next previous
Up: A new readout

 

2. A balanced bias with two square waves and a resistive load

The choice of a periodic bias results from the desire to avoid the contamination of the useful signal by low frequency noises of electrical origin, which is critical for instruments that cannot benefit from a high frequency optical modulation, such as the COBRAS/SAMBA project (Bersanelli et al. 1996). Unlike the systems proposed by Lange et al. (1996), which use a sine wave bias, the system presented here uses a square wave bias. This means that the electrical power dissipated in the bolometer is constant, and therefore the temperature of the bolometer does not oscillate. In consequence, there is no low frequency limit associated with the thermal time constant of the bolometer, and the well developed theory of DC biased bolometers can be directly used. A second original feature of this new system is that the bias is applied with opposite phases to the bolometer and to the load resistor. It is then possible to tune the amplitude of both bias voltages to obtain a signal near zero at the middle point and an optimal operating bias of the bolometer.

 figure248
Figure 2: Electric modulation system for bolometers with a resistive load (Diabolo, May 1994; left) and with a capacitance load (Diabolo, March 1995; right)  

In practice, two very stable square wave voltages of opposite sign (see Fig. 2 (click here), left) are applied to both ends of the bolometer-load resistor bridge, and the middle point is connected to the pre-amplifier. The value of the load resistor tex2html_wrap_inline1101 has to be very large in comparison with the bolometers resistance tex2html_wrap_inline1103 (tex2html_wrap_inline1105), in order to determine a nearly constant electric current (Mather 1982). It is possible to adjust the two bias voltages tex2html_wrap_inline1107 and tex2html_wrap_inline1109, in such a way that the bridge is well-balanced and that the bolometer is optimally biased for the current application. The voltage tex2html_wrap_inline1111 at point S is then equal to zero. When the bridge is not well-balanced, due for example to a change in the incoming radiation, the voltage at point S is given by:
eqnarray260
It is then possible to measure the small changes of the voltage at point S due to the small fluctuations of the bolometers resistor i.e., at first order:
eqnarray269
Since tex2html_wrap_inline1107 and tex2html_wrap_inline1109 are square waves, tex2html_wrap_inline1111 is also a square wave that can be amplified by a conventional pre-amplifier which includes as a first stage a JFET at low temperature (tex2html_wrap_inline1119) used as a follower. However, parasitic capacitances (C1, C2 and C3, due to the resistors and the wires; see Fig. 2 (click here), left) introduce transients (spikes) in tex2html_wrap_inline1111 at each step of the square wave, even when the bridge is balanced. For a half-period of the bias, the shape of these spikes is given by the following equation:
eqnarray279
where tex2html_wrap_inline1129 is the output voltage at point S and, tex2html_wrap_inline1131, tex2html_wrap_inline1133 and tex2html_wrap_inline1135 are equal to:
tex2html_wrap_inline1137, tex2html_wrap_inline1139 and tex2html_wrap_inline1141.

The resulting spike is minimum when the following condition is fulfilled: tex2html_wrap_inline1143. This condition is generally not verified, but can be approached by addition of a capacitance. In practice, it has been found simpler to control the shape of the bias voltage in order to minimize the spikes and the non linearities in the output signal (see the shape of the signal at point S in Fig. 2 (click here)). If a residual transient still exists, it can be discarded from the handled data by introducing a blank delay before integration of the output signal at each half period in the lock-in amplifier.


next previous
Up: A new readout

Copyright by the European Southern Observatory (ESO)
web@ed-phys.fr