Get Latest CSE Projects in your Email


Modeling the SPS Feedback and Feedforward Systems for Improved Performance

ABSTRACT

The Super Proton Synchrotron (SPS) is the last link in the chain of accelerators providing protons to the Large Hadron Collider (LHC). The SPS is currently the limiting factor on the maximum number of protons and thus collisions in the LHC. The SPS upgrade is under way to expand the discovery potential of the LHC. The accelerating system — Radio Frequency (RF) — is being improved. Models of the SPS RF feedback systems were developed. These models could assist with design choices, evaluating the upgraded system performance, and anticipate limitations and issues.

THE SPS

Figure 2: Left: the inputted sine wave from the system in figure 1. Right: The result of the noise being added to that sine wave

Figure 2: Left: the inputted sine wave from the system in figure 1. Right: The result of the noise being added to that sine wave

In an input-output system, where some perturbation or noise has been added to the input and distorted it in some way, a feedback system can be added to make the input and output values match. At its most basic, a feedback system works by finding the error between the output and desired value, i.e. subtracting, and adding that value to the input, making the next output closer to the desired. For example, consider the system in figure 1. A simple sign wave is added to a noise generator causing a distorted wave in the output figure 2.

Figures 5(left) and 6(right): With an added delay of 1 to the system, the output becomes unstable even with the gain lowered from 2 to 1

Figures 5(left) and 6(right): With an added delay of 1 to the system, the output becomes unstable even with the gain lowered from 2 to 1

However, as seen in figures 5 and 6, the introduction of a delay causes instability even for a gain of one. While higher gain reduces the time it takes a system to reach the desired value, too much gain will cause immediate instability. The SPS accomplishes this through an RF control system. An RF feedback system works to stabilize the effect of the beam traveling through the cavity. The beam stability is modeled in the time domain. It corrects the signal by comparing it to a desired value, finding the error and applying a gain and adding the correction back to the input.

THE FREQUENCY AND TIME DOMAINS

Figure 7: The phase of the proton bunches as they move through the cavity

Figure 7: The phase of the proton bunches as they move through the cavity

The ultimate goal of this project is to find the RF feedback system parameters that will lead to the best possible performance of the SPS. For our purposes performance is measured in two ways, the stability of the beam and the stability of the RF system itself. The stability of the proton beam is best measured in the time domain. A stable beam is characterized by a steady phase of the proton bunches.

RADIO FREQUENCY SYSTEM STABILITY

Figure 8: A typical Bode plot, the top is amplitude, or magnitude, against frequency and the bottom is phase against frequency

Figure 8: A typical Bode plot, the top is amplitude, or magnitude, against frequency and the bottom is phase against frequency

Bode plots are used to represent the frequency response of a system and offer two measures of stability. One measure is known as the gain margin. By plotting amplitude against frequency the response of the signal can be seen over all frequencies. The gain margin refers to the part of the plot where the phase is 180°. If the magnitude is greater than or equal to one at this point, then the system is unstable.

CREATING THE SIMULINK MODEL

Figure 10: The complete Simulink block diagram of the short cavity of the 200MHz feedback system

Figure 10: The complete Simulink block diagram of the short cavity of the 200MHz feedback system

The 200MHz RF system consists of 4 cavities, two short cavities and two long cavities. The figure above is the Simulink model for the short cavity. Each pair of cavities is modeled by a “macrocavity” with the same characteristics but twice the voltage. It is beneficial for verification purposes to build each model (short and long cavities) separately, to be combined later. The Simulink model does not work on its own, it requires a Matlab script to describe the variables.

MATCHING DATA

Figure 13: Example of matching the SPS data with the Simulink model for the short cavity. Blue is the SPS data while red is the output of the model, Vshort from figure 11

Figure 13: Example of matching the SPS data with the Simulink model for the short cavity. Blue is the SPS data while red is the output of the model, Vshort from figure 11

As always the same was carried out for the long cavity. The best fits for the short cavity for high and low pass gains were used from earlier analysis with good agreement. The long cavity had less success in gain analysis which will be remedied when the delay is analyzed. The data, for both cavities, won’t fully match until the delay is dealt with. Using a similar process with the delay, and a few more tweaks of the gain, the simulation will come closer to attaining an accurate replication of SPS.

SENSITIVITY ANALYSIS

Figure 15: Increasing the delay decreases the amplitude of the initial perturbation, while increasing the disturbances at the end

Figure 15: Increasing the delay decreases the amplitude of the initial perturbation, while increasing the disturbances at the end

There is also a delay that acts between the high and low pass gain in the RF feedback system. Like the loop delay, it is best used for matching the initial spikes and end shape of the response.

Figure 18: The low pass gain changes the shape of the response to the beam

Figure 18: The low pass gain changes the shape of the response to the beam

The low pass gain changes the shape of the response to the beam. Lower gain will create a steeper, more pronounced disturbance. Its best to analyze this in conjunction with the high pass gain to find the correct ratio.

CONCLUSION

The Super-Proton Synchrotron was successful modeled in both the frequency and time domains. Using the frequency domain model, significant gain and phase margins were estimated for the upgraded system. A time domain model was also created with Matlab and Simulink and has been validated with SPS data. Once the data is fully replicated with the simulation, these two models can be used in conjunction as a tool for optimizing the SPS facility, providing useful information to the system designers, and estimating the limitations of the upgraded system before it is even operational.

The time domain model was also used to check beam stability as a function of various operational parameters. The initial analysis presented here, seems to indicate that better beam performance can be attained with a different set of parameters, in particular with adjustments to the low pass gain.

Source: Calpoly
Author: Jake Hargrove

Download Project

>> Latest 50+ IoT based Security Projects for Engineering Students

>> Top IoT Projects using Matlab for B.E/B.Tech Students

For Free CSE Project Downloads:

Enter your email address:
( Its Free 100% )


Leave a Comment

Your email address will not be published. Required fields are marked *