What was the Filter Room?
It was June, 1940.
The Nazi war machine had conquered Continental Europe. Only Britain's Royal Air Force stood between the Nazi's and total victory. Alone and outnumbered, their only hope was a secret weapon - the world's first RADAR network.
The British called it Radio Direction Finding (RDF), but the principle was the same as RADAR: using radio waves to detect aircraft entering British airspace.
Although RDF was revolutionary, an individual station was too inaccurate to fix the location of incoming aircraft. So the RAF built a network, called Chain Home, around the coastline of Britain with each station overlapping its neighbors to prevent gaps in RADAR coverage. And the overlap meant that at least two stations could see any incoming aircraft, which RAF scientists realized would help them filter out the inherent inaccuracies. To do this scrubbing, they built the Filter Room, which became the heart of the entire system.
In the Filter Room a team of Women Air Auxillary Forces (WAAFs), scientists, and officers took inaccurate information that came in from the RDF stations, displayed it on a central map table, filtered out inaccuracies, and made critical decision about the incoming German raids. They passed this information throughout the defense network to coordinate responses to German attacks. The team had to work quickly, balancing the desire to wait for better information, knowing that each second meant less time for the pilots to get into the air, and reporting raids too quickly but missing critical information. Every passing second brought the whole system closer to disaster, but passing incorrect information risked lives.
Despite the odds, the men and women of the Filter Room succeeded. Their exceptional teamwork saved the free world.
HOW DID it work?
The Filter Room was a cramped, noisy, chaotic place. Despite this appearance, everyone in the room had a role to process and communicate information through the system.
Plotters
Plotters were stationed around the map table, each connected via a headset to the Operator at a RDF station. As the RDF Operators detected incoming aircraft, they reported the information to the Plotter, who would display it on the central map using a series of tokens. Plotters had to multitask. During attacks they were listening to an almost constant stream of information about the raid's location, strength and altitude, finding the correct tokens, and placing them at the right place on the map.
Filterers
As the Plotters filled the table with tokens, the Filterers turned them into a coherent picture. They were expected to make critical decisions about the information in less than two minutes because every minute they delayed cost defending pilots 4 miles and almost 3,000 feet of altitude.
Because the RDF stations were inaccurate, the Filterer could not rely on the information from a single station. They had to use plots from multiple stations to triangulate the correct location, and then decide which stations were reporting the most accurate number of aircraft and altitude. They made these decisions based on the position of the station relative to the incoming raid, but also based on their knowledge of the strengths and weaknesses of each stations and its operator.
As they were filtering the information they placed arrow shaped tokens on the table to track the raid's course. As they determined the raid's strength and altitude they displayed it using magnetic numbers on a small metal strip, which they placed on the map. By placing this "Raid Plaque" on the map, they were confirming that this raid was ready to be passed on throughout the defense network.
Teller
Tellers stood on a dais overlooking the map table. They passed information about filtered raids to group and sector operations rooms, using a standard "telling" format and the RAF Phonetic Alphabet for clarity.
Controller
The Controller sat on a dais above the Filter Room Map table and oversaw the operations of the entire group. The Controller also worked with the Liaison Section to ensure that flights from Coastal Command, Bomber Command, and other services were not identified as hostile.
Rebuilding the Filter Room
After 5 years of researching documents and interviewing surviving veterans, we have accurately recreated the Filter Room by combining hardware and a custom software suite.
The Hardware
Because working in the Filter Room was such a physical experience, we felt we had to get the hands-on equipment as close as possible to the real thing.
The Map Table
We spent years researching and redrawing the Filter map to scale, so that we could recreate the map table that everyone worked on. This was a challenge, not only because detailed descriptions of the map are rare, but also because the configuration of the map table changed frequently throughout the war, as technicians developed new techniques. This adaptability helped the RAF succeed, but it made recreating the system a challenge.
We chose, where possible, to focus on the table as it would have appeared in the summer of 1940. We redrew the map from primary docs, photographs, and interviews with veterans. We printed South East England, the most active section of the map, at 100% scale, mounting it on an "L-shaped" table, each limb measuring 10' x 5'.
Just like 1940, the finished map table showed the information used to process radar information, with a coordinate grid overlaying the map and radar arcs extending out from the center of each RDF station.
Recreating the Tokens
Like the map table, the tokens used to plot radar information evolved throughout the war. Based on primary sources and interviews, we narrowed down that three types of tokens were in use during the Battle of Britain:
· the circle, used to show location of incoming aircraft,
· the triangle, used to display the number of aircraft detected,
· the square, used to show the altitude.
We commissioned a production run of these tokens to match the size and color outlined in primary and secondary documents.
Recreating the Timing Devices
The Filter Room personnel had to process a lot of information and it quickly became apparent that when they were tracking multiple raids, too many tokens on the table paralyzed the work. To solve this problem, consulting scientists devised a timing device that the Plotters used to coordinate and remove their tokens every two and a half minutes. This device illuminated a number in sequence, 1 through 5, every 30 seconds. Plotters would use only the circle tokens that corresponded to that number. This timed their plots. When the number repeated, they removed the old numbered tokens from the map. This cleaned them up.
We built a physical reproduction of the timing device and incorporated a timing circuit and an LED array. We programmed a Raspberry Pi mini computer to control the sequence timing. When a raid starts, the server sends a signal to the timing device which illuminates number 1. Every thirty seconds, the server sends a new signal that advances the timing light. We also built a server-controlled Sector Clock. Every room in the Dowding System had a synchronized Sector Clock, divided by color into five minute segments. These were used to color code plots throughout the entire defense network, ensuring that everyone had the latest information. We refaced an antique GE Telechron and added a relay and a Raspberry Pi mini computer. We made controlling the clock seamless by adding code to the Start and Stop Raid buttons on the software that sends a signal to the controlling computer.
Recreating the RDF Stations
Each Plotter around the Filter Room map was connected by a headset to a different RDF station. To replicate the RDF stations for our simulation we used Raspberry Pi mini computers. Each RDF station is a Raspberry Pi, and they are all wirelessly networked to the server. These computers are mounted on the underside of the table, out of sight, and the participants plug a headset into them to listen to audio from the station.
The Software
The hardware of the Filter Room recreation is nothing more than a static display without data to process. To move from display to simulation, we created a software suite that controls the integrated hardware in the room. This suite includes:
The Administrator Tool
The Administrator Tool is the central control for the entire simulation. From here, the Facilitator can build up to 5 raids, selecting the type, number, and altitudes of German aircraft. The Facilitator then plots the course each raid will take, assigns a bombing target, and a defending airfield. The software runs calculations based on historical aircraft performance data, and for each raid determines a "time to last intercept." This is last possible moment that the RAF squadrons can be scrambled to intercept the incoming raiders successfully. This time to last intercept becomes a primary scoring factor for the simulation. Once the Facilitator launches the raid, the server passes the raid information to the next tool in the suite.
RDF Voice Synthesizer
Each client Raspberry Pi runs code that simulates the performance of that RDF station. Once the server passes the info to the client, the client calculates the true coordinates, factors in a historically accurate error, builds new coordinates based on the error, and determines whether the raids are visible to that station. If they are, the software assembles a dynamic voice over audio file to "speak" the raid information to the participant in the voice of a British female RDF Operator.
Click the Play button below to hear a sample of the voice synthesizer.
Presentation Map
The simulation can accommodate up to 12 participants, but can involve more people in the audience. For those people, we have built a dynamic, moving map that shows the location of incoming aircraft, the time left until the last possible intercept, and the status and progress of defending RAF squadrons. As the simulation progresses, this map is updated in real time by information from the server. Observers can watch the action on the physical table and compare to the "answer key" they are watching on the Presentation Map to get a sense of how the Filtering Team is doing.
Tally Board
To recreate the dilemma between solving the problems quickly, and waiting for more information to solve them better, we created a scoring system that factors in both accuracy and speed. To raise the tension about this with participants we created a software scoreboard that displays the current raids, a speed scoring bonus, and the raid's status (inbound, targeted, success, or failure). The speed scoring bonus starts at 200 points, and counts down as the simulation progresses, spurring participants to maximize the team score by solving the raid as fast as possible.
Simulation and Game Modes
We built the baseline software to be as historically accurate as possible, but during early pilots of the simulation, it was clear that we needed a simpler, more approachable version for casual participants. We created a game mode of the simulation, combining Chain Home and Chain Home Low stations, simplifying the audio, and replacing the Strength and Altitude tokens with a paper form. This makes it simple enough that players can perform the job after a 15 minute briefing while still experiencing the stress and tension of the full simulation.