Designing control force loading

[WJH308]

I believe I discovered how to do this. I am going to connect the flight controls to linear actuators, and they will be connected to the linear actuators with load cells inbetween. My L1011 yoke has 4 load cells in it, 2 for pitch, 2 for roll, one for each autopilot. Light control loading, the linear actuator moves with the pilots movements as sensed on the load cells. To increase control loading, you simply require the pilot to excert more force on the controls to make the linear actuators move with the pilots inputs. For pitch trim, as we all know, trim feels neutral at a certain airspeed, as trim is for airspeed. As the airplane picks up airspeed, more than what is trimmed, the linear actuator will move the yoke in the direction of pitch that the airplane is trimmed for. Pilot resists this, and the larger the airspeed spread to trim speed, the more force that is required to fight against the linear actuator. Simply move the trim wheels to relieve control pressure, which will also trim the airplane for the new airspeed, as it does in real life. I just ordered a microcontroller developement board and a 25lb load cell to experiment with this. Will probably make a mosfett powered H bridge motor controller that runs off the MCU to control the linear actuator. More to come when I find time between my flight instructor job that takes 6 out of the 7 days in the week...
I am pretty sure this is how the professional sims do it, and well, how the L1011 did it with the autopilots and the trim wheel. The trim wheel on the yoke of the L1011 didn't do anything when in CWS mode unless the pilot excerted 4lbs or more of force. I am going to rename the L1011 the "Enterprise" Freaking airplane was way ahead of its time.

[Jackpilot]

I do not want to be rude ..but in plain English ..how does it work?

[WJH308]

Blood, sweat, and tears

[Michael Carter]

I believe I discovered how to do this. I am going to connect the flight controls to linear actuators, and they will be connected to the linear actuators with load cells inbetween. My L1011 yoke has 4 load cells in it, 2 for pitch, 2 for roll, one for each autopilot. Light control loading, the linear actuator moves with the pilots movements as sensed on the load cells. To increase control loading, you simply require the pilot to excert more force on the controls to make the linear actuators move with the pilots inputs. For pitch trim, as we all know, trim feels neutral at a certain airspeed, as trim is for airspeed. As the airplane picks up airspeed, more than what is trimmed, the linear actuator will move the yoke in the direction of pitch that the airplane is trimmed for. Pilot resists this, and the larger the airspeed spread to trim speed, the more force that is required to fight against the linear actuator. Simply move the trim wheels to relieve control pressure, which will also trim the airplane for the new airspeed, as it does in real life. I just ordered a microcontroller developement board and a 25lb load cell to experiment with this. Will probably make a mosfett powered H bridge motor controller that runs off the MCU to control the linear actuator. More to come when I find time between my flight instructor job that takes 6 out of the 7 days in the week...
I am pretty sure this is how the professional sims do it, and well, how the L1011 did it with the autopilots and the trim wheel. The trim wheel on the yoke of the L1011 didn't do anything when in CWS mode unless the pilot excerted 4lbs or more of force. I am going to rename the L1011 the "Enterprise" Freaking airplane was way ahead of its time.

Sounds like an interesting project.
Pictures, as always, are welcome when you start your project. I'd really like to see this.

[Mike.Powell]

WJH308,

That's pretty much the way it works. One item you didn't mention was the position sensor. Your control electronics will need the control position as a parameter when calculating the force required of the actuator.

[Roland]

Indeed a very professional approach. You may want to contact Ian Hopper (buggies build for fun). He also experimented with control loading and toyed with the load cells idea.

This approach is a bit over my head, I'll stick with the servo-gain-offset system.

[AndyT]

Mike beat me to it.

You will need the position sensors. You should also have a strain guage connected to your sensors so that you can keep track of how much force you are actually imposing. I don't have the numbers, but I know that Pro sims have them all calculated and can dial up different control forces depending on the plane being flown.

For those of you that are wondering:
A load cell is like your bicep muscle. The more electricity you give it, the more it contracts. It's an analog system that can be controlled digitally. The less juice you give it, the more it can stretch out. While the newer type of load cells are electromagnetic in design, it's my understanding that the older electromechanical ones are the ones to use because of the magnetic interference created by the newer ones.

[warvet]

WHAT THE **** ARE YOU GUYS TALKING ABOUT??? *someone pass the nachos*
Tim

[AndyT]

Tim,

We are talking about making the yoke actually shimmy and shake with the air pressure hitting the control surfaces of the plane as you fly. Control loading will make it harder to fly unless you trim the plane correctly just like if you were really flying.

Ever hit a serious bump in your car? It makes the steering wheel jump wildly. Same kind of thing.

[WJH308]

WJH308,

That's pretty much the way it works. One item you didn't mention was the position sensor. Your control electronics will need the control position as a parameter when calculating the force required of the actuator.

I'll use the pots connected to the controls as the servo sensors


Mike beat me to it.

You will need the position sensors. You should also have a strain guage connected to your sensors so that you can keep track of how much force you are actually imposing. I don't have the numbers, but I know that Pro sims have them all calculated and can dial up different control forces depending on the plane being flown.

For those of you that are wondering:
A load cell is like your bicep muscle. The more electricity you give it, the more it contracts. It's an analog system that can be controlled digitally. The less juice you give it, the more it can stretch out. While the newer type of load cells are electromagnetic in design, it's my understanding that the older electromechanical ones are the ones to use because of the magnetic interference created by the newer ones.

I need to figure out how to interface to the load cells in my l1011 yoke, I'd rather not alter the yoke.



Indeed a very professional approach. You may want to contact Ian Hopper (buggies build for fun). He also experimented with control loading and toyed with the load cells idea.

This approach is a bit over my head, I'll stick with the servo-gain-offset system.
That is a method I will do some research on. Is it like the frascas where the elevator trim simply moves the centering spring back and fourth, and on the rudders to move the center position as to simulate an engine out on a multi engine aircraft?

[WJH308]

Tim,

We are talking about making the yoke actually shimmy and shake with the air pressure hitting the control surfaces of the plane as you fly. Control loading will make it harder to fly unless you trim the plane correctly just like if you were really flying.

Ever hit a serious bump in your car? It makes the steering wheel jump wildly. Same kind of thing.

Not so much as turbulence but when you fly a cessna or a piper, when I do some crazy unusual attitudes, and bring the airplane to a stall, I've had the roll axis on the yoke go completely limp as all airflow over the ailerons stopped, and as you pick up airspeed, the yoke gets much harder to move. And on the elevator, helps you set trim. In fact, I dare say, the pitch in most flight sims is so sensitive and hard to trim correctly simply because no one is using control loading on the controls. If anything, PC simulation should be much easier with proper control loading. This applies to Flight sims and car sims as well. Imagine if your playing R-factor and you can judge how well the front wheels are gripping based on the control loading. The front wheels start to skid, your control forces go down to nothing.
I am going to start out with a simple variant of control loading for the Spoiler lever in my L1011. In real life it is power assisted. If you disable the power assist with the little red button on top of the lever, "Considerable force is required to move it" So with it on, lets say it requires 4lbs of force to move it, and the servo keeps it positioned where I had it. Disable power assist, the loadcell now expects to see a 20lb force to move it, and it returns back to the forward position(Need to ask an L1011 pilot if that is accurate).
I might be able to get away with using a giant scale R/C servo for this. The servo will also drive the lever back and fourth for Direct Lift Control, and automatic ground spoilers.

[Jackpilot]

Like Tim, I have a hard time visualizing/figuring it out. Not the result, but how to achieve it with these neverheard-of-before gizmos.

I think nevertheless that if done , this "real flight" loading would be a terrific improvement (second to none but 3 axis mouvm.) and would make flying without it really "flat" !!

Worth explaining in more details and ,why not, some schematics ???

Very interesting.

[WJH308]

Like Tim, I have a hard time visualizing/figuring it out. Not the result, but how to achieve it with these neverheard-of-before gizmos.

I think nevertheless that if done , this "real flight" loading would be a terrific improvement (second to none but 3 axis mouvm.) and would make flying without it really "flat" !!

Worth explaining in more details and ,why not, some schematics ???

Very interesting.
I will be 100% honest with you all. As a pilot, I need a plan "B" and If I can make certain products for the flight sim community to sell to fall back on in case I ever lose my medical or do something stupid that the FAA does not like, I will NEED a plan "B". The other side of me is a huge champion of Open Source products, and also a champion of "Do It Yourself" projects. Being that I am just brain storming right now and have no products or technologies, just idea's, they are worth discussing so everyone can benefit from each other to strive for better experiences in all of our cockpits.
When I get a chance, I will make drawings in Solid works to post on the forum.

First things first, an MCU is a micro-controller, an integrated chip that is a miniature computer that you can program in either C or assembly language. I prefer the ATMEL AVR series. For this hobby, knowing how to program and implement microcontrollers into your cockpit is a huge benefit.

A load cell is a sensor that you can wire into a MCU that measures pressure. The load cell I purchased from Mouser is rated for 25lbs. Now I will know more when I receive it, but it puts out a voltage in proportion to how much force is applied to it, up to 5v which is standard. The MCU has Analog to digital converters and digital to Analog converters. It can measure the voltage output from the Load Cell and give it a numerical value that can be passed along to Flight Sim or drive other functions such as driving a linear actuators speed control.

A linear actuator is what your garage door opener is. A long worm screw that turns and drives a little trolley along a "way".

Now instead of driving your garage door, use it to move the yoke in and out. Take the yoke, mount it to the linear actuator, but between the linear actuator and the yoke is the load cell to measure force, and mounted to the linear actuator is your Potentiometer to measure location of the trolley. Hook all these things into a micro controller with some fancy programming and interfacing to flight sim, you have the basics for control loading.

Anyhow, the nice thing about programming is that it is FREE to get started.
C++ is fun. For 100$ you can start programming ATMEL AVR's. Buy an Atmel STK500 development board from either digikey or Mouser, a couple of MCU's, AND download the free AVR studio from ATMEL, and the FREE WinAVR IDE for programming.
avrfreaks.net is an excellent resource on ATMEL AVR's.

Other projects I am brainstorming include my own version of cockpit instruments like project magenta, but for the L1011 driven by OpenGL. Guys, if you can conquer this stuff and master it, there is absolutely nothing you CANNOT do for your sim.

[Mike.Powell]

Wittenstein Aerospace makes control loaders http://www.wittenstein.aero/170_580.htm as do Servos & Simulation http://www.servos.com/default.aspx and Moog http://www.moog.com/SimulationAndTesting/solutions/control_loading/

The sales literature may give you a few ideas.

Immersion http://www.immersion.com/ has a website that has lots of interesting tidbits about "haptics" another word for force feedback.

There had been an Asian company selling inexpensive load cells through Ebay. Some of the race car sim crowd were using them as force sensors for the car pedals. The company listed a broad range of load cells along with, I believe, postal scales. Naturally I can't find any Ebay listings now. I think the search term I used (successfully) in the past was "load cell". Perhaps they will resurface, and offer a low-cost source of load cells.

As an alternative to developing a control loading system using a force feedback loop, you might consider developing an open loop system which uses a force actuator. I wrote about such an approach here: http://www.mikesflightdeck.com/old_stuff.htm in the August 18, 2006 entry.

[Roland]

My method does not use springs, it uses a motor that drives the controls in a closed loop servo. By altering the servo gain, you will have more stiff or more sloppy control feeling. By altering the servo input offset, you can alter the zeroforce center. Another gentleman gave me a simpler idea, by driving the position pickup potmeter in opposite direction. works quite well.

Details can be found on http://www.simprojects.nl/diy_force_feedback_interfacing.htm

[ak49er]

I was perusing the Hagstrom electronics website today and it looks like they have a card for use with control loading. I didn't look into it much though, it may help.

[AndyT]

Two different kinds of load cells.
I was thinking of the artificial muscle type and you are using the sensor type.

Mine 'creates' the load and yours measures it. Both fit the design. what a weird coincedence.

[rsandor]

Hi everybody!

Just now joined Your forum, and already have to say something:
Load cells, are expensive by nature (those which are already conditioned and outputting voltage in range 0-5V). Unconditioned load cells are outputting extremely low voltage. A whole study required to eliminate problems related to temperature, and so on.
What if we mount a DC motor (with constant magnet as stator) as a constant mechanical load to pitch/roll mechanisms, and short circuit the rotor via a variable current regulator? It is relatively low mechanical load to existing mechanism if the circuit is open (current low), and a much greater load if the motor (a generator this way) is loaded by current. For now don`t know the the values of required motors, as it depends on a lot of factors, but this is an idea only.

Robert

PS: The motor may be a one with short circuited rotor, and a wound stator, but than the load can be regulated by voltage applied to stator.

[WJH308]

Hi everybody!

Just now joined Your forum, and already have to say something:
Load cells, are expensive by nature (those which are already conditioned and outputting voltage in range 0-5V). Unconditioned load cells are outputting extremely low voltage. A whole study required to eliminate problems related to temperature, and so on.
What if we mount a DC motor (with constant magnet as stator) as a constant mechanical load to pitch/roll mechanisms, and short circuit the rotor via a variable current regulator? It is relatively low mechanical load to existing mechanism if the circuit is open (current low), and a much greater load if the motor (a generator this way) is loaded by current. For now don`t know the the values of required motors, as it depends on a lot of factors, but this is an idea only.

Robert

PS: The motor may be a one with short circuited rotor, and a wound stator, but than the load can be regulated by voltage applied to stator.

I though about that, some frasca's use a linear actuator that changes the center position of the spring for simulating the trim setting.

[rsandor]

I though about that, some frasca's use a linear actuator that changes the center position of the spring for simulating the trim setting.

I`m not sure we are talking about same thing. Belove is a sketch. A DC motor free wheeling on the column mechanism gives a light load, but when the rotor of the motor (talking about a permanent magnet motor) is short circuitted by a switch, it will represent a much higher load. If we replace the switch by a current regulator modulated by the sim...

Robert
1487

[WJH308]

I`m not sure we are talking about same thing. Belove is a sketch. A DC motor free wheeling on the column mechanism gives a light load, but when the rotor of the motor (talking about a permanent magnet motor) is short circuitted by a switch, it will represent a much higher load. If we replace the switch by a current regulator modulated by the sim...

Robert
1487

That can also work, in fact it is how many exercise equipment vary tension for different levels. You will need to gear it though so the motor turns many revolutions to the yokes one.

[rsandor]

Exactly! How many times to rev motor, while yoke revs half? What is the minimal/maximal tension? How can be coupled hw and sw... A lot of questions.

[Mike.Powell]

Hi everybody!

Just now joined Your forum, and already have to say something:
Load cells, are expensive by nature (those which are already conditioned and outputting voltage in range 0-5V). Unconditioned load cells are outputting extremely low voltage. A whole study required to eliminate problems related to temperature, and so on.
What if we mount a DC motor (with constant magnet as stator) as a constant mechanical load to pitch/roll mechanisms, and short circuit the rotor via a variable current regulator? It is relatively low mechanical load to existing mechanism if the circuit is open (current low), and a much greater load if the motor (a generator this way) is loaded by current. For now don`t know the the values of required motors, as it depends on a lot of factors, but this is an idea only.

Robert

PS: The motor may be a one with short circuited rotor, and a wound stator, but than the load can be regulated by voltage applied to stator.


Hi Robert,

This scheme generates a force that is proportional to the speed of movement of the control. The faster you try to move the yoke, for example, the larger the resisting force would be. Unfortunately, this is not the behavior we look for in control loading.

In a real plane the resisting force varies with the position of the control. For example, pull back on the yoke and the control surfaces on the horizontal stabilizer deflect upward. There is an aerodynamic force on those surfaces pushing them down. This force will be translated through the rigging back to the yoke. As long as you hold the yoke back, you will feel this force.

Simulator control loading is designed to mimic the forces encounted during real flight. We need a scheme that can produce a force that varies with the position of the controls. This is why springs are used so frequently. Of course springs have shortcomings. They mimic forces over only a restricted set of flight conditions. Ideally, we would like to have stiff control feel at cruise, and increasingly mushy control feel as airspeed decreases. Then there's the whole world of turbulence, wing stall, etc.

[hercules]

iam not sure about the importance, but i found a cheap plug&play loadcell interface.

http://www.leobodnar.com/products/BU0836-LC/

It seems you can connect any load cell to it.

THomas

[nax228]

What about a hydraulic circuit system like the ones used to stear boats?
I think that way it would be easy to adjust torque and hook it up to a dual yoke setup.

[hercules]

i have a questions regarding Linear actuators. As i know they can move in 2 directions with a given max. force. So we can build a easy working trim system. But what is with the movements against the actuators with a force more than the max. force of the actuators? Is there a break trough possible, so we couls move the columns manuell?

In other words....does a linear actuator act as a variable spring with variable forces???

Thomas

[Mike.Powell]

There are many different kinds of linear actuators. They have different charateristics. Some could be applied as control loaders more easily than others.

What you describe, a spring with a variable spring rate, can be built. The general technique is to measure both the force and the displacement and use feedback to adjust the force as needed by varying the power to the actuator.

[hercules]

Mike,

ok..thx for the explanation.. But what i want to know is, does an actuator allow manuell movement? When i take a servo as an example....
As long the servo is supplied with voltage, you cant move it manuell. It also has a holding force and movement force! When you move it against the holding pressure, you will destroy the gear, but if you turn off the voltage, the servo will be easy moveable via hand without risk.

Is this the same with a linear actuator,or has it only fixed positions?

Thomas

[Mike.Powell]

Some do. Some don't. Some models of commercial control loaders are built very much like an RC servo, though much heavier. They have an electric motor followed by a gear train.

The term "linear actuator" is very broad. You really have to look at specifics to get specific answers. A servo designed for control loading can be over-ridden by the pilot because that's a design goal. A linear actuator designed to position a satellite antenna for receiving television is specifically designed to not be over-ridden by wind loading forces. Both may be called "linear actuator" but will behave in different fashions.

[hercules]

yes...thats answered my question. There are many types of Linear Actuators available with many different designs.

So it seems that the idea with control loading is not so far from reality, when we use linear actuators. I explored some on the website of trossenrobotics. They also offers force-sensor-controllers for a small money. But these sensors are not voltage based. They gives a variable resistance as output and the controller convert this resistance into a voltage (0-5V).

Thomas

[hercules]

HI all,

i found an interesting kind of spring! Is is a gasspring, that is varibable by a dc voltage from 0-4 Volts. In this range you can change the force from 0-300N! The responstime is within milliseconds and the dimensions are about 400mm. All parameters varies from type to type.

I requested a prototype for some tests and still waiting for an answer. The cost for one spring is about 300€.

I will inform you if i get them and toke some first tries.

Regards

Thomas

[ivar hestnes]

That was very interesting. Let us know when you have tested:)

Expensive parts but probably worth it if it is suitable for us.

[hercules]

I have more details of the spring now.

http://www.bansbach.de/easyerf/easyERF_engl.pdf

The sping has a complete parameter table, so it seems that you can adjust the force depending on the speed of movement.

Since i use opencokpit cards, i tend to buy the output card that can handle 0-50V per output, adjustable via SIOC in 127 steps. The input of the spring is from 0-4V. Thats not so nice, because i would have only 10 steps for change the force. (50V / 127 = 0.4V) (4V / 0.4 = 10).

What so you think about this plan? I could also use a stepper motor, that moves a pot of an adjustable power supply.

I think i will get the springs next week.

Regards

Thomas

[IanH1960]

Hi,

These look like interesting devices. It might be worth just getting a little clarification from the manufacturer of their function first, especially if it is a spring with an adjustable spring rate (or stiffness) that you are looking for.

They are described as adjustable "dampers" and the brochure suggests that it is the damping force (ie the element of force that is proportional to speed of movement) that is adjustable. The basic extension force is described separately and there is an implication there that the adjustable element is only part of the total force.

My reading would probably be that each size has a fixed spring rate but with an adjustable damping force component to allow more or less resistance to be added depending on speed of movement.

Perhaps worth checking - given the cost.

Ian

[hercules]

I think only one site (push) is adjustable i relation to the speed. The other site has a fixed extension force. But i will ask for further information.

Since the dempers has only one adjustable directions, you will need 2 of them for each flightcontrol (elevator, ailerons)..one for push and one for pull.
But when iam wrong and the dampers has 2 adjustable directions, one dampers could be enough for each fc.

Maybe my informations about real flightcontrols are not correct.
Is there a difference in force in relation to the speed of move? If not, these dampers might be useless.

THomas

[hercules]

Is the force of an linear actuator not always the same? When the actuator moves to increase force to the columns, it will move the columns. But when the pilot fight against this force, the mechanics will destroy, cause the actuator can´t be move against its motor direction. May i be wrong.

Or is it the torque of the actuator-motor that can be override?

Thomas

[IanH1960]

Maybe my informations about real flightcontrols are not correct.
Is there a difference in force in relation to the speed of move? If not, these dampers might be useless. Have a look a Mike's post here -

http://www.mycockpit.org/forums/showpost.php?p=73920&postcount=23


Is the force of an linear actuator not always the same?No - the force an actuator generates will depend on how it is powered - for example if it is an actuator driven by an electric motor the force characteristics will generally mirror the torque characteristics of the electric motor. If it is a hydraulic actuator the force will mirror the pressure characteristics of the hydraulic system supplying the pressurised fluid. This is a general principal only - the mechanical design of the actuator may also affect the force characteristics - eg a lead screw actuator may provide very large resistive forces if you try and back-drive it from the load end, whereas a ball screw actuator may not ....

As Mike said in an earlier post there are lots of different types of linear actuators with a wide range of force characteristics. There can be quite a lot to get your head around!

Ian

[hercules]

Hi all
i reactivate this thread, because i spend much time to the control load idea. I wrote Leo Bodnar. He made great interfaces for inputs like buttons and switches, pots but also load-cells.

He offer some load-cell amps, that could be connect direct to a analog-input off a joystick-interface. The value of this input could be read out by SIOC and could be (recalculated) send to a linear-actuator controller. If we stay on SIOC, you could use the DC-outputcard, that can handle max 50Volts in 127 steps.

In this way you can realize control-loading in an easy way.


Are this ideas the wrong way, or what do you think about it?

Regards Thomas

[cscotthendry]

I've been tossing around an idea in my head about force feedback.

From what I've read, the problem with most of the force feedback systems is that what you feel is related to what is providing the feedback force. One of the main issues I've seen discussed is the cogging of DC motors and how it makes the feedback feel unrealistic.

A solution I thought of (kick me if this has been suggested before) is a leadscrew type of control positioning device, but coupled to the controls with springs. I haven't tried this and I don't know whether the spring action would mask the motor cogging or not.

One of the problems with spring returns that I found is centering. If you have two springs opposing each other you have several choices:
a) Both springs have a bit more tension on them than what you want to feel. They should have enough tension such that when you move the control fully one way, the spring in that direction doesn't become completely slack. Unfortunately, I found that when you have springs working against each other this way, you don't get reliable return to center due to friction in the control mechanisms.
b)Instead of having the springs in line with the control movement, you have one mounted at a right angle to the control travel and mounted at the center of the control's travel. I've found that this arrangement also doesn't reliably re-center the controls.


Here's a possible solution. It has crossed springs with a center stop. (Top view)

The advantage of the design is that when you pull or push the slide, you will be working against one spring only and the springs never work against each other. Each spring only brings the slide back to the center stop. That way, both springs can have enough force to return the control to the center, but they don't need to have enough force to overcome the other spring as well. Also, you get a definite center position which you don't seem to get with opposing springs. The idea I had for the force feedback is to mount the rods and the center stop on a leadscrew mechanism. Hopefully the springs would give an even force and mask the motor cogging. The leadscrew would effectively move the center position so that for those aircraft where the controls droop when the plane is standing still this could be simulated as well as shakes and bumps.