden air that comes out of people's lungs (get a clear vinyl tube and blow through it--you'll see droplets of water forming within a few seconds. Blow into your palm and a little puddle will form.)

The pressure sensors people use have to be water-proof. The connection of the tube to the sensor has to be water tight because the chip is usually surrounded by electronics. Because I can't observe what happens when I use the breath controller, I don't know how much water builds up at the sensor connection. It doesn't seem to affect the response.

I don't know how well a Pitot tube would work in this environment. A Pitot tube appears to measure the relative pressure between ambient and whatever you're measuring. If the pressurized side collects water, I'm not sure if it will or won't work as well as a pressure sensor.

While the Pitot tube has a "dead end" just like the pressure sensor, its advantage appears to be that radius of the dead end is independent of the radius of the exhaust. With a pressure sensor, a large exhaust lowers the effectiveness of the sensor.

 

[jrmichler]

A hot wire anemometer can be kept at a temperature above the dew point of the air passing through it. The dew point of exhaled air cannot be higher than body temperature. As long as the hot wire is above the dew point temperature, water cannot condense on it.

A Pitot tube measures air velocity by measuring the pressure difference (differential pressure) between the velocity pressure and static pressure. If the differential pressure sensor is physically located above the Pitot tube, any liquid water in it will drain down into the tube.

 

[Freixas]

A hot wire anemometer can be kept at a temperature above the dew point of the air passing through it. The dew point of exhaled air cannot be higher than body temperature. As long as the hot wire is above the dew point temperature, water cannot condense on it.

It's a wet environment. Even if the anemometer wire can be kept above the dew point, there's nothing to stop a big, fat water drop from falling onto it.

A Pitot tube measures air velocity by measuring the pressure difference (differential pressure) between the velocity pressure and static pressure. If the differential pressure sensor is physically located above the Pitot tube, any liquid water in it will drain down into the tube.

One of the constraints I mentioned in the OP is the need to be able to operate in any orientation. Now, I know you will think "Can't you guarantee that the sensor is always above the tube?" And it's possible, but it eliminates some design possibilities.

For example, I'd like to use a melodica hose connected to a little black box that functions as the actual breath controller. This box would include any sensors and perhaps also a muffler. I would then velcro attach this box to a MIDI keyboard. In my case, I use a KORG microKey keyboard controller with an added guitar strap: I "wear" the keyboard and treat the whole system like a MIDI melodica. While playing I might move the keyboard in various orientations.

In case you're curious, my YouTube channel contains videos of me playing a melodica and playing various versions of a MIDI melodica. The later one show my latest muffler design. The advantage of using a melodica hose is that there is no need for a headset. If nothing else, I'd like to move the TEC card into a new container that would allow me to connect it to a melodica tube.

 

[Borek]

Use platinum wire and it will survive everything. And - as you want it thin - it won't be really expensive.

But there are several alloys used for resistive wires, like kanthal, nichrome, chromel - they are made of metals that are quite resistant to corrosion, so I won't be surprised if they can survive as well.

 

[Freixas]

Use platinum wire and it will survive everything. And - as you want it thin - it won't be really expensive.

But there are several alloys used for resistive wires, like kanthal, nichrome, chromel - they are made of metals that are quite resistant to corrosion, so I won't be surprised if they can survive as well.

Good to know, but I suspect the problem is a bit harder to solve.

A hot wire anemometer, from what I've read, determines temperature from the cooling effect of moving air. And it's just a short piece of wire. If the piece of wire is coated in water it is 1) unlikely to sense any air flow, 2) the cooling effect of the water would probably make it useless, and 3) surface tension might make it hard to shed any water drops (plus there are always more coming).

Corrosion seems like the least of the problems.

 

[Borek]

As it was already pointed out, wire will be hot enough to not allow any condensation on its surface. Condensing water will not cool it, quite the opposite - but even if, it will just shift the calibration which you have to do anyway.

But yes, water can be a problem in general, as it will condense on other parts of the duct. I wonder if keeping it heated to about 40°C would not keep whole thing dry, exhaled air has a dew point well below the temperature of the human body. Add few degrees to allow drying of saliva. It will again shift the calibration and it again doesn't matter.

 

[Freixas]

As it was already pointed out, wire will be hot enough to not allow any condensation on its surface. Condensing water will not cool it, quite the opposite - but even if, it will just shift the calibration which you have to do anyway.

But yes, water can be a problem in general, as it will condense on other parts of the duct. I wonder if keeping it heated to about 40°C would not keep whole thing dry, exhaled air has a dew point well below the temperature of the human body. Add few degrees to allow drying of saliva. It will again shift the calibration and it again doesn't matter.

Hey, thanks for continuing to think about this.

I've observed what happens in a clear vinyl tube as I blow into it. Water droplets form and are blown this way and that. If one has a hot wire in the path, no condensation will form on it, but I imagine that big, fat water droplets will fall on it...and get blown off, all in a random pattern that would defy calibration.

If you heat the whole tube, that's another story--and another complication. I have a CPAP with a heated humidifier hose, so it's clear it's doable, but I don't want to try to build one (nor would I want to try to adapt a CPAP hose).

I do know of one melodica player who has installed a water trap on his hose and is thinking about heating it. This gets way too DIY for my tastes. My dream device would be a small black box that connects with standard, off-the-shelf melodica hoses and mouthpieces.

By the way, contrary to what some people think, this condensation is water, not spit. One should not spit into one's instrument--and, yes, the term "spit valve" is a misnomer.

 

[Borek]

I know it is mostly condensation, but now and then a droplet of saliva is unavoidable. We do exhale them all the time, the stronger you blow, the more of them. Even if the tube is heated and there is no condensation such droplets would appear - and they have to dry out.

 

[Tom.G]

Hmm... I wonder if some baffling in the air stream would work... it would probably take some creative experiments though.

My thinking is the more massive water droplets would tend to flow in more of a straight line, hopefully impinging on baffles where gravity can do its work.

Another possibility is have the main airway a straight-thru path with the hot-wire in a (smaller?) parallel path; inlet side plumbed with a "T" and outlet with a "Y" to blend the two airstreams (and supply some pneumatic gain?). There could be a user-operated valve for calibration, etc. in the straigh-thru segment if desired.

Ah well, enough brainstorming.

Cheers,
Tom

 

[Freixas]

Hmm... I wonder if some baffling in the air stream would work... it would probably take some creative experiments though.

My thinking is the more massive water droplets would tend to flow in more of a straight line, hopefully impinging on baffles where gravity can do its work.

Another possibility is have the main airway a straight-thru path with the hot-wire in a (smaller?) parallel path; inlet side plumbed with a "T" and outlet with a "Y" to blend the two airstreams (and supply some pneumatic gain?). There could be a user-operated valve for calibration, etc. in the straigh-thru segment if desired.

Ah well, enough brainstorming.

Cheers,
Tom

I don't mind the brainstorming. My OP was kind of a theoretical question, although if someone were to give me an answer within my skill levels, I would indeed want to build this.

No solution can require a fixed orientation. Nor is it acceptable that it works "most of the time". If I am giving a live performance (even for friends), I wouldn't it to go screwy because a water drop found its way to the hot wire.

It's certainly an interesting problem. How do you run a hot wire anemometer in a super-humid environment? Heating the tube is probably the only thing that would really work.

You know, I thought I'd get more ideas along the lines of the pinwheel or flap ideas that I posted in the OP. The flap idea seemed particularly good, although I don't know how I measure the angle of the flap. I'm sure there's something. The flap would need to be on a spring and it would be great if there were a way to easily adjust the tension. It seems as though it would work at any angle and in a humid environment. It would certainly require a ton of prototyping.

 

[Borek]

If the flap were to react to the slow flow/low pressure spring should be rather weak, so the flap could easily dangle when you move the instrument.