Virtual srf cavity: testing srf cavity support systems without the hassle of liquid helium and klystrons *

Setting up and debugging SRF support systems, such as LLRF control, quench detection, microphonics and Lorentz-force detuning control, etc., often requires extensive time spent operating the cavities. This results in time consuming and costly operation. Early into the development stages the actual cavity system may not even be available. It is therefore highly desirable to pre-evaluate these systems under realistic conditions prior to final testing with the SRF cavities. We devised an FPGA-based "virtual cavity" that takes a regular low-level RF input and generates the signals for RF-power reflection, transmission and detuning that mimic the response of a real cavity system. As far as the user is concerned, the response is the same as for a real cavity. This "black-box" model includes mechanical modes, Lorentz force detuning, a field depended quality factor, quenches and variable input coupling and is currently being expanded. We present the model and show some applications for operating the quench detection, LLRF and microphonics control for 1.3 GHz bERLinPro cavities. The same system can be used for other cavity types, including normal conducting cavities. INTRODUCTION The unavailability of SRF cavities and all the associated ancillary systems for testing them is a big issue when designing and debugging LLRF control algorithms and related techniques like quench detection, detuning compensation, etc. In [1] an FPGA-based virtual cavity was presented which could be used to perform hardware-in-theloop (HIL) simulations of the mentioned systems. This system, based on off-the-shelf National Instruments hardware, takes the forward RF signal coming from the LLRF system and models the electrical behaviour of an SRF cavity generating the RF transmitted and reflected signals. In addition to this basic electrical behaviour, some more advanced features were introduced to the model such as a quenching, field dependant Q and mechanical response fed by Lorentz force detuning and microphonics, making the system more realistic and close the real world. In this paper all these components of the virtual cavity are summarized and a more modern and more compact hardware is presented, which allowed the introduction of one more feature: the simulation of the piezo tuner with its associated mechanical transfer function. The resulting system offers the operator the flexibility to choose between different quality factors, couplings, quench thresholds, mechanical responses while the FPGA calculates in real time the transmitted and reflected RF voltages for a given RF input. Figure 1 depicts the overview of the working principle of this virtual cavity, Figure 1: Overview of the virtual cavity operation scheme. NEW HARDWARE The hardware used in [1] was the main limitation to add more features to the virtual cavity as almost all the FPGA resources were used. Therefore the whole design was migrated to a newer and more compact device from National Instruments [2]. This device is composed by a real time controller in charge of the communications with the user via Ethernet and a Kintex-7 FPGA where the cavity model is implemented. The analog module was also changed, [3], in order to have more number of inputs and outputs, 16-bit ADCs and a bigger dynamic range. Figure 2 shows both devices. Figure 2: National Instruments NI-7935R FlexRIO controller, [2] with the NI-5783 Analog Adapter Module, [3]. ___________________________________________ * Work supported by German Bundesministerium für Bildung und Forschung, Land Berlin, grants of Helmholtz Association and partially supported by Basque Country PPG17/25 project. † pablo.echevarria@helmholtz-berlin.de 19th Int. Conf. on RF Superconductivity SRF2019, Dresden, Germany JACoW Publishing ISBN: 978-3-95450-211-0 doi:10.18429/JACoW-SRF2019-WETEB4 WETEB4 772 Co nt en tf ro m th is w or k m ay be us ed un de rt he te rm so ft he CC BY 3. 0 lic en ce (© 20 19 ). A ny di str ib ut io n of th is w or k m us tm ai nt ai n at tri bu tio n to th e au th or (s ), tit le of th e w or k, pu bl ish er ,a nd D O I. SRF Technology Ancillaries LLRF Figure 3: Graphical user interface where the parameter input, analog output selection and data visualization are depicted. ELECTRICAL MODEL The basic cavity electrical behaviour can be modelled using an equivalent RLC-circuit, [4], which leads to the following equation: � �cav = (−� / −∆� ∆� −� / )�cav + +( � / � / ) �amp (1) where � / is the half bandwidth, ∆� the cavity detuning, L is the load resistance, m is the transformer ratio (both dependent on r/Q, Q0 and the coupling factor) and �cav = (�cav , �cav�), �amp = (�amp , �amp�) are the real and imaginary parts of cavity voltage and the driven current respectively. The reflected voltage is calculated with the following equation: �ref = �cav − 0�amp (2) where Z0 is the impedance of the RF system (normally 50 ohm). Equations (1) and (2), after a proper discretization, have been implemented in the FPGA, where the in-phase and in quadrature components of the input are obtained through an IQ-sampling block. The user interface allows the introduction of the following parameters: r/Q, Q0, Qext, frequency, and detuning and the program calculates all the derived variables as QL, coupling factor, half-bandwidth, rise time, etc. as it can be seen in Fig. 3. The range of these parameters are limited by the fixed point representation inside of the cavity and depend on each other. For instance at 1.3 GHz the minimum QL is around 1.75×105 and the maximum around 5×1011. MORE REALISTIC FEATURES The utilization of an FPGA allows that the parameters in equations (1) and (2) are not only given by the user but also changed dynamically and in real time. For instance the amplitude of the transmitted voltage can be calculated and used to change Q0 to the minimum possible value when the field is above a threshold given by the user, simulating this way a quench of the SRF cavity. The amplitude can be also used to address a look-up table where different Q0 values are stored and fed back to the cavity model, so the system presents a field dependent Q0. Finally, the square of the amplitude value can be added to a signal coming from the host simulating microphonics and fed to a block that implements the mechanical transfer function of the cavity. Up to five mechanical eigenmodes can be set by the user defined by the natural frequency, � , the quality factor, , and the coupling factor, � . Each of these mechanical modes are defined by: � (∆� � ∆�̇ � ) = (−� � ) (∆� � ∆�̇ � ) + +(−� � ) (�cav +� ) (3) 19th Int. Conf. on RF Superconductivity SRF2019, Dresden, Germany JACoW Publishing ISBN: 978-3-95450-211-0 doi:10.18429/JACoW-SRF2019-WETEB4 SRF Technology Ancillaries LLRF WETEB4 773 Co nt en tf ro m th is w or k m ay be us ed un de rt he te rm so ft he CC BY 3. 0 lic en ce (© 20 19 ). A ny di str ib ut io n of th is w or k m us tm ai nt ai n at tri bu tio n to th e au th or (s ), tit le of th e w or k, pu bl ish er ,a nd D O I.