Open Source Projects

Low noise tunneling bias

The energy resolution of STM is largely influenced by the noise of the tunneling bias. In part, it comes from the thermal distribution of electrons, i.e. the electronic temperature. In many cases, however, the experimental resolution is significantly worse than the thermal smearing. In many cases, the actual potential is not free of noise caused by electronic noise of the voltage source, rf-signals picked up by the cables and voltage noise caused by induction due to time varying magnetic fields (typically from high power consumers running from the main power lines). Rf-noise can be easily filtered by dedicated low-pass filters but induction is harder to screen. Usually, all magnetic flux that crosses the loop between the grounded IV converter and the grounded voltage source of the tunneling voltage is relatively large limiting the energy resolution to an equivalent of several hundreds of mK. As the tunneling junction is the part with the highest resistance, the complete induction voltage drops here. To prevent this, we have developed an electronics with differential input, that resynthesizes the bias voltage locally next to the IV converter with the ground potential of the IV converter. This minimizes the induction loop to only the space between the voltage and current lines inside the STM chamber. Additionally, it provides low-pass filtering and the option to divide the voltage by a factor 10 or 100. This reduces the voltage noise to below 20μV rms over the full bandwidth of the measurement, or to an equivalent noise temperature of 69mK.  

The common rejection, i.e. rejection of the induction voltage at the output V(out2), is shown in the following two figures (left for a gain setting of 1:100 and right for 1:1). In any case, the rejection is better than 80dB.

    

To reduce high frequency noise on the bias voltage, the electronics is a second order low pass filter with a cutoff frequency of 10kHz. The transmission characteristics at V(out2) are shown below (left for 1:100, right for 1:1). -40dB represents the gain of 1:100.

    

Finally, the voltage noise density at the output is significantly reduced, especially at high frequencies and low gain. It reaches about 3nV/Hz-½ for a gain 1:100 (left) in the frequency window of interest. The noise is naturally larger for a gain of 1:1 (right). To record spectra with best electonics temperature, a low gain should be used.

   

SPICE models and KiCad circuit diagrams and pcb board files are available upon request.

Low-cost 8 channel resistive thermometer driver

During the CIVID pandemic, the supply chain for electronics was essentially down. Purchasing temperature control electronics for cryogenic applications was nearly impossible. Thus, we developed a low-cost alternative based on a Raspberry PI an 16 bit AD and DA converters. The electronics can read 8 resistive thermometers sequentially using a four probe method and a software lock-in amplifier. A programmable constant current source applies a current to the resistors in the range of 10nA-20μA and the voltage drop across the resistor is detected in the range of 10μV-20mV. The circuit is optimized for Pt1000/100 resistors (10Ω-kΩ) for temperatures above room temperature down to tens of Kelvin and Cernox or RuOx sensors (50Ω-10kΩ) for temperatures below tens of Kelvin to mK. The software is written in Python and communication for protocoling the data is via TCP/IP.

SPICE models and KiCad files as well as the rudimentary Python code are available upon request.
Low-cost accelerometer
Running an STM in a vibrationally noisy environment requires damping of vibrations. Accelerometers are used to detect vibrations of the floor, the mechanical frame of the STM or the cryostat. In many situations, you want to measure several axes (x,y and z) at several places at the same time. We developed an low-cost accelerometer based on MEMS chips that measure the acceleration over a wide frequency band (1.8kHz) with a noise floor of 50μgHz-½. It is battery powered and includes an amplifier to allow direct connection to a +-10V data logger (e.g. Nanonis electronics).
SPICE models and KiCAD files are available upon request.
Low-cost He recovery line monitor 
He recovery from experiments is sensitive to leaks. A leaky recovery line may not only lead to He loss but also to air impurities in the recovery line. Purging the low-temperature gas filters also contributes to He loss. The problem is to find the leaks in a He recovery system and monitor the impurities. With the high prices of He, this task becomes more and more important. Best is, to monitor the recovery line in every branch calling for a low-cost option. This was developed in the group within the frame of two Bachelor theses. The low-cost solution is based on a Raspberry Pi electronics that extremely precisely measures the sound velocity of the returning gas achieving a sensitivity for impurities well below 0.1%.
Currently, the project is not open source and we look for a distributor.
Open source STM electronics
We are developing an FPGA based STM electronics with the latest generation of AD and DA converters with a sampling rate of up to 2 MHz at 24 bit sampling. Once the project is finished, we intend to publish it and to invite the community to contribute. As the system has a very low overhead, relatively simple FPGAs can be used such that the electronics will have a BOM of less than 5k€. If you are interested in details, please contact us.