Full text loading...
Top: Schematic and photographs of a four-row code-division multiplexer. The dc flux offset is applied to one of the two SQUID switches in each SPDT switch. The row address lines (, , , ) then modulate the flux in each SPDT by switching between zero or flux. The result is that one half of the SPDT switch is open, while the other half is closed, so the input signal, , couples to the summing coil through either or . The row address currents are orthogonal Walsh functions as shown to the left of the schematic. The signals from all SPDT switches are coupled to the SA SQUID amplifier through a summing coil. Room temperature digital-feedback electronics (DFB) are used to servo the feedback current, , which keeps the SA output in its linear regime. The second dimension of CDM would come from common row addressing lines for multiple CDM columns (not shown, but similar to TDM). Bottom: Schematic of SPDT integrated with a TES, showing how a dc biased TES can be connected to an SPDT switch with the same components used in TDM systems: a Nyquist inductor, , to limit the TES bandwidth and a shunt resistor, , to provide a hard voltage bias for the TES.
SPDT switch curve measurements: normalized feedback current, , vs input current, . Positive (solid) and negative (dashed) polarity switch curves are shown for three inputs. The linear region between is where one SQUID switch remains superconducting, while the inflection points near are indicative of the SQUID switch critical currents and the transition to the normal state. These data are used to extract the switch-to-SA mutual inductance , the switch mutual inductance , and the consistency of the SPDT coupling for all rows ( for both slopes of three input channels). The independent integral linearity error of all SQUID switch curves was measured to be smaller than 1% between .
Demonstration of four-row CDM. Top: Raw DFB data, . The four sequential Walsh states are interleaved in time. There is an arbitrary offset in the feedback data; however, the relative dc levels enable extraction of the dc current for all switched rows. (States “c” and “d” have been shifted down by and , respectively, for clarity.) Bottom: Decoded signals converted into units of . The square-wave signal input to the nonswitched pixel (black) clearly shows more drift and pickup than the three switched signals (triangle, sinusoid, and ramp). The dc offsets of the switched signals are properly calibrated after decoding, while the nonswitched signal contains an arbitrary SQUID readout offset.
Noise measurements of a four-row CDM multiplexer. Each panel shows Fourier transformed data from one of the decoded Walsh functions, or rows, as specified in the legend. The polarity switching rows have strongly suppressed knees and pickup of parasitic lines compared to the nonswitching (top) row. Fits to the data below 500 Hz (dashed lines) indicate that the nonswitching row has a knee at (consistent with nonmultiplexed noise measurements), while the switching rows show no evidence of a knee above 20 mHz. The data also confirm that there is no SQUID noise aliasing degradation due to the number of multiplexed rows. The low-pass filtering apparent near is due to the PI term amplitudes in the feedback loop.
Article metrics loading...