DIY Retracts



When looking at the possibility of building the Welkin, one of the first areas of concern was the need for retracts. I was surprised to find that the readily available commercial units all seemed to be designed for sideways retraction whereas on the Welkin (and indeed on most twin-engined aircraft) the undercarriage legs retract rearwards. This of course means that fore and aft landing shocks are transmitted directly to the locking mechanism and my local model shop was not able to re-assure me that any of the units they stocked had locking mechanisms designed to withstand this.


Whilst contemplating this problem, my attention was drawn to a design for DIY retracts by Ivan Pettigrew. Ivan, whose plans I used for the Sealand and Rapide, has a reputation for simple, low-tech, workmanlike solutions to modelling challenges so, in spite of never having acquired any metalworking skills at school, I thought it was worth having a look.


The drawing on the right, drawn for Ivan by Hal Norrish, shows the principle of operation. (click on it to view it in full in a separate window)


The key to the mechanism is the operating crank which runs in a slot formed by making a hairpin bend in the top part of the leg. Indeed, making this hairpin bend is really the only metalworking challenge in the whole thing! If you look at the bottom three views on Hal's drawing you will see that, both in the retracted and fully extended positions, the crank arm is at right angles to the slot so that shock loads on the wheel cannot apply any turning force to the crank and hence to the operating servo.


The first snag was that the full size Welkin drawing shows a retract angle in excess of 150 degrees which is simply not achieveable with this mechanism. After a bit of doodling though, I was able to find a position for the pivot which would reduce the retract angle to 90 degrees and still preserve the scale angle of the visible part of the undercarriage leg.


Although the undercarriage leg was now longer, it looked as if it should still be possible to accommodate the retracted wheel in the nacelle, so I pressed ahead and built a rough mockup of this arrangement and, just to test it out, mounted a micro servo in front of the bulkhead to operate the crank.

The video on the right shows this mockup in operation.


The unit worked well but served to draw my attention to the fact that the price paid for this simple locking mechanism is that, at mid travel, the operating crank passes close to the pivot point of the leg meaning that the servo has to work quite hard at this point. If you do the geometry you will discover that this loss of mechanical advantage is directly related to the retract angle - which is no doubt why Ivan's original drawing shows a retract angle of 80 degrees rather than the 90 degrees one might have expected - back to the drawing board!

By this time I had decided to put my faith in the locking mechanism and stick with idea of a micro servo in each nacelle rather than the centrally mounted retract servo used by Ivan.


The original mockup had needed a counterbalance spring to cope with anything other than a lightweight wheel but the Mk2 units proved to be much more powerful, demonstrating the importance of reducing the retract angle.


This second video shows the two units operating together off a Y-lead. You can also see that the sideplates have gone, saddle clamps being used now for mounting both the leg and the operating crank. The collet on the upper portion of the leg is for a balancing spring but in the end it proved to be unnecessary - these units will easily lift double the required load without the need for spring assistance.

With the units working, attention turned to building the model until, as the nacelles began to take shape, the question of the u/c doors had to be addressed.

I read about many different approaches to this, most of which were beyond my competence, so I decided to try a simple approach first. I did though make my own hinges out of 1/32in ply. This was a tedious business but did mean that the doors could be temporarily hinged by pushing pins through the hinges, while the detail of the closing scheme was being worked out.

This video shows the first experiments whereby the door is pulled closed by the u/c leg encountering a loop of fishing trace. This simple approach worked surprisingly well, allowing the door closing operation to be delayed until the wheel is almost into the nacelle.

So - what's the verdict? Well, the system has so far proven to be robust and reliable in use. Making sure that you get the geometry right in adapting the system to your particular model is obviously crucial. In addition to that I would stress the following points:


IMG56231. Choose wire that is strong enough for the job - but not beyond your wire-bending capability! I used 3mm wire which has proved to be about right. Even so, making the hairpin bends was a bit of a challenge and so I made them a bit over-long and 'fine tuned' the end of the hairpin with an epoxy fillet as shown in the picture.


2. Saddle clamps work well as bearings. I did have to ease mine slightly so that the wire could rotate freely. Space them as widely as you can to allow the torque rod section of the u/c leg to work. Because access is limited, I fixed the saddle clamps with hex screws into captive T nuts to be sure that the leg could be removed should it need to be straightened.


3. If I were doing this again, I would use retract servos. The standard HS85MGs are plenty strong enough and the locking mechanism does protect them from shock loads. However, I now appreciate that there is more to the different servo types than that. For instance:


i) Standard servos are much faster than a retract servo, so you need to have a servo slow function on your Tx (and be careful to check the switch position before switching on!) otherwise things can get a bit violent.


ii) Retract servos drive at full power right to their end point, then stop. A normal servo 'homes in' on its target point and reduces power as it gets nearer. This means that the end point is less well defined when under load - for example when drawing u/c doors closed. This can also result in a higher residual current drain than with a true retract servo.


So that's about it. The full size Welkin has a retractable tailwheel but I drew the line at that! If anybody adapts this system for use with a steerable tailwheel or nosewheel, I would like to hear about it - After all, I still have a yen to build a bigger P38 one day!


You can read more about the retracts in the context of the Welkin build in the Project Diary