Recent Evolutions of the Microcalorimeter Technique

The microcalorimeter technique continues to be the key technique for the realization of the primary power standards in the high frequency, because at the state of the art, only it allows tracing the power standards to the direct current standard, that is a SI quantity. This traceability is obtained through the determination of the effective efficiency, usually indicated by € ηe, of a power sensor as a frequency function. Even though since long time the microcalorimeter is a highly explored measurement system, the first realizations being dated from the late of 1950s, still it is possible to propose improvements that should finally increase its accuracy. Actually this one ranges realistically from 0.2% to 2%, if a significant frequency band is considered, e.g. from the radio frequencies to the millimeter waves. Better values are obtainable below 1 GHz, but beyond 18 GHz a 2% still is a challenge, particularly for power sensors in coaxial line of 3.5 mm, 2.92 mm and 2.4 mm. Waveguide power sensors allow to obtain the mentioned accuracy more easily, but in some frequency bands they are no longer available and coaxial solution could be mandatory. In the paper we highlight the limiting factors of the coaxial microcalorimeter accuracy and how we worked around to improve the technique. INTRODUCTION Microcalorimeter was developed for bolometer sensors based on thermistors mainly, because these allow measuring the high frequency (HF) power by means of the DC power substitution method. Up to now thermistors have been working well, even if they are downward frequency limited, that is, not usable below 10 MHz or more realistically 50 MHz. However, power sensors based on indirect heating thermocouples have been successfully used as an alternative to the thermistors for realizing the HF power standard up to 26.5 GHz in 3.5 mm coaxial line. The upper frequency limit has been recently extended up to 40 GHz in 2.92 mm coaxial line and in a near future is expected up to 50 GHz by using the 2.4 mm coaxial line. The change in the microcalorimeter technique is more then an option, because thermistors are no more available in waveguide mounts and rumors exist that even the production of the coaxial versions could be suspended by the manufactures in favor of the more commercial thermocouples or diodes. Anyway, the use of the thermocouples in realizing of the HF power standards turned out to be an improvement to the technique, in some sense. Thermocouples are not downward frequency limited; they can accept as reference power both DC power and low frequency (LF) power; this sensor type, that still is a true r.m.s. sensor, is not sensitive to the absolute temperature like thermistors, a particular that make the device more efficient in the power standard transfer. Thermocouples have been used as traveling standards in an international CCEM comparison (CCEM.RF-K10.CL) in which the authors experimented for the first time a new twin-type coaxial microcalorimeter specifically optimized for such sensors. Using thermocouples as transfer standards, the classic microcalorimeter technique requires some small changes and the final result is a more simple calibration process. The limiting factor in the microcalorimeter technique is related to the losses of the insulating line that supply HF or both HF and LF to power sensor under calibration. The advances in the hardware are no more enough efficient because beyond 1 GHz the line losses may not be reduced up to be negligible, even through a more accurate fabrication process of the items. The right way is therefore that to determine the losses or mainly the effect they have on the measurand. In other words it is necessary to determine accurately the efficiency of the microcalorimeter by means of a calibration process. This is everything but not academic, because it requires changes in the hardware configuration of the microcalorimeter, with the possibility to measure irrelevant parameters for the power sensor under calibration. MICROCALORIMETER The Microcalorimeter is a relatively simple measurement system adjusted for measuring power ratios rather then absolute power levels [1]. A thermostat insulates a power sensor that realizes the thermal load of the microcalorimeter from the external environment while transmission lines with low thermal conductance supply the power sensor alternatively with HF power and DC or LF reference power levels. This power substitution allows the development of two different states of thermodynamic equilibrium inside the thermostat, while a thermopile detects continuously the temperature variations of the thermal load associated to each state change. Based on the experimentally verified assumption that the thermopile response is linear with the temperature, the requested quantity € ηe, the effective efficiency of the power sensor [2], is determined as ratio of two thermopile outputs, each one corresponding to different thermal equilibrium conditions. Several microcalorimeter designs were developed [1], but the twin type model, as schematized in Fig. 1, reveals to be the most efficient [3]. In this version the thermal load consists of a couple of twin power sensors, one of which is never supplied with HF-power because has to work only as thermal reference. Originally microcalorimeter was developed for bolometers, resistive power sensors needing a DC-bias. This fixes the working point of the sensors and allows measuring the HF-power by means of the DC power substitution method [1]. Bolometers probably will disappear from the market, but they may be substituted more efficiently by thermocouples that already have been successfully used in a CCEM key comparison [5]. Indirect heating thermocouple are not downward frequency limited like bolometers, furthermore they are not so sensitive to the absolute temperature that is a well appreciated feature for a standard. The use of thermocouples is itself an improvement to the microcalorimeter technique therefore. The accuracy of the technique is limited mainly by the parasitic losses on the insulation lines to the range (0.2 – 2.0)%, being the best value obtainable with waveguide mounts or only below 1 GHz about if coaxial power sensors are considered. The diminution of the losses and their consequence on the measurand are not possible by means hardware refinements only, especially in the case of the coaxial lines with small diameter. Therefore, for the increasing of the microcalorimeter accuracy, it is necessary to measure the losses and to correct for their effects on the measurand. All this consists in the microcalorimeter calibration, an operation not trivial because it requires the change of the electrical and the thermodynamic configuration of the system, with the risk to get irrelevant results for the power sensor under calibration. The authors have developed calibration processes that reduce the mentioned risk and along with a more careful data analysis should produce a significant improvement of the microcalorimeter accuracy [5]. In the following we present this process applied to the thermoelectric power sensors. MICROCALORIMETER CALIBRATION AND MEASUREMENT THEORY As reported in previous papers [1], [3], [6] the microcalorimeter behaviour is described by the electro-thermal equation: € e =αR K1PinS + K2PinL ( )HF (1) Fig. 1. Twin type microcalorimeter scheme. It follows as an application of the superimposition principle of the linear effects in our measurement system and € e is the thermopile output voltage, € PinS the total power dissipated in the sensor mount, € PinL the feeding line losses and