• Teensy++ 2.0 (Arduino compatible) Microcontroller board (www.pjrc.com or www.sparkfun.com)
• 20x4 LCD (any HD44780 compatible 20x4 LCD will work, $10 on ebay)
• Rotary Encoder with a builtin pushbutton (Panasonic EVQ-WTEF2515B, eBay)
• Detector circuit using a tandem match coupler and 2x AD8307 logarithmic amplifiers

Some display examples:

A20x4 LCD can be coerced into displaying four simultaneous bargraphs, each showing what effectively amounts up to100 bars.  This is achieved by turning each character position of the LCD display into six bars (one in every six bars is not visible).

Thisdisplay shows a continous carrier, at 103W:

Asticky bar can be implemented by showing a long and fat bar in themiddle of a character position which is not being used, thus showingan approximation of the highest level within a period of say, 1, 2.5or 5 seconds. This sticky bar is then of course with a much coarserresolution, or only 20 bars in total, instead of 100. Thesticky bar is only visible if that particular character position isnot being used to show the “regular” bars. This methodologyworks surprisingly well.

Thisdisplay shows a SSB voice signal. The highest power seen within theselected PEP period (1, 2.5 or 5 seconds) is 84W. The sticky barshows the PEP value. The highest power within the last 100milliseconds is 8.5W, as also indicated by the bargraph. Otheravailable displays can be selected with the rotary encoder, to depict average power and instantaneous power:

Thisdisplay shows 99W being transmitted into a very badly alignedantenna. The transmitter sees a good load, as it is in front of atuner that has been aligned for a good match. The meter ispositioned between the tuner and the antenna:

Below is a different display selected, indicating the exact same situation. Now we see that the forward poweris 211W and the reflected power is 112W, i.e. we have 99W out(apparently the SWR display was flickering between 6.4 and 6.6). Onthis display the bargraphs have been compacted down to 14 characterpositions each. This display has two power graphs for forward andreflected power, no sticky bar and no SWR graph:

Detector Circuit and Tandem Match Coupler:

(click on schematic to enlarge)

Here is an X-ray view of the PCB layout for the detectorcircuit.  Please refer to the bottom of this webpage to download an accurate size PCBlayout for the detector circuit:

Here is a picture of the detector board populated withcomponents for the 10 bit ADC option and hooked up for testing:

Below, the same detector board has been populated withcomponents for the 12 bit ADC option.  The main drawback of this option is the miniscule size of theAD7991 ADC chip (see the middle of the printed circuit board).  It takes a very good iron and a steady handto solder:

The detector circuit uses all surface mount components.  R1 – R6 are SMD1210, 1/2W.  Other resistors and capacitors are SMD0805.  Whether to include the AD7991 and itsassociated components or not, depends on which option is chosen. 

You may notice the lack of any slope adjustment circuitryaround the AD8307.  If necessary, anymismatch between the two AD8307 detectors can be automatically adjusted out bythe microcontroller firmware, using two separate signal levels, first in theforward direction and then also in the reverse direction.  Hence, no additional circuitry is requiredaround the AD8307 offset and output pins. See  Menu Option “Calibrate”,under the section on Meter Displays andConfiguration Menus.

A very comprehensive description of what a Tandem Matchcoupler does can be found here:

http://www.giangrandi.ch/electronics/tandemmatch/tandemmatch.shtml

I have no preferred way to build the Tandem Match directionalcoupler.  The main issue is a goodbalance and isolation between the two transformers.  However, in the end, it comes down toaccuracy and luck; how similarly you wind the ferrite cores, and how identicalthe cores themselves are. My own builds were only marginally successful at achievingisolation approaching 30 dB at 28 MHz and above.  There are many successful build descriptionsavailable  (ARRL Antenna Book,internet…).

Printed circuit board version of a Tandem Match Coupler:

Another version, this one built free style into a shieldedenclosure:

A third version, a combined Tandem Match Coupler and DetectorCircuit (10 bit ADC):

Circuit Description (Tandem Match and Detector):

The Tandem Match directional coupler consists of 2x 30:1transformers.  The ferrite cores I usedare Laird 28B0375-400 (Digikey 240-2296-ND). These cores seem to handle power levels of up to 500W (200W output hasbeen tested with high levels of SWR).  Other Tandem Match Coupler implementationshave successfully used cores such as T50-3, T68-2, FT114-61 etc.

The 50 ohm terminations for the coupler consist of 3x 150 ohm1/2W resistors in parallel.  The thirdresistor is actually two resistors in series acting as a voltage divider, 127 (1/2W)and 24 ohms  (1/4W).  Actually the 24 ohm resistor is in parallelwith 1100 ohm, which is the internal resistance of the AD8307.

30:1 gives a coupling of 1/30, or [20log(1/30)] -29.5dB

The voltage division gives a further attenuation of[20log(23.5/150.5)] -16.1dB

The maximum input power specified for AD8307 is 17dBm.  Then the maximum input power the meter canhandle is:

17dBm + 16.1dB + 29.5 dB = 62.6 dBm or 32.6dBW, which is1800W.

On the other hand, there is a trade off.  We are using a voltage reference of 2.56V or2.6V.  While this adds the benefit of asimpler circuit and higher measurement resolution (0.1 dB when using 10 bit ADC,0.025 dB when using 12 bit ADC), it caps or limits the max AD8307 input toapproximately 15.5 dBm, rather than the 17dBm as specified.  A very acceptable trade off for the highermeasurement resolution. 

In other words, the highest possible input power reading whenusing 30:1 transformers and a 127 ohm + 23.5 ohm (24 ohm in parallel with theinternal 1100 ohm in the AD8307) voltage divider is just over 1kW (1300W). 

To be able to measure power levels above 500W (or 200W inreality, with high SWR levels) you may need to use larger cores for the TandemMatch coupler. 

Using T68-2 or T68-3 cores and a 40:1 ratio, as described inthe ARRL Antenna Book, or perhaps better, FT114-61, will make the metersuitable for power levels up to 2kW, without any modification required of theresistors R1 through R8.

I like using surface mount components.  This is my personal choice, it is of little orno relevance at short wave frequencies, the 10 bit version of the detectorcircuit could of course also be built using through hole components.

Informative links to two other similar power meterimplementations, including 2x AD8307 and Tandem Match directional couplers:

http://www.arrl.org/files/file/Product%20Notes/2012%20Handbook/KAUNE.pdf

http://www.telepostinc.com/Files/phipps-1.pdf

The Controller circuit:

Teensy2++ from PJRC.com (Atmel AT90USB1286 microcontroller running at 16MHz).  The cheaper Teensy 2.0 cannot be used, as it doesn't have enough Flash ROM and RAM to support the firmware.  The current version of the firmware uses approximately 40 kB of Flash, 7 kB of RAM and 50 bytes of EEPROM.

Acheap 15PPR mechanical rotary encoder with 30 detent and a built-inpush switch (Panasonic EVQ-WTEF2515B) , available from severalsources on eBay.  Note that the 5V connection shown on the schematic is not used with the Panasonic encoder.

20x4LCD (HD44780 compatible), available from multiple sources on eBay

Here is the version using the internal 10 bit ADC:

This version makes use of the internal2.56V reference available in the AT90USB1286 microcontroller.  This gives 2.56V/1024 = 2.5mV per bit, whichequals 0.1 dB measurement resolution. (example power levelresolution:  ... 9.55W, 9.77W,10.0W, 10.2W … … 95.5W, 97.7W, 100W, 102W …)

Here is an alternate version tobe used with the AD7991 12 bit ADC:

The AD7991 version of thedetector circuit uses an external 2.6V reference for the 12 bit ADCs. Thisgives 2.6V/4096 = 0.635mV per bit, which equals 0.025 dB measurement resolution.

Note that two versions of AD7991 areavailable, AD7991-0 and AD7991-1.  Theonly difference between the two versions is the I2C address of the chip.  The firmware automatically senses which chipis being used.

Red LED for SWR alarm, green LEDto indicate whether serial output is active. 

Optional:  A resistor, a NPN small signal transistor (BC547or similar) and a relay to break the PTT path in case the SWR alarm has beenactivated.

If SWR alarm has been activated,then deactivate by a short push of the pushbutton.

The Section Meter Displays and Configuration Menus gives a detailed overview ofthe firmware functionality.

Both the 10 bit and the 12 bitversions of the meter are powered by 5 volts through the USB port, either usinga 5V USB wall wart , or directly from a Computer USB port.  Total current draw is approximately 200mA,including LCD and the detector circuit. If powered by Computer USB port, then the meter can also providemeasurement data to the computer via the USB port.

Here is a picture of an earlyversion of the Power/SWR Meter.  Thisversion used an external combined detector+Tandem Match Coupler circuit.  The meter is connected toa signal generator, approx 50mW input, 25 ohms (2x 50 ohms in parallel) as aload gives an SWR of 1.98 (should be 2.0):

Here is a picture showing theinnards of the controller (not much here). In the latest upgrade, the detector circuit has been included in the box:

Meter displays and configuration menus:

Serial output (USB virtual serial port):

Thecommand $ppollproduces two numbers, Power and SWR:

1234.12345678,1.02

Poweris in watts, lowest readable power level is 10nW.

$pinst    produces a human readable message, instantaneous power (nW-kW) and SWR

$ppk      produces a human readable message, peak (100ms) power (nW-kW) and SWR

$pinstdb  produces a human readable message, instantaneous power in decibel over milliwatt (dBm) and SWR

$ppkdb    produces a human readable message, peak (100ms) power in decibel over milliwatt (dBm) and SWR

$plong    produces a human readable message, example:

The command $pcont will produce the output of one of the above ($ppoll, $pinst, $ppk, $ppep, $pavg, $pinstdb, $ppkdb, $ppepdb, $pavgdb or $plong) continuously, 10 times per second.

For example the commands:       

will produce continuous reporting of the instantaneous power and SWR.

The basic functionality of the power measurement routines isto read the voltage output of the two logarithmic amplifiers (both forward and reverse direction)and calculate the resulting power levels in dB-milliwatt.  To do this, the meter needs two accuratesignal levels, 30 to 40 dB apart, as calibration points.  The calibration should be performed with aslow SWR as possible, using a 50 ohm dummy load.

The Calibration Menu requires the signal levels to be inputin decibel referenced against 1 milliwatt, or dBm.  To convert Watts to dBm, you can use thefollowing formula:

If the Tandem Match Coupler transformers are made with a 30 to 1ratio, then the meter should be accurate to within +/- 1 dB or so, using thepre-set calibration levels provided in the firmware. However, the calibrationprocedure can accommodate any other Tandem Match transformer ratios, and a calibrationagainst an accurate reference will result in an almost equally accurate meter.

Calibrating in the forward direction sets the calibrationpoints for both the forward and the reverse AD8307.

Calibrating in the reverse direction (switching the input andthe output) will only set the calibration point for the reverse AD8307.  Therefore, a forward direction calibrationshould always be done first.  A reversecalibration is useful to compensate for any slight variations between the twoAD8307.  The calibration routine automaticallysenses whether power is going in the forward or the reverse direction.

The single level calibration is sufficient for mostpurposes.  The first level is set byregistering the actual power measured using the dB-level dialed in with theRotary Encoder as a reference (default 40dBm = 10.0W, can be any level between10.0 and 53.0 dBm).  The second level isset at 30 dB lower, assuming that the AD8307 slope gradient is  25 mV per dB. The AD8307 datasheet indicates that the slope gradient is between 24 and26 mV per dB.

The first level calibration is normally not required, use thesingle level calibration instead.