When do you use exponential?

[Bob Cotsford]

Posted by Biggles' Elder Brother - Moderator on 28/12/2012 22:30:51:I've always said that having too much time to think is a dangerous thing - amazing what you can come up with when painting a window frame!

BEB

I knew there was a reason I fitted uPVC double glazing all round!

[steve-richards]

I learnt to fly without it, i think it helps you but thats only my opinion

[riverlandgirl.]

My gear is too primative, and my sloppy, undisciplined flying style dosn't require it!......

[Erfolg]

I have been thinking about the proposition that set points are used to approximate the exponential curve. Although it could be a way of achieving the objective, it does seem an involved method. I am suggesting that the issues are having a table where the set points are held, compiled by multiplying (the selected percentage exponential chosen, which is converted to the) constant values of the set point, with the values from the potentiometer, to produce a value which is then converted into a signal to the Rx. Which seems a bit of work.

From the start i must make it plain I am not a electronics engineer or even an electrical engineer. On the plus side, I am old, i have been around a bit. Over the years I have picked up and been taught a lot of skills and knowledge, which often comes from being older. My early life was as a technician apprentice, who did the night school bit, obtained a degree. In obtaining that degree I did study instrumentation and control. Which did include feed back potentiometers, amongst other transducer type devices. From school i had learnt about inverting functions, treating the function as a sausage machine, as my teacher said. This concept turned up in IC, as part of the a mathematical method of calculating system response etc. I also learnt Fortran, later Pascal, Basic and VisBas. The early programming was all about structure to make best use of limited processor power and storage issues etc. Perhaps just as important, I had some casual training on a Electronics package for designing circuits, component placement and finally auto routing a board. Strangely all these things have many things in common.

My proposition is based on this non electronic back ground. The other aspect stems from working and knowing many engineers and mathematicians, who trained in other countries. Almost unbelievably there is a common approach to most engineering mathematical systems, all remarkably similar way of doing things.

I have assumed that the Japanese engineers will have sat down with a piece of paper. They then will have asked themselves what do we want to achieve. Stemming from knowing what it is, that they want to do. The next question is how can it be achieved? I am sure that the idea of set points will have crossed their minds. Being engineers they will have known of the sausage machine mathematical principle, learnt at school, the idea rearing its head in the field of circuit designing. Where components did the sausage machine thing, rather than just logic (if this do that, type gate processing) . The next question would then be is there a component that is commercially available that has a exponential type function? I now assume that there is at least one.

From this they would quickly have written down, how would I ideally design a circuit to do what i want. I assume that they will have said I have a circuit which translates a potentiometer value, to an output at the aerial. The relationship being a linear value. All I need to do is use this chip, which modifies its value dependant on the variable given by the voltage of the potentiometer, to a value which the circuit outputs to the aerial. Now if the processor allows a constant value to be selected, which is selected by the radio operator (whilst setting his equipment), wow, I now have an instantly calculated value just at the cost of the electronic component.

Although my description is long winded, it requires in principle one component to do the necessary calculation. No storing of data, that needs processing anyway,

[Chris Bott - Moderator]

Erfolg, with a microprocessor in a Tx then the exponential calculation is very likely to just be one mathematical step in the chain that leads from stick to Tx output. I expect it uses a formula very similar to those of BEB's. It's probably very early on, after stick position becomes a digital number but before any mixing, end points, sub trim and trim are added or subtracted from that number.

[Erfolg]

I had thought that Chris, Just a long winded way of getting to the principle that a calculation via a processor is easier than tables with set points.

It was one of the aspects of Circuit Design Programmes that did strike me, that so many functions were available on one component (selected from the catalogue). To be honest I was impressed. Although speaking with some electronic people more recently, they are less impressed than me. I had thought that a non electronics engineer could design a basic board, with just logic, the programme doing the rest.

[Myron Beaumont]

Erfolg

I'm a "non electronics " engineer (retired).I have just read your window painting thoughts ! As a mechanical engineer my mind drifted towards (guess what?) a mechanical solution to making linear motion from circular motion -no electronic manipulation being involved.Anyone heard of a CAM ?The sort used to operate valve motion in IC engines .One could be used in the control connection between the servo and the control surface .As the rotary motion of the servo reduces the linear motion of the snake /whatever then arranging a cam midships would increase the amount of linear movement as the rotary motion(s) become less effective by the hump on it increasing the effective movement & if designed correctly giving you perfect linear motion as per signal sent out by linear motion of your Tx stick.Are you still with me? .Now then, How do you do that you may ask? Simples .

Just got to think about it for a little while .Now which window should I start on ?

[Myron Beaumont]

OK

First thoughts-- A tiny epicyclic gearbox type arrangement on top of the servo ie sun wheel etc etc -like an automatic gearbox configuration of years ago .Could probably be fashioned from servo gear bits .

I'll come up with something else eventually I expect .Something like a bell crank in -line arrangement perhaps ? Come on you engineers !

[Erfolg]

I am sure Myron that you will remember the linear servos. Where the output was via a rack that was driven by the output shaft. I guess they disappeared because that modellers so no advantage, or maybe they were more troublesome/reliable.

To be honest, I have no issues with standard rotary outputs.

I can see that for the same angular motion at say the 0 degree point compared to the 90 degree position, will differ. Is the difference worth worrying about?

I can also see that I think BEBs argument, is basically right, if the servo rotates 1 degree, so will the surface arm, when set to the same radius (ie 1 degree). Of course it will not be if set to another radius, although this will be the same for a linear output

Like many things, good compromises often are more effective in practice than theoretical perfection.

I remember going around the science museum, looking at models of steam engines, some very sophisticated, all believed that they would be more energy efficient. What struck me that the search for perfection mainly ended in failure, either to intensive in maintenance, or just not realising the theoretical gains.

KISS always comes to mind!

[Simon Chaddock]

Surely if the plane of the servo rotation is the same as that of the control surface hinge and the arm and horn are the same length then the movement is identical over the full range.

I dont use expo either.

If the plane is too twitchy I reduce the throws or the size of the control surface or just learn to live with it!

[Martin Harris - Moderator]

Posted by Biggles' Elder Brother - Moderator on 28/12/2012 22:30:51:

Posted by Martin Harris on 27/12/2012 19:08:30:

A wise man (can't remember who) once wrote that the point of adding negative expo (in Futaba parlance) was to nullify the positive expo introduced by the rotary output from the servos. I believe he made a good point.

This point has been rattling around in head since Martin posted it. Then Erf's post asking about the nature of the exponential function used got me really thinking about it all

Hi BEB - it did have me waving fingers in space before I posted as I realised there would be an opposite and equal effect at the control surface with equal length arms and my empirical reasoning lead me to think it might be a valid point. I believe that when you take into account that (other than extreme 3D models, we normally (or at least we should!) set a mechanical advantage to give around a 2:1 ratio of servo angular movement to control surface (except in the case of a closed loop rudder) and then there are linear irregularities introduced. It's a pet hate of mine when I find people using 20% travel because they've just chosen the outside hole on the servo arm because it looks neater..lessons learnt when setting up models before ATV was invented do result in maximising mechanical advantage/granularity!

Coupled with my (again empirical) assumption that small deflections are proportionately more effective due to larger ones distorting the aerofoil shape further from an ideal shape (do you think there's any validity in this - your earlier post seems to suggest that the aerodynamic force generated is simply directly proportional to the deflection) then I'd suggest that there may be a significant non-linearity in effectiveness.

[Myron Beaumont]

Simon

I think you've missed the point .

[Erfolg]

I think we all agree on that Simon, with servo arm movement, withe the control arm, at the same radius.

What we are now considering is that for each degree of rotation of the servo shaft, will the distance moved relative to the axis of the pushrod wtc. will it be a linear movement.

My contention is that it is not. Easy to prove just by using Tan equation, with radius for the adjacent.

Although my opinion is that the movement is not linear for each degree moved. At a practical level I do not think it matters, Not One Jot.

[Chuck Plains]

Posted by Myron Beaumont on 30/12/2012 14:27:07:

Erfolg

I'm a "non electronics " engineer (retired).I have just read your window painting thoughts ! As a mechanical engineer my mind drifted towards (guess what?) a mechanical solution to making linear motion from circular motion -no electronic manipulation being involved.Anyone heard of a CAM ?The sort used to operate valve motion in IC engines .One could be used in the control connection between the servo and the control surface .As the rotary motion of the servo reduces the linear motion of the snake /whatever then arranging a cam midships would increase the amount of linear movement as the rotary motion(s) become less effective by the hump on it increasing the effective movement & if designed correctly giving you perfect linear motion as per signal sent out by linear motion of your Tx stick.Are you still with me? .Now then, How do you do that you may ask? Simples .

Just got to think about it for a little while .Now which window should I start on ?

Seems to me that a cam type movement could work Myron. But wear factors might make it heavy, due to the type of material that would be needed for the friction surfaces.

I used to operate a 'cam' type auto lathe, and the ultra hard metal cams had to run bathed in oil. Perhaps that would not be necessary with modern lighter weight ceramics that are available. But with the addition of ceramic precision cams, plus a chassis that can take the required pressure without deflecting, there goes the price, up into the clouds.

[Erfolg]

Chuck, You have the wrong guy, try Myron. Apologies, I see it is Myron you are addressing.

I was also thinking of autos, I have not seen any for years. Yet i remember seeing whole plant areas filled with them, where operators, cautiously fead the bar stock and cleared components and swarf. I wonder if they are still used in large scale mass production. Or perhaps UK no longer is a player in this area.

Oh, I also agree with cams and the general increase in complexity.

Edited By Erfolg on 30/12/2012 18:51:04

[Myron Beaumont]

A couple of near cross postings methinks . Simon ,I apologise for my abrupt remark as your following post tells me you realised your fundamental thinking . I do it all the time & am probably going doolaly anyway .

Happy new year to you all ,may your God go with you .

[Chuck Plains]

Haha! Sorry Erfolg. :D I was answering what you said about cams, but addressed my remarks to Myron. Hi Myron! Have a cake.

Ok, every one can have a cake!

[Chuck Plains]

Posted by Erfolg on 30/12/2012 18:46:52:

I was also thinking of autos, I have not seen any for years. Yet i remember seeing whole plant areas filled with them, where operators, cautiously fead the bar stock and cleared components and swarf. I wonder if they are still used in large scale mass production. Or perhaps UK no longer is a player in this area.

Oh, I also agree with cams and the general increase in complexity.

Edited By Erfolg on 30/12/2012 18:51:04

Can machines are extremely good for producing consistent components of moderate error tolerance over high volume production runs. But the components can't be complex in nature if a finished item is required from a single operation. So, yes, they are still in use.

[Mike Rondel]

Interesting reading everyone's thoughts.

On the basis that linear output servos are not seen these days, (cant remember the last time I saw any advertised), and that rotary servos do not provide a linear output, no doubt this was the reason that manufacturers incorporated expo - no doubt cheaper than incorporating an extra stage on a servo to provide a linear output. I always use expo on all surfaces in varying quantities so as to obtain, if possible, a near equal response to any input on the Tx sticks. I'm surprised to see that less than 50% of pilots in this survey use this facility.

[Dino Tagliero]

I have a sensitivity problem in my left thumb - flying mode1 - which causes unwanted controls interference when using elevator or rudder...So I am using "full" deflections" on both controls with some more exponential on rudder, which I do not use as much as the elevator. This way I do not have to move the thumb a long distance, more thumb movement equals more unwanted commands!

Obviously I also use exponential on ailerons.

The result is smoother and more precise flying...my landings have become nearly always perfect.

[Steve Dorling]

It is (frankly scarlett) impossible to suggest any reason at all for not using expo to some degree (oh that'll cause howls from the naysayers...) and I challenge anyone to tell the difference between no exponential and 5% - in which case how about the differnece twxit no expo and 10% - or 15% etc etc!

There will be a sweet spot that works for you - start with 30% on everything and then tune up or down until you find your spot (which may even be negative or opposing expo)!

 

(I'm here to help - make friends and influence folks - can you tell...)

Edited By Biggles' Elder Brother - Moderator on 31/05/2013 21:38:11

[Steve Dorling]

PS - everyone is flying with some exponential movement somewhere - from throttles to ailerons - elevators to flaps, there will be an exponential factor to consider - unless that is, you've install the perfectly linear, 100% square and aligned linkage, hinge, output arm or disk and control rod, which is doubtful!