Wednesday, 20 June 2012

Topsky 2 DLG review and modifications

For the ultimate in flying fun, there's nothing quite like flying an F3F model in a good breeze from the top of a beautiful hill. There is just one little problem though: unless you are lucky enough to live next to a decent slope, a session with an F3F model will involve a bit of a drive.

Which is why I also love flying discus launch gliders. Long lazy flights, using nothing more than arm power and thermals to keep the model aloft - and all from the local park! It's flying at its purest.

Choosing the Topsky 2
My introduction to DLGs was the Longshot, a popular model from Hořejší model. It taught me the basics - like how to launch the model without hitting things... or ripping off a wing panel in the grass (something you only do once, trust me!). However after seven years of frequent use, it was time to retire the model in favour of something new.

Then a month ago, a friend turned up at the park with a brand new Topsky 2. Not only did it look pretty, but it comfortably outflew my Longshot. The glide angle was better, it handled better, and the wings were better finished. Even allowing for the age of my Longshot, I was impressed! An order was duly placed with the ever-helpful Hyperflight, and my Topsky 2 kit arrived the following day.

In this post I'll describe my impressions, as well as some mods including how to save weight by installing a closed loop linkage. I won't go into the build in detail as there's a very comprehensive thread on RC Groups (see links at end).

Author with Topsky 2 Disser at La Muela in Spain. 'disser' refers to the criss-cross parttern of carbon behind the D-box.  (photo by Sylvia Eldridge)

Adding Lightness
The Topsky 2 has a fairly standard DLG layout i.e. 1.5 meter span, foam/kevlar/carbon construction and a Drela wing section.

In order to capture those low-level thermals, my goal was to make the model as light as possible. The target weight is 260-280g, however after weighing the kit contents, and rc gear (by no means particularly heavy), it was clear that some weight-saving measures would be required.

So how to save some weight?

A key decision was taken to throw away the supplied tube and cable linkage. Not only does it spoil the looks of the model (the tube runs outside of the boom - ugh!), it is also stiff and - crucially - quite heavy. In its place, I considered using a cable/spring return arrangement, however I rejected that on mechanical and structural grounds. Instead, I would use a closed loop ('push/pull') linkage, like the one employed in the Longshot. This would save at least 25g - 30g, and also offer more precise centring.

The instructions offer a choice of fixed or removable tailplane. However, having opted for a closed-loop linkage, the only option was to fix the tailplane permanently, which was fine as it would save a few more grams. Some people worry about transporting the model with a fixed tail, but it's never been a problem for me (obviously you don't want it sitting under your lead sled!).

The Kit. Note the rod & tube for the tail linkage, all coiled up in the plastic bag - I junked it!

Home brew closed loop system. Note single take off point on servo arm, and turnaround made from ali tube. Cables are secured using brass ferrules pinched with pliers

Planning the closed loop
Unfortunately getting a straight run for the closed loop cables wasn't going to be easy, due to the configuration of the pod and boom. The conversion would require careful planning!

The first step was to make some full size drawings of the pod/boom. These showed that the servos would need to be short, and they would need to be mounted low. I chose Ripmax SD100's (= Dymond D47) as these seem to be the shortest of the popular DLG servos. Even with the servos hard against the bottom of the pod, a couple of cables are still brushing against the boom and pod (at the join), though not enough to cause any problem.

From the photo above you can see the turnaround posts located forward of each servo. These allow both cables to be connected to the same hole on the output arm. This offers a couple of benefits compared with a conventional closed loop setup (where the takeoff points are on opposite sides of the servo arm). First, it avoids a bias load on the servo shaft. Secondly, the cables for the elevator and rudder are on opposite sides of the boom, making it easier to avoid tangles.

Installing the closed loop
Installing the closed loop was extremely fiddly and required a certain amount of perseverance and low cunning. All turnarounds must be perfectly perpendicular to the cable runs otherwise the cables will tend to ride up or down. Everything must be marked out and drilled carefully. Needless to say, the cables are pretty tricky to thread. Patience, a caliper gauge and tweezers are really essential pieces of equipment if you're comtemplating this mod.

Kevlar thread was initially used for the control lines but I found out - the hard way - that kevlar has limited resistance to abrasion (fortunately it was the rudder not the elevator cable which broke!). I've now replaced the kevlar with a multi-filament fishing line which is much more durable in spite of being thinner. 

To pull the cables through the boom I used a very fine steel rod with a tiny hook bent at one end. The rod has to be maneuvered carefully to avoid tangles. The cables are terminated using crimped ferrules (see photo).

Exit for elevator line. A pilot hole was drilled, then opened up with a needle file.

Rear end mods for closed loop. Note turnaround for elevator line, made from carbon tube offcut.

Attachment point for elevator line. Line hooks round micro saw cuts in horn.

Other tips
A few other little mods... the instructions show an infill of epoxy and microballoons to support the wing bolts. I decided to substitute blocks of hard balsa (with the grain vertical), just because I prefer working with wood. Also, I drilled out the holes for the wing bolts oversized, and epoxied in some aluminium tube inserts. These provide better support for the wing bolts under lateral launch stresses. This was done before the centre section was carbon'ed up.

I used a bottom drive for the ailerons, partly out of laziness and partly because it's easier to control the geometry. The resulting linkage is solid and slop free though in retrospect it would have been nice to make the extra effort for an internal (top-drive) linkage.

A word of warning if you're building one of these: don't ignore the sheet of  lite-ply which comes with the kit. At first I thought it was a packing error. In fact, they build into alignment jigs for the tail surfaces. Go to the Topsky web site and you'll find a document which explains their  use. The jigs take all the guesswork out of the lining up the fin and elevator.

Tailplane and rudder aligmment jig

RC Gear
Here's the R/C gear I used:
  • Multiplex Cockpit SX M-LINK transmitter
  • Multiplex RX-5 Light M-LINK receiver
  • Rhino 360 2S lipo battery
  • OpenAltimeter with the optional switching voltage regulator.
  • 2 x Hyperion DS09 (wing servos)
  • 2 x Ripmax SD100 = Dymond D47 (tail servos)

The RX-5 Light receiver employs just a single rf circuit, and does not have the extra amplifier stage of the  'non-Light' receivers. In other words it's pretty basic, and as the model contains a lot of carbon, I was concerned about possible range issues. However, having since specked out the model at La Muela without any problems, I'm pretty confident that the receiver is up to the task. A 2.4-friendly pod is apparently available.

The OpenAltimeter is a splendid little gizmo. Not only does it continuously log your altitude (in its internal  memory), it will also beep the height of your previous launch, and/or your airborne battery voltage, in response to commands from your transmitter. Dead cool and genuinely useful! Mine has the optional voltage regulator since the DS-09 servos are not rated for 2S lipos.

Final word (for now)
Was the closed loop worth all the pain? Yes, definitely - though I should stress that the conversion is not for the faint-hearted or hamfisted. 

Thanks to the lighter linkage, my model doesn't require nose-weight. All up weight is 285g, 5g above the target although I suspect that the 280g recommended maximum applies to the lighter non-disser version. The CG position is currently at 80mm. Some people are running their TS2s at around 85mm so - assuming you're happy with reduced pitch stability - there's further scope for weight reduction e.g. by replacing the OpenAltimeter with a lighter regulator, stripping the receiver case or using a smaller Lipo.

So far I've had about forty launches with the new model and the performance is all I'd hoped. The model hangs on to weak lift like a limpet - it's much better at working low-level thermals than the Longshot. The model responds well to flap inputs. Handling in light conditions is good - the turns are well harmonised. The TS2 is also mercifully free of the porpoising which afflicts my Longshot in anything other than still air.

Any negatives? Well I guess the TS2  is more of a handful in rougher conditions, but then it's more lightly loaded than the Longshot. Also launch technique seems a little more critical than with the Longshot. Finally with full spoiler deployed I'm not getting the turn authority I'm used to with the LS1, but that's more of a tx programming issue.

Longshot v. Topsky
It's worth mentioning that the Longshot has been updated I built mine all those years back. Version 3 features similar tailplane and rudder construction with the Topsky, and has a strengthened pod. I believe that the wing is unchanged however, so performance should be similar to my earlier model (perhaps somebody will correct me). The original Longshot was designed for closed loop, although the newer ones have gone over to tube/cable. My guess is that the Longshot 3 is easier to build, but the Topsky 2 is the better flier.

Related links:


trebroN said...

Hi Mike,

nice and clean build - well documented!

I have one question though, as I do not get your sentence: "However having specked it out at La Muela without any problems, I'm pretty confident". Were did you pop out the antenna? In front, through the nose?

I am basically flying the same setup with my DLGs, this might be interesting to you - a stripped RX-7 only weights about 7 grams ( As my TX antenna is horizontally showing to the left, I let the RX antenna go vertically to the sky (guided through a small pipe (bowden control?)).

As I have a vario in one DLG, I can confirm a superb performance of the M-Link stuff in general:-)

RC Soar said...

Thanks trebron for your comments and link... The antenna on my Topsky pokes out sideways through a hole in the the pod. It would be nice to check the LQI but of course the RX-5 doesn't have telemetry - the stripped RX-7 looks a possible solution though with the OpenAltimeter space would be tight.

Anonymous said...

great post, loved your writing..could get a few more pics of the closed system?

Daniel said...

es imposible conseguir 260-280gr enun topsky 3, tampoco se puede ahorrar 25gr con el sistema pullpul,, 265gr pesa un blaster 3,5 con full tecnology

Unknown said...

Hi Mike,
I really appreciate your effort to create the F3F and F3J templates for Taranis - it makes setting up a model a breeze. I recently help a buddy set up a model on a slope in a very short time and he flew with no problems.
Now, I was wondering if you by chance are working on a similar template for a DLG. I would greatly appreciate if you could produce one, based on your F3J setup. Thanks in advance, Bill.
billbabin at

RC Soar said...

@Bill - Thanks for the feedback. A DLG setup would be fun to do, but I suspect it might be a little while - I'm still using the little Cockpit SX.

Unknown said...

I will be looking forward to it. Thanks in advance!

Unknown said...

Would it also be possible to use pull pull on the rudder and closed loop only on the elevator?
Would be half as complicated, but you would have a small weight gain due to the double control horn..
Great work! I am inspired to consider this option..

RC Soar said...

@Sean: I think pull-pull on the rudder and closed loop on ele is what is already described (or maybe I have misunderstood your terminology).