This article describes an instrument for measuring common mode current in a ham station HF antenna system.
Common mode current is a component of current flowing on an RF transmission line. It is a component of each line current in the case of two wire open line, and in the case of a coaxial line, it flows entirely and exclusively on the outer surface of the outer conductor.
In both cases, common mode current forms a standing wave on the transmission line, it varies in amplitude and phase along the line.
Because it forms a standing wave, measurement at a single point does not guarantee that a significant current will be observed even though there is significant common mode current, you may just be measuring at or near a current minimum.
Is a current minimum at the shack as good as no current? No, not at
all because associated with a current minimum is a voltage maximum,
very high electric field intensity exists in the region of a voltage
maximum (or current minimum). This is what gives rise to RF burns to
the lips from microphone grilles!
Common mode current can be measured for either coaxial or two wire open line by making a current transformer comprising the transmission line conductors as the primary and a secondary winding with low resistance load and a half wave detector driving a high impedance voltmeter.
A simple half wave detector with Germanium diode was chosen as having excellent sensitivity and least compression of the scale at low input levels.
In practical application, the instrument is likely to be used to assess common mode current in a situation where an common mode choke has been employed. Common mode chokes range in impedance, typically about an average resistance component of around 1000Ω. If such a choke could safely continuously dissipate say 10W due to common mode current, that current would be (P/R)^0.5=(10/1000)^0.5=0.1A RMS.
An instrument designed for Full Scale Deflection (FSD) of 0.1A RMS will have sufficient sensitivity to read upscale on currents that become an issue for typical common mode chokes. In practice, transmitter power would be reduced to minimum, and raised only sufficiently to obtain useful deflection. Common mode current measurements do not need to be made at maximum power out.
An option is offered for an additional switched range of 1A RMS FSD.
The scale on a simple half wave rectifier with capacitor filter is close to linear for input voltages that are much greater than the diode nominal forward voltage drop, but no so at lower voltages.
To achieve the desired sensitivity, this instrument will have a non linear meter scale. The FSD input voltage, source impedance to some extent, and all of the detector components affect the scale shape. The main factors are the FSD input voltage, diode characteristics, capacitor and load resistance.
A SPICE model of the detector is useful in demonstrating circuit
behaviour. Above is the detector circuit after the current transformer
modelled in LTSPICE. The current source is equivalent to 100mA RMS
input current via a 1:10 ratio current transformer, R2 represents a
50µA FSD moving coil meter and sufficient resistance to obtain
full scale deflection with 100mA input to the instrument. In practice,
R2 includes a pot that allows calibration of FSD with 100mA RMS of
Above, the LTSPICE waveforms with current drive of 10mA RMS, shown after the circuit settles. The green line is the RF voltage into the detector, the blue line is the output voltage, and the red line is the diode current.
We might expect the RF voltage across the 100Ω resistor (the green line) to peak at 10e-3*100*2^0.5=1.414V, but it is a little lower at 1.356V due to the loading effect of the detector circuit.
Diodes are often thought of using a very simple idealised model that current flows when they are forward biased, and it doesn't flow when they are reverse biased. The red line shows that at small signals, the behaviour is not so simple.
The voltage drop across the diode is the difference betwen the green line and the blue line. It can be seen that as the green line increases above the blue line, diode current increases slowly at first, but more rapidly once the voltage drop exceeds about 150mV, decreasing again as the voltage drop falls to zero.
Note that the diode current doesn't just fall to zero, but is
actually negative roughly until the green line reaches its negative
peak. This is caused mainly by the diode's self capacitance. The peak
diode current is 550µA, the average (DC) current is 50µA, and the RMS
current is 140µA.
The effect of the current waveform over time is that the C1 acquires charge and reaches an average voltage (there is a little ripple which cannot be seen on the graph) of 1.086V. R2 is adjusted so that the average (DC) current in the meter (part of R2 here) is 50µA.
At lower input voltage, the current waveform is less peaky, and the output voltage is smaller in relation to the input voltage.
The higher range of 1A FSD is almost linear down 0.1A.
A prototype was constructed and output measured for input at 10MHz from 0.1V to 1V, and the DC output recorded.
Above is a block diagram of the detector measurement setup. The computer controls the instruments via the GPIB bus, setting the HP8658B SSG power level and frequency, and measuring the detector output using the DVM option in the HP5328A.
(% of FSD)
(% of FSD)
The table above shows the deflection vs primary current for the 100mA range. The 1A range is taken to be linear.
The prototype was calibrated by measuring the current with a 7MHz transmitter adjusted to 50W and 0.5W (using a 20dB power attenuator) into a 50Ω dummy load.
In the prototype, the DC voltage developed in the load at 1A is 12.2V, and at 0.1A is
0.94V. This requires nominal load resistance of 244k and 19.2k
respectively. In the prototype, the high range uses a 220kΩ fixed in series with
a 50kΩ pot and the low range a 15kΩ fixed in series with a 5kΩ pot, with a
single pole double throw switch to select high/low range.
Above is a plot of the frequency response of the low range detector over HF.
Above is an illustration of the scaling, it is scale plate artwork
for a common MU-45 size 50µA meter. The artwork is available for download. (You may notice from
the artwork that it was created with Tonne Sofware's Meter program.
This is the ONLY piece of software that I have ever purchased that
cluttered the printed output with its own promotional message, so to
give balance, for that and many other reasons, it is without doubt the
worst piece of software for which I have ever paid money, I do not
Above, the snap on suppression sleeve. Dimension B of the 0443164151 is 13.5mm.
Above, the copper side of the veroboard. Eagle eyes will see there is a track broken where it should not have been and is bridged with the pigtail of a resistor... don't copy my mistake!
Above, the meter interior. The 100Ω resistor used is a carbon film 1W rated device, so the meter's continuous current rating is I=(P/R)^0.5/n=(1/100)^0.5/0.1=1A.
Above, the meter ready for use.
The prototype used a commercial snap on balun. Though the hinge is likely to fail with repeated use, it provides a good prototype instrument. The core is fixed to the plastic box using double sided foam tape, and then the secondary winding of ten turns of 0.5mm ECW wound tightly around both the core unit and plastic case via two 3mm diameter holes.
Whilst the meter would usually be supported by the cable being measured in practice, provision is made for hand holding the instrument at the end opposite the transformer where there is no internal circuitry, and isolation will be quite good.
Although the meter only measures the magnitude of current, by making three measurements, the magnitude of the differential and common mode components and the magnitude of their phase relationships can be determined.
Note that I1-2 must not be greater than I1+I2, if it is, then the measurements are in error.
The above form allows calculation of Ic (the common mode component of current), Idc (the differential mode component of current), θdc (the magnitude of phase difference of Id wrt Ic, θ1-2 (the magnitude of phase difference of I2 wrt I1). (The instrument does of course directly read Ic as I1-2.)
Note that external fields are principally due to 2Ic (or measured I1-2).
The calculator tries to use the number entered, negative numbers will be changed to positive, missing data 0, NaN means Not A Number (correct the input if needed), results will not be calculated when I1-2 > I1+I2.
The low meter range is the most useful range, as meaningful readings can be made at relatively low power. Keep in mind that Ic=0.1A flowing in each conductor in a common mode choke with an R component of Z (the effective RF resistance of the choke) of say 1000Ω means (2*0.1)^2*1000=40W of dissipation in the choke.
A correspondent described failure of a balun on his G5RV though arcing at 800W output on 40m.
Above left is a section of the balun winding that had arced and burned the covering PTFE insulation.
The balun is a 1:1 Guanella balun, "W2FMI type" rated at 5kW continuous, 10kW peak. The conductors were specified by the manufacturer as 200°C magnet wire, and they appear to have been wrapped with PTFE tape.
A G5RV on 40m will drive a VSWR of about 10:1 on 400Ω line, so at 800W for example, the worst case differential voltage will be (2*800*400*10)=2500Vpk. Evidently the balun insulation was not sufficient to withstand the applied differential operating voltage at 800W in this case, less than 10% of the manufacturer's peak power rating.
To better understand what was happening, the correspondent replaced the faulty winding with one of a twisted pair of PTFE insulated flexible wires, wound in exactly the same configuration as the original. Fig 11 shows the rewound balun.
Having rewound the balun, it now withstood the operating voltage without flashoveer, but the correspondent experienced serious overheating within a minute using just 800W tx output on a G5RV at 40m. Some other bands had the same problem, some not.
He constructed a similar clip on current meter, and measured I1=0.4A, I2=0.39A and I1-2=0.1A in the open wire feeder adjacent to the balun at 50W tx output.
Using the calculator, Ic is just 12% of Id which would be regarded by many as insignificant, yet the balun had been destroyed and needed to be rewound, and even after rewind (which had not solved the underlying problem), behaved badly at more than about 50W (notwithstanding that it was rated at 5kW).
If the R component of Z of the balun was say 1000Ω at 7MHz, then the power dissipated in the core would be 0.1^2*1000=10W, or 20% of the 50W tx output. Ten Watts might not seem much, but when the transmitter was operated at 800W, that would be 160W which increased the core to its Curie temperature withing seconds, rendering it ineffective as a choke and damaging insulation materials.
Total common mode current of 0.1A might not seem much, but it is enough current to drive very high losses in a medium impedance common mode choke (balun). The answer is to reduce the current further, even if increasing the choke R, until I^2R is within the choke's acceptable dissipation.
Some messages to take away are:
A better winding configuration is for this balun is the W1JR "cross over' configuration as shown in Fig 12. The winding provides for input and output on opposide sides of the core, whilst spreading the turns out to minimise self capacitance.
There have been a number of designs that purport to indicate transmission line current balance. Current balance means that the line currents at a point are of exactly equal magnitude AND exactly opposite phase.
Two examples of flawed designs that compare magnitude but not phase follow.
Such instruments reliably show unbalance, but they do not reliably show balance (equal magnitude currents is not sufficient to prove balance). The first of these links is to a product current in MFJ's catalogue at Jul 2011, and the second is to an article online at the same date and appearing in print in AR magazine in August 2009.
Try entering 1, 1 and 2 A in the calculator above. Both line currents are equal in magnitude at 1A, but the common mode current is 2A and this represents the worst imbalance, even though both line currents have equal magnitude. Both instruments mentioned would indicate perfect balance.
Plenty of hams have been tricked by these meters and have conducted quite detailed experiments which unfortunately have produced invalid results.
Another approach described by WZ5Q uses an oscilloscope to compare the amplitude and phase of the voltage on each wire of the two wire feed line at a point. He then tinkers with the tuner to obtain equal magnitude but opposite phase voltages. This procedure is also endorsed by WW8J. The procedure will only assure equal but opposite currents if the load is perfect symmetric, which is unlikely. Had he used current probes and adjusted for equal but opposite currents, he would have been assured of very low feed line common mode current and its attendant problems.
(Lofgren 1996) offers the following method.
For checking this or any other tuner, a couple of simple test devices are worthwhile. One is an "antenna simulator." This device is especially useful for making comparisons between various tuners (or tuner/balun combinations). The simulator consists of a pair of resistors of equal value whose total resistance approximates the expected feedpoint impedance at the tuner. (Several pair allow checking a range of impedances.) Connect the two resistors in series across the balanced output terminals on the tuner, and connect the junction of the two to the ground lug on the tuner. This *roughly* simulates an antenna system consisting of a balanced horizontal antenna over ground (such as a center-fed zepp or G5RV) and its feed line. To check the output balance of the tuner looking into the resulting feedpoint impedance, use an RF probe to measure the voltage drop across each resistor to ground while feeding a small amount of RF into the tuner (having adjusted it for a match). If the currents on each side are equal, the voltage drops across the resistors will be equal.
This method compares the magnitude of the voltage on each output terminal. In the test described with a symmetric load, the measure implies the magnitude of the currents, but does not measure phase and so does not assure the condition of equal magnitude and opposite phase.
(Butler 1996) offers measurements citing Lofgren's method, but they are flawed because the method is flawed. (Butler 2009) is based on the flawed method, (Butler 2013) adds and subtracts the RF current samples to properly indicated differential and common mode current.
(Duffy 2009) describes a simpler form to the design used by (Butler 2013).
There is a practice amongst hams to compare at one point on a line, the magnitude of common mode current to the magnitude of the differential component (or in some cases the current in one conductor), often giving the figure in dB. Some pundits suggest that -10dB is goodness.
This has very limited value, both common mode and differential current may be standing waves (common mode almost always is a standing wave), so single point measurements are of very limited value, and the differential current is not a meaningful reference.
The impact of common mode current will most often be best represented as the total current moment, ie the integral of common mode current over the length of the feed line, and the most useful reference would the the integral of radiator current over the length or the radiator. This gives a better indicator of the extent of feed line radiation relative to the nominal radiator in the system.
So, in cases where an antenna is fed with a very long feed line, the acceptable level of common mode current may be much smaller given the effect of it flowing over the long feed line. So for example, if a 40m long dipole was fed with 400m of feed line, the common mode feed line current would want to be much less than one tenth of the dipole current (at which there might be almost equal radiation from feed line as dipole), perhaps on hundredth (ie-40dB) would be a good target for this scenario!
If you choose to buy a ready to use product, be wary of products that might not actually work. There is such widespread misconception of the subject that hams are quite likely to recommend products that do not work.
It is possible that commercial products exist that do work. MFJ have at least two instruments that purport to measure common mode current, the MFJ-835 does not work. The MFJ-854 might work, but the author has not evaluated it so is unable to recommend it.
Some readers have suggested that Schottky diodes are more suited to this application than the Germanium diode used, and others suggesting a voltage doubler circuit is better. This suggestions have been made without any supporting evidence... other than that is what everyone else does.
No simple rectifier circuit is going to result in a linear meter scale at low current, there are no ideal diodes.
Fig 12 shows the DC small signal characteristics of the 1N34 diode used in this article, a 1N5711 Schottky diode and a 1N4148 Silicon diode. None are ideal, none have zero threshold, and none have zero dynamic resistance. Note that in this application circuit, the instantaneous peak diode current is less than 0.6mA (Fig 2) at FSD on the 100mA range, even less at the lower end of the scale.
The least suited to RF low voltage detection is the Silicon diode because of its higher knee voltage.
The Schottky diode also has a higher knee than the Germanium diode at very low current though it does have slightly lower dynamic resistance, but overall it is less suited to this circuit.
A voltage doubler circuit gives higher DC output voltage, but at the expense of greater sag with current (ie its internal resistance is higher). There is plenty of voltage to drive the meter using the Germanium diode in a half wave peak rectifier.
|1.02||12/10/2013||Added further section on small signal diodes.|
|1.03||18/01/2015||Moved to owenduffy.net.|
© Copyright: Owen Duffy 1995, 2017. All rights reserved. Disclaimer.