World's Smallest Geiger Counter
ELF Monitor with AGC and Hum Removal
Enrico Mathesar
This little gadget will simulate a 500 volt or less Geiger tube when driven by a signal generator of sufficient amplitude, typically 5 volts p-p or greater. The purpose is to generate a steady, precise number of counts per minute so that the Geiger counter's meter may be calibrated.
The circuit rectifies the signal generator's output to provide power for the circuit so no battery or power supply is needed. The generator's frequency is divided by 100 so that low counts per minute may be simulated with ordinary bench generators that typically only go down to 10 Hz (600 CPM).
The two input diodes may be just about any schottky type or even older germanium diodes. The generator should be able to supply over 5 volts peak-to-peak. Higher amplitude is fine and will stretch the pulse width slightly. I vary the amplitude to modulate the pulse width when experimenting. The pulse width is also set by the 2.2 nF; increasing the value will increase the pulse width. The width is about 200 uS with the values shown in a typical Geiger counter circuit operating at 400 volts. The 2N3440 was chosen for high breakdown voltage with a 22k resistor connected from base to emitter. The prototype's transistor breaks down at about 600 volts, making this device useful for simulating tubes operating at up to 500 volts. The breakdown voltage was tested at a low current. Since the divider is wired to divide by 100, the counts per minute is equal to the frequency of the generator multiplied by 0.6. Setting the generator to 100 Hz will give 60 counts per minute. To simulate a higher voltage tube, a higher breakdown transistor will be necessary. Make sure to test the transistor's breakdown voltage with the 22k from base to emitter. The breakdown will be higher than with the base open but lower than with the base connected to the emitter directly.
Construction technique isn't critical. The prototype is built into a plastic box with a BNC for the signal generator connection and miniature binding posts for the simulated Geiger tube output. Please don't notice the typo on the label! ('CMP' instead of 'CPM') Note that this circuit is for Geiger counters that ground one end of the Geiger tube (most do). It might work with other types if the Geiger counter case is floating relative to ground.
Enrico Mathesar
It probably isn't the smallest and it isn't really a Geiger counter (no Geiger tube and it doesn't count) but this really small radiation detector flashes the LED every time a particle of sufficient energy strikes the tiny PIN photodiode. The small detector gives about 1 pulse per second with a 2 mR source (using an old fallout shelter Geiger counter and test source as the reference). This sensitivity is enough to determine if a lantern mantle is radioactive or a mineral sample is uranium. Most importantly, its small size makes it inconspicuous! A typical thorium lantern mantle gives about one flash every two seconds. Thoriated welding rods give a blink about every 10 seconds and weak Vaseline glass marble gives only one count every 45 seconds so the detector becomes impractical for the weaker sources. A larger PIN diode is the simplest way to improve the sensitivity.
The circuit is designed to consume virtually no power so that the detector can operate for long periods without a power switch. The total current drain when the LED is not flashing is only 3 uA, giving an operating battery lifetime for the wimpy type 10A (30mA-Hour) of over a year. The random flashing of the LED due to background radiation might shorten that lifetime. A regular 9 volt rectangular battery would be hardly affected.
When particles hit the PIN diode, tiny positive pulses appear on the gate of the 2N4417. These pulses are amplified by the 2N4117 and the MPSA18 and applied to a two-transistor monostable lamp flasher. The 2N4117 is biased by selection of the 1 megohm resistor in the source such that the drain is at least a couple of volts above the source. The drain is DC-coupled to the MPSA18 and it should have a couple of volts from collector to emitter, too. The actual values aren't particularly critical as long as there is a little voltage across the first two transistors; the pulses are pretty small. The 2N4401 is biased by two, 62 megohm resistors connected in series and a 3 meghom to ground so that the flasher circuit is on the verge of flashing. If the 3 megohm is too large, the LED will flash constantly and if the 3 megohm is too small, the circuit will not be sensitive. Other high value resistors may be used but the values should be well above 1 megohm and the ratio should be chosen such that the flasher is easily triggered by the amplified pulses. The flasher doesn't draw significant current when it isn't lighting the LED so the quiescent current is due to the first two amplifier stages and is only about 3 uA.
This "Geiger" counter flashes the LED just like a Geiger tube detector and a crystal earphone connected across the LED and resistor will give clicks just like a Geiger counter but the detection mechanism is significantly different. The main difference of interest is that the PIN diode usually converts the entire particle's energy into a pulse so the pulse height may be used to determine the original energy in the particle. (I should mention that high-energy Gamma rays will plow right through the PIN diode without losing all their energy. The pulse will be big but not necessarily representative of the real energy. Gamma ray detectors often employ large blocks of material to capture the whole particle.) Sophisticated detectors sort the pulses by amplitude and can determine what element produced them but this simple detector simply looks for pulses above a certain level.
Certain precautions must be observed to make a working unit. First of all, the PIN diode is extremely sensitive to light and it must be kept in total darkness. I even had a problem with light coming in through the body of the LED even though it was mounted in a separate compartment! I eventually painted the back of the PIN diode with black "liquid tape" and added an internal light shield. The circuitry is also ridiculously sensitive to electric fields and the circuitry must be shielded. I decided that an old Minox spy camera film box was the container to use for a variety of reasons having nothing to do with good engineering! It is light tight but it provides no electrical shielding. The first step was to glue the PIN diode to one end of the case and to add internal shielding in the form of thin copper circuit board material. Solderable metal foil would also work.
The construction technique could be called the "ship in a bottle" method. No kidding; the construction of this gadget required a lot of dexterity! The battery holder was constructed first, using part of a spring from an AA cell battery holder and a piece of PCB material for the other end. The PCB material forms the positive connection and was made by etching a pad in the middle of the board with an abrasive cut-off wheel and drilling a hole in the middle of the isolated land for the red wire. The solder holding the red wire forms the positive terminal for the battery. Once the battery cavity was made, it appeared that the flasher portion of the circuit might fit in the small cavity directly above. And it did, just barely. This location seemed good since light can come in via the LED body. The PIN photodiode is to the left in the photo above and the FET and MPSA18 are in the same cavity. Not shown is the black "liquid tape" that I dripped on the photo diode and the two wires coming through from the flasher section to block more light. Also, an electric shield was added to completely shield the electronics by adding a metal plate above the electronics with a wire connecting it to ground. This shield is held in place when the lid is installed. The amplifier is so sensitive that the tiny exposed part of the PIN diode picks up electric fields! The unit will not operate properly near large electric fields from TVs, computers, etc.
The circuit illustrates how one can use a PIN diode to get Geiger counter like pulses and a more practical package arrangement will make construction easier! I plan to take this covert detector to Wall-Mart and other stores in search of radioactive lantern mantles, minerals, etc!
Enrico Mathesar
This highly specialized ELF/VLF receiver is for continuous monitoring of ELF signals primarily in the audible frequency range. The receiver is intended for a fixed installation in a location plagued by excessive line-related interference. Many components are selected for the local conditions and an optimized system can be difficult to achieve but the results can be well worth the effort. The unit is mounted on the back wall of a basement laboratory (photo to left). For experimentation purposes, the circuitry is built on a white protoboard permanently mounted in the case (right). The final unit has several variations from the picture and is easily reconfigured for additional experiments.
The circuitry consists of several sub-systems shown in the above block diagram. Each will be described separately.
The first component is the antenna interface and
preamplifier stage. The antenna is a single element about a meter long at the
end of a fairly long cable exhibiting about 2000 pF capacitance. Despite the
high capacitance, the preamplifier is designed with a high impedance to
accommodate shorter cables in the future. The antenna connects directly to a Lumex surge arrestor rated at 230 volts breakdown. This arrestor is paralleled
by a 51 megohm resistor that slowly dissipates any charge on the antenna that
has not reached the Lumex threshold. The antenna is coupled to the preamplifier
FET via a .01 uF, 3000 volt ceramic
capacitor which limits the energy that can reach the preamplifier without
significantly attenuating the desired signals. A NE-2 neon lamp is connected to
ground on the preamplifier side of the capacitor to further suppress voltage
transients. An 18 volt zener diode is connected to the positive supply voltage
to clamp any voltage transients generated by the firing of the Lumex tube,
possibly eliminating the need for the NE-2. The 9 volt reverse bias on the zener
ensures that it has little to no effect on the received signal. This surge
suppressor system survived a severe storm where charges were great enough to
cause the Lumex tube to repetitively fire in a 
relaxation oscillator fashion,
causing quite a squealing buzz. No circuitry was damaged as a result. The FET is
additionally isolated from the surge protection circuitry by a 10 k resistor that was
actually added to reduce the tendency of the FET to demodulate AM radio
stations. The resistor works in combination with the FET capacitance to
effectively roll off frequencies above a few hundred kilohertz. Originally, a
choke was used but it required Q-killing and the simple resistor works as well. A 10 mH choke
and two capacitors near 253 pF were added in series to ground to eliminate 100
kHz interference from Loran-C transmissions. The null is quite deep and
effective but the capacitance must be carefully set (the two parallel mica caps
at the top-left in the photo). The FET is an E113 but many substitutes are possible;
choose one with an IDSS of a few mA at least. The 2.5 k
source resistor was selected to bias the drain to about 5 volts, leaving about 3
volts across the FET. Make sure to select the source resistor to achieve at
least a few volts across the FET and the drain resistor. A good choice of FET
will have only one or two volts on the source. The source resistor is bypassed by a 22 uF
capacitor in series with a 330 ohm to give the stage some gain. The optimum gain
depends entirely on the level of AC hum at the receiving antenna and this gain should be
high enough to drive the AGC amplifier into significant rollback. Otherwise, the
AGC amplifier will be operating at such a high gain that the bandwidth of the
system will be limited by the gain-bandwidth of the AGC op-amp. As a point of
reference, this installation has a 60 Hz signal of about 3 Vp-p on the drain
and, after passing through a bridged-twin-T filter, the voltage is about 400
mV p-p which is large enough to drive the AGC amp into a nice gain region. Yes,
it is a bridged-twin-T in that the twin-T filter shown below is bridged by
a resistor to let a little of the hum through for the digital filter to use for
locking. The amount of gain in the FET and the amount of bridging are somewhat
arbitrary but lower gain in the FET seems best. No bridging resistor would be
used for external processing with a computer filter like Karen's PC Humnuller.
Remember, this circuit is specifically designed to deal with a large line
frequency signal and the AGC will crank the gain "way up" without a large AC
signal present. A simple fix for environments where the hum is low or varies a
lot would be to bridge the photocell in the AGC circuit (below) with a smaller
fixed resistor to keep the gain below some reasonable value (perhaps 4.7 meg
instead of 22 meg).
The twin-T notch is built from old-fashioned mica capacitors and inserted into a small plastic box (simply for convenience). Any capacitor type will work as long as the values are precise. Two, 270 k resistors are placed in series to achieve 540 k and two, 0.005 uF capacitors are paralleled to achieve 10 nF so only one value of each type part is needed. The bridging resistor, about 1 megohm, that lets a little hum pass this filter is located on the protoboard and is not shown in the schematic. The dedicated protoboard is beginning to earn its keep!
The AGC amplifier employs a CA3140 as the gain block with its gain controlled by a resistive type optoisolator. The LED/CdS optoisolator used in the prototype is a Raytheon CK2143 acquired from Charle's vast inventory of old stuff (charles@wenzel.com). Other similar parts or even an LED glued to a CdS photocell will also work. The signal level is determined by rectifying the audio with two ordinary silicon diodes. In this implementation, the diodes are referenced to the power supply and the rectified voltage drops down from there, requiring an op-amp that can "look at" the positive rail, typically a JFET type like the LF356 or some more modern rail-to-rail types. The diodes could be reconfigured to operate near ground, requiring an op-amp that works well with near-zero volts on the inputs. The LF356 compares the rectified voltage to the voltage on pin 3 and slowly increases or decreases the current in the LED to achieve balance. The 4 uF combined with the input 1 megohm give this AGC a very, very slow response and it can take a minute or two for the amplifier to stabilize when power is first applied so be patient. This very slow response prevents impulses from significantly changing the gain which would cause the hum filter to pass a bit of hum for several seconds. The 220 ohm resistor in series with the LED may need to be significantly increased, perhaps to several thousand ohms, if the AGC "hunts" which will appear as a slow increase in gain followed by a sudden decrease in a periodic manner. Increasing the LED resistor reduces the gain of the LED/photocell pair.
The values shown may be varied quite a bit, depending on available components but a few precautions should be observed. Bias the first op-amp (resistors on pin 3) to a voltage that is well within the common-mode input and output voltage range of the amplifier; the output will be swinging several volts. The AGC level is set to achieve about 4 volts p-p since the PIC Humnuller cannot tolerate signals above 5 volts on the input. The following block includes a rail-to-rail amplifier that runs on 5 VDC that further protects the PIC from excessive inputs but you don't want that amplifier to be clipping as a normal part of operation.
The limiting amplifier is simply a rail-to-rail voltage follower running on 5 VDC (TLC2272). The input of the follower comes from the audio output of the AGC amplifier directly above. This follower protects the PIC from excessive input voltage swing. The PIC provides a filtered version of the hum that is input into one side of the differential amplifier with the buffered signal being applied to the other input. The output of the differential amplifier is therefore the desired ELF signal with most of the line-related noise removed.
In the research laboratory, the filtered audio output feeds an FM transmitter so that the signal may be received at various locations throughout the facility without wiring. The 1 uF on the output is intentionally on the low-value side to roll off the tiny residual 60 Hz signal.
The tan board contains Karen's amazing PIC Humnuller. This receiver works great without the PIC Humnuller. Simply don't bridge the bridged-t filter with a resistor (not shown anyway) and use Karen's PC Humnuller program or Wolfgang's Spectrum Lab to filter out line noise. Actually, the TLC2272 could be left out with the AGC amplifier output going directly to the soundcard through a capacitor.
A sample of the filtered audio during an unusually inactive period demonstrates the effectiveness the PIC Humnuller. Half way through this recording, the white wire from the PIC is pulled out, eliminating the filtering. The difference is dramatic as is the horrible level of hum in the air around this facility! Karen's PC filter does work better but this unit does not require a computer. The PIC is easily bypasses by unplugging the wire from the PIC to the differential amplifier. A switch might be added there in the future.
Note that the filter is called a "bridged-T" in the photo but that is incorrect. I'm so used to using the amazing bridged-T filter in RF work that the name stuck in my head!
Direct questions and comments to charles@wenzel.com since the location of Area 50 is top-secret.
