From: Scott Stephens [SMTP:stephens-at-enteract-dot-com]
Sent: Monday, January 26, 1998 5:36 AM
To: Tesla List
Subject: Re: Voltage/Length EM-PROBES(fwd)
This is a file I found on a web site, I think.
Subject: DIY probes for checking E & M fields (long)
Organization: University of Waterloo, Waterloo, Ontario, Canada
Here's some info that I promised a while ago, about making
do-it-yourself E & M probes for general purpose use. Please check these
'plans' over fairly carefully if you are going to try building them. I
may have mistakes in this document, I don't want you to pull your hair
out trying to do something impossible! Send me any suggestions,
corrections, or even flames if you think these designs stink. They
aren't sophisticated designs!
You'll need a reasonable knowledge of electronics to design these
probes, and you need a signal generator and a 'scope. I don't have any
plans from which you can make up a good working probe from scratch!
Magnetic field probes, ELF (extremely low frequency)
These are the ones you would use for measuring 60 Hz line problems,
the field from the vertical drive coils on TV's, monitors, etc. You
should be able to get the coil to cover a response from 30 Hz to at least
a few KHz. If you are trying for really unusual response, you should be
able to extrapolate from my crude design.
The first thing you need is the pickup coil. It should have an air
core, and should be wound on a non conductive form. If you are using
high impedance inputs (1 megohm or greater), this coil will need to be a
few henries inductance. I have used a form that is about 1/2 square, and
wound about 2000 turns of #38 or #40 AWG insulated wire to get the
inductance I need. I made a crude coil winder out of a DC motor and some
mechanical junk, and ran it with a variable DC supply. An alternative to
this misery is to get one of those interchangeable relay coils, get one
for a 220 v or even better, 550 v small AC relays. These coils fit over
the steel core of the relay. I usually can find them in the electronic
'junk' stores. You shouldn't use ones with a steel core, although I'm
beginning to wonder if you can get away with it. The problem is the eddy
current loss at the higher frequencies - it will screw up the frequency
Now you've got your coil, the next thing to do is to flatten the
frequency response, and still get enough sensitivity. The coils' output
is proportional to the frequency of the magnetic field - that's not what
we want. My trick is to use the coils' own self inductance to flatten
out its response. By using a low value of resistor load, you'll get a
6db/octave drop that exact cancels out the coils 6db/octave increase.
You need to know the lowest frequency that you'll analyze.
Let's say your coil is 1 henry, and the lowest frequency you want to
look at is 60 Hz (the -3db point). You might find that a bit loose, so
if you want response to within 5% at 60 hz (-.5db), the -3db point would
be at about 20Hz. At 20 Hz, 1 henry has an impedance of about 100 ohms.
So you use 100 ohms as the load resistor! This probe is going to be
very insensitive. Maybe we'd better use 5-10 henries, and increase the
load resistor. If you use too fine a wire, the internal resistance of
the coil will cause attenuation. For reasonable sensitivity, the
inductance you need is defined by the lowest frequency that you want to
measure. The lower the frequency, the higher the inductance, and
preferably not with a series resistance much more than half or quarter
of the load resistor. Large and heavy coils might be necessary. This
is not a very elegant solution, but it works in a pinch, and is fast and
A better way is to put the coil onto the input of a low noise opamp,
and use a capacitor in parallel with resistor in the feedback loop. This
will give an integrating response to balance out the coils
differentiation output (Vcoil=K*df/dt,where f is the magnetic field). At
very low frequencies, the circuit will have very large gains. You have
to arrange the circuit to determine the maximum DC gain, and the point
where the op-amp no longer integrates. Pick an op-amp that does not have
severe 1/f noise, or 'popcorn' noise. The coil must be terminated with a
resistor, or there will be a peak in the response due to the resonance
within the coil of the inductance and stray capacitance.
To determine the best value of this 'damping resistor', put the coil
in series with a signal generator, and the damping resistor(pick an
arbitrary value). Measure the output across the damping resistor as a
function of frequency. At some point the response should peak, then
gradually get less. Lower the resistor to eliminate the peak, or
increase the value to get more output. If you use a square wave signal,
adjust the output for the best damped response.
| | | |
| | | |
R1 R2 R3 | C3 |
( | | | | | \ |
( | > | point A | >----------out
( | < | ____| /
( ----- > ----- | |/
L1 ( C1 ----- < R5 C2 ----- |
( | > | |
| | | | |
Here's a practical circuit. This is a low frequency pickup. There are
a few more components than discussed, that's because it also has a high
frequency coil that's switched in (not shown), and I wanted a reasonable
steep rolloff, so that the low frequency coil wouldn't pick up high
frequency stuff and the HF coil wouldn't pick up LF stuff.
L1=3.3 Henries, 4000 turns on a 1/2 in coil form
R1=1035 ohms (DC resistance of coil)
C1= 5pf(stray capacitance of coil) + 220 pf silvered mica capacitor
R5= 220 kohms (damping resistor)
R2= 100 KOhms (it and C2 form a low-pass filter)
C2= 820 pf silvered mica
R3= 100 Kohms (isolates low pass filter from feedback loop)
R4= 20 Megohms (determines maximum gain, about 200)
C3= 470 pf (polystyrene) determines sensitivity and lower freq.
The opamp is an LF356, there are others that are better for low
frequencies. Since this one was used for high frequencies (up 500KHz) as
well, this opamp seemed to be the best compromise.
The coil had a better high frequency response than I wanted, so
that's why I added C1. With L1,C1, and C2 I got an effective 3rd order
C3 is frequency compensating capacitor, it flattens out the rising
frequency response of the coil. R3 and C4 determine the lower -3db point
of the pickup.
Sometimes I put a fairly large capacitor in series with R3. I pick a
value whose reactance is R1+R2+R3 at a low frequency, maybe half of the
designed -3db point. I do this to reduce the wild signals you get when
you move the probe around, and the coil cuts the earth's magnetic field.
It can be annoying if you happen to have a maximum signal pickup, or
'track and hold' circuit, or you are scanning with the probe to locate a
field source such as a buried cable.
There is a high frequency limit to these probes, it's determined by
the wiring capacitance. You will find that there is a point where the
coil resonates, (especially if the damping resistor is too high) and the
output peaks, then drops rapidly. Poor winding techniques (not
increasing the diameter in layers when winding coil) and close proximity
to the shield will lower the resonant frequency.
Physically, these coils must be shielded from electric fields. They
will pick them up otherwise. When you wrap or enclose a pickup coil with
a metal case or conducting tape there are a few things to watch out for,
especially as the frequencies increase. The shield must NOT act as a
shorted turn, or else the pickup's calibration will be off. This is easy
to arrange, when the turns of foil or metal overlap, they must be
insulated from each other to prevent cirulating currents.
Higher frequency coils (several KHz to several MHz)
Higher frequency probes (from 10 KHz to 1-5 MHz) are not much
different to build, except don't need anywhere near the number of
______________| |__ to point A (remove R3 and stuff to right)
( | | | | of above circuit (lo-freq version)
( | >
( | < C10
( ----- >
L10 ( ----- < R10
( |C11 >
( | |
The above is the high frequency (10KHZ-1 MHz) version of the probe.
L10 is 81uH, 60 turns
R10 is 1500 ohms
C10 is .033 uf
C11 can be anywhere from a few pf's to several hundred, depending on
C10 is used to limit the response at the low frequency end, so that it
doesn't overlap the high end of the low frequency probe.
The high frequency probe must have very little stray capacitance if
you want to get the highest possible frequency response. The amplifier
would need to be very close to this coil. If you can tolerate (or want)
a lower top end, then there will be a capacitance to ground as a result
of shielded lines, or added capacitance. That capacitor is C11.
As was seen in the low frequency probe, R10 is used to dampen or
flatten the response of the coil, when the circuit gets to resonance
(L10 and C11).
The characteristics of the amplifier have have a large bearing on the
preformance of this circuit. Noise can be a problem because of the
larger bandwidths. The faster amps can be fussy about layout, some of
the FET, MOS or CMOS will oscillate at the drop of a hat.
High frequency Magnetic field probes (>10MHz)
For these probes, the mechanical construction is important. I use
anywhere from 1 to 20 turns of wire inside a brass or copper tube, and
run them into a twin-axial RF plug. In a twin-axial receptacle I mount a
balun to go from differential to single-ended output, and a BNC output
Take a piece of brass or copper tubing, shape it into almost a
complete circle (with no kinks). Cut it in half, so you have 2 almost
180 degree sections. Clean off the burrs. Join (leave a 1/8-1/16 in
space between the sections) the two pieces with fairly rigid heat shrink
tubing. Cut two holes or notches in a piece of short larger tubing or
the end of the twin-axial plug, and solder the cicular tube so that the
openings lead into the twin-axial plug or large tube.
Now you have a hollow circular tube, that cannot induce currents in
the loop, because you have an insulated break in it. Feed the wire into
and around the tubular loop (this can be a real chore). The two ends of
the wire are soldered onto the twin-ax pins.
----- piece of heatshrink to align
.--------- --------. cut sections and
/ .--------------. \ insulate them
/ / .----- ----. \ \
/ / / ----- \ \ \
/ / / \ \ - inside turns of wire
/ / / \ \ \
\ \ \ / / /
\ \ \ _________ / / / - brass or copper tube
\ \ \ | |/ / /
\ \ `| | / /
\ `------, ,-----' /
`----| | | |---'
| | | |
| | | | - piece of brass tubing
| | | |
|| | | ||
| | | | - twin-ax RF connector
\ 2 pins of twin-ax connector
Note that the round piece of tubing is broken at the top, to
eliminate circulating currents. The piece of heatshrink aligns the
tubing, and maintains a fixed gap.
For measurements up to 100-200 MHz I used 1 turn of the inner
conductor of some foam dielectric coax cable. The reason for this was to
space the inside wire away from the walls of the tube, and to minimize
capacitance (foam dielectric has considerably less capacitance than the
solid dielectric). For measurements from 2-20 MHz I used 20 turns of
wire-wrap wire inside the circular tube. It was good, since the
insulation is tough, and you can push it fairly readily around the loop.
The next thing to do, is to frequency compensate the coils, using
their own self inductance, or in stubborn cases using added capacitance.
The output of this device is normally terminated in 50 ohms (for most RF
measuring devices). You may have to fiddle around with the matching and
compensation networks to get a 50 ohm output, flat response and no
resonant peaks. These are done AFTER the balun so you have a single
ended signal. Before the balun you can do tricks like putting both wires
through some ferrite beads to reduce common mode signals (which are
obtained by wiring inconsistancies and external electric fields).
Here's a letter from Josef Bergervoet, he's got another design.
(I've edited this letter extensively).
[Part of letter.......]
You could use a simpler set-up, where the coil is at the same time
a balun transformer. This can be made entirely from coaxial cable:
Slit in outer conductor
Left side is connected .--------. --------.
all-together to the / .------|-------. \
inner conductor of / / .----' ----. \ \
the right side / / / \ \ \
/ / / \ \ - inner conductor of semi-rigid
/ / / \ \ \
\ \ \ / / /
\ \ \ / / / - shield of semi-rigid cable
\ \ \ / / /
\ \ `---,------' / /
\ `-----| ,-----' /
`-------| | ,----'
| | |
| | |
| | |
| | |
| | |
| | | - coax RF connector
If you want a frequency range as large as possible for a given loop size,
you could use a double slit antenna, as described in:
J.R. Bergervoet, H. van Veen, A large-loop antenna for
magnetic field measurements, Proceedings of the 8th
International Zurich Symposium on EMC (1989) 6B2.
[part of a reply to some questions from me......]
> Thank you for your reply and additional information. In your letter:
> This looks like a reasonable way to 'kill 2 birds with 1 stone'. Do
>you have any idea how well it rejects E-fields in close proximity to the
>source? I can see the balanced construction, but I'm wondering just how
>well it would work as an magnetic field probe when only 5-30 cm's from
>the source of radiation (with both E & M fields). I would expect unequal
>currents to be capacitively induced in each half of the loop when there
>is a high electric field sideways (not broadside) to the probe.
Unbalanced capacitive currents are the problem with any shielded
loop antenna. Also if you separate the function of the shield
and the loop-winding! The reason is that the shield itself also
is a winding and has a coupling factor of almost unity to the
actual receiver winding.
So why do we use an electrical shielding at all?
1) Because without it, any capacitive current means trouble. With
a shield, only the unbalanbced capacitive currents do.
2) If the number of turns of the pck-up turn is more than one, the
high coupling factor, mentioned above, still means a down
transformation of current. Without the shield, the full capacitive
current would enter the windings.
In conclusion: For a loop with only one winding, the shield can just
as well be exploited to construct a balun transformer.
[end of letter.....]
Directionality of probes
Both probe designs as shown above are directional, they behave as a
dipole antenna in such a way that the maximum sensitivity is broadside
to the turns of wire.
In order to obtain a non-directional (isotropic) response, you need to
have three probe coils, each 90 degrees to each other. The output is
amplified and linearly rectified and filtered. Then the three outputs
are put into a squaring circuit, so that you get M1^2 + M2^2 +M3^2 where
M1,M2,M3 are the three outputs from the orthogonally oriented coils. You
then take the square root of the sums of squares (easily done by putting
a non-linear scale on the display meter that coreesponds to square
A method that is used for vhf-uhf-lower microwave E-field measurement
can be used here too. They keep the voltage levels down and run the AC
signals into detectors (rectifiers) that are operated in the square-law
region. That means you can't exceed 30-100 mv input voltage. The outputs
will be DC level proportional to the voltage squared. This would be a
substitute for high frequency multiplier circuits. Take the outputs of
the 3 orthogonal coils, and sum them together. You can square root
electronically, or by using a non-linear meter scale.
If the magnetic field vector is pointing in the same direction all the
time, the above approach should work OK. If not, then you can't rectify
and filter, you must go straight into a squaring circuit. The catch is,
this squaring circuit must be accurate up the the highest frequency of
the probe. Good choices for this applications are analog multipliers,
the signal is fed into both the X and Y inputs. As an example, the MC1594
will give reasonable accuracy up to about 500KHz, the MC1496 will get
you out to 5 or 10 MHz at fairly poor accuracies (5-10%). The outputs of
the three multipliers are still added together, and then square rooted.
Filtering would be done after the addition, so that the signal phases
will properly add at the high frequencies. I'm not sure if you are likely
to run into something like this, although it is quite possible to get a
situation where there are fields from two sources, both at the same
frequency but different phases. The effective field vector would then
change orientation with time. In an AC motor, for example, the field
vector actually rotates at 60 Hz.
Although the meter can't follow fast changes like this, the numerical
value for the average value for a rotating field will be different using
the filtered sums of the squares than the sum of the square of the
filtered (rectified) signals. If I'm not mistaken, the worst case error
is about 30% for sinusoidal signals.
If the signal is pulsed, (as in video display terminals), you will
need the 'real-time' multipliers, since the rectified signals will not
give a realistic measure.
Calibration of Magnetic Field Probes
Within certain constraints, this is kind of easy to accomplish.
Basically you generate a known field by passing a current through a coil
of known geometry.
First determine the size of the calibration coil. For low frequencies
I use one that's about 30 cm. in diameter, and has 9 turns. It's
resonant frequency is around 200 KHz, so I can't use it above 100KHz.
For this coil, a current of about 33 ma. gives a field of 1 amp/meter
(about 1.25 microteslas, or 12.5 milligauss). It's impedance is a few
hundred ohms at 100 KHz, so I apply a signal generator to the coil in
series with 600 ohms, and set the current to exact 33 ma.
The calibration coil is big enough to give a fairly uniform field at
the centre of the coil. If you want to get really precise, you can make
two coils spaced apart (Helmholtz coils), and in between them is a
fairly uniform field.
At the high RF frequencies, self resonance will force you to use
quite small single turn, heavy wire (or tubing) coils. If you calculate
the inductance for them, a series resistor can be chosen to give more or
less constant current with a wide range of applied frequencies. You will
want to make sure the values don't get very large (not more than 500
ohms or so) or the RF will take 'sneak paths' through stray capacitance
and foul up the calibration.
If you have a tracking generator for your spectrum analyzer, you can
make quick work of calibrating and testing frequency response of your
setup. Put the output (50 ohm) of the tracking generator to the
calibration coil and series resistor, and monitor the probe output with
'H', the magnetic field strength, is simply H=(I*T)/L where I is the
current in amps, T is turns, L is length (around the coil).
For the probe, if you don't frequency compensate the output, the
voltage is related to the field strength by: V=H*2*pi*f*u*A*T
V=voltage,H is field strength,f is freq,u is permeability of free space
(1.257E-6 V*s/(amp*meters)), A is area of coil,T is turns. Watch out for
units!...especially area. If you know the frequency, then it's simple to
measure the field strength. This sometimes serves as a calibration
especially for tiny high frequency setups.
When you use the probes, don't trust the measurements when the coils
are close to metallic objects. At low frequencies, ferromagnetic
material may give you falsely high measurements, whereas at high
frequencies anything highly conductive will give you lower
Electric field or potential probes.
In order to measure the electric field, you must measure the potential
at two points, take the difference and divide by the separation
distance. This will give the field in volts/meter in the direction
defined by the line between the centres of the the two potential probes.
As you can see, this measurement is directional. There are ways to make
measurements using 4 potential probes that are aligned so that the sum
and differences when squared, summed, and square rooted will give you
measurements that are independent of probe alignment.
The problem of measuring potential, is that any measurement requires
current to be drawn, and that alters the potential measured. My crude
solution is to make a potential plate whose self capacitance is large
enough to swamp the effect of the 'sampling' circuit. I also use a very
small series capacitor to the sampling circuit, to isolate the measuring
components from the measured potential. This reduces sensitivity, but
The wiring to the measuring circuits must be as thin as possible so as
to not severely alter the electric fields. Your presence, and the meter
and your hands that hold it will affect the measurement, especially with
fields of high potential, but low field.
Here is a crude 'ASCII' picture of the probe and circuit.
| <> small cap (0.1-0.47pf) connected to centre conductor of coax
|----------------------------------------------------i thin coax
| plexiglass handle RG 178/u
| aluminum plate (10 cm diameter) connected to the small cap.
The effective circuit for this thing is below:
P1_______________||___________________o amplifier input
| || | |
| C2 | >
---------- ------- <
C1 ---------- C3 ------- > R1
| | <
| | >
| |_____|_____o amplifier common,case
| your hand, in many
'zero cases is close
potential' to 'zero
potential' but not
C1=self capacitance of probe plate, say 10-20 pf
C2=small series capacitor, 0.1-0.5 pf
C3=capacitance of cable and any extra capacitance, say 500 pf
R1=input resistance, say 10 Megohms
P1=potential voltage measured
The loading effect of the circuit on the main plate (C1) is small, it
can be kept to a few percent or less. There is also stray coupling
capacitance, this can be minimized by a very physically small C2, and
very thin wire. The centre conductor of the coax must be very short,
since it will pick up the electric field directly. You can extend the
braid of the coax to cover it, and part of C2 to reduce this pickup.
The input potential P1 will be attenuated by the value of C2/C3. There
will also be a low frequency cutoff determined by the time constant of
R1 and C3.
The accuracy is dependent on how close the amplifier common and 'zero
potential' are. This is usually very questionable. However, since we are
interested in FIELD measurements, we can make TWO probe circuits, and
measure the difference. Doing this makes the measurement independent of
where the 'zero potential' is.
The way I make a field probe is to use a common shaft, and mount two
plates about a few inches apart, so that the probe looks like the
| insulator |
| / |
|_____________ _____________| potential plate
|--\--------- \/ ------------|
| wire |) | |
| |( | |
potential plate |) |
|) |-probe handle (plexiglass or insulator)
(- coax, two cables twisted together
Before the wires go into the coax cable, I put in series capacitors
(about 0.1-0.5 pf) as before. At the amplifier end, you use a
differential amplifier connection. You can put a variable capacitor in
parallel with one coax line, and a fixed capacitor equal to 1/2 the max.
variable cap. on the other coax line. This will allow you to adjust CMRR
for the differential amplifier by adjusting the match of the
sensitivities of the two plates. The input resistance might be also
variable, to maximize CMRR at the low frequency cutoff.
Calibration of potential probes:
This is real easy. Hook up the plate to known voltage (potential) and
see if it's OK. For a better test, take a cardboard box, cover it with
aluminum foil, and make a hole large enough for the probe to fit
through. Connect the aluminum foil up to a voltage, and see if the
inside of the box is at that potential (it should be). The probe should
measure less, depending on how bad the loading is. For a large plate (6
in. diameter) and very small coupling capacitor (0.1 pf), there should
be a fairly small error.
Calibration of field probes:
This isn't too hard either. Take two plates (or cardboard covered with
aluminum foil, foil side toward probe) that are at least 2 ft x 2 ft,
space them apart a foot or less. The plates should be square, and the
dimension of one side should be at least 4 times the separation
distance. Apply a difference of known voltage bewteen the two plates.
The field is the voltage divided by the separation distance. If you
notice significant variation of the measured field as you move the probe
closer to either plate, then there is too much loading from the probe.
There! That's about it for a crude idea how to make the probes. I have
more designs kicking around, but these here seem to be the most successful.
If you have any improvements I'd sure like to hear about them, and I'll
add them to this file and send an improved document out on to the net
when it's done.
As an afterthought, here's a list of articles that might be useful. We
also have a bibliographic database of circuits that's available FREE by
ftp, from this site (gaitlab1.uwaterloo.ca, under /pub/circuits/main)
that will have additional references. It has a search program, the
database is in DBASE format so you can use your favourite searching
'An optically Linked Electric and Magnetic Field Sensor for Poynting
Vector Measurements in the Near Fields of Radiating Sources',IEEE
Transactions on Electromagnetic Compatibility,Vol 30,#4,Nov
'Variable Oscillator reacts to magnetic flux changes',Electronics,Nov
'A Wide Dynamic Range Portable 60-Hz Magnetic Dosimeter with Data
Acquisition Capabilities',IEEE transactions on Nuclear Science,vol
'Evaluate EMI reduction schemes with shielded-loop antannas',EDN,May
'Multipoint latches allow uniform door/cabinet seal',EDN,July
1984,around p 160. (This article discusses antennas, baluns etc.)
'Optically sensed EM-field probes for pulsed fields',Procedings of the
IEEE,Vol 80,#1,Jan 1992,p 209-
If you have any more good references, please send me some to add to
this file. You will be able to find it also on our ftp site
(gaitlab1.uwaterloo.ca, under /pub/em).