Thread Tools
This thread is privately moderated by BAF23, who may elect to delete unwanted replies.
Sep 09, 2015, 05:38 PM
The sky is the limit
BAF23's Avatar
Thread OP
Build Log

FMS Spitfire TR9 two-seater mod

June 2011 I purchased a secondhand 1400mm FMS Spitfire from a club member who had only flown it once and found it much too challenging for him. I wasn’t a hero either at that time, but I definitely didn’t want to miss this opportunity because on the scrapheap of the clubhouse I had seen an almost complete but total loss identical model for takes. Having performed aerobatics in the back seat of the real PV202 (G-TRIX) Spitfire two-seater during the Oostmalle warbird meet of 1993, I wanted to fly that as a model as well.

A flight by an RC shop owner confirmed it was very unstable in pitch and dropped a wing during the flare. Inspecting the model I noted the elevator to be anything but rigid. The foam couldn’t handle the torsion caused by the control horn only on the starboard side, and the aerodynamically correct outboard balance sections twisted much too easily, thus exaggerating any elevator input and making the model very twitchy in pitch. I resolved the problem by cutting off those outboard sections from the elevator and gluing them with some carbon rods to the fixed horizontal stabilizer surfaces. It thus didn’t alter the overall shape, but drastically reduced the torsional forces on and between both elevator halves.

Because the model continuously ended up on its nose, I also increased the forward rake of the retractable landing gear. I unscrewed the gear from its mount and added triplex plates under the aft attachment screws. By doing that the gear still pivots into the original wheel wells with only the gear-door plates at a slight angle instead of completely flush with the wing, but seriously moving the wheels forward as per real Spitfire pictures. The assembly also gains some lateral strength in the process.

The most striking change is the second canopy installation. The rare Spitfire post-war two-seat conversions for the Irish Air Corps (of wartime Mk9 types) came in two varieties, either a second cockpit streamlined behind the first one, or an elevated second canopy providing the instructor a better forward view. When I flew her, PV202 still had the low canopy, but after a fatal accident around the turn of the century she was rebuilt and still flies with a raised second canopy, much spoiling the elegant lines of the Spitfire. With an intact spare canopy at hand it was easy to choose for the early variant. FMS uses a removable front canopy on a foam frame, with a lip at the front to keep it (and the battery) firmly in place, and a magnet at the back to keep it shut in flight. With the balance in mind I decided to keep the back cockpit as light as possible, except that I wanted a bust of me under that hood (I couldn’t care less about “Goofer” in the front seat because nobody would pay attention to him on the model). After cutting some foam behind the original canopy assembly, I cut off the clear parts forward and aft of the spare canopy. The back part including the bowed foam bottom was then turned around and glued to that forward canopy bow, forming half of the clear part between both canopies. Using solvents to remove the paint on the canopy bows made the junction even more conspicuous, and by having both canopy bows on top of each other there ensures a tight fit of the assembly in flight. As usual I glued a carbon transverse rod which extends a bit out of the removable back of front canopy as to facilitate lifting it from the magnet.

After adding some nose weight to compensate, the model including a 4S3200 35c battery still only weighed 1600gr. With the stock 3648 kv520 engine and the 13x9 tri-blade prop and a 50Amp ESC with Bec, it still had ample power for scale flying. After changing clubs I started flying it on the tarmac runway but still didn’t like the stability. 2012 was the time Hobby King came out with a very cheap 3-axis gyro system that simply could be plugged in between my Spektrum AR500 receiver and the servos. This vastly improved the stability of the model but reduced the maneuverability. Most RC pilots wouldn’t like the trade-off, but it is perfect for me because the model now flies like a real Spitfire during airshows as opposed to unrealistic typical RC corners. Those mark1 gyros were very difficult to adjust and couldn’t be switched off in the air, sometimes causing serious walloping during the early adjustment flights. Once the pots had been turned to a correct value, it made flying a pleasure, but takeoff and landing still required much attention and work. You can see my Spit in action as from minute 3:15 during shots taken early 2014 amidst busy aerotow operations at our model field. The FMS B25 was also mine and was flown without gyro, but has been sold and replaced by a larger Hangar9 version.

FMS B25 and Spitfire model flying at EBZW (7 min 48 sec)

I flew it occasionally during the next years (avoiding crosswind operation) and thought of selling it during the summer of 2015 when I switched from Spektrum to Taranis/FrSky electronics. Having the transmitter DM9 module allowed me to fly it on Spektrum rx, but I got increasingly erratic uncommanded inputs in the air, accompanied by unsteady AR500 Rx light. I was unable to pinpoint the cause, but it was certainly not the transmitter. Having sold the FMS B25 I decided to keep the Spitfire and swap the onboard electronics for a FrSky X6R receiver (thus eliminating the need for the DM9 module on the Tx), and substituting an on/off switchable Orange MK2 gyro for the probably faulty Mk1.

For the first trial flight in that setup in August 2015, I asked another qualified club member to check with me on the correct control deflections, both in manual or in gyro response mode, because it is easy to make mistakes during the setup and adjustments. Everything worked well, he could see nothing wrong in the control responses, so I taxied out and positioned myself behind the model for takeoff because I had dialed in a throttle to rudder mix like on the Stampe. If those real taildraggers need almost full rudder during takeoff, why shouldn’t we program some in the TX to facilitate directional control on the ground? I decided to make the takeoff without the gyro, only switching it on in the air after the airplane had been trimmed on the new electronics. Testflying will reveal how best to takeoff in the long run: without gyro but with throttle to rudder mix, or with gyro and in that case without any throttle to rudder mix, but probably with some aileron to rudder mix to facilitate coordinated turns and help combat tipstalls at lower speeds.

To my surprise, it tracked completely straight during the first takeoff roll, even in 2-point attitude, but at about the time of liftoff I saw the tail quickly rise and decided to apply back pressure to get it airborne before the prop hit the tarmac. I needed almost full up to keep it in a shallow climb and couldn’t believe I had made such serious setup mistake. I initiated a shallow climb to downwind for a short pattern and elected to keep the gear down. In downwind I found it increasingly difficult to control it in pitch, the slightest command caused vigorous pitch changes and I had doubts about my ability to land it under those conditions. I already turned final when the aircraft was abeam the threshold, but glide slope control was nearly impossible and I even didn’t attempt to land it on the tarmac, but aimed for the grass alongside that stretched way beyond the tarmac. Once I found an acceptable glide, I didn’t touch anything anymore and let it descend without further touching the controls. Being 150meters behind the airplane I couldn’t judge the speed anymore and when it as only about half a meter from the ground, it tip stalled the right wing and ended up on its nose.

As I approached the model with its tail in the air, it was obvious something didn’t look right. One elevator pointed up, the other down, which normally was impossible because they are physically linked and operated by a single servo. The foam link was still there, albeit with a serious twist, but the FMS thinner foam portion used for control surface hinges had completely severed on the port (slave) side. The starboard elevator still worked perfectly and dragged the left one along by the thin foam junction strip. The trailing port elevator thus ended up flapping at a free angle above or below the horizontal stabilizer depending on up or down starboard elevator inputs. No wonder I had very little pitch authority. My club buddy confirmed the so called hinge still looked intact after the checks and during the taxi. Closer inspection against the sunlight showed the starboard elevator hinge and aileron hinges still intact, but the rudder foam hinge already showing interior cracks. The rest of the Spitfire seemed unaffected, except that the starboard wing had moved about 1-1/2cm out of the fuselage junction, probably during impact.

Back home I concluded that I just had been lucky during the previous years, and that the mishap occurred at slow speed. I still hadn’t tried out the receiver/gyro compatibility and definitely wanted to do that before I converted my DC3 and T6 to FrSky receivers with Orange gyros. The cheap Spitfire being half-disposable was the ideal trial airplane for those proving flights so I decided to repair it to get at least that job done, and maybe stay in the fleet because of its unique character. The time was ripe for an overhaul of the four year old foamie, so I installed it on the Robart stand and got a fresh scalpel out to start the surgery.

Those FMS foam “hinges” are about 3mm wide at their thinnest thickness, but when that was cut away and replaced by proper hinge mechanisms, that 3mm gap could not be eliminated without altering the smooth circumference of the elliptical wing and tail outline. This is not my preferred way of finishing control surfaces, but that will do for this foamie. All the hinges on the model got the treatment because I didn’t trust the factory solution anymore. During the setting up of the controls, I intentionally adjusted both ailerons to a mild reflex angle (less camber), hopefully curing the temperamental tip stall occurrences. The FMS “operating manual” is a joke and doesn’t show anything clearly, nor gives any adjustment values or part numbers. Having acquired the model already assembled, I had no clue as to how the wings held together, or how the tail had to be taken apart to remove the elevator. Luckily RC Groups had a 100+ page thread on this model with many good pictures of the individual parts and practical adjustment figures. The wings only being held together by a flimsy screw tightening a nylon strap around the hollow wing spar, I decided to use the infamous FMS glue to glue the wings to the fuselage on the bottom for safety.

The same glue must have been used by the guy who assembled the vertical tail into the fuselage, and that couldn’t be pulled apart anymore even after unscrewing the mechanical link (long screw from the bottom of the fuselage). The only practical solution thus was to first cut away the rudder at the hinge, then cut away the small block behind the elevator center foam junction. After cutting the remaining elevator hinge, I then was able to extract the elevator assembly from the model. It now became obvious the foam between both elevator halves had endured serious torture and wasn’t rigid enough anymore. I epoxied two carbon strips, one in front and one behind the foam link and extending a few cm into the elevator foam to reestablish torsional rigidity, but during the drying process the twist in the foam caused my elevator halves not to be perfectly aligned anymore so I’ll have to watch how it affects roll with speed. The picture below shows the surfaces joined by proper hinges but with the necessary gap.

Whilst on the tail I also made a cut through the fuselage top to access the erratic tailwheel steering mechanism and servo. It was obvious that the plastic gear teeth were the culprit so I took the (glued) servo out and replaced it with a metal gear one that was compatible with the unique long slotted steering arm of the model. When looking in the cockpit area I still saw vast empty areas and having 100 Amp capable power sensors prepared with a variometer signal adapter and twin EC5 plugs already soldered in a compact unit, decided to install one until I has the hard figures about power consumption for my timer setup. I reprogrammed my Taranis transmitter to get the appropriate voice call-outs of consumed power, position of the landing gear switch, and if the gyro was active or not. With all the loose equipment under the canopy I was afraid some might hinder the movement of the elevator and rudder servo arms, so I added a transparent plastic plate to keep them separated from the rat nest.

I never had been happy with the excess movement of the engine caused by the rather thin firewall plastic plate just being squeezed between a slit in the foam fuselage. Over time and probably following a fair amount of prop hits on the runway, the situation had deteriorated and this general overhaul was the time to get that tighter. To get there the plastic spinner had to come off, then the aluminum spinner removed so the prop could slide over the axle, then the plastic spinner’s base plate removed, and the forward plastic fairing between spinner and fuselage foam. It then became possible to unscrew the aluminum engine mount from the firewall with a long screwdriver to which I let a piece of rare earth magnet hold on so the screws would come out with the screwdriver instead of falling somewhere in the cavern. That simple trick already saved me much time and spared me from frustrations during previous builds. I then hung the model so the nose pointed straight up. That position allowed me to let expanding PU glue flow by gravity through a straw which I sleeved over the glue applicator. The coca cola type straw having a flexible part, gave me the possibility with a 45 bend to guide the glue all around the firewall base-plate so the glue could expand in the foam slit. After a few hours the assembly was solid as a rock and the engine could resume its place. It then became possible to check how straight the prop holder was after the impact on the nose. The deviation was about half a millimeter and thus acceptable for flying until a spare would arrive.
Here you can see a picture of all the nose parts, including the extra lead blocked by foam blocks, old and new props, and the plastic parts that make the shape and keep the prop in place.

The prop had had its best time (partial nose-overs), but I had a spare one. Putting that FMS 13x9 tri-blade on the balancer revealed a serious imbalance which I corrected by sanding thickness off the end of the heavy blade, and using a much thicker yellow paint layer on the tips of both light blades. Until the model has been fine-tuned during the following trial flights, I installed the old prop and put the model inverted on my Sig balancer to see where the balance was. On the RCG thread I had read that most used a 85mm from leading edge CG instead of the FMS suggested 100mm. To obtain that I modified my old lead compartment under the nose and found out I had to stuff 160gr extra lead to balance the Spitfire, even using a heavier 4S4000 battery. That CG point comes dangerously close to the gear position (even with the modified forward rake) and will render the model even more prone to nosing over. If it improves stability in the air I’ll keep it that way because I only operate the Spit from a smooth Tarmac runway. With a more forward CG I was a bit afraid of running out of (the reduced seize) up elevator, which previously had reduced throws for not being overly sensitive with the aft CG . To be on the safe side I increased that throw to the original value, by altering the positioning of the pushrod on the servo arm. With everything in place and balanced, the ready to fly weight became 2280 grams. A rough estimate of the wing area (how do you calculate that exactly on an elliptical wing?) ended up around 34 square decimeters, resulting in a wing load of 67gr/dm2, identical to the one of my Svenson Stampe and slightly less than my Art Tech T6. On this picture you can see the new more forward CG mark, the wings having been hot-glued to the fuselage (only at the bottom), and the forward tilted gear assemblies that cause the gear doors to rest at an angle against the front of the wings.

The trial flight after all these mods was a bit of a rush. National television reporters were on the field and took special attention to us, the woman flying in the club. Whilst they were busy interviewing my friend preparing her Su37 for her maiden flight, I thought I could safely use the runway to get the Spit in the air for a feel of the results of the changes. The camerawoman noticed me taxiing out and promptly abandoned the static shoot to follow me to the runway. I told her this was a test flight but she initially didn’t care about the airplane and spent most of the time just filming my face, she must have noted my body contortions and grimaces. I always fly like that, not because of the camera but because I often allow my body to make some natural moves when having to control tricky flight phases. I tried to ignore the camera but that was impossible for the landing because she took the head-on shots of me from a position in between the airplane and me.

I started without gyro and although takeoff tracking was ok, I again had to react very quickly when I saw the model abruptly wanting to nose forward halfway the run. I analyzed that as too much thrust during transition between two and three point attitude with insufficient airflow over the elevator to compensate the down pitch with the wheels so close to the CG. The solution will be to keep the tail down longer, but not as long as to have the airplane fly off in a 3-pont, risking a tip stall before it attains controllable speeds. That phenomena is certainly due to the minimal horizontal stabilizer area relating to the large wing. Once in the air it didn’t need much trimming so at altitude I engaged the gyro and gradually built the speed up. My meticulous ground adjustments paid off, the pot settings for all three axis were perfect without any oscillations. A few loops, rolls and stalls confirmed the gyro settings were perfect. Without cameras I would have switched the gyro off and performed CG dive tests and minimum speed handling to confirm respectively the balance and results of the aileron reflex settings, but I noted her turning around (everything went smooth so no funny body nor facial expressions and thus not interesting to get on tape) and beginning to film the plane. I came down from my safe altitude and made some flyby’s, but she asked me to fly some lower passes which were beyond the scope of this test flight, but I obliged and made 3 passes and felt both happy and confident about the changes I had made to the Spit. The only way for me to get off the hook was to announce I would try to make some approaches and if aligned, land out of it.

I told her I would come in for a pass lowering the gear for a visual check, then flew to downwind for a first final turn. That seemed very docile and I made a go-around keeping the gear down. In the meantime she must have noted I became more tense and completely abandoned the Spit and pointed her camera on my face again. Next approach being steady, I 2-point landed the Spit like I never did before, with a steady roll-out without noticeable roll tendency on the narrow track undercarriage. While I told her she better had filmed the airplane, she asked me to make another landing so she could film that. With weak crosswind I wanted to takeoff the other way around (to save on battery power) but apparently they always want the same clothes, movement direction, voices etc for ease of cutting and assembling scenes. By that time her sound man also had joined (I had been wired with a personal mike in the beginning of the day) and they were already formulating all kind of scenes, but I took control and clearly told them that because the Spit still had to be flight tested more, and the low battery power level, I would just takeoff and after a short pattern land to a full stop and that was it, they better get it right from the first time. I didn’t experience the nose-over problem anymore (gyro on probably helped), but although the 2-point touchdown was smooth, during the last part of the rollout I got some roll oscillations without ever losing directional control.

Lessons learned:
-I could handle a test flight without too much influence from disturbing external factors
-The midlife update I did on the FMS Spit had eliminated the shortcomings of the ARF model
-Selling the model is not an option anymore, it finally responds without vices to my inputs.
-FMS makes visually appealing models for relatively cheap prices, but they often need essential personal attention and tweaking before becoming completely docile and reliable to operate on a regular basis, and a thorough inspection of hinges and engine mount play after every flight is essential.
-Like many other (A)RTF models, they are excellent value for money, but flying them with factory CG and without (cheaply) fixing production driven shortcomings, will sooner or later bring you into trouble. It is obvious that most (pseudo)scale ARF models are marketed after having been test flown by 3D model pilots who have no clue about the longitudinal stability requirements of (model) airplanes. I’ve had so many factory stock models where the recommended CG was so far off, that I now calculate a 25% MAC CG position and balance it as such before the maiden, regardless of what the manual suggests. Even RCG forum figures are often unreliable, because a lot of those pilots like to fly aft CG positions so they can fly their typical RCmodel maneuvers (much beyond the maneuvers the real aircraft was capable of) but never have flown full scale aircraft and not aware of how the real thing flies and reacts. That might be ok for them, I don’t care if they fly loops and inverted with Boeing737 or B25’s, but if you buy a semi scale model, it should be flown as such and behave as such. There are sufficient other kits on the market (SBach, Pitts, Xtra etc) on which the CG can be further back to allow them to fly their “crazy capitalistic maneuvers”, but the factory recommended CG of my Spit (and elevator setup) didn’t even allow to have this model flown in a realistic way and made it so tricky to fly that this charismatic model is unpopular and hardly sold and seen anymore. Test flying for real aircraft is no Sunday fun job, the producers of models should realize that hiring pro’s to get the glitches out of their prototypes and manuals before putting them into production could substantially increase their production and profit figures . Freewing models started going that route since 2014, also replacing cheap hinges and components by sturdier ones, and model adepts noticed and are responding by paying that extra money for the extra quality. Just one more desire, the option to buy white (uncovered) models so we can decorate them ourselves.

Sign up now
to remove ads between posts
Apr 22, 2016, 11:34 AM
Build to fly, fly to fix


Small world! The father of a very good friend of mine from our RC Club had 90+ combat sorties in PV202 in WWII, before transferring to Tempest V's.

Quick Reply
Thread Tools