tomcl / addie Goto Github PK

Is your feature request related to a problem? Please describe.

The DC value of a voltage source cannot be changed from the Properties tab. If a different value is required, the Voltage Source has to be deleted and re-added.

Describe the solution you'd like
Voltage Sources should support instant value/type change from the Properties tab, as all other components do.

Describe alternatives you've considered
N/A

Additional context

Screenshot:

The simulator does not work with a simple integrator circuit. An integrator contains 1 capacitor, 1 op-amp and 1 resistor. The error message says there is an unknown error in the circuit.

The simulator can normally support 1 capacitor/inductor/diode. I don't see why this circuit breaks it.

There is an op-amp in there so that might be the problem. When replacing the capacitor with an inductor or a diode, the transient analysis works. Only the capacitor seems to break the simulator in this case.

• OS: MacOs 13.2
• Version 1.0.0

Frequency Response currently works by calculating the magnitude and phase at 20 points per decade (equally spaced in the log axis) starting from $\omega=1$ up to $\omega=10^7$. This covers all the common scenarios and offers relatively good precision.

However, the following two improvements would significantly improve the frequency response offered by ADDIE:

• More points around corner frequencies by inspecting the phase.
• Variable start and end frequency to cover only the regions of interest

Explanations:

1. Phase remains constant in the regions far from the corner frequencies, and changes abruptly near them. This is a characteristic that can be exploited for the purposes of AC Analysis, in order to apply an adaptive step size, instead of a constant one (20 points per decade). Specifically, the algorithm will start again from 1 rad/s with the existing step size (20 points per decade), and then when a change of more than $1^\circ$ is detected between the two previous frequency points, the step size will become 2-10 times smaller (depending on the precision needs). Of course, in order to be able to detect the end of the region of interest, the specified difference between two points that the algorithm looks for will have to be reduced an equal amount of times (2-10). Then, when the difference between the last two points is larger than the updated difference, the step size will be set back to 20 points per decade.
2. Large corner frequencies ($\omega>10^7$) would not be depicted using the current algorithm. Also, if a circuit has only one corner frequency at $\omega=10^6$, seeing the constant magnitude and phase for small frequencies $\omega<10^3$ does not provide any useful information.

Sheets in ADDIE cannot have underscores in their names. When creating a new sheet ADDIE lets yu create a sheets with underscores in their name.

A warning should be given during the creation of a sheet, when the name is invalid.

ADDIE's error messages currently originate from a simple error-checking function that checks for the most common errors in circuits, such as the absence of ground, short-circuits, parallel Voltage Sources, etc.

This could be improved by adding a function that analyzes the nullspace of the MNA matrix, and is able to deduct from there which nodes and/or voltage sources make the circuit not solvable. Then, it should give this information to the users using an informative notification.

Something crucial that is currently not appropriately handled is checking that the opamps are used with negative feedback. Opamps with positive or no feedback can cause erroneous results when used with the current MNA algorithm.

This should be checked by the error-checking function and give appropriate warnings to the users.

Currently, ADDIE supports only diodes when it comes to non-linear components. This is done using the iterative Newton-Raphson method along with Modified Nodal Analysis. More information on the algorithm can be found in the Addie Wiki. The function performing the simulation of non-linear components is called newtonRaphson and is in Simulation.fs.

This issue aims in ADDIE supporting other non-linear components such as MOSFETs and transistors in a similar way as it is done for diodes (using the linear companion models). A good starting point for the linear companion models of MOSFETS and transistors is this paper.

Before implementation, the newton-raphson with MNA method would have to be further evaluated to guarantee that a solution can always be obtained, or explore what should happen when a solution cannot be obtained. Also, a more
advanced problem that will be encountered when dealing with more complex circuits, is finding all
the operating points (DC Solutions) of a non-linear circuit. This is a fundamental problem in the
computer-aided design of analog circuits, and this paper presents a SPICE-Oriented Method for Finding Multiple DC Solutions in Nonlinear Circuits using a deflation-based technique .

PENDING

1. Allow not-zero initial steady-state
2. Transient analysis should explain more clearly how it works – e.g. input voltage is always given a transient 0 -> 1 and is assumed steady-state 0 before that

There is a type of circuit that will not give you any warning messages when trying to do transient analysis, but the output is blank.

If this is due to the simulator not supporting this type of circuit, there should be some sort of error message.

I tried circuits with 1 diode or 1 capacitor and the analysis works. It seems to break down when both a capacitor and a diode are used in a circuit.

When trying a diode and an inductor, I got an error message:

As stated in the error message "Time Simulation currently supports a maximum of one Voltage Source and one Capacitor or Inductor". This circuit has an inductor and a diode and it generates that error message.

I believe that more than one capacitor/inductor/diode or any combination of those will break the simulator. The error message appears in some of those cases but it does not appear in some others (like the one in the first picture).

• OS: MacOs 13.2
• Version 1.0.0

Describe the bug
In the file src/Renderer/UI/SelectedComponentView.fs for a value between 0.1 and 0.99 a maximum value of 0.099 is returned even though it should be 0.99:

|v when (v>=0.1 && v<1) -> 0.1,0.099,0.01

(line 217)

When creating a sheet with a long enough name, the toolbar will move to the right enough to get some buttons off screen. I discovered this while making an opamp circuit sheet.

Here, the ground component is missing from the toolbar.

As a demo of this point, I tried with a longer name...

As seen here, almost all components are off the screen.

The components can still be found in the catalogue on the right hand side, but the ease of the toolbar is lost.

• OS: MacOs 13.2
• Version 1.0.0

In the graph area, when a transient simulation is run, the input and output are labeled as x and y.

This should be changed so that the legend contains the voltage source label (for inputs) and node number (for outputs).

ADDIE should have proper tests that check all of the supported features to ensure that the correctness of the algorithms doesn't break when making changes or improvements. Given that the simulation requires the use of NodeJS packages, the tests cannot run on the .NET environment (as done in ISSIE).

The interface to run the tests is already set up and this is done via the Tests tab on the RHS pane which appears only when ADDIE runs in dev mode (i.e. does not appear when published as a desktop app).

An example of a test exists in Simulator/Tests.fs. To add a test, follow a similar pattern to testCase1 in Simulator/Tests.fs.

The slider used to fast-change the values of components from the Properties tab does not work properly when it is used with very small values (e.g. 40nF for a capacitor). This should be fixed.

Also, the usability of the slider could be improved by:

1. When the slider is used, the value in the box gets changed as well (currently only the on-canvas value changes)
2. The slider should start from the midpoint with its original value and can then go (linear) from 0X to 2X this or maybe (log) 1/10X to 10X.

The current build scripts do not work on macOS.

./build.sh exits with error message:

Script is not valid:
        unknown (1,0)-(1,0): Error FS0193: The specified file name or path is too long, or a component of the specified path is too long.

build.log

The existing build scripts can be updated as in ISSIE to allow build on macOS, i.e. use dotnet fsi build.fsi instead of dotnet fake.