Accurate altitude measurements?

[Speakerguy]

Hi,

Does anyone know of a method to accurately determine altitude up to about 1/2 mile? I've seen GPS modules that measure altitude but they are only accurate to +/-35m, which is problematic when trying to land something This would be on an RC plane for an auto-pilot like device.

ETA: Actually I could use two methods, one for high altitude stuff and the other for landing (where the accuracy is super important but it doesn't need to be good to a 1/2 mile).

 

[3v0]

The model rocket people have been measuring altitude for years. Their contests are often based on maxium altitude I would suspect they have what you need.

 

[Speakerguy]

Excellent - thank you!

 

[jpanhalt]

You don't have to be accurate with respect to altitude above the ground when you are 1/2 mile up. The critical phase for landing is within 100 to 200 feet for full scale and a lot lower for models.

A barometric altimeter (±50 ft) will not suffice for landing. It's a lot worse when the altimeter reads high.

Down low, you need radar, lidar, or something fast. For models, sonar may actually work; although, I have not seen any such units for that purpose.

John

 

[Speakerguy]

What about for stunts, like dive-bombing and then pulling up?

I'm probably going to break a lot of expensive planes messing with this. If anyone remembers my inertial nav system idea from a few months back this is a resurgence of an interest in something like that.

 

[jpanhalt]

Barometric altimeters do not respond fast enough for dive bombing. They indicate high, which makes the pilot into an unintentional Kamikaze.

Stunts are done by visual reference.

John

 

[dknguyen]

What scale are these planes? Perhaps these can be of help. I've been looking for a plane for a while that doesn't involve sanding or skinning of the wings but isn't realy foam either:

https://www.acesim.com/rc/index.html

PLus wing warping is just cool. I actually have a Carbon Prime on order right now. SHould survive a bad landing quite well. But I agree...you kind of need either radar or eyes for dive bombing. Depending on how fast the plane is moving you might be able to attach a very long range IR proximity sensor (or ultrasonic if you can get around propeller wash and wind interference issues) to the front so that if an object is detected it pulls up. Obviously this requires that you limit the diving speed the plane or increase the range of the sensor (which probably means radar or lasers).

THe longest range IR sensor I know of 5.5m should be plenty for a small "floaty" plane:
https://www.acroname.com/robotics/parts/R302-GP2Y0A700K0F.html

But for non-time-of-flight EM range sensors, large max range means large min range so you'd have to use something else for landing (but a forward facing sensor isn't too helpful for landing anyways). THe standard for hobbyist UAVs is to use sonar due to it's long range and cost, but for a small plane like the above IR should work.

Remember to test for the sun's interference!

 

[ericgibbs]

Hi,

Does anyone know of a method to accurately determine altitude up to about 1/2 mile? I've seen GPS modules that measure altitude but they are only accurate to +/-35m, which is problematic when trying to land something This would be on an RC plane for an auto-pilot like device.

ETA: Actually I could use two methods, one for high altitude stuff and the other for landing (where the accuracy is super important but it doesn't need to be good to a 1/2 mile).

hi speakerguy,

Attached pdf's for alt and opa, the student has been using the alt/opa in rocket work and is getting good results.

 

[Oznog]

https://www.sparkfun.com/commerce/product_info.php?products_id=8128

Senses down to a difference of 9cm in still air!
Has ADC on board with a noise filter. Vibration noise is very significant in many sensors, in fact due to the weight of the sensing diaphragm other devices may read off by 100ft if flipped upside down from their initial calibration. Not so with this one.

BUT... this depends on getting a decent "static air" source and sealing that path around the device. Ram air (pitot) pressure of a forward-facing port can add hundreds of feet at high speed and will vary based on speed. A static air port is placed so it will give atmospheric pressure unaffected by speed. In fact the difference between pitot and static port pressure is how an airspeed gauge works.

 

[dknguyen]

https://www.sparkfun.com/commerce/product_info.php?products_id=8128

Senses down to a difference of 9cm in still air!
Has ADC on board with a noise filter. Vibration noise is very significant in many sensors, in fact due to the weight of the sensing diaphragm other devices may read off by 100ft if flipped upside down from their initial calibration. Not so with this one.

BUT... this depends on getting a decent "static air" source and sealing that path around the device. Ram air (pitot) pressure of a forward-facing port can add hundreds of feet at high speed and will vary based on speed. A static air port is placed so it will give atmospheric pressure unaffected by speed. In fact the difference between pitot and static port pressure is how an airspeed gauge works.

I wouldn't expect the ground to be level either.

 

[Speakerguy]

Man, you guys are awesome. So I'm gathering this historically has been somewhat of a difficult thing to do with great accuracy. And dknguyen, you have me seriously looking at those planes. Warperons are awesome.

Scale is of no importance. A 1lb payload would be nice if I wanted to put a good camera or something on it, but that's about it.

 

[Speakerguy]

The IR looks promising. I've also been looking at inclinometers on digikey to combine them with (distance to ground perpendicular fuselage axis + inclinometer = actual distance to ground). Any links to sonar based UAV's?

 

[dknguyen]

The IR looks promising. I've also been looking at inclinometers on digikey to combine them with (distance to ground perpendicular fuselage axis + inclinometer = actual distance to ground). Any links to sonar based UAV's?

Regular inclinometers won't work on a plane since you are always moving and thus accelerating which adds into the acceleration of gravity. What you need is a 3-axis gyroscope (and all the associated calibrations and possibly accelerometer to control the drift which is $$$$$), or get an IR horizon sensor.

FMA DIrect sells one called the co-pilot for quite cheap, considering the cost of thermopiles.

 

[Speakerguy]

So something like this won't work?

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=551-1005-2-ND

 

[dknguyen]

Nope, not unless you can guarantee zero acceleration when you decide to take readings. Not so hard for a ground vehicle since you just stop...it's a bit harder for an aircraft, plane or helicopter.

It is better than nothing though, but you would have to low carefully pass filter the signal and take the readings with a grain of salt (ie. if all control surfaces are at neutral and the throttle has been unchanged for a certain period of time, the chances that the plane is accelerating are less than if they were being jarred around.

BTW, just to point out that if you do decide to go this route, you want to take the direction vector of gravity and use that and ignore the magnitude (since the magnitude is different if you are actually propelling in any direction other than horizontal). The lift of the plane also throws in an additional acceleration against gravity during angled flight which I don't know how to account for...so yeah.

 

[Speakerguy]

Gotcha. Now that I understand there's no way I'd trust the inclinometer.

 

[Oznog]

Once you're calculating angles and such, the math can become more intensive and typically requires 16-numbers. And you almost certainly want some form of digital filtering at some point. So a DSP core may be necessary and seeing as the 33F series is dirt cheap and low power that's not a problem.

This guy did a nav system project:
http://forum.microchip.com/tm.aspx?m=323311

 

[dknguyen]

There are 3 huge advantages of horizon sensors over gyros.

1. The first is that they provide an absolute attitude reference so there is no unbounded integration error and things are much easier to deal with- no accelerometers or extra sensors or algorithms needed to keep the error bounded.

2. The second huge advantage is that the math is very simple and the simplest PIC could do it. This is different from the extra algorithms needed for point 1 as Oznog has pointed out, because even if you had perfect gyros, you would still need a some complicated coordinate rotation math to keep track of the ground (since the the rotational axis of measurement from one moment to the next is constantly changing because it rotates with the plane and you are only interested in keeping track of the ground in most cases- one rare exception is the gyro that holds the tail of a helicopter stable.

3. The third is cost!

The disadvantage is that their response time is slower than gyros so you can't use them by themselves on very nimble aircraft or helicopters, or at low altitudes where there are surrounding structures. They can be used with gyros though to help correct for the bias.

If you do decide to pursue this, let me know. Digikey has some i2C thermopiles, but they are much much cheaper when you order them in bunches of 25, and though I can find uses for about 12 of them, finding uses for 18 or 25 is pushing it.

 

[Speakerguy]

What are you using the thermopiles for?

 

[Speakerguy]

These look pretty awesome, but the price is ridiculously scary:

https://www.analog.com/en/other/multi-chip/adis16350/products/product.html