---+ Strong Electrothermal Feedback
Introduction
I am trying to simulate a bolometer using Simulink (an addon to MATLAB).
I start from equation (3.6) in Trevor's thesis, which can schematically be written as. (I'm not sure about the derivative and I make some assumptions about G=const)
%BEGINLATEX%
\begin{equation}
\left( P_{sky} + \frac{V_{bias}^2}{R} \right) \frac{1}{1 + i \omega \tau}= G \Delta T
\end{equation}
%ENDLATEX%
where
G and
C are constants and all other variables can change with time.
So this equation gives a way to relate inputs
Psky and
Vbias to temperature
T and TES resistance
R. An additional equation relating
R to
T is required: the shape of the transition.
The arctan function will do for now. The {
R ,
T } closure relation is then:
%BEGINLATEX%
\begin{equation}
\frac{R}{1 \Omega} = \frac{1}{2} + \frac{1}{\pi} \mathrm{arctan}\left( \frac{T-T_c}{T_0} \right)
\end{equation}
%ENDLATEX%
where
T0 is some parameter determining the steepness of the slope.
Or choosing to work in terms of resistance, we can solve for T in order to substitute it into the {
P ,
R ,
T } relation.
%BEGINLATEX%
\[
T = T_c + T_0 \mathrm{tan}\left( \pi \left[ \frac{R}{1 \Omega} - \frac{1}{2} \right] \right)
\]
%ENDLATEX%
Substituting:
%BEGINLATEX%
\begin{equation}
\mathrm{Low Pass} \left( P_{sky} + \frac{V_{bias}^2}{R} \right) = G \left[T_c + T_0 \mathrm{tan}\left( \pi \left[ \frac{R}{1 \Omega} - \frac{1}{2} \right] \right) \right]
\end{equation}
%ENDLATEX%
This equation has no closed form for
R. Two options are:
- Use a root finder to solve the equation for R at each timestep.
- Approximate the transition by a linear function (Taylor expansion about an assumed stable point) and solve for R analytically.
Parameters: First TES Simulation
I used order 1 parameters for frequencies, voltages and power values to avoid numerical issues. The transition used is the arctan function with Tc=0.05, Tc=1 such that the maximal alpha=40 occurs at R=0.5 and T=Tc=1.
Results
The Simulink simulation looks as follows:
where the "Scope" looks at the bolometer resistance. Varying the strength of the sky signal results in the following plots.
So the bias voltage is keeping the TES at
R >= 0.5 and the sky signal can linearly alter the resistance up to
R >> 0.5 where the TES saturates at R = 1.
Bolo I-V curve
The current as a function of carrier voltage amplitude is shown in the following plot. The axes should be:
- X axis: Carrier Voltage (V)
- Y axis: Current (A)
Note the inverse relation between I-V at low voltages, the turnaround and the linear behaviour when the bolometer resistance saturates at 1 Ohm. At the lowest voltages, the input impedance of the current meter takes over and the relation becomes linear at a resistance of 0.1 Ohm +
Rbolo (
T = 0 )
The feature at the high V end of the plot is an artifact (from when I started the simulation).
Amplifier Input Impedence
The input impedence of the amplifier is a non-ideality. The above I-V curve has
Rinput = 0.1 Ohm. The following I-V curves have
Rinput = 0.2 Ohm (top) and
Rinput = 0.4 Ohm (bottom).
The
I_*_V =
const part of the curve is the interesting part, since that is where the power applied to the bolometer is constant. For large input impedances, the curve deviates from
I_*_V =
const behaviour. That deviation implies deviation from strong electrothermal feedback.
-Tijmen
This topic: ColdFeedback
> WebHome > StrongElectrothermalFeedback
Topic revision: r9 - 2011-11-17 - TijmenDeHaan