The North American AT-6 was the result of a series of aircraft developed and supplied to the US Army Air Corps in the early 1930’s for advanced training. Significant modifications to the airframe, including an all-metal fuselage and retracting undercarriage, led to the true AT-6 series. The prototype of which first flew on February 11th 1938. Supplied to the US Navy as the SNJ-1 variant and to the RAF as the Harvard, the production volume resulted in an entirely new factory being built in Dallas, Texas which led to US pilots naming the AT-6 the TEXAN.
Approximately 17000 were built and it is estimated that around a million people learnt their craft on this machine before progressing onto Spitfires and Mustangs.
Less frequently modelled, not particularly elegant on the ground, but with a most pleasant planform once aloft, the AT-6 Texan appealed to Ted as an attractive prospect for his next large model project. The prospect of a 33% scale model with adequate power for an aerobatic flight envelope of prototypical manoeuvres he found irresistible.
Ted’s excellent model on landing approach.
Length: 9ft 8″
Engine: 3W 140 Vegas
Wing loading: 3lb/sq. ft
Power loading: 6lb/hp
Ted’s wanted the model to embody the full-size wing and tailplane aerofoil sections with the objective of simulating the characteristics of the full-size. Performance around the stall would be ‘softened’ by a light wing loading. To replicate the characteristic ‘tail high’ sit of the aircraft in-flight the same incidence angles would be used. This meant a semi-symmetrical wing root aerofoil section NACA 2215 at +2 degrees, and at the wing tip the undercambered NACA 4412 at 0 degrees. The tailplane would use a symmetrical section at +1 degree. No engine thrust offset was to be employed, but the fin/rudder offset used on the full-size would offset engine torque.
Scale three-view drawings, from two apparently authoritative sources, were obtained only to discover there were major discrepancies around the parts of the engine cowling in both profile and key dimensions. The tailplane, elevator and rudder outlines also differed. A visit to the Duxford museum and photographs of a full-size was used to help establish which drawing to adopt.
The initial dimensions of the model were decided upon as:
42 ft 1/4″
28.2 sq ft
29 ft 11 7/8 inch
9 ft 7 3/4 inch
11 ft 8 1/2 inch
3 ft 10 1/2 inch
15 (3W-140 Vegas)
20.3 lb/sq. ft
3 lb/sq. ft
12 ft 11 5/16 inch
4ft 4 inch
The basis of the design was to achieve a wing loading on the 28 sq. ft wing area of only 3lb/sq. ft. This would ensure a docile performance around the stall. Also, a power loading of around 6lb/hp would give a spirited performance of prototypical manoeuvres. From this the estimated dry weight of the model was 86lb (39kg).
Having determined the dimensions it was time to estimate the weight in detail. From some experience and enlightened guesswork the estimates of weights were produced. The table opposite was produced to give the estimated weight of 86lb.
The plan was to achieve a similar flight performance to the full-size. A good rate of climb with the ability to fly around a large loop was wanted. Also, Ted wanted a gentle stall and no tip stalling. It was therefore crucially important to avoid weight build-up behind the centre of gravity and hence reduce the required nose ballast. It can be seen that allowing for the engine, undercarriage, batteries, propeller etc, 40% of the weight was already accounted for.
Weight in lb
3W-140 engine and ignition module with battery
Wing centre section
Outer wing panels
Tailplane and elevators
Fin and rudder
Glass cloth and resin
Cellulose primer and paint
Allowance for nose ballast 10%
The basic structure needed to provide a solid mount for the 3W engine with a 70lb static thrust. Transfer of this thrust through the structure needed thought as did the need to provide a rigid mount for the wing and tailplane. The approach of adopting ‘best practice’ rather than a thorough analysis of the applied loads was chosen. Even with Ted’s own considerable experience, he found the additional experience of LMA members to be a help where ideas were exchanged for the future benefit of all.
Haigh Hall March 2002 with the basic fuselage structure complete.
Ted at Haigh Hall in March 2003 with the main construction completed.
The engine is mounted directly to a birch ply firewall with a doubler behind the fixings holding six captive nuts. The wooden box contains the ignition module and battery and the 12V 2400mAH battery for the landing lights.
A ‘ladder’ type crutch is locked into the firewall and extends the full length of the fuselage to the sternpost. Eight foot of 9 x 18 mm cyparis was used. Side plates, made from 6mm ply reach up to the tailplane mounting tubes (20 gauge, mild steel, 16mm diameter) with a stern post erected for subsequent construction of the fin. Cyparis stringers were used in 8ft lengths. Sheeting was with 3mm balsa.
The 3W engine can turn a 30″ x 10″ prop at a static speed of 6100 rpm. Assuming that propeller slip and rpm gain when airborne negate each other a 10″ pitch will provide a forward speed of approximately 60 mph. A disadvantage of this propeller is that the tip velocity will approach the speed of sound producing a lot of noise at full-throttle which can sound great, but is definitely not recommended for the local flying sites! So with less acceleration at take-off a 28″ x 12″ propeller was a better selection.
The tailplane halves have no spars. They are built from 3mm ribs erected upon the trailing edge. Epoxy glass tubes were bonded into the ribs to accept the mounting tubes. False leading edges locate the top and bottom 3mm balsa sheeting. A 9mm balsa leading edge was added to complete the construction.
The elevators were built-up and fabric covered with ribs at the scale spacings and functional trim tabs installed. Each elevator servo is installed in the bottom of the tailplane half and exposed, although flush to the surface, to allow thorough pre-flight checking. The elevator horns and servo output arms were made to achieve full-size deflections of +30 degrees to -20 degrees. The elevator servo is required to cope with the blowback force produced from a 70 mph airflow. An 11kg-cm servo was selected. The boost tabs (see picture opposite) are connected to assist the servo, thereby reducing current drain and increasing effectiveness. The conclusion of much experimental work elsewhere was that the boost tab is maintained parallel to the airflow throughout the range of the control surface deflection.
The elevators, rudder and ailerons were fabric covered with some rib tapes added.
Tailplane joiner rods.
The wing panels, outboard of the dihedral break, are removable. This enables the option of transporting the fuselage standing on its undercarriage in the centre section.
To be suitable for aerobatics Ted envisaged it was required to withstand +6g. The wing centre section is an I-section beam constructed from 3mm birch ply webbing and 9mm x 12mm cyparis. Secondary spars top and bottom, fully webbed with 3mm ply, were used to support the aft attachment screws and extend the aileron hinge points. Wing skins of 2mm liteply added to the rigidity of the wing panels. Attachment of theouter panels to the centre-section was achieved using stainless steel bolts.
Wing rib profiles and aileron linkages were drawn and cut. Slots were added for the spar and drilled for a carbon fibre torque tube that would operate the flaps.
For the outer wing panels a pair of inner ribs were cut from 6mm liteply to NACA 4412 and its profile polished with an old piece of 400 grade wet/dry paper. The outer ribs (NACA2215) were made from 3mm liteply and also cleaned-up. Two degrees of washout was incorporated. The other ribs were cut from 1/2″ pink foam. The wing was covered in 2mm liteply.
The ailerons were cut from blocks of white foam placed in the aileron’s position on the wing panels. The leading and trailing edges were balsa and the assembly veneered. The end result looked great, but it took seven attempts to cut them correctly! Hinging is a simplification of the ful-size using a frise design. There is 30 degrees movement up and 15 degrees down. Pushrod driven boost tabs are fitted. Hinge pins produced from machine screws lockedinto one component and turning incaptive nuts on the other side.
Flaps were installed on the centre-section and each outer panel. Using 2mm ply liteply sheet and a profile doubler with liteply riblets they were quickly assembled. Four pushrods, equally spaced, prevent blowback. Each flap is driven from a carbon fibre torque tubes and 20kg-cm servo which was deemed sufficient for 40 mph deployment.
After considering the landing load, attachment of the landing gear, distribution of the forces to the fuselage and an appropriate location for the attachment bolts, the wing centre-section was fitted out with the requisite air bottles, sequencing valve, servos/linkages and wiring. Plugs and sockets were taped over to keep them clean.
Working from the 33% drawing of the wing aerofoil, the oleo leg, wheel axel and retraction hinge point were over laid. A simplificationof the units was made concentrating on the accurate wheel placement relative to the airframe and its centre of gravity position. Most of the weight of the model is supported on a stout alloy angle bracket. Backward retarding forces which would tend to twist the main spar do so via a relatively light Z-shaped brackets which act as weak links.The wheels have liteply hubcaps and are 10″ diameter trolley wheels with pneumatic tyres. The tread was removed leaving a final diameter of 9 1/2″. The undercarriage system can be pressurised to 100psi and allows for several cycles of the system with the legs being locked at each end.
The oleo legs have internal springs taking up to +3g per leg before they are fully compressed.
Making the wheel root wells moulding was a serious threat to Ted’s sanity. Three view drawings gave sufficient detail of the compound curves and dimensions. Eventually, some foam blocks were satisfactorily completed and two layers of close weave cloth applied. The profile of the wheel doors were also marked out onto liteply.
The tailwheel was built from the Ted’s scrap box using a plywood trailing arm, an M5 caphead bolt with a gaint servo arm attached for the closed-loop cables. A 5mm collet acted as a top bearing, an M5 captive nut for the bottom bearing and a suspension unit from an oleo off an old 1/6th scale spitfire. A 100 mm lightweight wheel was used and all this is driven by a 3kg-cm servo y-leaded to the rudder servo.
The canopy is removable for access to the R/C equipment, petrol tank, wingbolts etc. It is a major feature of the model. None of the panels slide to make it easier to protect the inside from rain.
Formers in 6mm liteply were used from the firewall to the aft end of the canopy. These were fretted out to allow for the cockpit detail. A liteply and balsa frame was fitted to the fuselage opening and canopy frames laminated from 0.8mm ply.
Three main sections of 0.5mm clear sheet overlay each other and are glued to a laminated wooden frame structure. The windscreen is a single curvature and readily fitted whereas the aft cockpit was a nightmare to get right. The small side panels are flat, but the moulding was curved. The mould was made from a balsa block and each piece of clear sheet was hot moulded helped with flat ply positioned on the top of the plug and pulled down using elastic. Careful use of a heat gun enabled the shape to be slowly formed and the elastic kept pressure on the mould. Finally, they were cut out and glued in the frame with contact adhesive.
The near circular cowl was made from liteply rolled into a cylinder, fitted to a foam block which was cut into a ring and using a profile template to check, shaped and sanded . It only took about three hours to complete. One layer of glass cloth was added over the entire structure. Alloy brackets enable it to be attached to the firewall.
This was made by Mike Jackson and comprised dual expansion boxes feeding a third box exhausting in the scale position. Cut outs between the cylinders of the dummy radial allow some cooling air to the muffler parts while close fitting cut-outs for each cylinder of the 140cc engine allows a blast of cooling airflow through its fins. The outlet is three times the input area.
A duel receiver system, is a requirement for the model as it was over 20kg. The assignment of the channels can be seen in the table below.
A six volt system was used with two 600mAh packs acting as the supply and backer for the receivers with two 2000mAh packs for the servos. The receiver batteries are installed adjacent to the receivers in the central cockpit area. Servo batteries are installed in a foam lined box immediately behind the firewall. Opto-isolators were also incorporated to prevent any interference affecting the system which the long leads could be vulnerable to picking-up.
The engine ignition module and its battery, with a 12v 2000mAh pack for the landing lights (2x 12v 50W halogen bulbs), are installed in a foam lined box and bolted to the engine mount.
The receiver aerials were doubled in length and a 22pF capacitor added at the join to aid noise suppression.
Finishing and Painting
Ted could not find a museum exhibit or a photofile of a Texan in US Navy SNJ colours which was natural with bands, cheques and markings that was his first preference. Instead a magazine was loaned him that featured a AT-6 in US Navy training colours which had been refurbished and was operating out of Shoreham in the UK. The owner/operator kindly assisted Ted with some photographs and a summary of the paint colour. Ted decided not to add surface detail, but to aim for a high gloss finish and let the outline and contours be the principal static appeal of the model.
Light weight filler was used on the balsa sheeting to prepare it for glass cloth and resin. As per standard practice, an initial thin coat of resin was used to secure the cloth with a second coat applied to fill the weave and achieve a smooth finish. An orbital sander with wet and dry paper finished off the surface. Motor factor primer was used with substantial rubbing down between coats. Three sprayed coats of cellulose gloss were applied and rubbed down between each coat. This resulted in the gloss finish Ted wanted. The wing and tailplane was treated in a similar fashion.
The US stars and bars and other lettering were then sprayed using masks from low-tack masking tape and newsprint.
Finally, to achieve the centre of gravity required, 4lb of lead was added to the firewall.
After the model had had its final inspection by the LMA examiner and the ‘Exemption to Test Fly’ received the first visit to the flying field was made to set everything up and check. Flap linkages needed to be adjusted, transmitter ATV’s set, aileron and elevator linkages centered, dual rates set-up for the rudder and exponential applied to the elevator to name a few things that needed to be completed.
An air leak was identified in the retract pneumatics, but could not be found. The legs could be locked down though and so engine testing and fail-safe operation could be set. A check of any loosening of fixings due to vibration was also made.
The ground range of the radio was explored, with and without the engine running. 150m could be obtained with only one section of the transmitter aerial extended.
Some ground-hops were made to increase confidence in the position of the centre of gravity. Transfer from the tailwheel for steering to the rudder operating with the tail up was not satisfactory and less tailwheel movement had to be set. A bit of air under the wheels before throttling back enabled an initial feel of the elevator sensitivity to be made.
A few more fast taxis and the first real problem occurred when a weld failed on one of the legs and a wheel came off. No major damage was done as the model dropped onto its wingtip.
So that was that, time to get back in the workshop correct the things found to need attention and then back for its first real flight.
Remembering not to haul the model off the ground in less than 100m, power was progressively applied and the main wheels eventually left the ground. On its fine pitched prop it began a shallow climb, then the first problem…… only one wheel retracted. So just the one circuit, both wheels were lowered, but at least some initial testing could be made. A few fast and slow circuits were made in a five minute fight. The first landing was made without flap and fairly fast onto the main wheels.
The wheel that had struck down was tested and seemed to be OK. So off again. All was well this time and the undercarriage worked fine. Climbing to a safe height the stall proved to be a non-event with no wing drop. Release the elevator and the nose dropped into a shallow dive, throttle-on and resume flying. A wingover at full power demonstrated that the model had a good rate of climb and then a loop with the Texan floating over the top. This took 8 to 10 seconds and was very satisfying. Little rudder correction was needed, the scale offset of the rudder proving to be just right. Now for some figures of eight and then a roll. Just a gentle one with a bit of rudder and elevator as expected.
Landing this time was with flap and another touchdown on the main wheels performed. After the flight a good inspection of the linkages servo mountings, hinges etc was made before the rest of the testing was continued.
Since then the model has completed its test flying program and was displayed at several shows in its first season in 2003. There have been some undercarriage problems, especially if having to make cross-wind landings, but the flying is superb and gives a fine scale performance, just what Ted wanted.
Mike Jackson – Muffler design and construction
Alan Dew – landing gear
Len Gardiner – machined components
Steve Holland – wheels
Stuart Solomon – Liteply and Cyprias
Ted at Longhorsley in 2003
Shop online with the LMA
For event tickets, merchandise and more visit our online shop.