H. R. HOLBO   31 Jan 2010
Comstock Instrument, Albany, Oregon

The hygrometer design described here uses the salt-solution, phase-transition point of lithium chloride (LiCl) salt to measure the dew-point temperature, or dewpoint, of the air. The design is based on the fact that LiCl is very hygroscopic, and that its conductivity changes abruptly when heated to a temperature where the salt is dry and in balance with the atmosphere around it. The first dewpoint hygrometer developed to exploit this characteristic was introduced in about 1945 by Foxboro2 called a DEWCEL™. (Please see the Endnotes section for more information about the numerically superscripted items.) As originally developed, the DEWCEL's™ sensing element was heated by conduction through the salt on the surface of a cylindrical bobbin. In contrast, in the design described in this article, the surface of the sensing element attains the necessary temperature by using a small electronically-controlled bobbin heater where the salt's phase-transition is detected by two small electrodes; no heating current passes between the electrodes. Because it is much smaller, it operates using substantially less electrical power. This makes practical long-term measurements in the field using a battery for power. Experience has demonstrated that this LiCl dewpoint hygrometer design has exceptional measurement accuracy that does not change in over extended periods of deployment at remote field sites.


This is the story of an effort that began at a time when, regardless of the type of humidity sensor chosen from the marketplace, neither long-term measurements nor accuracy was assured. It is also a story typical of proponents who rejected many so-called hygrometers, the operating basis for which could not be derived from physical principles. It is also a story of necessity, in that funding sources for agricultural and forest meteorology was limited, yet the need to acquire certifiably good humidity data provided a strong motivation to overcome obstacles.

Humidity is a difficult atmospheric property to measure accurately over extended periods. Obtaining reliable, long-term humidity records at unattended, remote locations which lack mains' power is a chronic problem for field meteorologists. Recently the situation has improved because environmental interests have increased, leading to increased sponsorships, and with the deployment of many hundreds of remote weather stations for hydrologic, air quality, or climatic monitoring purposes. It is also the case that this kind of instrumentation would not have been possible were it not for space, military, and entertainment developments in electronics' miniaturization and in telecommunications.

As a part of these recent developments, several technologies have begun to fill the gap for accurate, low-power, field-deployable humidity instrumentation. One of these is a company by the name of Vaisala Instrument. Their hygrometer products are the HUMICAP™ and the DRYCAP™, the names suggesting that capacitance change is employed to make the measurement happen. This approach probably uses the change in dielectric constant within capacitor's plates brought about by specific concentrations of water molecules in the air. Another is Xentaur, with a product using what they call HTM™ (representing Hyper Thin Film) that may also be capacitance-based.

Infrared hygrometry is another new technology. It is based on the infrared attenuation by water vapor along an optical path to derive humidity measures. This type is fast responding, suiting it for eddy correlation estimates (the average of the product of a property times its flow rate vector) of water vapor flux (evapotranspiration). Even more recently, lasers now make better infrared sources than do incandescent lamps, in that they can be chosen for discrete wavelengths, and so discriminate among various atmospheric constituents, e.g., H2O, CO2. This affords measurements, not only of water vapor concentrations, but also of carbon dioxide concentrations. It is probable that optical methods are better suited for atmospheric fluxes of those entities to be estimated than for evaluating the absolute quantities of these properties.

Dewpoint hygrometry is a long-established measure of atmospheric humidity. It is based on fundamental physics (condensation on your glasses stuff), and requires only one temperature measurement. The approach developed for automating dewpoint measurements continuously involves thermoelectrically cooling a mirror to the temperature at which a film of dew is just maintained upon it, i.e., the temperature of the gas-to-liquid phase transition, or dewpoint. The mirror permits optical detection of this point, which is used to provide feedback to its circuitry, and the resulting mirror temperature defines the atmospheric humidity. Unfortunately, the superb fundamental nature of this type of dewpoint hygrometry is difficult to use at remote sites where electrical power for cooling the mirror is not available, and dust and other contaminants can interfere with dewpoint optics. The physical size of this instrument also mandates ductwork to bring air from the desired measurement level into the instrument. This also means maintaining the walls of all ductwork used above dewpoint.

An alternative form of dewpoint hygrometer, marketed by Foxboro2, was called the DEWCEL™ (Hickes, 1947). Instead of cooling a mirror to the point of condensation, it utilized heating to the point where moisture was kept from adsorbing into a salt. This was done by using a fabric-covered cylindrical bobbin upon which two wire electrodes are spirally wrapped. The fabric is doped with a saturated lithium chloride (LiCl) solution and low-voltage AC electrical power was applied. Current then flowed between the electrodes through the partially conductive salt solution, directly heating the bobbin until the temperature of liquid-to-solid phase transition is just attained, at which time the current decreased, reducing the heating rate. Temperature was measured using a thermometer inserted within the bobbin. The equilibrium temperature of the bobbin is directly related to the ambient atmospheric humidity in a manner similar to the temperature of a cooled mirror dewpoint hygrometer.

The relationship of dewpoint (or vapor pressure, say millibars, mb) to bobbin temperature is specific for pure salts. Nelson and Amdur (1965) found that measuring dewpoints using salt-solution, phase-transition hygrometry can be as precise as ±0.15°C. This is an exceptional specification for any type of hygrometer. Additionally, the elevated temperature of the bobbin tends to promotes air flow past it, and reduces concern about spurious thermal loads.

Tanner and Suomi (1956) and Fritschen (1965) successfully used DEWCELs™ for agricultural meteorological studies of evapotranspiration, quantifying humidity gradients ΔTDz, where TD are two or more measured dewpoints over prescribed distances z, above field crops. However, the large physical size of a DEWCEL™ limits its deployment within plant microclimates. And, at more remote field sites, such as inside forests and rangelands, the DEWCEL's™ requirement for a substantial amount of electrical power make long-term studies using batteries impractical. In addition, because DEWCELs™ are heated by current passing between the electrodes, that current must be alternating (AC) to avoid their polarization.

Campbell et al. (1971) made great progress in overcoming these problems by developing a small version of the heated LiCl dewpoint hygrometer in which the functions of bobbin heating and phase-transition detection were separated. They supplied a small alternating current signal between two electrodes, only used to detect phase-transition, not to provide heat. This signal was then used to electronically control a larger flow of direct current in through the bobbin heater. Its power requirement was obviously much lower because of the smaller overall surface area in relation to the DEWCEL™.

In 1972, I deployed several "Campbell" LiCl dewpoint hygrometers (purchased from Logan Scientific, Logan, Utah) at remote meteorological stations in the forest. Initially performing satisfactorily, they lacked long-term reliability and accuracy, and had to be removed from service within a few months. As appealing as the design was, improvements were still required, and I set about further development. The design described in this report is the result of that work, including the changes in physical and electrical characteristics. Many years of successful field use in mountain, desert, forest, coastal environments support its recommendation for further use.


The derivation of a heated LiCl dewpoint hygrometer's operation is based upon the fact that salt solutions act to reduce the equilibrium vapor pressure in the nearby atmosphere below that which would obtain were no salt present. To achieve the same vapor pressure over a salt-plus-water solution as over pure water requires raising the temperature of the salt solution. This can be seen in Figure 1, which shows the relationships for several saturated salt solutions: (sodium chloride, NaCl; magnesium chloride, MgCl2; and lithium chloride, LiCl); and pure water, H2O. Notice that the temperature contrast between the LiCl solution and water is greatest. And, since LiCl requires the most heating to come into balance with ambient conditions, its usefulness as a heated dewpoint hygrometer would extend over a wider humidity (vapor pressure) range than other salts. Of course, the intermixing of other salts in the solution must be avoided, as this would alter the relationship and cause measurement errors.

Figure 1 Vapor pressure versus temperature curves.
Fig. 1. Vapor pressure - temperature relationships for water, H2O, and for saturated solutions of NaCl, MgCl2, and LiCl in water. The interrelationship between dewpoint and bobbin temperature of an LiCl dewpoint hygrometer is drawn for a situation where the ambient humidity is 20 millbars.

Utilizing this interrelationship, plus the fact that salts go through a dramatic change in electrical conductivity as their hydration state changes, one can understand how temperature control of the Foxboro2 LiCl dewpoint hygrometer (DEWCEL™) works. Because of that dramatic conductivity change over a very narrow range of relative humidity (Nelson and Amdur, 1965), any current flow between electrodes embedded in such a salt would increase or decrease as needed to achieve the equilibrium temperature with respect to the prevailing ambient humidity, essentially automating DEWCEL™ operation, and require no external electrical control to do so; just a source of alternating current applied to the coiled electrodes. Direct current cannot be used because that result in electrode polarization.


The relationship between bobbin temperature of a heated LiCl dewpoint hygrometer at equilibrium with ambient humidity (vapor pressure) can be inferred from Figure 1. And, even though the vapor pressure-temperature relationships are exponential, one can derive a linear equation of the form y = mx + b, where y is the dewpoint temperature and x is the temperature of the dewpoint hygrometer salt solution defining the interrelationship. For example, values for the slope m and intercept b could be obtained by calibration of a hygrometer in a closed system humidified using saturated salt solutions, which humidity was measured using a laboratory dewpoint instrument. Or, it could be derived using published equations describing those saturation vapor-pressure, eS, curves. The equation provided for water by Murray (1967), whose formulation follows the so-called Teten's approximation of the equilibrium vapor pressure eS over water (left-most curve in Figure 1), is described by Equation 1:

eS = 6.1078 exp [(17.269TD) / (237.3 + TD)] (1)

In Equation 1, the coefficient has units of millibars (mb); at equilibrium, water temperature and dewpoint temperature, TD, are equivalent.

A companion relationship describing the shape of the saturation vapor pressure curves over a saturated salt solution is also required. This curve is assumed described by the same basic relationship, modified by a coefficient, hTS, as shown in Equation 2:

eS = hTS 6.1078 exp [(17.269TS) / (237.3 + TS)] (2)

In this equation, hTS is the equilibrium relative humidity expressing the proportional depression in water vapor pressure resulting from the salt solution at temperature TS, or TSENSOR. Clearly, the value of hTS chosen is critical to the outcome, and a variety of values for it have appeared in the literature. The experimental determinations by Acheson (1965), which seem the most carefully done, have been used here to calculate hTS . A word of caution; because LiCl undergoes hydrite transition near 18°C, this derivation strictly applies only to hTS values above 20°C. This is not regarded as a serious limitation in most environments because an 18°C solution temperature, which will also be bobbin temperature, TS, as well, corresponds to a dew point of about -12°C and is outside of the usual range of meteorological interest (industrial applications excepted).

The interrelationship between TD and TS is clarified by setting Equation 1 equal to Equation 2 and solving for TD to obtain:

TD  =  {237.3 [(17.269 TS / (237.3 + TS)) + lnhTS]}
{17.269 [(17.269 TS / (237.3 + TS)) + lnhTS]}

Collecting terms, the tidy y = mx + b relationship that results is:

TD = 0.743 TS - 25.76°C (4)

Table I provides a basis for comparing this relationship with others. Insofar as possible, the values given pertain to LiCl solution temperatures above 20°C. The disparity in values demonstrates the difficulty in establishing dewpoint hygrometer calibrations. Unfortunately, the tabulated range of values would suggest, when one does the arithmetic, that dewpoint estimation errors could be as large as 4 to 6°C.

 Table I  A comparison of heated lithium chloride dewpoint hygrometer calibration relationships that are in, or calculable from, published reports.

Slope m, °C/°C

Intercept b, °C

TS Range, °C

TD Range, °C



20 to 100

-10 to 46



-7 to 68

-30 to 24



20 to 65

-10 to 23



20 to 65

-11 to 22



20 to 65

-11 to 22



20 to 65

-13 to 22

Sources: A Tanner (1968) and Tanner and Suomi (1956). B Hedlin and Trofimenkoff (1965). C Wexler and Hasegawa (1954). D Yellow Springs Instrument 9101 probe (1974), Yellow Springs Instrument Co.4, Yellow Springs, Ohio. E This report and Acheson (1965). F Logan Scientific, pers. comm. (1972).

The only table values obtained using the same type of dewpoint hygrometer design as described here were estimated from data supplied by Logan Scientific. This distinction is probably important because earlier versions operated in a self-heated (heating rate depended upon electrical conduction through LiCl doping) rather than an indirectly-heated, constant resistance mode (Campbell et al., 1971). Self-heated dewpoint hygrometers may be more easily influenced by ambient temperature and wind. In fact, the equilibrium temperatures of self-heated dewpoint hygrometers have been found to depart from theoretically predicted values (Tanner and Suomi, 1956), and they are apparently less sensitive to humidity (i.e., the slope m of the relationship is smaller).

It is believed that Equation 4 probably best describes the sensor/dewpoint temperature relationship of indirectly heated dewpoint hygrometers such as the design presented here. Although the danger of undetected calibration biases always exists, in this case dependent upon Acheson's (1965) data for the values of hTS used, and on Teten's approximation for the shape of the curves. However, many repeated calibrations comparing different thermometric methods have revealed no detectable departures from the slope m and intercept b constants adopted here.

Controlled dewpoint hygrometer operation is possible only when the ambient air temperature is lower than the equilibrium temperature necessary to cause phase transition of the salt solution so that heating is necessary. For saturated LiCl solutions, dewpoint hygrometer control is not possible (that is, phase transition would not require any further bobbin heating) when the relative humidity is below 11%. Instead, bobbin temperature would simply respond to changes in air temperature rather than humidity.

Thus, in its controlled operating mode, the dewpoint hygrometer is above air temperature and heated to maintain it at the phase-transition temperature. Resultant additional power losses due to convection, conduction, and radiation will always be present. So, to assure proper operation of the sensor and its control circuitry, the designer must anticipate these losses and compensate for them. Tests of the "Campbell" version suggest that these losses, which can be quite large, may not have been adequately considered. For example, radiative losses may reach 860W/m²; convective losses in a saturated airstream moving at 1m/s could approach 1800W/m²; and conductive losses to the mounting bracket are potentially huge. Thus, the surface areas of both the sensor and its mounted configuration (e.g., support wires) are critical to sensor performance.

Hygrometer assembly

The sensor fabrication of this hygrometer (Figure 2) physically differs from the "Campbell" version by having a 40% smaller surface area, largely due to the smaller bobbin heater chosen (an Ohmite3 U-5818), finer-gauge platinum wire electrodes, and higher-denier bobbin covering. Before attaching the electrodes, a layer of fine mesh fiberglass fabric (sold in hobby shops for model airplanes) was wrapped around the bobbin to hold the salt. I also clipped the original 16awg resistor lead wires short, and soldered on 20awg support wires to reduce conductive heat loss by that route. One can select from a variety of thermometer types (thermistors, thermocouples, platinum resistance, or diodes) to insert into the core-space of the bobbin, always being mindful to avoid sizes that would increase useless heat losses. [Compensated small gauge thermocouples may be easier to deploy than thermistors if funds allow.] Transistor heat-sink grease (common today between a CPU and its heat sink in computers) is good for making the thermal coupling between the bobbin core and the inserted thermometer.

Figure 2 The hygrometer assembly Fig. 2. The dewpoint hygrometer is built upon an epoxy-coated, vitreous-enameled power resistor (Ohmite3 U-5818) as the bobbin heater and base, 'A', for mounting these other components: 'B,' fine weave fiberglass fabric (sold for covering model aircraft), tightly wrapped on 'A'; 'C', two turns each of small (30awg) platinum wire for electrodes, ends twisted together with smooth-jawed needle-nose pliers, and made tight near the middle of the bobbin, the electrodes are then gently pushed apart with the shaft of a dissecting needle; 'D', thermometer lead wires; 'E', resistor core filled, using a hypodermic syringe+14ga needle, with transistor heat-sink grease (leave no air pockets) to provide good thermal coupling between thermometer and bobbin and to protect the thermo-meter element. As illustrated, an octal-base relay case, 'F', serves as the overall support and enclosure, but other mounting bases types may also serve. - When ready for use, 'dope' the fiber by applying a "small" drop of saturated, reagent-grade LiCl in H2O solution in the gap between the electrodes. If an oil film prevents wetting, flush bobbin with acetone, followed by distilled water, then allow to dry. - After use, store in seal-able jar with fresh silica gel desiccant.

Before assembly, the bobbin heater, a vitreous-enameled, wire-wound power resistor, should be coated (dipped) several times with clear epoxy paint (NOT the quick-setting type) and then heated after each coat. This coating further prevents the LiCl solution from contacting metal within the bobbin when the fabric is doped. Power must be kept on once the sensor is doped and put into operation, otherwise hygroscopically-absorbed water will mobilize the salt solution, which will eventually contact the metal. This leads to contamination of the LiCl salt (a yellow color is the sign that iron or nichrome has migrated from the power resistor) corrosion into the heater, not only spoiling sensor calibration, but destroying it.

Following hygrometer operation, the best practice is to place the sensor assembly in a seal-able glass 'desiccator jar', along with a supply of functioning silica gel to absorb moisture. This prevents self-destruction and keeps it ready to use. In the desiccator, I like to use 'indicating' silica gel granules (J. T. Baker Chemical 3401). The granules display a vivid blue color when 'fresh' and change color (to violet, then pink) as moisture is adsorbed. The color can be seen without opening the jar. When blue, the silica gel dries the air and lowers its dewpoint to about -40°C. Baking used-up granules in a 200°C oven (shield the granules from any radiant heating elements) will 'freshen' them for re-use.

If you elect to store the hygrometer without using a desiccator jar, you must first remove all traces of LiCl with hot water rinses, then distilled water. You should then test the sensor by seeing if it heats up when placed in a humid atmosphere; a successfully 'un-doped' sensor will not heat.

Wind gusts can cause spurious fluctuations as the dewpoint hygrometer seeks equilibrium with the atmosphere, and aerodynamic buffering of the sensor is recommended. However, since the sensor is above ambient temperature, buoyancy will likely serve to maintain a small airflow around the sensor, effectively assuring the continual sampling of atmospheric air, even within a mostly-closed protective shroud. Reliable atmospheric sampling is thus aided.

If possible, avoid using aspirated ductworks (e.g., to draw air in from the roof) unless precautions are taken to assure that inner surfaces are always to above dewpoint (radiative cooling can quickly result in below-dewpoint surfaces), which could cause temporal errors owing to the arrival times of wetter or drier air masses from those surfaces.

Building the sensor by hand limits the degree of miniaturization that can be obtained. Eventually, newer technologies, because of the fundamental appeal of acquiring true 'dewpoint' measurements, may be employed to permit further size reduction and consequently lower than 800mW worst case power requirement of this hygrometer (see Figure 4).

Sensor power requirement

To determine the power requirement of the sensor assembly, one first has calculate how much temperature elevation of the sensor above ambient will be required to achieve LiCl phase transition. The maximum elevation will be in vapor-saturated atmospheres, where the air temperature and dew point are the same (100%RH). For example, if saturation occurs at 25C°, one can apply Equation 4 to get:

TSENSOR = (25°C + 25.7°C) / (0.743°C/°C) = 68.2°C (5)

or a sensor temperature elevation of 43.2°C. In other words, the total power dissipation by the sensor bobbin must be sufficient to sustain that temperature difference despite all thermal losses due to radiation, conduction, and convection. Figure 3 shows required temperature elevations (TSENSOR  — TAIR) over commonly encountered air temperatures and dewpoints. When constructing this graph, I drew diagonals at 5°C intervals, beginning on the right at, and parallel to, the (heavier) 11% RH line, ending on the left at the (heavy) 100% RH line.

Figure 3 Bobbin temperature requirements
Fig. 3. Requisite sensor temperature elevations (Tsensor — Tair) for a heated LiCl dewpoint hygrometer over a range of conditions.

I used an experimental approach to determine the maximum amount of bobbin heater power required. After deciding on a sensor assembly and its mounted configuration, I built a sensor prototype (e.g., bobbin heater, heatsink grease, and thermometer). I also provided another thermometer to measure air temperature. I connected the bobbin heater to a stable voltage-variable power supply and varied the heater voltage up beyond whatever value produces the requisite 43.2°C temperature difference between the bobbin and ambient air temperature. The total of all thermal losses can then be estimated by measuring the power (voltage times current) required to sustain that temperature difference. This power requirement establishes the upper control limit required from the heater and electronic circuits. Then, depending upon the intended voltage supply (12 volts usually is convenient for field operations), I was able to objectively select the best ohmic value of the bobbin heater (from within the values offered by the manufacturer).

Figure 4 Evaluating bobbin power need
Fig. 4. Bobbin power requirement test results on prototypes of the dewpoint hygrometer of the sizes evaluated for this report. The power limitations of the various bobbin heater resistances are shown using a 12V DC supply. Based on worst-case scenario requiring LiCl phase-transition to obtain in saturated 25°C air (required temperature line), the conservative (safe) choice is a 150Ω heater.

Figure 4 shows the results of this power requirement test in 'still' and in 'moving' air. To be on the safe side, considering both the possibility of greater thermal losses in the field environment and the certainty that battery supplies drop in voltage output during discharge, a value of 150 ohms was judged best for this design when operated from a 12-volt battery. This value assures that sensor temperature can increase to the phase-transition point under the most demanding humidity and wind conditions. Regulating the power supplied to the sensor at lower temperatures will not be a problem for the control circuitry. In the field, power consumption depends on local atmospheric conditions, being higher when wet and lower when dry. For this design, the maximum power consumption will be about 2AH (amp-hours) per day [(12V ÷ 150Ω) × 24 hours], or 60AH of battery capacity per month; a fully-charged, deep-discharge group 27 can do this.

Operating-point control and measurement

The circuit shown in Figure 5 was designed to control the operating temperature of the bobbin. It senses the conduction status of the LiCl doping with 2 electrodes. The signal imposed on the electrodes is generated with an oscillator, which supplies a square wave to the sense electrodes (X) and a triangular wave as a reference level (Y). Both signals are compared. When their difference is large, transistor Q1 switches heating current through the bobbin heater; when small, the current is switched off. The triangular waveform causes the comparator to toggle on and off continually, with duty cycles varying from 0 to 100% as needed to servo the bobbin temperature precisely at the phase transition point. The measured temperature of the bobbin is then used to get the dewpoint according to the relationships discussed above.

Figure 5 Hygrometer heating-control circuit
Fig. 5. The control circuitry used to maintain bobbin temperature at the phase-transition point of the LiCl solution. A1 and A2, are both TO-99 style 301A-type operational amplifiers, have TO-99 pin 4 connected to common and TO-99 pin 7 connected to the positive side of the 12V DC supply (these pins not explicitly shown on the schematic). A1, used as an oscillator, presents a 12V square wave at 'X' feeding one electrode, and a 6V triangle-shaped waveform centered about a 6V reference potential at 'Y' on A2. The proportion of the square wave (rectified by the 1N914) present at 'Z' on A2 depends upon the resistance of the LiCl doping between electrodes (rather like a volume control). A2, is used as a comparator of the signals at 'Y' and 'Z', thus adjusting the duty cycle of the heating current switched by driver transistor Q1, a 2N3053. Gates G1-4 are all part of a 4001A C-MOS NOR-gate device, which here serve as wave shaping inverters. The 3.3µF capacitor prevents any DC current from passing between the 2 sense electrodes. Socket pin numbers #4, #5, #3 and #6 refer to Figure 2. This circuit may be connected directly to the socket or be attached through a cable.

Figure 6 is a strip chart recording over time of 2 warm-up episodes. The first is of a freshly doped bobbin, the second beginning after the already-doped dewpoint hygrometer has cooled. A thermistor was used for a bobbin thermometer (see Figure 7). A few minutes of off-scale heating occurs as excess water is driven off at start-up, after which the bobbin under-shoots and then returns to a proper reading. Thermal inertia of the bobbin may result in some initial searching or instability, but bobbin temperature, in correspondence with ambient humidity, is typically attained within 20 minutes. If longer warm-up times are observed, you have either too much doping solution or too low LiCl concentration. When warm-up is complete, the sensor will track changes in humidity within a few seconds.

Figure 6 Hygrometer start-up and run
Fig. 6. An example of a strip-chart record during two warm-ups of a dewpoint hygrometer as described in this paper. 'A' shows when power was first applied after doping the bobbin with LiCl, and 'B' points to when warm-up was complete. 'C' is a later power-up; 'D' shows the smoothing effect of adding a shield over the hygrometer. 'E' shows when room air conditioning was switched on, drying the air. And, 'F' shows further drying after the door of the room was closed.

Any type of electrical thermometer may be chosen for the hygrometer, subject to the heat-loss precautions discussed previously (smaller is better) A thermistor was a convenient choice here, because it is easily interfaced to a strip chart recorder like the Rustrak 5 288 using a bridge circuit (Figure 7), which was designed to be adjustable to zero current output when the thermistor is 27.8°C, and 50µA output when the thermistor is 61.5°C. These bobbin temperatures correspond to dewpoints of -5 and +20°C, respectively. This corresponded to the markings on the strip chart (Style I chart paper, p/n C2256), and could then be read directly in dewpoint units (as in Figure 6). Typical measurement precision, using a YSI 4 44005 thermistor in this circuit and a Rustrak5 Model 288 recorder, is within 0.5°C over this range. This type of recorder is well suited to examining the characteristics of slowly changing processes such as this. For field deployment, one would select from among the many analog-to-digital recorders, which may require development of an interface circuit for it and deciding on what sampling interval Δt to use.

Figure 7 Bridge circuit to get a dewpoint signal
Fig. 7. The bridge circuit that was used for driving the 50µA galvanometer (observe hookup polarity!) in a Rustrak5 Model 288 strip chart recorder. A YSI4 44005 thermistor was used as the sensor immersed in heat sink grease within the hollow core of the bobbin heater. Bridge excitation was provided by a 1.35V cell, famous for its stable output over time (and, owing to Hg concerns, is likely no longer available). The resistor values in this circuit have been selected to provide a 0 to 50µA current through the galvanometer over a dewpoint range of -5 to 20°C based on the relation TD = 0.743TS - 25.7°C, (Equation 4 in the paper) so that the chart scale can be subdivided as shown in Fig. 6 (y axis). For field deployment of this circuit, all fixed-value resistors must have low temperature coefficients. With this in mind, temperature compensation for the copper-wire-wound galvanometer coil (5400Ω) is provided by another thermistor (YSI4 44003). Socket pin numbers 2 & 7 relate connection points as shown in Fig. 2.

Calibration and field performance

Salt in saturated solution is well known as a method of creating and maintaining an atmosphere of fairly constant relative humidity, RH. And, while useful for roughly establishing a measure of atmospheric humidity control, dewpoint calibration work requires referring to a better standard, as is evident from Figure 8, which shows the dewpoint ranges obtainable with several different saturated salt solutions.

Figure 8 Humidity sources
Fig. 8. Dewpoints obtainable from several different saturated salt solutions; dewpoint variation over temperature is clear. A reference hygrometer, such as an EG&G6 880, is required to measure the actual dewpoint of the calibration atmosphere.

The humidity calibration apparatus used here (Figure 9) is a closed recirculation system with an airstream passing successively through the chosen saturated salt solution, a cooled-mirror reference hygrometer is the calibration standard (EG&G6 Model 880, apparently no longer built), and a PVC-pipe manifold, into which up to 4 dewpoint hygrometers may be fitted for calibration. Repetitive passes of the calibration atmosphere through the selected salt solution is used to achieve stability.

Before and during calibration procedures, take care to maintain a uniform and constant temperature throughout the system; both insulation and thermal shielding are recommended. The use of strip-chart recordings (previously discussed) during calibration provide a check that system humidity has achieved stability. Use of non-water-absorbing plumbing materials will help in this respect, but a long period of time is always required for the system to come into a new equilibrium with a new salt solution. Large pressure drops in the system must be avoided. These are particularly trouble-some when operating near saturation humidity values, where condensation can easily occur, and also can affect the accuracy of the reference hygrometer.

Because I had the space to leave the calibration apparatus set up, I have performed repeat-calibrations on many dewpoint hygrometers of this design when they returned from service in the field. Rarely did any depart from its expected calibration, i.e., conformity to Equation 4. Dewpoint accuracy discrepancies were seldom greater than ±0.5°C, and agreement between hygrometers were about ±0.1°C.

Accordingly, depending upon the accuracy demands of an application, it may sometimes be feasible to bypass calibration and depend solely on Equation 4 when one is primarily interested in dewpoint trends or patterns. User control of other factors, such as the accuracy of the chosen thermometry, its associated interface circuitry, and the method of data acquisition and recording can mask the hygrometer's expected performance. It is recommended that users always inspect the warm-up behavior of each hygrometer before field deployment.

Figure 9 Hygrometer calibration apparatus
Fig. 9. A humidity generation apparatus for dewpoint hygrometer calibration suitable for calibrating four hygrometers at a time. Always have a valve open when the pump is started. Be sure to insulate flasks from rapid temperature changes. Solenoid valves are convenient for switch-selecting amongst salt solutions. Filter the air entering the EG&G6 880 reference hygrometer to prevent contamination of its mirror. Avoid large pressure differences and maintain reference hygrometer near 1000mb.

Of course, some problems in the field were encountered, particularly arising from bad battery management. All personnel have to become familiar with how to insure adequate battery capacity between visits, i.e., one needs to know the battery's amp-hour, or AH, capacity to predict when field visits have to be scheduled. Not having this expertise will result in not only data loss, but in fatal damage to the hygrometer by corrosion. Even when a dewpoint hygrometer appears to be yielding correct readings and seemingly function properly, any corrosion-caused-contamination of the LiCl that occurred during a power loss may rendered any subsequent data useless. In some instances, personnel may try to help by re-doping. Symptomatic of this practice is the growth of large, needle-shaped LiCl crystals between the electrodes. Sometimes this also caused erratic operation by loosening the electrodes. This loosening may also happen due to the shrinking of the protective epoxy paint on the bobbin, emphasizing the need for a lengthy heating period to cure the epoxy and to minimize later dimensional changes that loosen electrode contact with the fabric.


The dewpoint hygrometer design described here has been successfully used in both long- and short-term meteorological studies. All applications have exploited its low power requirement in remote areas using battery supplies. Only rarely has calibration accuracy not been maintained, and then owing to circumstances not related to its operating principle, that is, to battery failure. Its dewpoint response, i.e. temperature-humidity calibration, agrees with a relationship derived by combining Murray's (1967) and Acheson's (1965) results, as derived here. These attributes, in a readily constructed humidity sensor, provide the DIY meteorologist with an inexpensive, yet reliable and accurate, humidity measuring instrument.


Acheson, D.T., 1965. Vapor pressure of saturated salt solutions. In: A. Wexler (Editor), Humidity and Moisture, Vol. III. Reinhold, New York, NY, 687 pp.

Campbell, C.B., Grover, B.L. and Campbell, M.D., 1971. Dew-point hygrometer with constant resistance humidity transducer. J. Appl. Meteorol., 10: 146-151.

Fritschen, L.J., 1965. Accuracy of evapotranspiration determinations by the Bowen ratio method. Bull. Int. Assoc. Sci. Hydrol., 2: 38-48.

Hedlin, C.P. and Trofimenkoff, F.N., 1965. An investigation of the accuracy and response rate of a lithium chloride heated electrical hygrometer. In: A. Wexler (Editor), Humidity and Moisture, Vol I. Reinhold, New York, NY, 687 pp.

Hickes, W.F., 1947. Humidity measurement by a new system. Refrig. Eng., 54: 351-354, 388.

Murray, F.W., 1967. On the computation of saturation vapor pressure. J. Appl. Meteorol., 6: 203-204.

Nelson, D.E. and Amdur, E.J., 1965. The mode of operation of saturation temperature hygrometers based on electrical detection of a salt-solution phase transition. In: A. Wexler (Editor), Humidity and Moisture, Vol. I. Reinhold, New York, NY, 687pp.

Tanner, C.B., 1963. Basic instrumentation and measurements for plant environment and micrometeorology. Soils Bull. 6, Univ. Wisconsin, Madison, approx. 320 pp., mimeo.

Tanner, C.B. and Suomi, V.E., 1956. Lithium chloride dewcel properties and use for dew-point and vapor-pressure gradient measurements. Trans. Am. Geophys. Union, 37: 413-420.

Wexler, A. and Hasegawa, S., 1954. Relative humidity-temperature relationships of some saturated salt solutions in the temperature range 0°C to 50°C. J. Res. Natl. Bur. Stand., 53: 1-26.


1. This paper, which I updated in January 2010, is based on my article (Holbo, H.R., 1981. A dew-point hygrometer for field use. Agric. Meteorol., 24: 117-130.) Paper 1495, Forest Research Laboratory, Oregon State University.

2. The Foxboro name and DEWCEL™ are registered by Invensys Systems, Inc. of Foxboro, Massachusetts. http://ips.invensys.com/en/products/autocontrols/Pages/Foxboro.aspx. And, while DEWCEL is a more convenient term-of-reference than 'heated lithium chloride dewpoint hygrometer', its continued registration make that 'nickname' inadvisable here.

3. Ohmite no longer lists the type U-5818, but does list a Series 200, within which the type B5J150E looks like it might be a suitable replacement. http://www.ohmite.com/cgi-bin/showpage.cgi?product=200_series. Testing of the type B5J150E should be conducted to verify that it can attain the temperature required for LiCl phase-transition.

4. Yellow Springs Instrument components, and the YSI brand, have been transferred to Measurement Specialties. http://www.meas-spec.com/product/t_product.aspx?id=5409. The YSI part numbers listed here can be purchased from Measurement Specialties.

5. Rustrak Model 288 recorders and chart paper rolls are now sold by ISE, Inc. http://www.iseinc.com/Rustrak%20288%20Recorder.htm

6. The EG&G instrument brand appears under URS banner. http://www.urscorp.com/ . But there is no mention of a Model 880. I've not located a comparable chilled-mirror dewpoint instrument. Yankee Environmental Systems, Inc.(YES) has several, though none seems convenient for calibration laboratory use. http://www.yesinc.com/products/met-hyg.html


Mention of trade names or commercial products in this publication are for the edification of the reader, and do not constitute endorsement or recommendation for use.