We demonstrate multiplexed readout of 43 transition edge sensor (TES) bolometers operating at 90 mK using a frequency division multiplexing (FDM) readout chain with bias frequencies ranging from 1 to 3.5 MHz and a typical frequency spacing of 32 kHz. We improve the previously reported performance of our FDM system by two important steps. First, we replace the coplanar wires with microstrip wires, which minimize the cross talk from mutual inductance. From the measured electrical cross talk (ECT) map, the ECT of all pixels is carrier leakage dominated. Only five pixels show an ECT level higher than 1%. Second, we reduce the thermal response speed of the TES detectors by a factor of 20 by increasing the heat capacity of the TES, which allows us to bias all TES detectors below 50% in transition without oscillations. We compare the current-voltage curves and noise spectra of the TESs measured in single-pixel mode and multiplexing mode. We also compare the noise equivalent power (NEP) and the saturation power of the bolometers in both modes, where 38 pixels show less than 10% difference in NEP and 5% difference in saturation power when measured in the two different modes. The measured noise spectrum is in good agreement with the simulated noise based on measured parameters from an impedance measurement, confirming that our TES is dominated by phonon noise.

We have characterized and mapped the electrical cross talk (ECT) of a frequency division multiplexing (FDM) system with a transition edge sensor (TES) bolometer array, which is intended for space applications. By adding a small modulation at 120 Hz to the AC bias voltage of one bolometer and measuring the cross talk response in the current noise spectra of the others simultaneously, we have for the first time mapped the ECT level of 61 pixels with a nominal frequency spacing of 32 kHz in a 61 × 61 matrix and a carrier frequency ranging from 1 MHz to 4 MHz. We find that about 94% of the pixels show an ECT level of less than 0.4%. Only the adjacent pixels reach this level, and the ECT for the rest of the pixels is less than 0.1%. We also observe higher ECT levels, up to 10%, between some of the pixels, which have bundled long, parallel coplanar wires connecting TES bolometers to inductor-capacitor filters. In this case, the high mutual inductances dominate. To mitigate this source of ECT, the coplanar wires should be replaced by microstrip wires in the array. Our study suggests that an FDM system can have a relatively low ECT level, e.g., around 0.4% if the frequency spacing is 30 kHz. Our results successfully demonstrate a low electrical cross talk for a space FDM technology.