Beitrag geschrieben von:

Beitrag in der Kategorie:

Technik, Tipps & Theorie

Mechanische Lösungen:

Ein Foto einer RDS-Anlenkung zu zeigen wäre sinnlos: Sie verschwindet vollständig im Flügel. Lösungen dieser Art sind schon einige Zeit im Gebrauch (siehe Harley Michaelis’ RDS); die hier beschriebene (leicht unterschiedliche) Version zeichnet sich durch besondere Festigkeit und Spielfreiheit aus. Ein sehr alter Artikel in englischer Sprache
An Alternative Aileron Linkage for Model AeroplanesAs pushrods and quicklinks cause considerable drag, even when covered, and are sometimes damaged during landing, I looked for a new kind of linkage. It should be ruggged, precise and it should vanish completely in the wing. Fig.1: Basic principle Fig.1: Basic principle

What came out looked quite close to a linkage which was already in use for several years and was published first as “RADS” (Rotary Aileron Driver System by Harley Michaelis.A RADS mechanic has some interesting advantages: It can be built as flat as the servo, is as steady as the servo’s gear and its precision is limited only by your skill. Its drawbacks: It requires some effort to build and you need modern RC-equipment, because mechanical adjustments are not possible.

As my version differs in some details from Harley Michaelis’ I describe it here and give some hints on how to build it without looking for hard to find material or costly equipment (you should be able to braze, the rest is standard for aeromodellers).


The mechanics are easy to understand: Figures 1 to 3 show the basic principle.

Overview Version 1 Fig.2: Version 1 for larger aeromodels I have built version 1 into the wings of my Multiplex ACRO Star (a great bird:-) which offers enough space even for a medium sized, high torque servo near the main spar. The bent end of the rod (the pin) is nearly perpendicular to the hinge, this making it easy to dismount the control surface if it should be necessary. This version is also shown in Fig.1.

In some respects the mechanics are comparable to a torsional link. The two bearings for the rod, the one near the knee (B) and the other one in the servo (A), allow the rod to turn, and the pin/slot linkage in the flap (D) allows all degrees of freedom except translation in the Z-direction, thus driving the flap. Obviously, to get maximum precision, a bearing should be used at location B which doesn’t allow the rod to move vertically. When enough space is available a ball bearing may be the best choice; for applcations in very light aeromodels or when the wing is very flat a simpler bearing may be necessary. The bearing in the servo is viewed to be “good enough”. The slot is the critical point but it is not too complicated to build this part with acceptable precision.

Version 2 may be used for very small aeromodels. As I finished my Sambal XXL (a 2m-glider, the suffix “XXL” is absolutely misleading) at the time of writing this article, I can show photos of this version below. The two ball bearings inside the servo are used. A link of this version can be dismounted. As described later this version cannot withstand high lateral loads and is therefore only to be used when no room is available for an external bearing. In the special case of the Sambal XXL I simply trust the housings of the two MPX-FL-servos – we will see… Overview Version 2 Fig.3: Version 2 for small gliders


Weight, lift and, in case of (hard) landing, inertia forces of the control surface act upon the rod and the bearings. The main loads on the link may be viewed as follows:

  • At location D the upward or downward (lateral) force is FD. So the lateral load on the bearing at A is FA =FD·BD/AB and the load on the bearing at B is FB= -FA-FD. The distance AB should be as long as possible, to protect the servo. Note, that for version 2 the lateral load on the bearing at location B is about 4 times the load on the pin (see Fig.3 and the measures given in the example 2 below)!
  • The maximum bending load appears near the bearing at B. The distance BC should be as short as possible. The bending load on the pin is comparable to the load on a conventional rudder horn. For large, heavily loaded versions the rod should not be tapered near the knee, as the bending momentum increases considerably from D to B.
  • Torque loads are forwarded unchanged to the servo; its gear must withstand all such loads (comparable to conventional links). An elastic twist of the rod would totally spoil the quality of the whole link and must not occur. This mainly dictates the material and the diameter (strength) of the rod to be used.
  • The lateral load on the hinge is comparable to that of a conventional link where the rudder horn is as long as CD. This makes it possible to construct a linkage which is very easy on the hinge.

A link with a rod of 4mm steel and a 2mm pin is good enough for my ACRO Star and survived 2 summers with 2 heavy crashes without damage. Please note, that these steadiness estimates are not applicable on Harley Michaelis’ version.

A Short Estimate of the Kinematics

Fig.4: Kinematics estimate Fig.4: Kinematics estimate A complete analysis of the kinematics is a bit complex. The pin of the pin/slot link moves along the surface of a cone which is defined by the pin, and the pin/slot link moves over a cylinder surface which is defined by the hinge and distance CD. Some simplifications make the whole thing much easier to understand: 1. Replace the cylinder surface by a plane “E” perpendicular to the rod (at location D); the pin coincides with E on an arc “k”.  2. Ignore the distance (d) between the hinge and the knee (set d=0). Both simplifications cause only a small error if the distance d is small (say d/CD<0.2).

Now we can enumerate the values which influence the flap’s deflection:

s, the turning angle of the servo, is the main value which (intensionally) moves the flap. For usual servos s is in the range of -45° to +45° (Note: As standard servos are neither symetric nor linear, these values are almost never exact). We can plot s at the center of the circle “k”. k, the “knee angle”, obviously limits the control surface deflection. The deflection of the flap can’t be geater than k. A view on plane “E” shows that sin (s)* sin (k)=  sin (j), the deflecton. A rough estimate: for the usual +/-45° the deflection j is about +/-0.7*k. This means that for a knee angle of 30° we can get about +/-20° deflection. Fig.5: Simplifiedkinematics estimate Fig.5: Simplified kinematics estimate

Some further remarks:

  • The nonlinear behaviour is not as desired: Dj/Ds becomes a bit smaller when the deflection increases; this should be compensated for in the RC transmitter by selecting a (greater) value for the exponential characteristic. This effect is not very distinct for, say, |s|<45°.
  • With ailerons the hinge is usually located above the knee (d>0), this increasing the upward deflection, a (very small) desirable differentiation effect.
  • The difference between version 1 and 2, the angle between the rod and the hinge (in top view), is negligible as long as k isn’t much greater than 30°.
  • The aerofoil normally has a curved centre line; this means that the rod must be turned such that the pin is directed downwards a little when the flap is not deflected. So the centre position (neutral flap) corresponds to a small negative s in the sketches above. In practice you will mostly not succeed to install the slot absolutely symmetrically in both flaps; to adjust the centre position is the only possibility to correct this.

Some Remarks on the Pin/Slot Link

The slot link needs some degrees of freedom:

    • Obviously the pin moves in spanwise direction and rotates (“rolls” and “yaws” 🙂 in the slot. 
    • If the knee is not in the hinge line (d¹0) the pin also must be allowed to be pushed into or pulled out of the slot (no problem) and it must be allowed to cant a bit. See Fig.6: For an upward deflection the centre line is plotted:

1) If it couldn’t cant (move against the flap with respect to the X-Z-plane – I just used the graphics program to rotate the flap and the pin): The knee would have to move from C to C’; this is of course not possible. The hinge would wear out. Harley Michaelis obviously solves this problem by allowing the rod/pin to move vertically (he doesn’t use a ball bearing).2) Canting a bit towards the hinge: The cant angle is about atan(d/CD)-atan(cos(j)·d/CD). The table lists values of the cant angle for several d/CD and j. The values are rounded up; of course only the two maximum values for your configuration for up and down deflections are of interest. For the configuration in Fig.6 a d/CD=0.2 is assumed. For 35° upward deflection a cant angle <3° (toward the hinge) is to be taken into account; additionally for a downward deflection of, say, 20° a cant angle <1° (away from the hinge) occurs. For usual configurations the cant angle is small. But if no attention is payed to it, the friction drag of the link will increase considerably.

Fig.6: Canting pin Fig.6: Canting pin

j d/CD
.4 .3 .2 .1


The 1st example, the animation at the top of the page, was copied from a real design (I took the measures from DesignCAD 3D). The 2nd example is the Sambal XXL:

Item Symbol Example 1 Example 2 Fig.7: Example data (Sambal XXL)
Knee angle k 30° 40°
Vertical distance d 4mm 4mm
Length of the axis AB 92mm 8mm
Distance from bearing to hinge BC 4mm 14mm
Length of the pin CD »16mm »11mm
  d/CD »0.25 »0.3
Centre position s0 10° (down)  
Servo motion®Flap deflection s j +45°®24° up  -45°®19° down  (see Fig.1) see Fig.7

How to Build Such a Linkage

There are two critical parts: the rod/pin combination and the (precise) slot. I shall describe the rod first.

The knee should be “sharp” to allow a short distance BC; the knee angle k should of course be equal for the left and the right aileron.

Version 1 (see Fig.8): As already mentioned, I used a 4mm steel rod for my ACRO Star. The first step was to bend the rod by about 25°; afterwards I cut and smoothed it “outside” the knee such that the resulting plane had an angle of 30° against the straight part of the rod (a). Then I milled a groove into it (b) and brazed a slightly tapered 2mm steel rod (c) into it (d). Of course this isn’t easy because you need at least 3 hands to do this until you fix the parts otherwise during brazing – see below. Fig.8: Knee assembly Fig.8: Knee assembly
Fig.9: Knee assembly, small version Fig.9: Very small version Version 2 (see Fig.9): The straight part of the rod is simply a screw, fitting into the shaft of the servo (3mm), thus extending its axis. A thinner rod (2mm steel) is brazed (b) into a groove, milled into the screwhead (a). Afterwards the knee is smoothed (c) as needed. Part d of the sketch shows the finished part, screwed into the servo shaft and fixed with a nut and a no slip washer.

It is impossible to fix the two parts and to bring the solder into place without a positioning aid (see photos): It consists of 3 metal strips, two of them fixing the parts and the third serving as a grip (1). In the case of my SAMBAL the screw is fixed with a banjo nut and secured with a nut (2) on the first strip, the pin is fixed with a washer on the other strip (3). The fixing aid can be dismounted such that you can take out your finished part. Use a simple template, just consisting of two pieces of balsa on a plate, to mount everything together (4). Photos: Details for building the link The slot is a bit more complicated. Of course the basic idea is to use 2 pieces of the same steel as the pin as references (R) to define the wideness of the slot. Hereby you may not simply press the two outer parts on the reference rods, because during brazing the red hot steel becomes a bit weak and thus the slot becomes too narrow. Instead, fix it as can be seen in photo 5; fix the one end tightly with some wire (A) and rather weakly near point B in order to avoid bending; look that the rods are parallel. At first braze at point C, then remove the wire and braze at point A. Cut and smooth it as necessary. After cleaning the pin should be moveable without drag and should be allowed to cant as necessary.

Photo 6 shows a finished slot. Of course the two outer rods (O) may be abraded outside, such that the slot fits into the flap. Keep a part of one of the reference rods (P), it will help positioning the slot into the flap. Use a tiny piece of balsa strip (B) to ensure that the slot is fixed in the right direction; remove it later. I used only a bit of 5-min-epoxy until everything was ok – the piece P also allowed me to remove a slot to correct a fault :-). Fig.10: Slot istallation Fig.10: Slot istallation

Photo 7 shows 2 servos with rods and pins. Note that with this configuration the servos turn into the same direction when both flaps shall be deflected into the same direction, thus nearly eliminating asynchronous movement due to nonlinearity in the servo control electronics etc.; nevertheless the cables are on the inner sides.

 The coupler: For version 1 I brazed something like a fork onto the rod and used a servo output wheel like in photo 8, such that in case of an overload the fork can jump off the servo; it didn’t till now. Harley Michaelis’ coupler is possibly easier to build and comparably precise and steady. For Version 2 no coupler is necessary, just secure the rod with some screw fixing glue.

Possible Improvements

  • You might wish to replace the long straight part of the rod by a piece of carbon tube to save some weight; I never tried this, because I’m afraid that the joins are not steady enough.
  • An elasic link could be a life assurance for the servo in case of a hard crash (fence-ing or tree-ing:-). I tried to fix the fork (photo 8) with springs onto a half open servo output wheel, but it didn’t work. When I (or you?) have a good idea or even a ready solution, it will (should) be described here or a link should be placed below.
  • Currently I am thinking about an extreme version of this linkage, suitable for small gliders. The idea is to fix the pin directly in the servo shaft and mount the servo directly before the hinge (see Fig.11). Up to now I have no idea, how to drill the hole for the pin into the shaft – exactly. If enough room for a srcew is left, the pin etc. can be dismounted without removing the flap (!). Another challenge is to mount the servo in a way such that the wing is not weakened there. We will see…
Fig.11: Extreme version Fig.11: Extreme version


  • The aeromodeller Charlie Fox was reported to have used such a mechanic in the 1970s, but he didn’t publish it.
  • Harley Michaelis described his 1st implementation in September 1993 in “Model Aviation”, and publishes his latest RDS version in April 1998 in his RDS website.
  • Dr. Martin Hepperle published an article on the drag of pushrods.
  • Multiplex’s ACRO Star, a heavy 2222mm wingspan acro glider isn’t sold any longer and I am really glad that I still have one.

Please help me to get the reference section as complete as possible.

Thank you, Roger Ogle and Mike Garton, for your help!

Legal Information

Copyright © 1999 by Helmut Stettmaier. All rights reserved. The information in this article is presented “as is”. No warranty, expressed or implied, can be given that it is free of errors or that the contents are usable for any purpose. No liability can be accepted for any damage which may result, directly or indirectly, from usage of the information in this article. The information in this article may be used only on a private, non profit basis, at your own risk.