SMX

Electrical Interface for Watermark™ or Gypsum Block Sensors.

http://www.emesystems.com
Watermark sensors are available from Irrometer Company, Riverside CA 909/689-1701
<http://www.irrometer.com>. These are "granular matrix" devices, and consist of a fine
aggregate mixed with gypsum crystals, held inside a permeable membrane and a per-
forated stainless steel sleeve. This device is buried in intimate contact with soil, and
reaches an equilibrium with the soil moisture. Stainless steel electrodes are embed-
ded in the granular matrix and the the electrical resistance measured between the
electrodes is the signal of interest. Electrical resistance decreases with increasing soil
moisture (and also temperature, which can be compensated). The purpose of the gyp-
sum is to buffer the measurement from ions that are found in uncontrolled amounts in
the soil. Delmhorst gypsum block sensors are similar: Stainless steel electrodes are
embedded in a gypsum block with proprietary additives. The Watermark formulation
claims to have a wider range of response and a longer field life.

The bottom line is that the readings are related to plant stress in specific crops. The
Irrometer web site offers links to a number of practical and academic papers evaluat-
ing the sensor.

A special circuit is needed to measure the electrical resistance of the blocks. DC cur-
rents must not be allowed to flow through the wet part of the circuit, or else irreverse-
able reactions occur on the metal surface that spoil the readings. AC excitation avoids
these problems, by reversing the polarity of the current many times per second, so that
no net reaction takes place at either electrode.

A related issue is galvanic currents. If two devices are immersed in the same environ-
ment (say two moisture sensors near each other in the soil), any differences in the
environment between them can generate "ground loop" voltages and currents. The
coupling effect between the two sensors both degrades the reading and exacerbates
corrosion. It is important to avoid DC "sneak" paths between different sensors. The
same consideration applies to other objects in the environment: Metal tanks, ground
rods, salinity gradients. Any or all of these can contribute to underwater potential dif-
ferences that are continually driving current
through the wet medium, and through sensors
immersed in the medium.

The SMX circuit provides the AC excitation and

galvanic isolation required. The output signal is a
frequency, or alternatively a current or a voltage,
that depends in a relatively simple manner on the
resistance of the sensor. Two wires connect to the
watermark sensor, and the other wires produce the
voltage, current or frequency signal that is routed to
a data logger or controller.
Figure 1: SMX module

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 1
Specifications:
• Supply Voltage: 4–15 VDC
• Frequency output 50 hz @ ∞ ohms
10 khz @ 0 ohms
open collector square wave (needs pullup
resistor to read out frequency.)
(as high as 12.5khz with sensor short circuit)
• Current output (also supply current)
0.2 mA dry to 1.0 mA wet
(as high as 1.5 ma with sensor short circuit)
• Voltage output
0.2 volts to 1.0 volts typical
(as high as 1.5 V with sensor short circuit)
• less than 0.01% per Volt supply variation.
The SMX is a potted module (0.825" sqaure x 0.25"
thick) for placement in the field near the sensor. There
are two wires for connection to the Watermark or gyp-
sum block sensor, and four wires for connection to the
data logger or other equipment.
• connections (see diagram, below)
red: + 5 to 15 volts DC
green: signal frequency, needs pull-up to +V
white: voltage signal
black: common
blue wires: soil moisture block
• Operating Temperature: -0°C to +70°C
no meaningful signal below 0°C

• http://www.emesystems.com

Wiring the SMX to the data logger

Figure 2 shows how to connect the sensor for digital frequency output. The resistor (4.7 kohms, value
not critical) can pull up to any voltage from 3 to 7 volts DC. The output signal is a square wave, and its
frequency varies from 50 hz when the sensor is bone dry, up to 10000+ hertz when the sensor is soaking
wet. This output can be measured using a COUNT or PERIOD function on the data logger. Note that
the white and the black wires are connected together

Figure 2: connection for 50hz to 10,000 hz frequency output

Figure 3 shows how to connect the sensor for voltage output. The green wire should be connected to the
white wire. The output signal is a voltage that varies from 0.2 volts when the sensor is bone dry, up to
1+ volt when the sensor is soaking wet.

Figure 3: connection for 0.2 to 1 voltage output

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 2
Figure 4 shows how to connect the sensor for two wire current output. The black wire should be con-
nected to the white wire. Leave the green wire disconnected (and insulated). The output signal is a cur-
rent that varies from 0.2 milliamp when the sensor is bone dry, up to over 1 milliamp when the sensor is
soaking wet. The current on the two wire circuit may be converted to a voltage at the input of the data
logger. A 1 kΩ resistor will convert the 0.2–1.0 millamp current into a 0.,2–1 volt signal. The power sup-
ply voltage must be a high enough voltage to sustain the maximum expected voltage across that resistor,
plus the line loss in the wiring, plus the 4 volts required by the SMX module itself.

Figure 4: connection for two-wire 0.2 to 1 ma current output.

Notes on reading the sensor signal.

Connect the SMX as in figure 2, for digital output. This program is written in PBASIC for the BASIC
Stamp, using the command COUNT to determine the SMX oscillation frequency.

' simple test of the SMX operation, count input.

x var word
loop:
count 12,1000,x ' count on Stamp P12 for 1 second
debug ? x ' show the result in counts
goto loop

Alternate, using the voltage input version. Connect the SMX as in figure 3, for voltage output, and con-
nect the voltage to the analog to digital converter input of the OWL2e data logger, or to your converter.
The ADread subroutine is not shown.

' simple test of the SMX operation, voltage input.

raw var word
loop:
gosub ADread
debug ? raw ' show the result in millivolts
goto loop

These results can be used directly if you have a table of frequency and voltage versus resistance and
wetness. (See page 5 of this document.) Or, the raw values can be transformed to resistance, and then
the temperature compensation applied, and that converted to moisture units. These routines will be
expanded in the next revision of this document.

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 3
Interpret the readings.
Documents from Watermark suggest the following rule relating the sensor AC resistance to soil moisture
tension. This should be compared in practice to a reliable instrument, such as a tensiometer, an
Irrometer or the "Watermark Digital Meter" from Irrometer, and observations of actual plant stress. Soils
vary greatly in their characteristics.
The ohm reading should be compensated for soil temperature, at approximately 1% per degree Fahrenheit deviation
from 75 °F. Ohms75=OhmsdegF * (1+ (degF-75)/100). The "<<<<<" marks the typical irrigation range. Higher
in the table is wetter. 1000 ohms is around 4 centibar. 3300 ohms is around 21 centibar.

SMX sensor ohms kHz (green wire) µAmp, millivolts (white wire)
100 10.75 1320
330 8.04 1026
1000 4.59 668 <<<<<< ~4 cb
3300 1.90 390 <<<<<< ~21 cb
10000 0.724 270
33000 0.260 222
100000 0.118 207
∞ 0.048 200

The graphs two pages hence illustrate the response of the circuit in terms frequency and voltage outputs
as a function of sensor resistance in ohms.

Irrometer Corporation has published a table of electrical resistance values in relation to soil moisture in
kPa. The resistance ranges from 550 ohms in saturated soil, 0.0 kPa, to 27950 ohms in very dry soil, 199
kPa. That is at 75 degrees Fahrenheit, 24 degrees Celsius. This table is the basis of the readings pro-
duced by the Watermark Meter model 30KTCD-NL. Crop irrigation typically takes place in the range of
15 (sandy soils) to 60 (clay soils), depending of course on the crop. That is a resistance range of 2000 to
10000 ohms sensor resistance. The Irrometer meter should be considered the standard. But it is informa-
tive to understand where the standard comes from.

The table is based on a careful study of the Watermark sensors carried out by Shock, Barnum and
Seddigh ("Calibration of the Watermark soil moisture sensors for irrigation management", Proceedings of
the 1998 Irrigation Association Technical Conference). They set up a temperature controlled environment
in a sandy loam soil, with 24 model 200SS Watermarks along with standard tensiometers and tempera-
ture sensors. The main part of the Irrometer calibration chart is based on the average of those readings,
with interpolation. The experiments were done at both 25 degrees C and at 15 degrees C, so temperature
compensation is validated over that range. The experiments covered the range of roughly -2 to -75 kPa.
The range from -10 to -75 kPa is in fact the most important for agriculture, because irrigation schemes
typically maintain tension in that range. Shock et al came up with the following equation as the best fit
to their data over the -10 to -75 range.

kPa = (3.213 * kohms + 4.093) / {1 - 0.009733 * kohms - 0.01205 * Celsius)

It is important to note that the fit outside that range is quite bad. For suction less than -10 kPa (very wet
soils), the curve drops precipitously. Also the study did not cover higher levels of suction, because
tensiometers do not function above about 80 or 90 kPa. The Irrometer Watermark calibration chart
extends up to -200 kPa. There is reason to believe that this is a simple extrapolation of the experimental
data, but the extrapolation is not linear (Rick Allen) .

I think the Irrometer meter uses the calibration chart as a lookup table, so as to cover the very wet and
and very dry end of the range that are not empirically correct by the above formula.

The calibration data is not too far from linear over the typical irrigation range. Thus, for many purposes
(and given the difficulty of cross calibration and variations due to soil type), a linear or piecewise linear
approximation may be quite adequate.

The simplest formula recommended by Irrometer is the following, to calculate soil moisture from sensor
AC resistance. This formula is an approximation valid in agricultural soils near the point where irrigation

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 4
will be required:

kPa = ohms / {135 *(1-[degF - 75]/100 }

(kPa is the same as centibar)

That is a simple linear response of 135 ohms per kpa, when the soil temperature is 75 degrees Fahren-
heit (24 degrees Celsius). The ohm reading has to be compensated for soil temperature, as the ohms per
kPa is effectively lower at higher temperatures. The compensation recommended by Irrometer Corp. is
1% per degree Fahrenheit deviation from 75 °F (24 degrees Celsius).

For example, suppose the sensor resistance is found to be 3000 ohms at a temperature of 68 degrees
Fahrenheit. The kPa value from the formula is about 21 kPa. The same resistance found at 86 dgrees
Fahrenheit would estimate 25 kPa.

Another slightly more complicated formula stated by Irrometer and found in the literature is:

kPa = (ohms - 500)/ (176 - 2.3*Celsius)

kPa = (ohms - 500)/ (217 - 1.28*Fahrenheit)

The response is 127.5 ohms per kPa at 21.1 degrees Celsius (70 degrees Fahrenheit). This formula pre-
sumably applies over a wider range of moisture levels.

These formulae are also plotted on the graph, in comparison with the standard Shock calibration curve.
An on-site calibration of an individual sensor could be had by comparing the Watermark output to a ten-
siometer at at least two different points. This on-site result will be better than any approximation.

Finally, there is the issue of how the resistance of the sensor is measured. Quite a number of readout
meters are on the market that acquire readings from the Watermark sensors. The operating principle
vary widely, in terms of the frequencies, voltage levels and waveforms applied to the measurement elec-
trodes. We are currently investigating these effects, in order to qualify the readings from the SMX mod-
ule in relation to these others.

The bottom line is to emphasize again and again the importance of on-site calibration in relation a spe-
cific soil, irrigation scheme or study question.

Please refer to http://www.emesystems.com/smx.htm for links to the scientific and commercial literature
and further discussion of these calibration issues.

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 5
 SMX: Frequency Output vs Resistance

12000

Full immersion in water

10000

8000
Frequency Output (Hz)

6000

4000

10–30 centibar typical irrigation range

2000

very
Dry
0
1.00E+01 1.00E+06

1200

Full immersion in
water
1000

80
0
Voltage Output (mV)

60
0

40
10–30 centibar typical irrigation range
0
1000–3000 Ω

20
0 ~80cb dry
very Dry
soil10,000
Ω

0
1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06
Resistance (Ω)

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 6
Technical information:

Figure 4: SMX schematic

The power supply voltage is regulated at 3.3 volts DC by the micropower regulator. The power supply +
input can be as low as 4 volts and as high as 15 volts. Filter capacitors are provided for stability and
averaging of the supply current. The LMC555 timer operates in its direct feedback mode, with a square
wave on the totem pole output from pin 3 charging or discharging the 0.1 µf polyester film timing capaci-
tor through the network of fixed resistors in series/parallel with the moisture sensor. When the sensor is
dry, the 150kΩ resistor sets a minimum oscillator frequency of 50 hertz. When the sensor is wet, or
short circuited, the 390Ω in series with the grid limits the upper frequency to about 12 khz. The current
through the sensing grid is AC. Nonpolar ceramic capacitors isolate the circuit from the sensor, to fore-
stall galvanic interactions. The output frequency is transmitted to the logger from the open collector DIS
output pin. Normally a pullup resistor will be provided to give voltage transitions at the logger. The cur-
rent drawn by the circuit varies linearly with the frequency due to the charge and discharge cycles. The
supply current is proportional to wetness, a voltage signal can be taken from across the 1kΩ resistor.
There will be a small AC component on the DC output signal, that can be averaged in software. A 10µf
capacitor in parallel with the output resistor will reduce the AC component to <5 millivolts.

©2002 EME Systems, 2229 Fifth St., Berkeley CA 94710...tel:(510) 848-5725 fax:(510)848-5748 SMX pg 7