Expo - And The Little Grey Cells

[scallybert]

Here's something to think about. With no expo, if you circle a joystick, the output (eg if the Aile & Elev channels drove an X-Y plotter) will be a circle. That is, it will follow the sticks.

If you have significant expo applied, it's no longer a circle; but becomes more like a curvy cross, the more expo there is. The shape may well change, depending on the size of your circle.

Does this matter ?

I'm not sure, but I've just seen some postings referring expo and circling sticks - so maybe it does.

[trillian]

No, it doesn't matter what theoretical shape is created, at the end of the day it's two independent channels sending a certain number of pulses to the RX.

If you're thinking in terms of FBL controllers having global expo values, then it depends entirely on how the FBL unit deals with all that. Sometimes the overall resolution is better with no expo in the TX and only using expo in the FBL unit, sometimes it works fine turning expo in the FBL controller off and using only the TX. In theory it may not be good compounding expo but there are so many variables that it's not possible to make a sweeping generalisation.

[dillwhacker]

I would think it does make a difference.
Consider a piroflip. As the tail piros 360 degrees, to keep the disc iniline and in a smooth flip, you need elev, then ail, then negative elev, then negative ail, all 90 degrees apart i.e. like sine and cosine, so the total cyclic at any point is constant all the way round.....
Probably...

[scallybert]

If you wanted to make a cyclic input of constant magnitude, but rotating around the heli; then it would matter.

I don't know if you'd want to, but a piro manoeuvre might need this.

This almost isn't the point, though. It's more that expo affects the cyclic controls in ways that aren't so obvious.

[coolice]

Hey.

It's a good topic of discussion, but are we being led along by something that doesn't truly exist?

In theory, as long as your aileron and elevator expos are the same, no matter where you are in the travel range and as long as you keep a constant circle eg the cyclic stick the same distance out from centre throughout the prescribed circle, then the cyclic input will be the same.
This assumes;

1. You can keep moving the stick in a circle the same distance from centre. Add a mechanical cyclic ring to the Tx for this.

2. Your expos, cyclic travels and model geometry is all correct and the same as each other to keep everything linear.

The problem is there is so many variables to a achieve a perfect chao's that I don't think one thing can be attributed to problems easily.

I'm not suggesting too much cannot make some manouvers more difficult, but I don't think it would be detrimental just because it's exponential. You adjust your stick inputs to account for variables in individual model setup.
.

[scallybert]

Would you expect, say, that if pushing the stick forward a certain amount tilted the rotor disk/swash by a certain amount; and that a similar stick movement sideways, gave a similar tilt; then that a similar stick movement at 45 deg (kind of North-East), would give a similar tilt in that direction ?

Expo will mean this certainly isn't the case.

[Planehazza]

I'll have to think of it in more detail but I guess it would matter. Never thought of it until now. Say you want to circle the sticks. With 0 expo, it should be a linear 1:1 output. You stir the sticks, the swash should stir the same. Now ad 20% expo (-20% for us Futaba lads). If you stir the same the circle on the sticks, at 3/9 position (only aileron) then the output will be reduced slightly. Now move the stick around the circle to 10/4, so now you have both aileron and elevator (albeit less aileron now). Now you've got a 1:1 out put on elevator, even though the stick movement is the same as before.

OK so without doing any math (because I suck at it) I'm starting to think that a pure circle on the stick would be a circle with slightly dented corners on the swash. IE, the corners where you're mixing aileron and elevator are being softened twice.

Hope my waffling makes sense. For this, it may now be better practice to have 0 expo in the Tx so that the physical stick circle matches 1:1 the output and let the Vbar etc do the expo? I really think regardless, it is a bad idea to have expo on both.

TL;DR version:

Would you expect, say, that if pushing the stick forward a certain amount tilted the rotor disk/swash by a certain amount; and that a similar stick movement sideways, gave a similar tilt; then that a similar stick movement at 45 deg (kind of North-East), would give a similar tilt in that direction ?

Expo will mean this certainly isn't the case.

What I'm trying to say, is with expo, if you think you're making the swash tilt in a circle (deflection starts aileron left, then goes north to elevator forward, then East to Aileron right....) by moving the stick in circle then actually you're not. So to answer the OP's question then yes, this would affect PFs/Chaos moves.

There is probably a lot of pie related math that is too difficult for my blonde brain. Providing you move the stick in a true cirle (the radius from centre is constant) then you need a formula to convert X and Y axis amounts to a circular motion. Expo with change this output like the OP said, resulting in more softening at the corners where a mixture of both inputs occur.

[Mark_T]

In theory, as long as your aileron and elevator expos are the same, no matter where you are in the travel range and as long as you keep a constant circle eg the cyclic stick the same distance out from centre throughout the prescribed circle, then the cyclic input will be the same.

I started out with the same thought as well, but now I'm not sure.

The problem is that if the exponential has turned what was a straight line into a curve then you will not get a perfect circle as a result. The max and min positions for elevator and aileron will be points on your theoretical circle, but in between them you will not track along the circumference of the circle as the individual components of the input are at different points on the exponential curve so they are changing at different rates.

Now to what extent that makes a difference I'll leave to those like yourself qualified to answer.

[Planehazza]

I started out with the same thought as well, but now I'm not sure.

The problem is that if the exponential has turned what was a straight line into a curve then you will not get a perfect circle as a result. The max and min positions for elevator and aileron will be points on your theoretical circle, but in between them you will not track along the circumference of the circle as the individual components of the input are at different points on the exponential curve so they are changing at different rates.

Now to what extent that makes a difference I'll leave to those like yourself qualified to answer.

I figure the less expo the lesser the effect. I guess it would be interesting to know how expo is applied via the FBL controller? The problem is that a swash plate doesn't work on X an Y - only the servos, pistons or whatever do. We think of heli control as Yaw, Pitch Roll and Collective, but mechanically a swash plate is not that simple. Even if you split it degrees (0-360) in which it's pointing, that would then be 360 individual controls.

I think the answer is if you want linear control you have to fly linearly - ie without any expo.


Now after all this, it doesn't matter to me as I cannot do an pirouetting maneuvers anyway ha!

[dillwhacker]

Pythagoras folks. X squared plus Y squared = 1.
So at 45 degrees north-east stick position, aileron = 0.707 of max and elevator = 0.707 of max, and total deflection north-east = 1. That is without expo.
P.S I can't do piro flips either.
Or maths...

[scallybert]

I'd already done some simple maths as a sanity check.

As a simple model for expo, square the stick input. [This isn't totally crass, as y=x**2 is quite a reasonable 'expo shape']. The scaling goes wrong, but that doesn't really matter, as it's consistent.

Now, simulate moving your sticks through a circle of radius '2', and consider 3 points on the circle: (2,0) ; (sqrt[2],sqrt[2]) ; (0,2)

Now, put these through your expo function, and you get: (4,0) ; (2,2) ; (0,4)

For the output to be a circle, all three should be the same distance from the origin. Those distances are: 4 ; sqrt[2**2+2**2]=sqrt[8]= ~2.8 ; 4

So, it's quite a bit not a circle.

[scallybert]

BTW, please don't assume we're talking about V-Bars.

This is far too pure for them...

It's more a bit of thinking, and something that might not have occurred to people.

[coolice]

Hey.

I see the logic here, but assuming all control surfaces move the same distances and the settings are the same on expo for both cyclic controls, the controls steps will be the same. This is the whole idea of let's say eCCPM, 3 servos all moving at the same speed and control steps.
Also, does this then go against the cyclic ring facility both mechanical and electronic? Perhaps Mikado add in the small amount to counteract the control ball moving within an arc on the servo arm/horn.

Are we now looking at model control design geometry flaws rather than electronic induced differences? A well setup/designed model should give the same NE swash plate deflection as a NW for the same amount of stick input.

A good check would be someone with a mechanical swash plate ring on their Tx, a well levelled model/table and a digital pitch gauge. Check the blade pitch at various circular points with and without expo added, dual rates could be used to simulate a smaller stirring circle. Although this won't perhaps work on a FBL model due to the way the FBL unit's work.

BTW, please don't assume we're talking about V-Bars.

This is far too pure for them...

It's more a bit of thinking, and something that might not have occurred to people.

Hello Mate.

Ah no, I refer to all models/types overall.
.

[scallybert]

I reckon, that if the CCPM is 'ideal', then moving the cyclic stick in a circle should mean that the normal (perpendicular to the plane) to the swash or rotor disk plane similarly moves in a circle. (Certainly the swash.) [This does disregard quite a lot.]

ie N, NE, E, SE, S, etc tilts are all the same.

Expo means this is no longer true; simply because the cyclic values output by the Tx are affected. It's like you have different expo curves depending on cyclic compass bearing. N expo curve <> NE expo curve

[coolice]

Yeah, I see what you mean now I think. Just stuck the tracking on the AutoCAD and followed the circle. As you move around the arc, say you start with forward elevator and move clockwise, you're using less aileron which then increases as you go further round the 1/4 arc until your stick is at the pure aileron point.
Interesting.

In my mind the mid way points, at the middle of each quarter section, the controls have moved the same amount, it's the in between parts that do not.

Question is, at what point does too much expo become more of a problem to these interim points? I'm not talking about 100% expo as that's silly, but moderate 30~40% and below.

I reckon, that if the CCPM is 'ideal', then moving the cyclic stick in a circle should mean that the normal (perpendicular to the plane) to the swash or rotor disk plane similarly moves in a circle. (Certainly the swash.) [This does disregard quite a lot.]

ie N, NE, E, SE, S, etc tilts are all the same.

Expo means this is no longer true; simply because the cyclic values output by the Tx are affected. It's like you have different expo curves depending on cyclic compass bearing. N expo curve <> NE expo curve

Ah, you just beat me to the post. Yeah so I'm on the right path, the main compass points if you like are correct, but the arc in between is the area that can be effected by the expo.
Question is, at a specific stick stirring action, does this matter? From what I've seen in my escapades of piro stuff no, only when my fingers timing goes wrong do I see a problem.
.