Angular rate sensor vs Gyroscope

[danielsmusic]

I've been browsing the net long time looking for a reliable gyroscope for my project. However, I have not been able to find one, at all. All I can find are angular rate sensors and these just will not do. Even items described as an "angle sensor" or a "gyroscope" turn out to be an angular rate sensor!

So is it even possible to obtain a relativity cheap IC sized gyroscope? If not, would using an angular rate sensor and calculating rotation based on angular force be at all reliable?

I am attempting to build an auto-leveller for my RC plane. So it would have to be immune to vibrations.

 

[Boncuk]

Hi,

you are talking about the same device described with different words.

They are also called accelerometers. Check out accelerometers at Analog and Maxim.

Any one will do to keep your model aircraft straight and level. Vibrations don't change angle and therefor you don't have to worry about that.

If vibrations cause the components of the board to start separating from it you should think about soldering techniques.

Boncuk

 

[dknguyen]

Gyroscopes are angular rate sensors...

To get rotational displacement you have to go through integrate the angular velocity output of the sensors to get angular displacement and go through some math to convert these into rotations for the plane's coordinate system. Errors do build up because perfect gyros don't exist and integrating produces a continually accumulating and unbounded error. Getting around this is, in general, very tricky and good hardware, or good software and signal processing (like filtering). If you also want to keep track of an absolute reference (like the ground) then you will need to augment the gyros with accelerometers or horizon (IR) sensors that do not drift over time but have other drawbacks (response time for IR sensors, and influenced by the plane's acceleration for accelerometers).

For stabilizing the plane in it's current orientation when controls are released or slowing down it's response rate (to a plane easier to fly easier to fly or make a very agile plane flyable by a human), gyros alone are fine since it is assumed the plane won't be left alone for too long so the accumulating error won't get too big because the zero point is continually being reset to the current orientation every time you release the controls which resets the accumulating error to zero.

But if all you want is an auto-leveller, just use IR thermopiles as horizon sensors. Compared to the other solutions (gyros or gyros augmented with other sensors), IR horizon sensors have the simplest hardware, lowest processing power, and simplest math. You only really need to augment them with gyros if your plane reacts faster than the thermopiles's response time which is unlikely if your plane is not an aerobatic plane or a helicopter.

**Accelerometers alone DO NOT WORK.** THe plane speeding up or slowing down will make it think that the plane is a different orientation than it really is. Accelerometers with gyros, if you have the processing power and know the math, will.

 

[MikeMl]

...
But if all you want is an auto-leveller, just use IR thermopiles as horizon sensors. Compared to the other solutions (gyros or gyros augmented with other sensors), IR horizon sensors have the simplest hardware, lowest processing power, and simplest math. You only really need to augment them with gyros if your plane reacts faster than the thermopiles's response time which is unlikely if your plane is not an aerobatic plane or a helicopter.
...

Do you have a link to a suitable IR thermopile?

 

[dknguyen]

The basic concept:
https://www.electro-tech-online.com/custompdfs/2009/10/TaylorBristol2003.pdf


THis is one part and is available at Digikey:
**broken link removed**

And here are other much cheaper thermopiles that will also work as well as sensor PCBs to mount them onto if you do not wish to build and pay for your own custom board. They could be the exact same as above thermopile for all I know, except they are being sold at high quantity prices which this place seems to do:
Halfbase - your benefit

This store also offers this unit at high quantity prices (a bit more than 50% lower cost):
http://shop.halfbase.com/product_info.php/products_id/56
This IS a calibrated accelerometer and gyro unit and is very cheap for what it is. But like I said, no gyros are perfect and you will still need to augment the accelerometers readings with the gyros to correct for long term accumulated error. This is just the calibrated 6-axis sensor, no sensor fusion or orientation math is done on board and has to be done through your own processor and software.

 

[danielsmusic]

Ok, well here's the catch. A mechanical gyroscope, is a spinning disk just like the little toys. The disk stays where it is and everything else moves around it. Providing a reference point.

An acelerometer or angular rate sensor work in a much diffrent way. They only detect forces, usually a weight and a spring.

So in my book, a gyroscope is much much diffrent than an angular rate sensor. So my question has been answered. There is no solid-state solution for getting a physical reference. I am wondering, however, if I could modify one of these toys to do what I want. My plane is quite a stable and high-lift platform so weight is not too much of problem. Just as long as i'm not stapping lead weights to it.

The other solution I can think of is a pressure sensor on the tip of each wing, that of course depends how sensitive they can be. As the wingspan is only 1.6 meters.

The IR horizon detector is a very good solution, however I would rather not have devices sticking out and looking ugly, as this would have to do.

I will have a quick look into the proffesional gyros for helicopters, I think they are just angular rate sensors that just apply opposite force to dampen movement. So they are probably not very good. I can just experiment and see what they actually are.

 

[Russ Hensel]

Stick with MikeMl and dknguyen, they have pretty much nailed it.

 

[Mr RB]

Ok, well here's the catch. A mechanical gyroscope, is a spinning disk just like the little toys. The disk stays where it is and everything else moves around it. Providing a reference point.
...

Except that they don't... The problems with gyros is that they drift, both types, the old spinning ones and electronic solid state ones. It's all a question of how much they drift over how long a time.

You MIGHT be able to get away with a clunky solution like just use a accelerometer (to measure tilt angle) and really damp it's output with a big filter. That will remove short term accelerations like movements of the 'plane and get closer to providing a pure gravity sensing output. If you coupled that with software that keeps the plane at steady engine rpm and control surfaces straight and steady (so it's not banking etc) I think you would get a pretty good tilt reading.

 

[dknguyen]

Ok, well here's the catch. A mechanical gyroscope, is a spinning disk just like the little toys. The disk stays where it is and everything else moves around it. Providing a reference point.

An acelerometer or angular rate sensor work in a much diffrent way. They only detect forces, usually a weight and a spring.

So in my book, a gyroscope is much much diffrent than an angular rate sensor. So my question has been answered. There is no solid-state solution for getting a physical reference. I am wondering, however, if I could modify one of these toys to do what I want. My plane is quite a stable and high-lift platform so weight is not too much of problem. Just as long as i'm not stapping lead weights to it.

Your personal definition of gyroscopes is semantics as far as things are concerned. It's gotten to the point where there are enough kinds of gyroscopes around that gyroscope now just means "a sensor that provides rotational information". All solid-state gyros use either the coriolis effect or laser interference to provide angular rate information- rate as in velocity, not absolute rotational displacement.

The other solution I can think of is a pressure sensor on the tip of each wing, that of course depends how sensitive they can be. As the wingspan is only 1.6 meters.

Won't work. I've already run through the math in an attempt to make a more effective rudder to yaw the plane so it would always fly straight into the airflow without a ridiculously long tail to help automatically minimize drag and adverse yaw by using differential pressures on either side of the rudder. The pressure difference produecd by 100km/h differential airflow is just barely though to register on the most sensitive pressure sensor I could find. Let alone somethign as subtle as the difference in altitude between two wing tips.

I will have a quick look into the proffesional gyros for helicopters, I think they are just angular rate sensors that just apply opposite force to dampen movement.

Yes, they are angular rate gyroscopes. As for your original question about whether calculating angular displacement from angular rate gyros is reliable, well...no one uses mechanical spinning gyros anymore because they are have a lot of complex moving parts which makes them unreliable, expensive, heavy, and big (nowhere near IC size. THe smallest I have seen is about the size of a bowling ball). It's more obvious how they work and they do obtain the data rotational displacement data you want more directly, but that doesn't mean they are more accurate. They are still subject to the time drift problems of solid-state angular rate gyros and you would still need to augment it with an absolute orientation sensor working off an external reference (like accelerometer on gravity or or IR thermopiles on the horizon) to keep the gyro error in check.

Of course, you could have a sensor of sufficiently high quality that the error does not get out of control before your flight time was up...but let's face it, that's not gonna happen.

 

[danielsmusic]

I realize this may sound a bit stupid, but what about a circular acrylic tube filled with a dark oil, and a see-through float (maybe even a bubble) inside of it. Using infer-red detectors and and emmiters. You could get a crude measure that way. It would be pretty vibration proof, but would not, obviously have a high resolution. Which i'm not looking for anyway. I don't know, just throwing out ideas now. As I think this project is going to go into the dump to be honest.

 

[Boncuk]

I realize this may sound a bit stupid, but what about a circular acrylic tube filled with a dark oil, and a see-through float (maybe even a bubble) inside of it. Using infer-red detectors and and emmiters. You could get a crude measure that way. It would be pretty vibration proof, but would not, obviously have a high resolution. Which i'm not looking for anyway. I don't know, just throwing out ideas now. As I think this project is going to go into the dump to be honest.

That's just a flaw right from the start. If the aircraft curves (coordinated) any liquid - may it be old engine oil or water - will make the bubble travel to a position exactly aligned with the vertical axis of the aircraft, meaning it won't travel at all.

Here is a cheap - not perfect - solution:

Get yourself a high speed motor and connect a stepup gear to have a moderate weight (500g) and 100% balanced flyweel be driven at - say 60,000 to 100,000rpm. Use needle and quartz bearings for the gyro.

Align the gyro with the aircraft vertical axis and cage it. Then start spinning it until it reaches full rpm. Disconnect (mechanically) the drive and the gear (remove completely for weight reduction) and uncage the gyro.

Connect the linkage for the ailerons accordingly and you'll be able to enjoy five to six hours of straigth and level flight. Leave enough freedom for the gyro to maintain orientation. Toppling a gyro will be desastrous for your aircraft with the gyro connected to controls.

You might try out toppling your (hand held) vacuum cleaner out of it's direction with the motor running. It requires quite some force to overcome the gyro force, but eventually you'll win. And observe how the "gyro" reacts. That's toppling.

Don't forget to use an extra R/C channel to engage and disengage the gyro to the control surfaces for - may be - a landing.

Happy landings.

Boncuk

 

[Russ Hensel]

After 5 hours of straight flight, perhaps someone will find a note on the plane and tell you where to come and get it, or you can build another plane.

 

[danielsmusic]

Should the bubble not float to the top?

 

[dknguyen]

What you are talking about exists in the form of a electrolytic tilt sensor that uses a heavy viscous electrolyte to dampen vibration. They are fragile, expensive, and temperature sensitive though (but can be compensated for), and obviously still affected by acceleration. THey don't beat out the simplicity, size, cost, and reliability of IR thermopile horizon sensors.

THe project being a wash because the best technical solution available is not the one you you're most warmed up to is a shame. I used to think the way about IR sensors too...until I did a lot of detailed research into getting the other methods to work. THere are few other places where you will find such an elegant and effective solution that is so simple. Many people seem to prefer accelerometers (which absolutely require augmentation with gyroscopes to dampen out accelerations of the aircraft and the associated complexity) over horizon sensors because they don't like how horizon sensors tend to orient the aircraft away when around tall buildings or mountains. Personally, I see this as a plus, especially for tall buildings- mountains that are so large they fill the horizon even when they are far away aren't such a big deal.

You do realize that the protrusions from horizon sensors are not nearly as big as you make them out to be right? Best case for aesthetics is if you run out the thermopiles on long wires resulting in three pairs of holes 5mm at most on opposite sides of the aircraft fuselage or wingtips.

More commonly (and electrical ideal) is a small PCB holding all the thermopiles about 2cm x 2cm sitting flush against the bellt or roof of the fuselage. Did you look at the photos of the plane equipped with them in the PDF I linked?
https://www.electro-tech-online.com/custompdfs/2009/10/horizon_sensing_autopilot.pdf
Certainly nowhere near as ugly as an AWACs plane:
Airborne early warning and control - Wikipedia, the free encyclopedia

 

[danielsmusic]

I would like to try infer red sensors, however sourcing them is quite difficult in the UK

I will do more research, however.

 

[Russ Hensel]

It may be of interest to note that General Relativity is based on the idea that locally acceleration and gravity cannot be distinguished.

 

[adamey]

OK, so you said this is for an RC plane.

Why aren't you considering the vast array of gyros already available for RC helicopters? Surely it'd be much easier to adapt one of these (since they're already designd to drive servos) then come up with something completely new.

Here are a few....

**broken link removed**

 

[dknguyen]

He probably just wants it as an electronics project rather than the final result.

And anyways, he wants an autoleveller which heli gyros will not do (They stabilize only or hold current orientation only- they will not track nor find an absolute reference like the ground.).

This would be more suited if he just wants to buy one (yes, it's IR based):
**broken link removed**

 

[danielsmusic]

OK, so you said this is for an RC plane.

Why aren't you considering the vast array of gyros already available for RC helicopters? Surely it'd be much easier to adapt one of these (since they're already designd to drive servos) then come up with something completely new.

Here are a few....

**broken link removed**

No, these are not gyros. Even though, of course they are labeled as them. They are angular rate sensors, and only detect movement. I don't want that, I want a fixed reference. I don't really want this as much of a project either. I would rather buy a module that is already capable of this and feeding it into my PIC. An auto-leveler is only the tip of the iceberg, I know there are modules that can level an aircraft already. However I want to do much more than just that.

 

[Mr RB]

It may be of interest to note that General Relativity is based on the idea that locally acceleration and gravity cannot be distinguished.

But gravity has a MUCH longer time constant, especially in RC planes.