in this episode of Machining and Microwaves, I'll
explain in detail how the Great Seal Bug worked and show you the design of the replicas I made
there'll also be a bit of ... and even some ... as a sneak preview of how I machined the replicas.
As I explained in an earlier episode, the famous Bug was found in the U.S Ambassador's residence in
Moscow in 1952. it had been eavesdropping on the top secret conversations in his study since the
spring of 1945. The Bug has no battery, no power supply and no active components. Let's have a look
at what the bug looks like. There's a cylindrical metal body a little under an inch in diameter and
three quarters of an inch long with a thin round rod sticking out of the side the throttle has
a threaded end and fits through a plastic bush with an internal thread. The rod's two millimeters
diameter and the threads are 0.4 millimeter pitch in Imperial that's about 64 teeth per banana. On
one end there's a grille with eight pierced holes around a central round hole on the other end
there's a plane cap under the grille there's a thin metal membrane fixed to a threaded bronze
ring. Inside the cylinder there's a part with a central post that's formed into a wide threaded rim
at one end and a small flat grooved disc at the other. The threaded rim fits into an internally
threaded section inside the cylindrical body The post's two millimeters diameter and the disc
is six millimeters across. The internal thread is an M20 by 0.5, so about 50 TPI. Apart from the
plastic bush and a flat ended metal disc on the inside of that long two millimeter rod, that's all
there is. No active parts, no wires, no trickery of any sort. To an untrained eye how on earth could
that be a bugging device? Have you ever seen the demonstration where a singer belts out a loud high
note towards a crystal wine glass? If the note's exactly correct, the glass vibrates very strongly
as it absorbs energy from the vibrations created in the air by the singer's voice. It can even
shatter if the singer is loud and exactly in tune That's a mechanical vibration. The exact note where
the sympathetic vibration is at a maximum depends on the density of the glass, its dimensions
and its stiffness. How sharp the resonance is and how long the glass continues to resonate
depend on the material the glass is made from Individual atoms in the glass don't move much
but the bulk effect of all those tiny movements at atomic level combine together, and the result is
quite a loud ringing at resonance. Now if you take a thin metal rod hanging from a fine insulated
thread and expose it to a radio wave at just the right frequency, it'll pick up energy from the
electromagnetic field of the wave. There isn't any need for air to be there, it's not a mechanical
vibration, it's electric and magnetic fields changing intensity very fast. The rod intercepts
those fields in a similar way to how the wine glass intercepts the sound waves, so instead of
a physical vibration, electrical currents start to flow in sympathy as the electrons in the
rod are pushed around by the changing field They don't move far, same as the atoms in the
glass don't move much. As the electrons start to move they generate a magnetic field. Now the
combination of changing electric and magnetic field in the rod means it behaves just like a
radio transmitting aerial, and it will radiate energy But wait! Where did that energy come from? Well, it
was supplied by the incoming electromagnetic field. So surely they cancel out? What actually
happens is the two fields are superimposed, but they don't cancel. There might be cancellation at
some specific points, but let's imagine you send a radio beam up a narrow street of tall houses. You
suspend your aerial rod vertically in the centre of a crossroad junction. The rod gets immersed in
the electromagnetic field of the radio beam. Those houses are made of material that soaks up radio
energy which is why I could never get a cell phone signal inside my house which is made from damp
bricks held together with custard. (Creme Anglais) for those living in civilized parts of the world.
The rod absorbs energy from the electromagnetic field and generates its own field. As a result,
the rest of the radio wave energy disappears straight on from the junction and eventually... Oh
wait. If there are any flat earthers (or in these political correct times perhaps I should say
terrestrial globularity deniers?) watching please cover your ears. Eventually it dissipates
or it goes into deep space as it carries on while the curvature of the Earth drops
away. OK, uncover your ears Flerfers! Hello ???? Now let's see what happens if the rod isn't
there. There's still a bit of diffraction at the junction and a little of the incident
radio signal leaks around the corner into the side street, but it's significantly less
and there's some deep nulls with no signal Comparing the steady state fields with
and without the rod shows clearly that the rod's making a significant contribution to
the signal down the side street as we'd expect Now think about what happens if we set up a
radio receiver down the road that crosses at the junction. It can't see the signal from
the transmitter but it can see the rod and it can receive some of the energy that the rod
re-emitted. OK, ok, in the real world you'd get some reflections and some direct signal but
if nothing in town is moving, all the reflections and direct leakage will superimpose. They might
look like this on a graph of power against time. All different amplitudes and phases. By some
trigonometric magic, a combination of sine waves with the same frequency but different amplitudes
and phases will combine into a single signal I have a wonderful mathematical proof of that but
this margin is too narrow to contain it. The phase of that combined signal will be fixed so long as
nobody moves anything, so if you collect some of it on a separate antenna and adjust the delay and the
amount of signal you could cancel out most of the direct signal you see on the main receive antenna.
It's a similar idea to the way noise cancelling headphones work, except they do all the maths
and fiddling for you. In a perfect world where the transmitter's hidden from the receiver, but the
rod is visible to both, you can adjust your receiver to hear mostly the re-transmitted signal. Not much
use so far but imagine for a moment that you could stretch and compress the rod so the resonant
frequency can be moved up and down a little As you change the length of the rod, the amount of
energy it harvests from the radio wave changes, as does the resulting re-radiation. Now if you could
change the length of the rod thousands of times a second you could use it to apply audio modulation
to the re-transmitted signal. There's a problem with a simple straight rod, though it's not
sharply resonant. Unlike a wine glass, it resonates poorly over several percent of its central peak.
Part of that's because it re-radiates the energy it receives very effectively, assuming there are
no other losses. In reality, all practical materials do have a bit of loss, but it's negligible for a
copper or silver rod in free space To make a really sharp and efficient resonator
we'll have to move to a different arrangement. If you've got a short cylinder, blanked off at one end,
and you fix a thin rod to the end plate, you can make a resonator that doesn't radiate its energy
away. If you shorten the cylinder and rod it'll resonate at a much higher frequency. One way to
move the resonance back down is to fit a disc to the free end of the rod and fit a plate over the
open end of the cylinder. The small gap between the disc and the plate can store electric charge and
then release it. It does that by concentrating the electric field much more tightly than where the
rod's just sitting in the open end of the tube. The effect is called Capacitance. Capacitors
are used in almost all electronic equipment but they usually have tiny gaps and large surface
areas. Often they're made from foil sheets folded or rolled up with insulating material between
them to concentrate the field even more. Using a shortish cylinder and rod and adjusting the gap
to be REALLY tiny, that extra capacitance can pull the resonance down to a fraction of what it would
otherwise be. Shortening the cylinder also reduces conduction losses, and as a result, the cavity
can have an extremely narrow resonance bandwidth, hundreds of times sharper than the rod in free
space. Sadly a closed cavity is no use to anyone. We need a way to get some electromagnetic wave
energy INSIDE the resonator so it can ... resonate So how about we drill a hole in the side and poke
our suspended rod into the cavity so the rod end is fairly close to the central post. A bit of the
electric field from the free end of the rod will couple with the post, but it won't be a very strong
coupling, so let's put a flat disc on the end of the rod to increase the area between it and the
post. Now more of the energy in the rod can couple into the cavity. The energy swills back and forth
like water in a bathtub, making larger and larger waves, but of course the energy is also coupled
back to the rod and excites a larger oscillation in it. As before, most of the energy that's
coupled into the cavity resonance is re-radiated from the rod, however the effect of coupling the
rod to the cavity means that the sharpness of the resonance in the rod is increased enormously,
while that of the cavity is flattened out a bit. If you change the gap between the post and
the disc on the end of the rod, the amount of coupling can be adjusted. Now unfortunately
that also changes the tuning of the cavity and the rod, so getting the gap, the length of the
rod, and the gap at the free end of the post all adjusted is hugely fiddly. SIX hours and
FIVE coffees it took me the first time I tried! So far we still don't have anything useful.
We need to find a way to adjust the resonance using a sound wave vibrating something like in
a microphone. Spookily enough there's a type of microphone which uses a stretched conductive foil
spaced a tiny distance from a plate with a cavity behind it. As sound waves hit the foil it vibrates
back and forth in sympathy with the air molecules It's known as a "Condenser microphone". Condenser
is an old name for a capacitor! How convenient. Remember we've got a capacitor formed by
the gap between the top of the post and the plate at the open end of the cavity? Well,
what would happen if we remove the plate and stretch a very thin foil across the end instead?
Assuming we stay very very quiet and the foil's a good conductor at a thousand megahertz like
the plate was, nothing will change. The Q factor will be the same, the resonance will be the
same, the re-radiated signal will be the same. Now, if the foil moves a little towards the disc on
the resonator post, that increases the value of the capacitance at the end of the cavity. The resonant
frequency falls a little because of Physics . If the foil moves away from the disc,
the resonant frequency rises a little Imagine we ask our opera diva to sing a note at
the unfortunate wine glass a tiny fraction of a semitone too high. The glass won't resonate as
much as when they're on the correct note. If they move the note up a tiny bit more, the amount of
the glass resonates will be even less. Conversely, if they change their strident yelling down to a
slightly lower note, the glass will vibrate like billy-oh. Interesting. Let's tune our Bug to exactly
a thousand megahertz and check the response over a few hundred kilohertz either side. It looks like
the Q Factor is about 1000, with about 1 Megahertz bandwidth at the half power points. Apply
a steady radio signal at 1000 Megahertz and check we're at the peak. Now let's tune the bug
300 kilohertz down in frequency to 999.7 Megahertz The amplitude's about half what it was at the peak
frequency. When a sound wave arrives and increases the pressure on the diaphragm above average, it
pushes it towards the disc on the resonator post That reduces the resonant frequency down to
perhaps 999.6 Megahertz and the amplitude of the oscillation falls a little more as we slide
further down the slope of that resonance curve. A thousandth of a second later though the
air pressure drops below average and the resonance shifts up to maybe 999.8 Megahertz,
so the amplitude of the resonance increases. That variation in the amplitude of the resonance
varies with the sound waves arriving at the foil. The re-radiated energy also varies in the same
proportion. The effect of the variations means that re-radiated signal has the audio signal
impressed on it as a few percent of Amplitude Modulation, just like a really terrible sound
engineer might produce on a broadcast AM radio station, You know, the type we have here that
plays BOTH types of music. Country AND Western! The re-radiated signal also carries some phase
modulation. For hugely complicated mathematical reasons that's a Good Thing. Now instead of a town
with a crossroads, let's install our transmitting equipment in a building over the road from the
Ambassador's residence at 10 Spasopeskovskaya Square in Moscow, and the receive equipment
in a different building off to the side. We'll arrange for some washing (laundry) to be hung out on
the balconies regularly and make the place look as normal as possible. The transmitter setup isn't
documented but it should have been simple enough, just a plain continuous carrier at around 1 GHz
with nothing clever apart from good frequency stability, a stable power supply, and
careful control over amplitude and phase noise. In 1945. In Moscow. During a long and hard war. It
could have been an injection locked Klystron or perhaps a UHF tube above its normal limits. I
don't see a suitable vacuum tube in any of the list of parts made by Svetlana or the other
makers in Russia, so I'll have to defer to those with more knowledge of thermionic device
history in 1940s Russia to fill in the details, The receiver was probably a Homodyne. If they
picked up a sample of the illuminating signal from a sensing antenna as I described before, it could
certainly be used as a coherent local oscillator and, mixed in with the receive signal, tweaked a lot
in phase and amplitude to get the best audio response I guess they'd have a cabled intercom
to the transmitter site for talkback liaison or perhaps a telephone to get
the transmit frequency and antenna alignment optimized, as well as fiddling
with the receiver settings and antenna setup It's certainly possible to use a modern AM
receiver but local oscillator phase noise and frequency stability would have been big
issues back in the 1940s. The homodyne might lack sensitivity, but you just need to make
up for that by using a bit more power at the transmit site, or higher gain antennas. Helicals and
Yagi-Uda antennas were certainly known at the time. In a homodyne, the mixer diode operates in
its square law region of forward conduction, performing a multiplication of the wanted signal
and the unmodulated carrier that results in some components at the original audio frequency
being produced by the diode. You need to pass the mixer output through a diplexer and low-pass
filter to extract the audio signal that carries useful voice traffic. Perhaps from 200 to 3000
Hertz. You also need infinite patience to adjust all the different parameters as they change
with humidity, temperature, people moving in the room, foliage moving in the wind, noises from
pipe work and all sorts of other metal objects. Right that's a very simplified but reasonably
accurate description of what's going on in the Great Seal Bug system. If you're still with me,
take a deep breath and give yourself a pat on the back. Now a proper YouTuber would give you a
Zen moment with an ASMR ambient soundscape and floaty pastel images of wildflower meadows and
nectar-tipsy bees. However it's still March here in Yorkshire. The wildflower meadow is muddy. It's
raining and blowing a hoolie out, there so I'll give you a mug shot of one of my Chihuahuas
instead. She's a very nice Chihuahua but I understand completely if your Meridian Response
isn't being very Autonomously Sensory right now. As a quick teaser for the upcoming Machining
and Metrology episode here's a bit of calming lathe work making one of the end covers, I made up
some precise gauge pins to get the fit absolutely perfect. The original findings from the naval
laboratory and FBI led to the conclusion that heavy press tooling was used to form the end
covers. Any half decent clockmaker or machinist would have been able to turn them, but as we don't
have the originals we aren't going to find out. As I mentioned, the thickness of the original
membrane was reported variously as anything from 7 to 75 micrometers. Now at one gigahertz,
the skin depth where RF currents in silver drop to 1/e or around 37 percent
is two micrometres. Nickel is a terrible conductor of RF as it's ferromagnetic. The
RF field lines are forced even closer to the surface than they would be in silver and
the bulk resistivity is also considerably higher, so the skin depth is tiny and the
RF resistance losses are huge. To get best performance, around six skin depths of highly
conductive plating is needed. At six skin depths the current is reduced by a factor of 0.37 to the
power 6 - about 0.2 percent of that at the surface. That rather indicated that the silver layer
would need to be around 10 micrometers, so a thicker nickel foil would appear to make sense ,
but it's less than ideal from an acoustic and mechanical perspective. I decided to go with a 10
micrometer copper foil, stretching it radially to work-harden it on a special jig. see links in the
description to the membrane stretcher videos I found that the best results were around
950 megahertz but because of licensing restrictions for transmissions in the UK, I had to
do the demonstration at 1.3 gigahertz. That meant making a slightly shorter model for
the demo to maintain the same performance. The external antenna rod is supported in
a threaded polystyrene insulating bush that keeps it rigid, as well as being a
good dielectric insulator. There are some interesting questions about how the material was
sourced. I haven't got any definite verification of this, but I suspect it may have come from IG
Farben who, amongst other things, made Zyklon B. They were involved in technology transfers to
the Russian State during the Molotov-Ribbentrop non-aggression pact that ended on 22nd of June
1941 at the start of the Great Patriotic War The material I used was from a
donation by a subscriber - thanks Pete It dates from the early 1960s, but should
have identical properties to the original I made the first bodies out of brass rather
than copper because I wanted to check how tricky the machining would be. The threads
are 0.5 millimeter pitch and proved to be very straightforward to cut in brass. The
bug I used for the live demonstration was made from horrible gummy C101 copper, which
is a little more... challenging to machine The original was polished internally and silver
plated. I reasoned that the performance would be almost as good with no plating but that it would
degrade over time, so for the purposes of the demo I didn't bother plating the bodies or resonators.
You could argue that a silver plated surface would allow a better electrical connection
between the resonator thread and the body as compared with bare copper, but in reality there's
a large capacitance between the two parts across the threads, and the impedance is very low as a
result, whether or not there's a good DC contact I would imagine that in the original the
plating was made such that it gave a good tight fit and with silver being relatively
soft and slightly self-lubricating like gold, it would probably have made rather a
good running fit with zero clearance. The half millimeter bleed hole from front
to back made perfectly good sense because if you slam a door in the room where the bug's
installed you don't want that sudden pressure wave to overwhelm the diaphragm and short it out.
I used a tiny long-series drill to make the hole, but they probably used a spade bit. Perhaps one
day I'll make a 0.5 millimeter spade bit to try. The long narrow hole effectively creates a sort
of mechanical audio high pass filter, so the response below a few Hertz will be severely
reduced and a sudden surge wouldn't be able to compress the diaphragm very far unless
it was very very sharp and instantaneous. Closing a door doesn't usually create a huge over-
pressure anyway. Tests have been done to measure the over-pressure from doors closing, and there
are components up only as far as a few Hertz in a large furnished room like the study at Spaso House.
So let's think about the movement of the diaphragm, With our one kilohertz signal at 40 dB sound
pressure level, the air molecules are moving back and forth with about 10 nanometers of peak
amplitude. The capacitance between the resonator and the diaphragm varies according to the
reciprocal of the spacing. With a 25 micrometer Gap - that's about one thou (or one mil) - and a six mm
diameter post, the capacitance is epsilon nought times the area over the spacing. That's only
approximately true where the discs are the same size. "With a small disk near a large membrane
the analytic solution is much more complex because of asymmetric equipotential lines, although given
how rough your measurements usually end up, it'll probably do" She's not wrong! However as I'd removed
almost half of the surface to a depth of half a millimeter, the result is probably closer to seven
picofarads, including some fringing effects. The CIA report from 1955 included some
measurements carried out on a copy of the Bug. They sliced the body in half and fitted a
polystyrene insulating ring into the annular gap and then readjusted the resonator for the same one
mil or 25 micrometer spacing from the diaphragm. They used a Boonton 160A Q-meter at a frequency
way below resonance. I think they're rated to 75 MHz and measured a capacitance of 10 pF,
although of course they called them micro-microfarads. It's not clear whether they
zeroed out the capacitance of the new gap when setting up the Q meter. If the polystyrene was a
sixteenth of an inch thick, say 1.6 millimeters, the capacitance would be Epsilon naught times the
relative permittivity of polystyrene which is about 2.6, times the surface area of the cut face,
divided by the thickness of the ring. That's about 2.2 picofarads and rather suggests that their
measurement could indeed include the strays from that gap. I'd have removed the diaphragm
and zeroed the bridge so the measurement only included the membrane to resonate to capacitance,
but it isn't clear whether that's what they did. I'd have recorded my methods too, to remove any
ambiguity. Sorry that's Professional Neil inside my head, getting all fired up about techies and their
terrible documentation skills. Deep breath Neil. Needless to say, in a future video I'll be
slicing one of my replicas in half and testing the capacitance using a modern Vector Network
Analyzer. HEY! if you subscribed and enabled notifications you'd be the first to find out when
that's published! Of course if you really want to get on the inside track of what I'm working on,
you could consider joining my Patreon page and receiving the twice monthly newsletter and some
previews and outtakes, or get access to my private Discord discussion server to help me with ideas
on which projects to do next. The Link's in the description and on a card at the top right of the
screen. You can view those cards at any time during the video. Right! now we know the capacitance of
the resonator to the membrane, how do we find the resonant frequency? Well, any tuned circuit has
a resonant frequency inversely proportional to the square root of the capacitance times the
inductance. The exact equation is on the screen. There is an analytic solution to find the
inductance of a thin rod in a round coaxial cavity, so we could start with that to get a
ballpark figure for the inductance. We need to take into account the changing diameter of the rod
and the end effects, but it's a good starting point If you've got a coaxial line of length a with a
central rod with an outside diameter of little d in a tube with an internal diameter of Big D the
inductance is just mu nought times the length over 2 pi times the natural log of the ratio of Big D
to little d. The CIA report shows this incorrectly by the way. Their equation would show a negative
inductance, which is a wild concept that makes my brain rattle. Plugging in the values with a length
of 15.5 millimeters and a resonator shaft diameter of 2.2 millimeters and the body ID of 19.53
millimeters the inductance works out at 2.9 nanohenries Taking the end effects and strays into
counts it's probably more like 2.5 nanohenries Putting that value into the equation for
the resonant frequency with 10 picofarad capacitance suggests a thousand and six
megahertz as the result plus or minus a lot for the fiddle factors. My original test
seemed to perform best at 960 megahertz which is spectacularly close to the calculated
values, but it certainly wasn't by design I was still influenced by all the disinformation
I'd read (!), so I had no idea,. Still like a good Engineer, I recorded my findings even if I
thought they were Wronger than a Wrong Thing Now the CIA lab did some careful checks to see
how the resonance changed with the spacing of the diaphragm. They marked 64 divisions around
the back face of the resonator and put a fixed mark on the body then they tweaked the resonator
in tiny steps of a quarter of a division, always in the same direction to remove any
backlash. That's 256 steps per rotation The original and my replicas have half
millimeter thread pitch which is 50.8 threads per inch, but the model they were using
had a 48 TPI thread which is reasonably close. Each quarter division represented a
resonator spacing change of about 80 micro inches. Their graph shows the variation
of response to a fixed signal versus spacing I marked out 60 increments on one of my resonators
using a rotary table one of my ticks represents 8.3 micrometers of resonator gap change, so with
it resonated at 960 MHz, it was at three ticks from the contact point. I used a nano VNA and
a pickup loop to look for movement of the dip at resonance. Half a division of rotation moved
it from 960 to 1030 megahertz sort of roughly 70 megahertz from a 4.16 micrometer change so 17
megahertz per micrometer as a measured result Assuming the spacing's 25 micrometers the
capacitance changes by a factor of 26/25 or 4 for that one micrometer move as the resonant
frequency is inversely proportional to the square root of the capacitance it'll change by about
two percent or 19.6 megahertz I think that's good enough given the huge uncertainties and
approximations involved. The resonance varied with temperature and I had to use some thread
lock to keep it completely stable against physical movement. I imagine the nkvd team would have to
get used to varying frequency with temperature and humidity changes, but with a telephone link
between the transmitter and the receiver it would have been simply enough to tweak things for
a maximum performance at the start of each session if the story about the caretaker's empty room is
true then perhaps he would know when someone's in the study and informed the operations team
that it was time to fire up Zlatoust. I'll try to include the fine detail of my measurement setup
in a future video on this channel or perhaps I'll simply upload the raw footage and commentary on
my second Channel Machining and Microwaves Plus The links in the description as proper
YouTubers would say. Thanks for watching, and thanks to my excellent Patreon supporters
for helping me to make more videos. Next in this series will be the Great Seal Bug Machining
and Metrology episode, which will be up THERE
