Blueprints were introduced in the early 1840’s to make the replication of drawings somewhat easier. The process used a ferric compound which was impregnated into paper. The process was normally that the original drawing would be made on Cartridge Paper. This was a very heavy paper made especially for rifle cartridges in the days before brass replaced them.
The drawing would then be traced onto a tracing paper with black Indian Ink. This would be placed on top of the treated paper and sandwiched between glass plates. The plates would then be paced in the sun for around 2 minutes (allow 10 minutes for a cloudy day…) and the copier could see the exposed paper turn blue (Prussian blue). When the desired depth of tint had been achieved the copy would be removed and washed. The areas underneath the ink would have remained white, the washing process would remove any remaining ferric compound and preserve the drawing. This would make any changes difficult to conceal and thus the “blueprint” of any plan would form the basis of construction/manufacture and seal whatever contract had been struck.
In order to celebrate the completion of the basic design, we have created a “Blueprint” of the Heritage Flight Simulator in AO size. Printed, this will create a most attractive poster. It’s a fairly large file of 16Mb. Feel welcome to head over to our shop and purchase it…for free!
Happy to announce that all the controls, other than the flying controls (Pitch/Roll/Yaw) have been checked and redone where necessary. This in order to ensure that they operate correctly with DCS-BIOS and to increase their robustness.
The Rudder Trim falls in the latter category. The original Spitfire rudder trim system had one turn from end to end, whereas the elevator trim had four turns. The easy way then was to directly connect a potentiometer to the knob. The risk here however is that the potentiometer can easily be damaged in the heat of battle through over-enthusiastic application. The updated design now works through a gear with a positive stop.
New Rudder Trim Base
Rear of Rudder Trim
Geared mechanism
Another component which has been updated is the Chassis Control Mechanism (undercarriage lever).
We are replicating the somewhat idiosyncratic pneumatic action of the undercarriage as accurately as possible by mechanical means. In order to raise the gear, the pilot was required to “in one deliberate movement”, pull the lever down and to the left to remove it from the gate and without pausing, move it up to the top of the movement. Here he had to hold it until the pnuematics took over and plonked the lever across to the right and back into the gate. We achieve this through a set of clever (we hope, still to be proven!) springs, scissor press-plates, traps and triggers. Thus the pilot will pull the lever down which depresses the press-plate, move it across to the left which sets the trap, move it up then hold it for a few seconds and then release the lever. This lets the lever jump to the right which in turn triggers the trap which pushes the lever back into the gate.
Thats all very interesting but how does that affect this particular update? Well, for the undercarriage action DCS-BIOS will accept a toggle switch, pushbutton or rotary encoder. The length of time that the action of working the lever takes, say 7 seconds or so, would require immediate triggering of the action in the simulator when you start moving the lever down out of the gate. It was not possible to achieve this with two microswitches, I tried in a number of combinations, normally open, normally closed, parallel, series etc. It would only signal at the end of the cycle, which was 7 seconds too late. So I then devised a geared system for a rotary encoder, attached to the shaft. The clever bit here was to allow initial movement in a reverse direction without triggering the action incorrectly, ie. when you move the lever down out of the gate you don’t want the sim to think you are moving the gear lever down. So the wheel attached to the shaft has a slot which allows free movement for the extremities and only moves the gear on the central 60 degrees of the movement. This solves the problem and gets the gear moving in the sim at the appropriate moment.
Ghosted image of the gear indicator mechanism
The gear indicator mechanism in place
The final update was to the Remote Radio Control unit. Here the internals were completed and the whole front face now incorporates the engravings and switch retaining mechanism. There are three switch units:
Power on/off
4 interlinked latching switches for channel selection
a 3-way switch to select radio mode
View of Radio Remote internals
What remains now on the controls is to redesign the main flight controls. They currently work directly against spring tension, which is not the most elegant way of doing things and provides poor feedback and centering. I wish to change this to a cam action, potentially with in-flight adjustable spring tension to provide force-feedback.
Melexis is a well known and reputable Belgian semiconductor manufacturer. When they recently announced that they were bringing out a new Tri-Axis magnetic sensor in analog format with standard, through-the-hole wiring, we believed this would provide an excellent consumer oriented product. At a reasonable cost, it offered great accuracy and a very compact footprint. It was the new MLX90371GVS-BCC-100-SP-ND.
It was therefore with great excitement and anticipation that we ordered twelve of these units from Digi-Key. We changed the designs of the Gunsight Base and Range settings to incorporate the new sensor and the Elevator Trim was also designed on this principle.
And then we tested it. Solid output of 2.48V, no matter how we moved the magnet. Refer to the oscilloscope diagram in the featured image above to see it flatline. Well it took many emails and to-ing and fro-ing between Digi-Key and Melexis before we obtained an answer. The units are pre-programmed to have both the lower limit and upper limit of the analog output at 50%! This effectively makes them useless unless you also happen to own a Melexis programming and daughterboard, available at the small sum of around $2000 🙁
I am waiting for an answer from Melexis why they have programmed them in this way, effectively making them unusable for the general consumers, hobbyists and makers. And why there is no explicit statement on this in their literature. I have yet to receive an answer.
Anyway, one has to move on from these little setbacks. We have updated the designs for the Gunsight rings and Elevator Trim with mini potentiometers. For the more critical controls of roll/pitch/yaw, we will be using the Bi-Tech 6127 Rotary Hall Effect Sensors. These come pre-programmed in a selection of operating angles which is great. Many thanks to Sokol1 on the SimHQ forums for pointing this one out. It is simply a drop-in replacement for potentiometers which makes installation and link-up extremely simple.
Some images of the changes that had to be made on the Elevator Trim design:
Elevator trim redesigned for a mini potentiometerThe elevator trim as it was with the Melexis MLX90371 sensor
Just as a bit of a progress update, we have redesigned the elevator trim to provide 4 turns, utilising the same spiral principal as per the original.
In the original, the smaller diameter section of the base contained a cog in order to move the chain which in turn pulled the elevator trim cables to and fro. This section is empty in our implementation.
The interesting bit is the larger diameter section of the base. The original had a spiral with a follower which moved the cable to the indicator on the instrument panel, thus providing a mechanical means of showing the elevator trim position. The follower also limits the movement at either end of the spiral, resulting in 4 turns without being able to stress the elevator trim system beyond its limits.
We have taken this principal but instead of the follower moving a cable, in our case it rotates a spur gear which has a magnet attached to it. The movement of the magnet then gets interpreted by a rotary hall sensor mounted on the rear base cover.
Spiral, follower and spur gear in the elevator trim baseThe Hall sensor mounted in the base coverAssembled elevator trim unit
The panel design has been finalised utilising our now established control philosophy. The changes were as follows:
Navigation light and Gunsight light switches updated to the Air Ministry replicas. These utilise NNK rocker switches with 3D printed base plates and toggles.
Undercarriage Indicator redesigned. It now incorporates coloured translucent inserts behind which LED lights may be mounted and the central hook now disguises a microswitch which in-sim will toggle the day/night screen. This was a little screen which could be pulled down over the indicator lights at night in order to dim them.
Magneto switches changed to more substantial 12mm panel mounts.
The Starter and Booster button covers now incorporate hall effect switching sensors which tell the sim whether they are open or closed.
The panel is now ready for manufacture. Something we will start on once all the design refinements have been completed.
I received the following questions today from Jan Svoboda and my email response failed to send for some reason. So Jan, I hope you get this and I hope you don’t mind me responding publicly as I am sure there are others with the same question.
Dear Jan,
Many thanks for your interest in our venture. In answer to your questions the following:
Do you have any timeline when the simulator parts will be done?
Our planning is to complete the designs and prototype build by the end of the year.
Will you supply leather parts, canopy and other parts in the shop?
No, we don’t plan on that at the moment but this will depend on what we find during the building of the prototype.
We intend providing full plans, including construction methods, for all of the build.
However, where there are more difficult components to the build, as for instance the Malcolm Hood (bubble canopy) could be, we will consider keeping a number in stock. Bear in mind though that the shipping cost for such a bulky item will be high, therefore we will endeavor to find an easy way to make these for builders.
Leather patterns will be made available. These can be cut and stitched easily by your local leatherworkers. The same will apply to laserjet cutting of aluminium and CNC routing of plywood. Most major centres in the world have these services available and shipping things which can easily be done locally does not make financial sense.
3D Printed SLS Nylon components will be available from our Shapeways shop. (We do not make the shape files available for these for self-printing)
And I would like to know if it will be possible to use simulator also for commercial purposes?
The purchase of a plans set will license the builder to build one example of the simulator cockpit. This may be used privately or commercially, for instance by museums or training organisations. The software that forms the basis of the simulator, for instance DCS World’s Spitfire Mk.IX, will be subject to their licence agreements and the builder/operator will need to make their own arrangements in that regard, although we are always available and happy to provide assistance or advice.
Let me know if that that helps and feel free to ask more questions. Be sure to follow our progress on our web page and we also always appreciate any “LIKE”’s we get on our Facebook page 😊
We are working through the painstaking process of confirming the correctness of our controls. We have tested all the possible permutations of selected control sensors with Arduinos, DCS-BIOS and the DCS World Spitfire Mk.IX. We are able to confirm that the result is fantastic.
Composite control view
Now we are back-designing the instruments to incorporate Tri-Axis Hall Effect Sensors in the place of potentiometers. Not only are these much more accurate and steady, they are also very compact and have a significantly lower cost to good grade potentiometers.
Here is a full listing of the available controls in our Spitfire. All the controls are fully functional.
We have a design which replicates in exact detail the cockpit of the Supermarine Spitfire Mk.IX. We have 75 controls which can be wiggled, poked and jiggled.
How do we connect this into the virtual world?! We need something that translates our physical actions and presents it to our simulation in a manner that it can understand.
We need an Interface.
Firstly, it needs to be clear that our product is being designed in such a way that all wires from switches, encoders, sensors and potentiometers are easily accessible on wire terminal blocks, and can be connected to any interface the builder prefers to use.
That said, the last few weeks we have been doing intensive research on how we will be implementing our own design while also getting to understand other available options.
The interface consists of the necessary hardware to gather and interpret the data and then the software to present that data to the simulator in a manner it can understand.
Hardware
Those of you familiar with the Maker Culture will have come across the term Arduino. It is a wonderful open source microcontroller that has made man-machine interaction so much more accessable through its low cost and simple interface.
The Arduino was a simple choice if it could be shown that it could handle the 75 control inputs required and that it could do so simply. We knew that the Arduino Mega has 54 digital and 16 analog inputs, but it does not simply show up as a joystick controller when plugged in to a PC, so what would it take to make it talk to the simulator?
Another option would be a ready made solution which operated simply as a plug-and-play joystick. Here the most well known and best supported option would be the LeoBodnar BU0836X with 32 digital and 8 analogue inputs. Other commercial cards offering this functionality include the GP-WIZ40, DSD Designs with a 64 button controller and the Brydling B256A13 joystick controller which offered an array of 256 buttons and 13 axes. However the current support and availability of these are unclear.
The plug-and-play option has the advantage that the Spitfire simulator can be used in conjunction with any number of simulation software programs which portray the Spitfire in one of its incarnations. For VR enabled versions these include IL2 Stalingrad, MS Flight Simulator/Prepar3d and X-Plane. However, none of the versions in the sims mentioned have the fidelity or depth of systems found in the DCS World Spitfire. The P-n-P option also requires quite a bit of controller setup when used with any of these sims, including DSC World.
Software
When working with the Arduino it is no longer a simple case of plugging in your joystick. It requires a program which does the interpretation of the control inputs into a form that the simulator can understand.
Many programs have been created to achieve just that. Some of these communicate directly with the simulator and are quite specific to the simulator being run while others turn the Arduino into a plug-and-play joystick. Most of these are free to use in non-commercial applications, while some are fully open source and may be applied in any manner. These are all factors which needed to be considered in our final selection.
A selection of the software considered follows:
SimVim: As with the others, this is Arduino based. It is a very expandable interface, however there is a somewhat higher degree of complexity and it is focussed on X-Plane. It is also limited to personal use.
MMJOY2: Turns the Arduino into a plug-and-play joystick. Certainly an option for those wishing to save on a Leobodnar card and prepared to follow instructions closely on how to flash their Arduino. Limited to personal, non-commercial use only, so not suitable for museum or training type applications.
DCS-BIOS: Some very enterprising individuals have directly mapped all the interfaces of a growing number of DCS World aircraft and translated this into a simple set of instructions which run of an Arduino. No setting up in the simulator is required once the program has been loaded onto the Arduino, and it is available for use whenever the simulator is started. It is fully open source. It is however restricted to DCS-World. It also only caters for the clickable items in the cockpit. As such the main flying controls (Rudder, Elevator, Ailerons) and a few specifics such as the primer lock and pump action need to be controlled separately.
Conclusion
There are many potential solutions depending on your preferences and specific requirements. Our selected route considers that the turn-key simulator will be used in a high-fidelity commercial setting with DCS World’s Spitfire Mk.IX. As such we will be taking the approach:
Two off Arduino Mega’s running DCS-BIOS and catering for all the functionality that this provides.
One Leobodnar board to allow for the non-clickable controls.
We’ve been hard at work to research the very best simulator control options. Our selection criteria was based on a balance of the following attributes:
the best accuracy
the easiest installation and setup
low cost
There are just such a plethora of options available that it became important to understand what the benefits of each are and work from there. The control options were categorised along the following lines:
Inputs – the sensors providing information to the control devices
Control Devices – the controller hardware which receives the Input information
Control Software – the software that is required for the Control Devices in order to interpret the Inputs and provides the simulator with information it can understand.
As indicated in various previous posts, there is no need to read information from the simulator to control external monitors, gauges and the like. Given that our simulator is built around VR interaction, our controls only need to provide a one direction feed to the simulator. In other words, we are creating a giant joystick interface.
The following is a summary of our findings on means of input. We will post separately on the control devices and control software.
For inputs we differentiate between Digital (as in switching) and Analog.
Digital Inputs
The Digital inputs can be:
Physical –
Toggle switches (Lever and Rocker),
Pushbuttons (momentary, latching and interlocked latched) and finally
Rotary switches.
In our design we use both Toggle switches and Pushbuttons but not Rotary.
Hall Effect – This is switched by the proximity of a magnet. They can be:
Latching (switch stays in position even when the magnet is removed, and can be either
pole sensitive where you need say a South Pole to activate and deactivate the switch, or
Non-pole sensitive, where any pole will do.)
Non-latching (switch flips as soon as the magnet is removed). Again, these can be:
Pole sensitive – examples are the Allegro A1101/2/3/4/6, which switch on South Pole only
Non-pole sensitive – example the UTC UH8104
In our design we use exclusively the UTC UH8104 non-latching, non-pole sensitive switches for the proximity sensing of the following:
switch cover is open or closed
fuel cut-off position
gunsight dimmer slider position
chassis control lever position
(the incorporation of these are currently undergoing redesign)
Finally it should be noted that Hall Effect Sensors come with through-the-hole (SIP3) wiring or as SMD (Surface Mounted). The latter makes it a pain to connect so we have stuck with units that are available in through-the-hole.
Encoders – These send on/off signals or pulses through two channels. The signals between the two channels are offset in a certain pattern depending on which way the encoder is being turned. It is then up to a controller to determine the direction and speed and provide that information to the simulator. It really provides multiple button or key presses and they are ideal for controls where there is no set number of rotations. An example would be setting a clock or altimeter. We have settled on a Bourns PEC11R which has a 12mm threaded through panel mount. They provide between 12 to 24 pulses per 360 degrees.
Linear Hall Effect Sensor for Brake
Analog Inputs
The Analog inputs serve axes of movement, such as trim, rudder, elevator, throttle etc. They can be linear, i.e. slide forward and backward, or rotary/angle sensors, measuring number of degrees turned. Most often the role for these were fulfilled by mechanical potentiometers, which slid across a resistive material thereby providing an indication of its position. These are being largely replaced by solid state Hall Effect sensors.
The advantage of Hall Effect sensors is that they provide a very steady, accurate signal, whereas potentiometers very often provide a bit of a quivering signal. A potential problem with Hall Effect sensors is that they send data in serial bitstream, which adds complexity and increases the real-time processing requirement to interpret these signals.
In the last few years however these have become more and more friendly, some providing an analog output based on a variation in voltage, typically from 2,5V to 5V. These can then be simply used as straight-up replacements for potentiometers.
So for our design we have decided to keep it simple yet very accurate by replacing all rotary potentiometers with the Melexis Tri-Axis MLX90371 (Gen III). These are tri-axis – they provide linear and rotary on-axis or off-axis. They also come in SIP3 Through Hole and provide an analog output. It should be noted that they measure 360 degrees, so for instance for the elevator trim where 4 turns are required end to end, we will be designing printed gears.
There are a few applications where a linear Hall Effect sensor will be used, eg. the brake lever. Replacing the original design, which had a very expensive Bourns 3046 linear pot, has simplified matters greatly through the use of an Allegro A1324 linear Hall Effect Sensor, again with analog output.
Tri-Axis sensor at the bottom of the Mk.II Gunsight
