Electronic yaw sensor

weirdjim

weirdjim

Ejection Handle Pulled
Joined
Jul 8, 2008
Messages
4,171
Location
Grass Valley, CA (KGOO)
Display Name
Display name:
weirdjim
Is there a cheap and easy way to detect small values of yaw (NOT rate of yaw, which is easy), say a maximum value of one end of the black marble in the glass tube to the other?

The only thing I can come up with is a glass tube with a marble in it and leds on one side and photodetectors on the other side.

Thoughts?

Jim
 
weirdjim said:
Is there a cheap and easy way to detect small values of yaw (NOT rate of yaw, which is easy), say a maximum value of one end of the black marble in the glass tube to the other?

The only thing I can come up with is a glass tube with a marble in it and leds on one side and photodetectors on the other side.

Thoughts?

Jim
Click to expand...

You seem to be thinking of slip/skid, not "yaw" per se. And for slip/skid, which is what the "black marble in the glass tube" detects, an easy electronic way to sense it is using the very low cost accelerometers modules you can get today. You check which way the current 'g' vector (i.e. gravity) falls relative to the aircraft's vertical axis, and you can get very high resolution values (which you'll likely need to filter since it's very sensitive). The modern modules do much of the math for you, so it's actually very easy to implement.
 
Couldn't you use a cheap auto yaw sensor? Type they use for stability control.

I'd imagine an aviation accelerometer would be quite expensive.
 
RotorDude said:
You seem to be thinking of slip/skid, not "yaw" per se. And for slip/skid, which is what the "black marble in the glass tube" detects, an easy electronic way to sense it is using the very low cost accelerometers modules you can get today. You check which way the current 'g' vector (i.e. gravity) falls relative to the aircraft's vertical axis, and you can get very high resolution values (which you'll likely need to filter since it's very sensitive). The modern modules do much of the math for you, so it's actually very easy to implement.
Click to expand...

Correct for slip/skid. Error in nomenclature.

Point me to the accelerometer module data if you would. Seems that an accelerometer would give me RATE and not slip/skid. I could bewrong.

Thanks,

Jim
 
Starting out with where the ball is, which gives you so much skid or slip, maybe you'd just have to find a way to measure how much the wings are off level and then do the math. The ball could be off to the side without any yaw. The slip or skid could be all induced by roll. Find a way to measure the tilt of the "wing" in the atitude indicator, compare it to how far off of center the ball is, and maybe there is a formula that will tell you how many degrees the longitudinal axis is off of the flight path of the plane through the air. I'm assuming thats what you mean by "values of yaw (NOT rate of yaw, which is easy.)
 
Hold it. I got it. Get one of those vane type Angle of Attack indicators, rotate the thing 90 degrees and the install it on top of or the bottom of a wing. Or the fuselage if it's a twin
 
weirdjim said:
Correct for slip/skid. Error in nomenclature.

Point me to the accelerometer module data if you would. Seems that an accelerometer would give me RATE and not slip/skid. I could bewrong.

Thanks,

Jim
Click to expand...

Here is one example of several available modules.
This example can do much more than what you want, the slip/skid is just a small subset.
And you are right, it does intrinsically sense (mostly) rates, but the internal processor also integrates those rates into "poses", or attitudes in aviation-speak, and does a good job of self-calibration (unlike older generations where you had to that externally).
I have done this (including the slip/skid functionality) and flight tested it, so it definitely works very well, all for peanuts in cost.
 
I have yet to meet a really good yaw damper
 
Let me try again.

You want to do the same thing as the ball in the turn and bank? All that is is spanwise acceleration. The only issue is signal to noise - you are dealing with fractions of a G mixed with a lot of vibration.
 
Last edited:
Capt. Geoffrey Thorpe said:
You want yaw as in nose not pointed in the direction of flight as opposed to yaw as in I ain't flying in the direction I used to. Right?

That rules out rate gyros and integration. Accelerometers would tell you nothing. I think the solution has to be aerodynamic - like the yaw string on a glider. 'cept it would have to be outside the propeller slipstream. Another complication would be spanwise flow if you try to mount something to a wing.

But, before I waste too much effort - is my assumption correct?
Click to expand...

"Nose not pointed in direction of flight" = slip/skid.
And slip/skid is easily measured by integrating accelerometers, which is what the modern low cost modules can do.

Hint: all existing low-tech slip/skid indicators in airplanes (aka ball inside tube) don't measure aerodynamic parameters.
 
RotorDude said:
"Nose not pointed in direction of flight" = slip/skid.
And slip/skid is easily measured by integrating accelerometers, which is what the modern low cost modules can do.

Hint: all existing low-tech slip/skid indicators (aka ball inside tube) don't measure aerodynamic parameters.
Click to expand...
My mind wandered down the wrong path for a moment.
You don't need to integrate. It's just straight "horizantal" (relative to the wings) acceleration. To exactly duplicate the ball function you would also need to measure the vertical (again, relative to the airframe) acceleration and fine the inverse tangent.
 
RotorDude said:
"Nose not pointed in direction of flight" = slip/skid.
And slip/skid is easily measured by integrating accelerometers, which is what the modern low cost modules can do.

Hint: all existing low-tech slip/skid indicators in airplanes (aka ball inside tube) don't measure aerodynamic parameters.
Click to expand...
OK, so we have a difference of opinion from two sources. One says integrating rate gyros doesn't work. The other says that integrating rate gyros (i.e. accelerometers) is the way that it is done. You two fight it out, and may the best yawyer prevail. And you don't need to be a lawyer to be a yawyer.

Jim
 
Capt. Geoffrey Thorpe said:
My mind wandered down the wrong path for a moment.
You don't need to integrate. It's just straight "horizantal" (relative to the wings) acceleration. To exactly duplicate the ball function you would also need to measure the vertical (again, relative to the airframe) acceleration and fine the inverse tangent.
Click to expand...

You are getting warmer. Yes, you don't need the integration to sense the primary "g" vector input. But you still need it to maintain a stable frame of reference relative to which that "g" is measured. The modern modules do both for you: they give you the "g" vector, as well as maintain the stable reference, using quaternions. The bottom line is that it's very easy to do, since the module's built-in processor does all the heavy lifting for you.
 
weirdjim said:
OK, so we have a difference of opinion from two sources. One says integrating rate gyros doesn't work. The other says that integrating rate gyros (i.e. accelerometers) is the way that it is done. You two fight it out, and may the best yawyer prevail. And you don't need to be a lawyer to be a yawyer.

Jim
Click to expand...
A rate gyro is not an accelerometer. The first measures rotation about an axis, the second measures acceleration along an axis.

The ball measures acceleration, the needle measures yaw rate.
 
Capt. Geoffrey Thorpe said:
A rate gyro is not an accelerometer. The first measures rotation about an axis, the second measures acceleration along an axis.

The ball measures acceleration, the needle measures yaw rate.
Click to expand...

Right rudder, left aileron, allow to come to equlibrium. Lots of ball, no acceleration OR yaw rate.

Jim
 
Jim
How about a metal marble in a curved tube with the sides wrapped in filament to act as a potentionameter. Measure the resistance thru the marbkw and each side. The resistance change as marble changes position gives the yaw value?
 
weirdjim said:
Right rudder, left aileron, allow to come to equlibrium. Lots of ball, no acceleration OR yaw rate.

Jim
Click to expand...
An accelerometer will measure a the lateral force in this case - that's the force that cause the ball to move.

You be sitting on yea olde butt right now. Gravity is trying to accelerate your butt downwards at 32.2 ft/sec^2 (or 9.81 m/sec^2 if you prefer) and the reaction force from the chair is preventing any actual acceleration. If you swallowed an accelerometer it would read that 1g force downwards. (OK, you could hold it or set it on your desk, you don't have to swallow it.) Same as being in a slip - there is the force of gravity pulling somewhat sideways (relative to the the aircraft coordinate axis) balanced by aerodynamic forces. An accelerometer mounted on the lateral axis will measure this - just like the ball.

An accelerometer just measures forces. Think of a mass mounted on a spring and you measure deflection.

Solid state rate gyros are built somewhat like tuning forks (not a spinning wheel), but as they move back and forth along one axis any rotation about an axis perpendicular to the motion results in a force on the third axis that can be measured.
 
Capt. Geoffrey Thorpe said:
An accelerometer will measure a the lateral force in this case - that's the force that cause the ball to move.
Click to expand...
is operating

I'm still not understanding ... my BS is in Semiconductor Physics '67, so I'm still pretty well steeped in Newtonian mechanics ... you know, that antique stuff that says that a body in motion remains in motion unless acted upon by an outside force.

So I put yonder aireromachine into a skid and hold it there...just hold it there. Yes, a force needed to be applied to get it there, but once there, what force is detectable (as in the ball to one side) ....

Wait a minnit. There NEEDS to be a force to hold that ball to one side or the other against its will. So is THAT the force we can detect with an accelerometer?

Jim
 
Just what exactly is it you want to measure? Is it the angle between the relative wind and the longitudinal axis of the airplane?
 
weirdjim said:
is operating

I'm still not understanding ... my BS is in Semiconductor Physics '67, so I'm still pretty well steeped in Newtonian mechanics ... you know, that antique stuff that says that a body in motion remains in motion unless acted upon by an outside force.

So I put yonder aireromachine into a skid and hold it there...just hold it there. Yes, a force needed to be applied to get it there, but once there, what force is detectable (as in the ball to one side) ....

Wait a minnit. There NEEDS to be a force to hold that ball to one side or the other against its will. So is THAT the force we can detect with an accelerometer?

Jim
Click to expand...

Couldn't it just be gravity? I'd guess centrigugal force is probably a part of it when you are yawING. But what about once you are stabilized in the slip/skid and there is no more "acceleration?" Like if you pull a turn coordinator out of the panel, hold it in your hand and rotate it. The ball is going to move to the side. No acceleration, no slip, no skid.
 
RotorDude said:
You are getting warmer. Yes, you don't need the integration to sense the primary "g" vector input. Butt you still need it to maintain a stable frame of reference relative to which that "g" is measured. The modern modules do both for you: they give you the "g" vector, as well as maintain the stable reference, using quaternions. The bottom line is that it's very easy to do, since the module's built-in processor does all the heavy lifting for you.
Click to expand...

I did a LOT of reading on quaternions, which is a Hamiltonian process. I remember calculating RF fields using the Hamiltonian back in junior year Fields & Waves and that was not intuitive in the least -- but just build one or two antennas and measure them. All of a sudden the theory becomes understandable.

Manipulating 3-d vectors in 4-d vector space has the same gut feeling ... the math isn't hard; the gut feeling and visualization doesn't give me the warm and fuzzy of an Einstein cartoon running one way on a train platform with the train approaching the speed of light moving in the opposite direction.

But that begs the question of the practical visualization of the process. If the processor is programmed to do the heavy lifting, then first we find out if it will give us the results and then perhaps the results will lead us to a deeper understanding.

Or not.

Thanks,

Jim
 
weirdjim said:
I did a LOT of reading on quaternions, which is a Hamiltonian process. I remember calculating RF fields using the Hamiltonian back in junior year Fields & Waves and that was not intuitive in the least -- but just build one or two antennas and measure them. All of a sudden the theory becomes understandable.

Manipulating 3-d vectors in 4-d vector space has the same gut feeling ... the math isn't hard; the gut feeling and visualization doesn't give me the warm and fuzzy of an Einstein cartoon running one way on a train platform with the train approaching the speed of light moving in the opposite direction.

But that begs the question of the practical visualization of the process. If the processor is programmed to do the heavy lifting, then first we find out if it will give us the results and then perhaps the results will lead us to a deeper understanding.

Or not.

Thanks,

Jim
Click to expand...

Yes, I did quite a bit of reading about quaternions, since they were not prominent in my Physics courses. It turns out they were hugely popular at some point a hundred years or so ago, then went out of fashion for some reason. Now they are back in big time again, and are the main mechanism used to convey "poses" for aircraft attitudes (they avoid the so-called "gimbal lock" from which the plain "Euler angles" suffer). But all in all, the modern 10 DOF (degrees of freedom) accelerometer/gyro/compass/barometer sensor modules do an amazing job of sensing and computing attitudes, altitudes and rates, and essentially give you a very simple and easy solution to aircraft orientation in a tiny and inexpensive package. Your requested slip/skid functionality is just one of the many cool features that they easily provide you.
 
weirdjim said:
is operating

I'm still not understanding ... my BS is in Semiconductor Physics '67, so I'm still pretty well steeped in Newtonian mechanics ... you know, that antique stuff that says that a body in motion remains in motion unless acted upon by an outside force.

So I put yonder aireromachine into a skid and hold it there...just hold it there. Yes, a force needed to be applied to get it there, but once there, what force is detectable (as in the ball to one side) ....

Wait a minnit. There NEEDS to be a force to hold that ball to one side or the other against its will. So is THAT the force we can detect with an accelerometer?

Jim
Click to expand...
How about simple piezoelectric scales with the ball sandwiched between them?
 
RotorDude said:
Yes, I did quite a bit of reading about quaternions, since they were not prominent in my Physics courses. It turns out they were hugely popular at some point a hundred years or so ago, then went out of fashion for some reason. Now they are back in big time again, and are the main mechanism used to convey "poses" for aircraft attitudes (they avoid the so-called "gimbal lock" from which the plain "Euler angles" suffer). But all in all, the modern 10 DOF (degrees of freedom) accelerometer/gyro/compass/barometer sensor modules do an amazing job of sensing and computing attitudes, altitudes and rates, and essentially give you a very simple and easy solution to aircraft orientation in a tiny and inexpensive package. Your requested slip/skid functionality is just one of the many cool features that they easily provide you.
Click to expand...

You gave me this URL:
https://cdn-learn.adafruit.com/downloads/pdf/adafruit-10-dof-imu-breakout-lsm303-l3gd20-bmp180.pdf

which explained in very brief terms what the device did. I didn't see any further references that I could learn from. Can you point me to further readings on this series of devices, perhaps in more detail?

Thanks,

Jim
 
GlennAB1 said:
How about simple piezoelectric scales with the ball sandwiched between them?
Click to expand...
That's a macro version of one of the many microscopic sensors embedded on the accelerometer chip, and the latter also includes the needed (very non-trivial) calibration and processing. When you hook up one of those cheap modules (typically <$50) to a low cost microprocessor like Arduino or Raspberry Pi, you'll get almost a direct readout of almost everything you need to control an aircraft in IMC (or drive a drone with high precision), including attitude, altitude, slip/skid, rate of yaw, magnetic heading, and more.
 
weirdjim said:
You gave me this URL:
https://cdn-learn.adafruit.com/downloads/pdf/adafruit-10-dof-imu-breakout-lsm303-l3gd20-bmp180.pdf

which explained in very brief terms what the device did. I didn't see any further references that I could learn from. Can you point me to further readings on this series of devices, perhaps in more detail?

Thanks,

Jim
Click to expand...

Jim, here is another link to some Adafruit hardware and related documentation that should get you more info, but there is lots of other material you can google, esp. on the "fusion" aspect of the devices (the idea being to fuse together input from different modalities to obtain a reliable and stable output), e.g. using Kalman filtering (which along with its different variants is a huge topic in itself, like quaternions).
One cool thing about the latest smart devices is that they do much of the filtering and calibration onboard, using proprietary algorithms running on their embedded processors, so the output is already clean enough to use with no additional filtering (except perhaps minor averaging).
There are several of these devices on the market, and in general I'd recommend using a "breakout module" as opposed to the individual chips. You should decide on your "host" platform, e.g. Arduino or Raspberry Pi, and the specific model within those families, although for development purposes that's not critical, though bus and voltages do matter, so be sure you have the correct I2C support (there are some 5V vs. 3V issues that must be addressed).
As I mentioned above, I have done this myself and got it all to work with minimal fuss. BTW, you can get a lot of this to run on a standalone Android device using its built-in sensors (which I have also done), but at least in my experience those sensors were less accurate and/or reliable.
 
Last edited:
RotorDude said:
That's a macro version of one of the many microscopic sensors embedded on the accelerometer chip, and the latter also includes the needed (very non-trivial) calibration and processing. When you hook up one of those cheap modules (typically <$50) to a low cost microprocessor like Arduino or Raspberry Pi, you'll get almost a direct readout of almost everything you need to control an aircraft in IMC (or drive a drone with high precision), including attitude, altitude, slip/skid, rate of yaw, magnetic heading, and more.
Click to expand...
That's pretty cool. I knew that from aircraft INS systems training, just generalities training, just hadn't retrieved it from memory. And I recall eddy current based accelerometers on the T-38 wing center box "strain guages" back in the '70's.
 
You must log in or register to reply here.
Back
Top