air speed sensor for rc plane

[Greg Needel]

i am helping a friend with an autonomous plane project where he is using an array of senors and completely control the plane. The only major problem that we are finding is figuring out a good way to do air speed. Most major aircrafts use pressure but for low altitudes this will not be accurate enough. other sensors use a heated element but within the confounds of this project for low power consumption and layout that wont work either. we have been playing with ideas with strain gages but are looking for an easier solution. any help would be appreciated.

requirements
low amperage draw
small (less then 2 inches)
low weight (less then 1/8th lb)
proven technology (we arn't trying to re-invent the wheel)

any ideas?

[Tristan Lall]

What sort of control system is on board? Is there any way to interface with an RS-232- or USB-based GPS system, for example?

[sanddrag]

Well, take for example an ordinary fan. The faster you spin the motor, the faster the air flow is coming out of it. Now reverse the idea. Instead of a motor, put some sort of encoder on the fan blade. The faster you are moving through the air, the faster the fan will spin (and the encoder will read this). I'm guessing it is a linear relationship but I could be wrong. In any case, with a little trial and error you should be able to come up with a simple equation that relates fan RPMs to airspeed. The only thing you need to make sure of is a very low friction encoder or encoding system (slotted wheel and photogate perhaps). I'm willing to bet you could do it with a relatively small fan blade, something like what's in the fan on top of a Victor 883. Just make sure everything is low friction so it spins freely.

Now, remember this will only give you your speed relative to the air, not to the ground. If there is wind, the readings will be off.

[mechanicalbrain]

Originally Posted by sanddrag

Well, take for example an ordinary fan. The faster you spin the motor, the faster the air flow is coming out of it. Now reverse the idea. Instead of a motor, put some sort of encoder on the fan blade. The faster you are moving through the air, the faster the fan will spin (and the encoder will read this). I'm guessing it is a linear relationship but I could be wrong.

How about a volt meter? The faster the motor spins the higher the voltage! And it can act as an alternator to charge the battery! However as to the aerodynamicness (yes i know its not a real word! ) of it all....

[Tristan Lall]

Originally Posted by mechanicalbrain

How about a volt meter? The faster the motor spins the higher the voltage! And it can act as an alternator to charge the battery! However as to the aerodynamicness (yes i know its not a real word! ) of it all....

You're thinking of a ram air turbine; those were used to drive the ECM pods on EA-6s and can be found as emergency power devices on many large aircraft (they deploy if all engines fail, for instance). On the EA-6B, I believe that combat radius decreased by 5% for each RAT-equipped ECM pod carried, so at least in that application, they were non-negligible. Plus, conservation of energy dictates that you're just going to waste energy by trying to charge the battery, if the battery is what's causing the forward motion of the plane (it's different story if it's powered separately, such as with a gas engine). And then, of course, there's the issue of charge rate; I doubt you can charge the battery fast enough to make up for a significant portion of the power output (again assuming it's an electrically-propelled aircraft).

Also, a plane moves in 3-D space; if there is pitch or yaw relative to the axis of the turbine (e.g. high-angle-of-attack or flat-turn maneouvres), it might not read as expected. In any case, I'm guessing that the relation between fan motion and airspeed has something to do with the drag force on the fan blades, which ought to be roughly proportional to the square of the velocity. If this is the case, though, C_d changes with the angle of the fan's axis relative to the direction of travel, so you'd need additional equations to model that dependency. (Yuck!)

If you don't like GPS, or a turbine, what about INS? Couldn't you set up some sort of 3-axis accelerometer, and continuously integrate its outputs over some short time periods? Actually, it seems like I'm thinking, once again, of groundspeed, rather than airspeed. I'd have to say that groundspeed is much more useful for navigation, anyway....



Edit: Here's an interesting idea I just dug up: GPS and INS. However, it sounds bulkier than you'd anticipated, difficult, and computationally intensive. Give it a few years:

Some sensor designers even foresee that by 2010, it will be possible to obtain navigation performance of 0.01 deg/h using a MEMS IMU only 2in

³in size.

[mechanicalbrain]

Originally Posted by Tristan Lall

You're thinking of a ram air turbine; those were used to drive the ECM pods on EA-6s and can be found as emergency power devices on many large aircraft (they deploy if all engines fail, for instance). On the EA-6B, I believe that combat radius decreased by 5% for each RAT-equipped ECM pod carried, so at least in that application, they were non-negligible. Plus, conservation of energy dictates that you're just going to waste energy by trying to charge the battery, if the battery is what's causing the forward motion of the plane (it's different story if it's powered separately, such as with a gas engine). And then, of course, there's the issue of charge rate; I doubt you can charge the battery fast enough to make up for a significant portion of the power output (again assuming it's an electrically-propelled aircraft).

Also, a plane moves in 3-D space; if there is pitch or yaw relative to the axis of the turbine (e.g. high-angle-of-attack or flat-turn maneouvres), it might not read as expected. In any case, I'm guessing that the relation between fan motion and airspeed has something to do with the drag force on the fan blades, which ought to be roughly proportional to the square of the velocity. If this is the case, though, C_d changes with the angle of the fan's axis relative to the direction of travel, so you'd need additional equations to model that dependency. (Yuck!)

If you don't like GPS, or a turbine, what about INS? Couldn't you set up some sort of 3-axis accelerometer, and continuously integrate its outputs over some short time periods?

uh no, not what i was thinking at all. Im assuming the plane is gas powered with a battery for the remote control (unless it uses some type of alternator). Also i was just thinking of a PC fan with some type of housing though the ECM pods sound interesting and i think ill look them up since i enjoy that type of thing. Also having worked with accelerometers for NASA's SLI rocketry competition i can tell you that its more hassle then its worth to do it.

[ahecht]

I didn't see low cost as a requirement, so I will recommend that you buy a commercial pocket anemometer (such as this one) cut off the unnecessary bits such as the processor and display, and wire it directly to your data logging device. It will probably take a bit of experimentation to calibrate the device, but it shouldn't be too hard (I assume that there is essentially a generator inside which creates a voltage, but it might be something more exotic like an optical encoder).

[Gdeaver]

A Freescale pressure sensor with the proper ADC can easily give altitude with in 5 ft. To measure air speed use a tube perpendicular to the airflow and measure the pressure drop within the tube. I forget the name for this device it begins with a p. The MPX4115a is the most used sensor for these apps.
There are GPS units about the size of a quarter that interface with a PIC by serial. There are allot of sources on the web on how to decode the packet. A 3 axis digital compass would also be a nice addition but they are expensive. You should be looking on the RC airplane forums. You'll find allot more specific how to's.

[Greg Needel]

thanks all for your input we had talked about the fan/voltage idea but the problem in that case is drag. for prototyping we are using an RC plane but the final design and circuit are going to be integrated in something like this





our sensor net work uses the following
basic stamp
2 accelerometers 90 deg from each other
a digital compass
custom servo drivers and receiver
and sonar for takeoff and landing.

the plane that we are using right right now is about 2 ft long

[Chris Hibner]

Originally Posted by Gdeaver

A Freescale pressure sensor with the proper ADC can easily give altitude with in 5 ft. To measure air speed use a tube perpendicular to the airflow and measure the pressure drop within the tube. I forget the name for this device it begins with a p. The MPX4115a is the most used sensor for these apps.
There are GPS units about the size of a quarter that interface with a PIC by serial. There are allot of sources on the web on how to decode the packet. A 3 axis digital compass would also be a nice addition but they are expensive. You should be looking on the RC airplane forums. You'll find allot more specific how to's.

The perpendicular tube you mention is called a pitot tube. You determine airspeed by measuring the difference in pressure between the impact air pressure (at the pitot tube) and the ambient pressure (same source you use for your altimeter). Any fluid mechanics book has good explainations on how pitot tubes are used to measure fluid speeds. You will need two pressure transducers.

Someone earlier mentioned that a drawback to the fan (and pitot tube) is that it only measures speed relative to the air and not the ground. This is NOT a drawback. An airplane flies relative to its surrounding air - NOT the ground. The airplane does not care whatsoever what is happening with the ground unless the plane is within ~1 wingspan above the ground (i.e. only takeoffs and landings).

I think you should get rid of the accelerometers and replace them with angular rate sensors (i.e. solid state gyros). Acceleration is not important in an aircraft and can lead to false information. It is the gyroscope that allows pilots to fly in the blind. I guess one accelerometer can be useful as a slip/skid indicator, but that's not important for an autopilot.

You'll have to let me know how it goes. I've long been wanting to develop an RC autopilot. My big goal is that I wanted to do an auto-land feature.

[Josh Joseph]

hey guys I'm actually the friend greg posted the question for - I decided to just register so I can give some feedback on this...

The main problem with an anemometer is that it will be for a micro air vehicle (like the picture greg posted) so the added drag of a device like that is too much.

Also I've considered a pitot tube but I'm afraid because the craft is so small the airflow will be not be nearly uniform enough to get an accurate reading. I could definitely be wrong here seeing as how I've never tested this.

It's all electric powered and is driven by a brushless motor. I've talked to a couple of EE friends of mine and they didn't seem very convinced I could relate current draw to velocity very accurately.

I'd be interested if anyone has tried a pitot tube even on anything as small as a typical RC airplane...or had some luck relating current draw to velocity..or anything else...

I'm currently trying to relate forward/backward acceleration to velocity but seeing as how it's on one of these college project budgets the sensors are anything but high end and at first glance it's seeming like I'm not going to get the accuracy I'd need.

thanks guys

[Alan Anderson]

Consider an acoustic airspeed detector. Put ultrasonic transducers in the airstream a calibrated distance apart and measure the time it takes for a "chirp" to get from one to the other. Moving air will change the speed of the sound wave relative to the plane.

[Tristan Lall]

Originally Posted by Chris Hibner

Someone earlier mentioned that a drawback to the fan (and pitot tube) is that it only measures speed relative to the air and not the ground. This is NOT a drawback. An airplane flies relative to its surrounding air - NOT the ground. The airplane does not care whatsoever what is happening with the ground unless the plane is within ~1 wingspan above the ground (i.e. only takeoffs and landings).

When I mentioned groundspeed vs. airspeed, that's exactly the situation I had in mind; operating near obstacles. Before he'd more fully explained the concept, I was envisioning something like the IARC, where small autonomous aircraft are operated near buildings, even (in the case of helicopters) inside of buildings. In that sort of situation, you have to know where you are in absolute terms to follow a clear path, or otherwise, know the position of all obstacles at all times, and know how to take evasive action when you do see one, while still heading for your destination.

If it's just going to fly around, airspeed is quite sufficient; it's just that you need to also know wind speed and direction at all points, in order to follow a ground-based course using velocity as a reference.

[Al Skierkiewicz]

Greg and Josh,
Since speed is going to be relatively slow and I am guessing that this competition will take place indoors there is not likely to be much variation in barometric pressure and little in cross or head winds there is an easier way to do this. I would just use a reflective photo detector to count prop revolutions. You can find a number of reflective devices from Digikey. Add a little piece of audio splicing tape, (from Radio Shack, silver adhesive backed tape) to the back of the prop and you will not affect it's balance or drag coeeficient. Attach the device to the side of the motor and calibrate by running the airplane over a fixed distance to determine prop speed vs. disstance traveled. Unless a strom comes through while flying, the ambient pressure would stay relatively constant and regardless of battery voltage, the prop RPM would still give a good indication of air speed.
It's hard to tell from the picture, but you may be able to add something to the motor shaft on the side opposite the prop and count revs that way as well.

Al

[Chris Hibner]

Originally Posted by Tristan Lall

When I mentioned groundspeed vs. airspeed, that's exactly the situation I had in mind; operating near obstacles. Before he'd more fully explained the concept, I was envisioning something like the IARC, where small autonomous aircraft are operated near buildings, even (in the case of helicopters) inside of buildings. In that sort of situation, you have to know where you are in absolute terms to follow a clear path, or otherwise, know the position of all obstacles at all times, and know how to take evasive action when you do see one, while still heading for your destination.

If it's just going to fly around, airspeed is quite sufficient; it's just that you need to also know wind speed and direction at all points, in order to follow a ground-based course using velocity as a reference.

Ah yes. If your purpose is to follow a groud track, you will need to reference the ground. If you purpose is controlling the performance of the airplane, airspeed is what you need.

In either case you will still need an airspeed sensor or else with any wind you either stall the plane or break it apart.