Yoke control loading?

[GodAtum]

Just saw this released by FSC http://www.flightsimulatorcenter.com/Product.aspx?lang=ENG&ID=116b36bf-6d6b-45ac-bcc5-81fd47ad3b02&CODE=390455#ccc

Looks amazing but very expensive. is there any way to this with a DIY approach? I guess the hard part will be programming and calibrating custom software for PSX?

[Hardy Heinlin]

I don't understand the product description. Are these pitch and bank actuators for the autopilot system, to move the controls? If so, why is it called "control load"? Otherwise, if it controls the feel pressure, why does it move the controls?

If it's about elevator feel pressure: The required network output from PSX is already available (elevator feel pressure in psi units). PSX computes it by a special algorithm using the same principles applied on the real 747, including system malfunctions etc.

Details:
Aerowinx manual page 265
http://aerowinx.com/board/index.php?topic=2514.msg24654#msg24654


|-|ardy

[John H Watson]

Are these pitch and bank actuators for the autopilot system, to move the controls? If so, why is it called "control load"? Otherwise, if it controls the feel pressure, why does it move the controls?

It seems to have 3 functions (for the B737): Autopilot, manual control with hydraulics (with elevator feel, I assume) and manual control without hydraulics (possible on the 737). i.e. Autopilot control and two manual flight control methods.

They perhaps chose a name which represented most of their programming efforts? (and a name which wasn't already covered by copyright).

[Hardy Heinlin]

If it's also able to provide a centering force, then perhaps it works like this: There are centering springs on either side, and the actuator moves the entire spring system to the center when the pilot is not deflecting the elevator. When the pilot pushes the elevator forward, the actuator moves the entire spring system backwards by a certain degree; the more the pilot deflects the elevator, the greater the actuator movement. This increases the opposite force, while keeping the elevator on the desired side. The force is promptly removed when the stick is centered. Vice versa when pulling. I don't know if it works like this. It's just a thought. If it does, it's a clever, universal solution, I think.


|-|

[farrokh747]

Thats quite a price!

you could try http://www.ebay.com/itm/Boeing-747-Aircraft-Feel-Computer-Assembly-401-13151-04-/231438822587?hash=item35e2d42cbb:g:kp0AAOSwwTlUpI4d&vxp=mtr

:-)

dynamic force feedback on the Pitch and Bank channels,

AFAIK the bank load is static ...?

Earlier, I used to use this: http://northernflightsim.com/dual-linked-flight-controls/ - no dynamic force or movement in auto here, but gave a pretty realistic loading....

fc

[GodAtum]

I found the specs for a 737

The specifications for the real aircraft state that the control column has a 37 pounds push/pull value +- 4 pound, while the roll pressures are 12 pounds +- 3 pound.

What is the equivilent for a 747-400?

[John H Watson]

The specifications for the real aircraft state that the control column has a 37 pounds push/pull value +- 4 pound, while the roll pressures are 12 pounds +- 3 pound.

Under what circumstances? Hydraulics on/off? Airspeed? Stabiliser setting? How far away from neutral? Push may be different from pull. Stating one value is pointless (unless the website states the preconditions).

[GodAtum]

The specifications for the real aircraft state that the control column has a 37 pounds push/pull value +- 4 pound, while the roll pressures are 12 pounds +- 3 pound.

Under what circumstances? Hydraulics on/off? Airspeed? Stabiliser setting? How far away from neutral? Push may be different from pull. Stating one value is pointless (unless the website states the preconditions).

It was from this blog: http://www.flaps2approach.com/journal/2012/12/10/genuine-b737-control-columns-a-closer-look.html

I'm not sure what the conditions are.

[farrokh747]

The specifications for the real aircraft state that the control column has a 37 pounds push/pull value +- 4 pound, while the roll pressures are 12 pounds +- 3 pound.  These pressures can differ from aircraft to aircraft, but fall within the published specifications. To replicate the push, pull and roll forces as accurately as possible, four heavy duty springs have been fitted to the column mechanism.

This is a well documented and cool project  - however, the description seems a bit simplistic -

the 73 elevator feel system (like the 744) is controlled by the feel computer, in series with the feel unit - the feel computer will send a metered hyd psi to the feel unit's actuator which will give resistance to the aft elevator quadrant via a cam/roller system - the feel computer takes inputs from the stab position as well as the pitot static system to calculate a load, in PSI, which is between 150 and 2100 psi - This variable hyd psi will eventually determine the load felt on the stick - additionally, the way the cam is cut, the force on the fwd and aft movements are different -

This site had some very good info on this: http://www.737flightsim.com/737center.html

Additionally, the 73 has a mach trim actuator, which the 74 does not -

The roll/aileron forces are constant, and do  not vary with the flight envelope (as is the rudder)

PSX has an output variable for hyd feel psi, after taking into account the above values -

FC

edit: further reading:

http://www.google.co.in/patents/US4477044

http://www.google.co.in/patents/US7559510

https://www.google.com/patents/US4477044

[John H Watson]

Additionally, the 73 has a mach trim actuator, which the 74 does not -

The 744 Freighter has a mach trimming system (for manual flight)

[Hardy Heinlin]

The 744 Freighter has a mach trimming system (for manual flight)

That just moves the stab trim (a tiny little bit), doesn't it? (Electronic stab trim commands like in the "speed trim" function, without special actuators.)


|-|

[farrokh747]

AFAIK, it's a mode/computation and not an actual physical actuator...... but i could be wrong....

c,

fc

[John H Watson]

That just moves the stab trim (a tiny little bit), doesn't it?

Correct.

To make things even more complicated, the 737NG also has a trim tab system on the elevators.

The 737NG Mach Trim is controlled by the Flight Control Computers. The FCC's provide an input to the elevator Power Control Units through the Feel and Centering Unit.

Some 737NGs have a stall warning system which affects the elevator feel system during a stall.

[John H Watson]

The 737NG Mach Trim is controlled by the Flight Control Computers.

This is with the A/P (or F/D) on or off

[Mark]

Is there any way to this with a DIY approach?

Definitely, I don't see any significant technical challenges with this. The cost of getting decent linear servos & servo drives is probably the biggest cost in DIY. Decent linear servos with proper feedback (I'd say both quadrature position/speed feedback and strain gauge torque feedback would make life easier), paired with appropriate servo drives & a bit of calibration makes the programming job relatively easy (easy for me anyhow).

[GodAtum]

Maybe with something like this? http://www.robotshop.com/uk/rugged-precision-brushless-robotic-servo-motor-large-aerospace.html

[Mark]

Hypothetically...
(I would need to do more than a quick 5 minute search of google before embarking on a project like this)

I am thinking more along the lines of this.
(the thrust capability is stupidly high but it's also the fastest linear actuator they do with enough thrust)
The actuator outputs a position signal (I would possibility consider a 2nd position feedback sensor of some kind).
I would then put a load cell in the linkage between the servo and the yoke mechanism, like this or this.
With both mechanisms of feedback, I can get position and the force that the pilot is applying and vary the load as required (& move the yoke when in autoflight modes).

[Hardy Heinlin]

I'm starting to doubt that a single force vector can reasonably simulate symmetric feel pressure.

That single vector must counteract a fraction of 100% of the human force. That's certainly possible.

But getting the actuator's reaction speed 100% in sync with the human action will be impossible, I guess.

Example ... pulling on the yoke:

Human force: 1.0 N
Actuator force: -0.9 N

Human force: 2.0 N
Actuator force: -1.8 N

If you change the human force slowly, it may work. But, for instance, if you now release the pulled stick quickly from 2 to 0 N, the actuator's counterforce might decrease to 0 a bit slowlier and actually push the stick forward for a fraction of a second while the human force is already released.

Similar effects might occur when the pilot pushes or pulls very quickly. The increase of counterforce from the actuator may occur too late, leading to unusual force oscillations within the feel system.

So I think it's not possible to do both jobs -- autopilot and elevator feel -- with just one actuator.

The elevator feel pressure should always be symmetric and should never move the stick off the center, not even for a millisecond.


Cheers,

|-|ardy

[the mad hatter]

Mark.. I am thinking a trip to TX is in order to put this to the test naturally it's all on my penny :-)   we have good meat and wild woman galore .....when works for you ?  yesterday would be great for me ;-)

[Mark]

But getting the actuator's reaction speed 100% in sync with the human action will be impossible, I guess.

Everything you said is an effective description of control lag, which can be resolved with faster control loops and careful attention to reducing any delay sources (like phase response of anti-alias filters, depth of data buffers, digital filter topology [if used], speed of motor response, low impedance wiring etc).

Have you ever had a ride on a Segway? That's a classic example of where the human perception isn't quick enough to detect any delay in response - when you lean forward, the Segway immediately (and intuitively) responds... or that's how it feels anyway. In reality, it probably has a control loop in the 200hz to 5khz loop rate region.

Sure, if you apply an impulse shock (I'm not sure a human could be quick enough), you might not get the required loading for a few milliseconds but I don't think that's something that happens in normal flight conditions.

Human force: 2.0 N
Actuator force: -1.8 N

Is feel pressure constant (or at least; slowly variable or rare step change due to virtual malfunction)?
That simplifies things; once the hypothetical system has been correctly calibrated (not an easy task, involves PID loop tuning, input scaling calibration & more) then the control loop (when not in any autoflight modes) just needs to maintain a constant resistance load (easily measured from the load cell) against human input for any yoke input away from a neutral position.
What I'm not yet clear on - what happens if you push the elevator to full deflection and then let go? Your response seems to suggest it would maintain position.

The elevator feel pressure should always be symmetric and should never move the stick off the center, not even for a millisecond.

I'm not saying it'll be easy but with appropriate hardware it's not impossible with a single actuator.

Maybe step one is to put my theory to the test with a scaled down proof of concept? Unfortunately I don't think we have the space to put a system like this in Gary's sim.