A 50 mK test bench for demonstration of the readout chain of Athena/X-IFU

The Entropy GmbH (L-series) cryostat¹¹1https://www.entropy-cryogenics.com/ is based on a double stage pulse tube refrigerator providing cooling power at 50 K and 3 K, where a double stage Adiabatic Demagnetisation Refrigerator (ADR) is located, with a gadolinium gallium garnet (GGG) stage at 500 mK and an ferric aluminum alum (FAA) stage at 50 mK. The thermometry of the cryostat is operated by an AVS 47-B resistance bridge, to control the different stages temperature. An additional LakeShore 372 resistance bridge has been installed, dedicated to thermal regulation of the FAA stage during experiments.

A focal plane assembly from NIST and GSFC, called “snout”[6], is attached to the 50 mK stage (Fig. 2). It is based on a 1024-pixel TES array[7] from NASA/GSFC. Two columns of 32 pixels of this array are connected to their associated cold readout electronics[8, 9] provided by NIST. Each pixel is read by a first stage Superconducting Quantum Interference Device (SQUID). These SQUIDs, referred as MUX SQUIDs, can be turned on and off, or addressed, via Flux Actuated Switches (FASs).

The Time-Division Multiplexing (TDM)[10] allows to read sequentially each row of all columns simultaneously. This snout allows us to perform a 2$\times$32 multiplexing which is sufficient for a validation of the X-IFU baseline readout electronics. The multiplexed signal of the first stage SQUIDs is amplified by an array of SQUIDs (SQUID AMP), situated on a cold PCB electronics (3 K card) attached to the 3K plate.

Because TES and SQUID devices are highly sensitive to magnetic field, the snout is enclosed in a niobium shield, superconducting below 9 K, that sets a very low uniform ambient magnetic field needed to run ground experiments on TES microcalorimeters[11]. A field coil has been placed at the level of the TES array. It produces a magnetic field perpendicular to the TES array, set to null the remaining field at the detectors level along this direction.

The characterisation of the thermal performances of the cryostat has been checked with the Goddard and NIST cold detection chain installed:

• -

  the temperature stability is around 5 $\mu$K rms, under optimisation;

• -

  the hold time for one ADR recharge is 15 h at 55 mK. It allows spectra with high enough counting statistics to determine the resolution to a few times 0.01 eV;

• -

  the residual magnetic field at the focal plane level, inside the Nb shield, is around 1$\mu$T when the cryostat is enclosed in a $\mu$-metal shield during cooldown.

All these measurements are in good agreement with all values found during our cryostat full characterisation[12].

The “Tower”[9] (NIST), on top of the cryostat (Fig. 1 and 3 left), is connected to the 3K card through a vacuum feedthrough by a set of flexes thermalized at 50 K. It includes low-noise amplifiers for the TES signals, provides TES and SQUIDs biases, and feedthrough for feedback signals and for row addressing signals from the multiplexing electronics.

The TDM electronics[13] (NASA/GSFC) (Fig. 1 and 3 right) is made of commercial-off-the-shelf (COTS) electronic components. It is composed of the row electronics box which controls the row addressing by sending the sequencing signals to the FAS, via HDMI cables to the Tower. The row electronics box is connected via BNC cables to the column electronics box for clock synchronisation. The column electronics box de-multiplexes the TES signals, for up to 2 columns, and sends back the flux-locked loop (FLL) signals that allows the first stage SQUIDs to stay in a linear zone of their response.

In order to perform a functional validation of the X-IFU instrument prototype warm readout chain, a satisfactory value of 3 eV or better FWHM energy resolution at 5.9 keV for a multiplexed acquisition has been chosen. Optimisation of the signal to noise ratio is performed through a careful EMI (Electro-Magnetic Interference)/EMC (Electro-Magnetic Compatibility) analysis and control implementation:

1. 1.

   As done on NIST and GSFC systems, a strict grounding scheme was implemented on the 50 mK test bench:

   1. (a)

      All measurement electronics and cryostat control electronics are grounded through a single large copper braid;

   2. (b)

      A metal cable path tray is installed between the cryostat structure and the measurement electronic rack.

2. 2.

   An isolation transformer allows to isolate the measurement electronics from any power supply disturbance.

3. 3.

   High-Frequency filtering is implemented at cryostat feedthroughs:

   1. (a)

      The magnet power-supply is filtered with Shaffner Single-stage Filter FN2410/FN2412;

   2. (b)

      LEMO plugs for thermometry are filtered with 560 pF capacitor EEseal filters performing a low-pass filter around 100 kHz.

Ground loops have been removed from the system, such as those created by the cable of the pressure gauge or the shielding of the field coil cable. These paths are removed permanently or just disconnected during experimentation. Further improvements are ongoing.

In order to characterise our end-to-end present detection and readout electronics, a radioactive ⁵⁵Fe source is placed in front of the down looking cryostat on-axis X-ray window. It emits X-ray photons in the Mn K$\alpha$ complex at 5.9 keV. A remote controlled filter wheel selects a specific absorber (various Mylar and Al films) to obtain a typical count rate of 1 photon/pixel/sec.

The optical path, allowing X-ray photons to reach the focal plane assembly inside the cryostat, is composed of a series of apertures at the bottom of each thermal shield. To block visible and infrared photons, each aperture has been filtered:

1. -

   the Nb shield aperture filter is a 25 $\mu$m Al foil;

2. -

   both 3 K and 50 K aperture filters are aluminized mylar foils (6 $\mu$m mylar coated with 200 nm Al);

3. -

   the 300 K window, sustaining the differential pressure between vacuum and atmospheric pressure, is a Luxel “LEX-HT” window providing a good X-ray transmission between 0.2 and 12 keV.

The alignment of the TES array center, the cryostat optical apertures, and the ⁵⁵Fe source, has been checked. The deviation from the cryostat mechanical axis was determined to be smaller than 1 mm for all optical items, excluding all possible vignetting.

We present here the first validation of the full NASA/GSFC and NIST detection chain as implemented in the 50 mK test bench, after performing the EMI/EMC optimisation detailed above. A dedicated integration (more than 6 hours) on a 1$\times$8 pixels TDM subset setup allowed to detect at least 20.000 photons per pixel. Due to some false pulse triggers and multiple-pulse rejections, the number of final records were rather around 12.000 photons per pixel.

In order to retrieve the photon energy thanks to pulse records, we apply optimal filtering technique[14] using the NASA/GSFC X-ray data processing software, that implements all the protocols and algorithms needed. This analysis includes several processes such as the correction of the drift of the energy baseline during acquisition, the lag-phase correction and gain calibration. Spectra of the Mn K$\alpha$ complex were finally generated to assess the end-to-end energy resolution:

1. -

   a 2.8 eV FWHM energy resolution has been measured for a single given channel read in TDM (Fig. 5);

2. -

   a 3.1 eV FWHM energy resolution has been measured for the combined 8 TDM multiplexed pixels of one column (1$\times$8) (Fig. 4).

These results have to be compared to the performance of this detection chain in a GSFC cryostat: 2.1 eV for a single channel and 2.6 eV for a 2$\times$32 TDM acquisition[15]. We assume that the difference might be due to still specific lines close to the sampling rate of our TDM frequency, visible on power spectrum density of our output baseline, together with the thermal stability of our system. We are presently performing tests to quantify the effect of bath temperatures variations on the energy resolution. We also observed that 2 of our 8 pixels have relatively lower resolution than others (3.4 and 3.5 eV). We suppose that these might have reduced the overall energy resolution and caused those “oscillations” on residuals in Fig. 4.

However, the energy resolution in the IRAP-CNES test bench is very close to the 3 eV requirement in TDM for our demonstrations, and further EMI/EMC actions, as well as investigations to improve out thermal stability are in progress. This demonstrates the validation of the 50 mK IRAP-CNES Elsa test bench for performing the end-to-end demonstration of the X-IFU prototype warm readout electronics.