Sometimes it is needed to indicate presence of a RF signal, or to roughly measure/compare its relative strength. For this, a number of simple and less simple field strength meter circuits are floating around the Net.
This circuit was used as the basis of the meter, and enhanced with some additional functionality.
A different Schottky diode than the original-circuit one was chosen, on the basis of availability in local retail stores.
Several problems were encountered during the initial phase of the design. The initial circuit had very poor performance at 2.4 GHz, where the meter was intended to be mostly used. It was determined that the effect is a combination of the capacitance of the chosen Schottky (several pF, way too much) and the likely stray capacitance of the 8 µH coil in the biasing circuit (originally there was a commercial choke used there) caused unacceptable losses.
The diode was changed for the BAT15-03 Schottky, a sensitive detection diode with very low junction capacitance, good up to 12 GHz.
The coil was made from a small ferrite toroid with five evenly spaced loops of wire, to minimize stray capacitance.
The BAT15-03 has however one significant disadvantage; it dies often. When this happens, the meter needle will become unresponsive to even the coarsest zero adjustment attempts, and tend to go to either positive or negative extreme. The skull-and-bones symbol in the schematics indicates which diode is the damage-prone one.
The whole detector is placed in a shielding enclosure cut and soldered from a tin-coated steel sheet salvaged from a can of sweetened condensed milk.
The device input is done via a pair of parallel connectors; one BNC connector for connection of generic signals by generic means, one SMA connector for work in the 2.4 GHz band (most wifi antennas have this connector and many antenna pigtails have it too).
Inside view, RF can opened
RF input stage, amplifier
RF input stage
There are some chokes and capacitors placed in the power lines, mainly to avoid coupling of noise from the other parts of the circuit to the detector-amplifier stage.
After watching the waveforms from the detector on an oscilloscope, it was decided to add an audio amplifier, so the short pulses can be heard. A LM386 amplifier was chosen for this task.
Two connectors, 2.5mm and 3.5mm audio jacks, were chosen as outputs, wired in parallel, in order to facilitate connection of all common headphones and cellphone handsfree earphones.
The 2.5mm stereo connector turned out to be incompatible with the 4-contact 2.5mm jacks used by handsfree units, as the ground pin did not have contact. A 4-pin female connector, with the ground spring in a slightly different position, had to be used. (Alternatively, the headphones can be connected in series, left-to-right.)
There are small chokes in series with the outputs, in order to filter out eventual RF signals that would couple into the headphone leads.
A small speaker was later added to the enclosure in order to facilitate acoustic monitoring without the need of headphones. Originally a small 8 Ω speaker salvaged from an old computer was used, but that turned out to be too quiet. It was replaced with a 32 Ω speaker, which turned out to have satisfying loudness.
An interesting problem was discovered during debugging. When the audio output was turned to high level, the whole receiver was feedbacking wildly. After an evening of chasing ghosts, the culprit was found to be a magnetic field coupling (from the chokes) to the coils in the input stage, through(!) the metal shielding box. The problem was solved by relocating the amplifier board far away from the sensitive part of the circuit.
Output amplifier with LM386
Audio amplifier board location
To facilitate the observation of even short pulses, and to see the energy distribution of the signals (e.g. a high proportion of high energy signals and low energy noise with little to nothing in between indicates a strong signal well above the noise background), the analog needle meter was augmented with a logarithmic bargraph indicator spanning 60 dB, based on a pair of LM3915 chips.
In many cases very short pulses are encountered and have to be reliably indicated. The bargraph will show them, but the human eye will not be able to see them, especially under the conditions of significant ambient light. An additional bargraph with a bank of pulse stretchers was therefore added.
The pulses are tapped from the outputs of the LM3915 chips and stretched using a classical circuit with an RC timing element and a pair of digital drivers. 74HC245 chips were chosen as cheap, compact and easily available banks of drivers, 8 in each. It was found that the output level on the LM3915 is too high when low, and the 245s do not register it as L. A dirty trick was used; the ground pins of the two 245s were lifted by about 0.5-0.7 volts with a series silicon diode. This also lifted the low-high threshold of the chip inputs to the level matching the LM3915 output levels.
To stretch the range of the pulse-stretched bar, the lowest 3 LEDs used pairs of signals to trigger.
The voltage fed to the RC networks is adjustable by a trimpot and sets the length of the pulse stretch. This allows choice between longer, easier seen pulses and shorter, somewhat headache-inducing but better indicating rapidly blinking ones.
To achieve higher contrast, the square red LED sides are painted with two layers of color; white inside, to reflect the light back and achieve higher brightness for no cost, and black outside, to prevent shining into the neighbouring LED.
It was found that the bargraph level adjustment is somewhat challenging. The output signal from the amplifier was referenced to Vcc/2, and going negative. After some only partially successful attempts for a simple solution, it was decided to use an inverting level shifter and a pair of noninverting amplifiers to provide the appropriate levels for the low and high range bargraph chips. A trimpot is provided to adjust the level shifter to proper zero.
There is a switch allowing bargraph coupling for both DC and AC. DC-coupling is useful for precise zero adjustment and for seeing the absolute magnitudes of the signals. AC-coupling takes out the uncertainities introduced by zero drift and shows only the AC component of the detected signals. Only changing signals are shown, high-levels of sustained unmodulated EM - detected as DC offset - won't be displayed well. As the DC and AC levels are different, a trimpot is provided to trim the AC level to match the DC one.
The high-side bargraph half however tended to trip its lower LEDs before the low-side bargraph reached its top. Its zero was therefore made adjustable with a trimpot connected between stabilized low-rail (used as reference for the bargraph low) and the low-voltage input.
The fourth op-amp in the chip was used as a battery-low indicator. It was connected as a comparator between the stabilized voltage line (moving relatively to the ground with the battery voltage) and a reference voltage set by a 1-meg trimpot.
Bargraph module, diodes side
Level adjustment board
Level adjustment board
A standard plastic box was used as the project enclosure. The holes for the meter and the bargraphs were cut with a hot knife. The boards and other elements that weren't screwed in with nuts were affixed with hot glue.
Box with cut out hole for the meter and bargraph
Box with cut out hole for the meter and bargraph
Populated outside panel
Back side of the panel
Lighted up box