"Agent Hammer's" English Translation of Robin Lobel's TEMPEST paper
Note from the translator: "I just don't have a mastery of the English words and don't know this stuff well enough to make them up. I think you will be able to decipher my crazed translation though. It doesn't look like they got all that they wanted out of their experiment...rather they weren't able to finish it. I get the impression they had to turn in their report before they were done. Its clear to me that these were students in a professional technical high school/ community college type setting."

Introduction:

1.            Definition and basic information

When an electronic device is used, it sends out electronic waves that can stretch several meters out into the surrounding environment.  When these waves are captured, they can be used to reproduce the information contained within them.  These waves are called “jeopardizing waves” because they put the information that is contained within them into jeopardy.  This is true for all kinds of electronic devices.  These waves can theoretically be captured and allow us to read even the most secretive information.  Nevertheless, the amplitude of the waves diminish quickly, making it difficult to capture them for more than a few centimeters from the initial signal, thus it is difficult to capture signals from most devices.  Computer screens, however, send out signals 500 times stronger than the initial image the video card sends out, thereby sending out waves at a an amplitude strong enough to easily capture them.

2.         Proof that the phenomenon exists

If a computer screen is plugged into a central unit with a non-reinforced cable, an echo effect will take place as well as a delayed reproduction of the original image on the computer screen.  The cord acts as a receiving antenna, capturing the waves from the antenna and transforms them into electronic waves sent to the screen.

3.            Objective

The objective of this report is to prove that the phenomenon of “jeopardy” exists, and to attempt to understand under what conditions and at what cost it is possible to reproduce images on a computer screen

I.          History of TEMPEST

A historical summary of TEMPEST is presented.  They summarize how and when Tempest began, previous names of what is currently called TEMPEST and discuss attempts to get declassified information about TEMPEST.  Most attempts have failed to produce satisfactory results.  The latest attempt in 1999 produced severely censored documents about TEMPEST.  There is very little detailed information available about this system.

II.         Theory of Screens

1.                  Deconstruction of an image

All colors can be broken down into three fundamental colors: red, green and blue.  Using variations of intensities in the combinations of these colors, any other color can be created.  An image is considered a complex assembly of colors through the use of a pattern of “pixels”.  A pixel is a point composed of three colors, red, green and blue.  By increasing the density of pixels in a single area, it is possible to recreate accurate images.  The resolution of an image is represented by the formula X*Y with X being the number of horizontal pixels and Y being the number of vertical pixels (ex:  640*800, 800*600, 1024*768…)

2.                  Reproduction of an image on a screen

A screen is composed of several modules.  The cathode tube is what reproduces the actual image.  An electron beam scans a fluorescent layer at an extremely high speed, creating the image.  The scanning goes across the entire screen from left to right and from top to bottom at a frequency of 50-100 Hz.  As the electrons pass through the fluorescent layer, it sends out a light.  This layer also becomes phosphorescent in that it continues to send out a light after its initial stimulation for approximately 10-20 ms.  Its brightness is determined by the debit in the electrons, which is regulated by a “wehlnet”.  The beam then passes through two bobbins that determine its trajectory through electromagnetic forces, and an image is then scanned onto the screen.

3.                  Coding of the video signal

The video signal passes through several channels:  6 channels for the video signal itself.  Meaning, the Red, Green and Blue channels as well as their respective masses; 2 synchronization channels for the horizontal and vertical scanning and the communal mass of synchronization signals.  The synchro signals are simply the difference in a few voltage potentials.  They take place 70 times per second for vertical synch (for a 800*600 resolution screen cooling at 70 Hz) and 70*600=42000 times per second for horizontal synch.

Video signals are at a voltage of 0 V to .7 V, which defines the brightness at the point where the scanning takes place (this voltage tends to change depending on each new pixel color.  For an 800*600 res screen with a cooling of 70 Hz, the changes in voltage can go all the way to a frequency of 800*600*70= 34MHz, or 34 000 000 times per second). 

III.               Theoretical Expansion on Circuits

1.   Circuit Demands

Earlier, we learned about the nature of the electrical signals that, through amplification, drive an image toward a screen.  As a result of this amplification, the“jeopardy waves” that we are attempting to capture, are created.

For every difference in potential that is created at the exit of the amp circuit, an electromagnetic wave of proportional amplitude is emitted.  The amplitude of this wave diminishes as the electromagnetic energy spreads across the front of the spherical wave.

(graph)

An oscilloscope shows us that the image on the right is deformed by the absorption of the wave linked to the horizontal synchronization signal (center); note that the signal oscillates on the y-axis because of disturbances in the supply.  We want to capture the video signal at left.

Based upon what we have demonstrated earlier, this image is not directly exploitable on a screen because we need a positive signal whose voltage is between 0 V and .7 V. 

The solution must allow us to cancel out the signal created by the synchronization signal and the supply while amplifying the signal.

  1. Signal filtering

In an effort to eliminate parasitic signals, we create an open (???) circuit between the receiving antenna and the screen.  There are two types of circuits, high band and low band, which will allow high and low frequencies through them.   In our case, we need a high band circuit because the video signals (several 10’s of MHz) are higher than the synchronization signals (several 10’s of KHz for horizontal synchronizations).

Drawing

The drawing, through the use of a condenser and resistance, creates a high band filter.

Essentially, all signals can be considered as the sum of the sinusoidal signals.  Consequently, the high band filter can “suppress” the components whose frequencies are less than the frequency of the breaker (??).

When the high band filter is exposed to a sinusoidal voltage (tension), the condenser takes a charge.  Then, when the sinusoid changes variation and direction (??), the condenser discharges.  However, if the voltage period is superior to the charging period, the condenser will react like a circuit breaker and impedes the signal’s passage.  One can vary the frequency of the breaker by adjusting the values of “C” and “R”.  Suppose t=RC.  If t increases, the charging period on the condenser increases and therefore the frequency of the breaker diminishes.

On can deduct from this that the frequency of the breaker fc is inversely proportional to t=RC.  We then have:

fc = 1/2pRC

One can then deduct how the voltage will appear as it exits the device relative to the voltage as it enters the device.

If the frequency at entry fe is higher than the frequency of the breaker, the condenser takes a positive charge then a negative charge.  On can thus write the equation of the voltage at the circuit terminals RC:

                                                             -t/t

Us = sin (fet).(1-e   )

On can therefore trace the voltage at the circuit terminals RC relative to the frequency of the voltage at entry.

(Graph)

As such, if the voltage at entry is the sum of two (or more) signals, one of which is higher in frequency than the breaker and the other of which is lower than the breaker frequency, all that will emerge will be the frequency signals that are higher than those of the breaker:

Graph

One notes that the exiting signal (yellow) is “almost” the same as the frequency signals that are higher than the breaker signal (blue).  It is slightly deformed.  In green, the entry signal – the sum of two different frequency signals

  1. Amplification

To have an exploitable signal, we need a signal that is between 0V and .7V.  It must therefore be amplified, but the proportion of the difference between signals must be preserved.  To do this, we must therefore multiply the voltage exiting from the RC circuit by a factor of k.  For this, we chose to use a circuit based on an Operational Amplifier (O.A.), referred to as an “inverser”.

Drawing

An inverser circuit that amplifies the entry voltage by a factor of k

In such a circuit, we see that Ohm’s Law applies to the resistance R2.  This then reduces the voltage entering the operational amplifier; the voltage Ur is therefore less than Voltage Ue.  The Voltage Us is therefore proportional to the value of the resistance R2.    The resistance R3 is in diversion with the OA; the more its value increases, the less current will cross it and therefore more current will cross the OA, knowing that i1<i2.  On deducts from this that the exit voltage is proportional to the value of R3’s resistance.

One notices that the set up is reversed (entry on the negative terminal and grounding on the positive terminal).  The exit signal is therefore multiplied by a negative coefficient, hence one can deduct that:

K = - R3/ R2

By regulating the values of R2  and R3, we are able to vary the voltage at the terminals of entry on the Operational Amplifier.  The OA will then multiply the voltage of the entry signal (through stable feeding from the V+ and V- terminals).

  1. Synchronizing of Signals

The recuperated signals are de-synchronized.  The screen cannot therefore recreate coherent images.  We must then, send artificial synchronization to the screen.  To do so, we can use two methods:

§         Generate a signal with the use of two GBF (one for the horizontal synchro, the other for the vertical)

§         Capture signals emitted by the graphic card in a functioning computer

The first option appears to be most appropriate because it allows for changes in order to adapt the synchro signals to the received signals.

  1. Expected Results

Because of the fact that a wave is only emitted for each different voltage in the screen, the obtained image cannot be an exact replica of the first image.  But it will allow access to the information posted on the original screen.  In theory, one would obtain an image something like this:

Picture

IV.              Experiment

1.   Choice of values for the high band:

As we saw earlier, we would like to eliminate the effects of the synchro signals.  These signals repeat themselves at a frequency of around 70*600 = 42KHz (for a screen of 800*600*70Hz, taking into account the horizontal synchronization).  We choose a slightly higher value for the breaker frequency of fc to assure a margin of safety and eliminate a maximum number of parasites.  One therefore chooses a neighboring breaker frequency of 160 KHz:

ƒc = 1/2pRC

RC = 1/2p ƒc

RC = 1/2p.160000

            –6

RC = 10   s

We must therefore choose an RC relationship around 10 –6s.  At this point, the condensers represent their maximum values in micro Farad (mF).  The resistance employed must therefore be represented by a kilo Ohm (kW).

  1. Choices for amplification values

a.      Choice of Operational Amplifier models:

For our experiment, we need an operational amplifier capable of supporting significant frequencies nearing 50MHz.  To be safe, we have chosen an OA that can handle closer to 60 MHz.  We are therefore in the realm of OA’s that are uniquely created for video production.  The model we selected is the AD844AN.  This model needs a stable feeding of 5V and a maximum voltage entry of 5 mV.

b.      Choice of resistance values

Our Operational Amplifier choice necessitates strong resistance values such that the entry voltage can be 5 mV.  To obtain such a large resistance value, we assert the following:

A voltage is a difference in electrical states, or the difference between two potentials.  The weaker the gap, the weaker the voltage!

Drawing

Our goal is to choose a resistance that allows us to have a voltage between 3 and 5mV.  The wire on the positive terminal is connected to the ground.  Its potential is therefore zero.  We will therefore need a potential of 5mV from the wire on the negative terminal.  This value is extremely weak near an entry potential (about .1V).  When adding a large resistance, we obtain a strong voltage at the resistance terminal (Ohm’s law) and therefore a weak potential at the lesser terminal.  We will therefore choose a resistance variable, R2, of 1 W (Ohm).

We witnessed the amplification relationship earlier.  Knowing that we want to amplify in a block of 1 to 100, we will need a resistance, R3, somewhere between 10 and 100 W.

Photo

The original signal (green, -3V) and after amplification (+5V).

V.  Results

At the point of writing this thesis, our experiment has not yet been completed (although we do have all the necessary components to make it happen).

Our most encouraging results that we have obtains thus far are below:

2 Photo

The original image is on the left.  At right the 2 central pics, captured electro magnetically, correspond to the beginning and end of the white band.

The signal on the right was obtained thanks to a high frequency filter, which demonstrates its effectiveness.  We now have simply to amplify this signal and transmit it to a second screen that we will synchronize with the help of 2 GBF’s to create a ghost image, similar to the original image.

We had several problems with the operational amplifier (most notably the resistance) but these were eventually resolved (as is evidenced by the image on the previous page.

back