Examining Datasets

Datasets are containers where the results of a simulation are stored. The results are stored in Variables that can be viewed in tabular form in the dataset. Examine the data in the dataset by opening the dataset (double-click it). You can also add new variables to a dataset, used for sweeping or just for examining the data. This section walks through using datasets.

Open Linear1_Data and click the S variable on the left to see (drag the right-hand divider at the bottom of the window up a little to see the text summary):

In the display above the left-hand pane shows all of the result variables (including F, the frequency or independent variable). The right hand pane shows whatever data piece you have clicked on in the left pane. The upper left-corner box is the units of measure (MHz down and dB for the values). The lower right pane (which is usually collapsed) displays a summary of the variable information.

Each type of analysis creates a different dataset. Linear analysis (as seen above) creates data with F (frequency), S, ZPORT (port impedance) and CS (noise correlation matrix).

In addition to seeing the simulation results, Datasets can have short equations to help you analyze and diagnose issues with your circuits.

1. Add a variable (right-click the white area on the left and select Add New Variable...).
2. Type ang(S[2,1]) for the formula, then click OK.
   
3. To get...
   

Creating Datasets

Datasets are created when Analyses run. Some analyses (particularly SPECTRASYS) create more than one dataset. Within the dataset are the fundamental result measurements created by the simulation. Also, a blank dataset can be manually created from the workspace docking window via the "new item" button.

Each dataset is used to contain variables, which can be matrices, vectors, or scalars. These variables are either created by simulation runs or manually by the user. Note that when a dataset is created by a simulation, the data within that dataset is always in MKS. You may display the data in a unit of your choice, but the actual data values are MKS values.

The data within a dataset is also determined by settings on the Analyses. CAYENNE and HARBEC and SPECTRASYS all let you limit which data is created during the simulation run. This can reduce the size of datasets significantly and also reduce their complexity.

To examine a dataset, open it ( alternate: double-click it in the tree ).

Here's a simple HARBEC dataset. Freq is a list of frequencies that the harmonic balance run analyzed. PPORT, VPORT, and ZPORT are (in this case) 6x2 arrays of power, voltage, and impedance at the two ports of the circuit.

Click VPORT on the left to get the tabular display on the right-top. It shows that the frequencies analyzed were 0,1.2,...6 MHZ. The port voltages are shown in columns. The display on the right-bottom (which is usually collapsed) shows the type and size of the clicked data. The single grid-cell (top left corner of the grid) that says MHz:V shows that the units for Frequency are MHz and the voltages are shown in volts (V).

The above is a minimal harbec dataset.

If we rerun HARBEC but enable all of the output options we get this:

Here we can't even fit the entire dataset contents on a page. This dataset has the current and voltage at all nodes and ports (internal and external) as well as waveform data for each node.

Although more complex and intimidating there are many cases where more data is better than less. File storage requirements go way up with this kind of data, however.

Creating Variables

Why add variables to a dataset?

1. Add a variable to examine more closely a piece of data (such as ang(S[2,1]) to examine S21's angle). Don't forget that all measurement data is fundamentally in MKS units.
2. Add a variable to propagate it during a sweep (enable the propagate option in the sweep and it will sweep the variable along with the rest of the measurement data).
3. Add a variable to use in an optimization.

How to add a variable to a dataset

1. Right-click the white area on the left and select Add New Variable...
   
2. Enter a formula into the entry field
   
   here we've added a simple formula to show the approximate second harmonic distortion (V[3]/V[2]) for a HARBEC port voltage measurement (which is seen in the first image at top).
3. For most formulas, the Unit of Measure and Independent Variable will fill themselves in once the formula is parsed.

How to delete a variable from a dataset

• Right-click the variable and select Delete

Using Dataset Variables

You can create variables and analyses will create variables when they run.

To graph a Dataset variable

• Right-click the variable and select a Graph style from the pop-out menu.
  To duplicate a Dataset variable
• Right-click the variable and select Duplicate

To edit a Dataset variable

• You can not edit Measurement variables (variables created during a simulation run). You can edit variables you create. Double-click the variable or right-click it and select Properties from the menu.

To delete a Dataset variable

• You should not delete Measurement variables (variables created during a simulation run). You may delete variables you create. Right-click it and select Delete from the menu.

To view a Dataset variable

• If the variable is an array, click it and the right pane will fill with the array values. If the variable is a scalar the value should be shown in the list on the left.

To export a Dataset variable

• Right-click the variable and select Export. This will export it into an XML data form.

Using Datasets

Datasets are extremely useful for comparing different circuit configurations. You can run a simulation, save the data, then change some parameters, rerun the simulation and compare the two sets of data easily.

Normally, an analysis has the dataset name stored within it. You might set that name to a formula based on the parameters, but it's simpler to just Snapshot or Checkpoint the dataset.

To Checkpoint a Dataset

• Right-click the dataset and select Snapshot. Another dataset named mydata_Snap is created. This Snap dataset contains the numerical data from the  first dataset (all formulas are parsed and converted to data and the formula text is stored in the variable description).
  

To compare PPORT[2] for the two datasets just enter two measurements in a graph or table like this:

The HB1_Data_Snap.VPORT entry says to use the VPORT variable from HB1_Data_Snap.

Note we use db() here because the data in the dataset is in MKS and we want dBV for display.

Importing Variables

Variables can be imported to the dataset from any text file. Access this feature by right-clicking in the variable block of the data set and choosing  "Import Variable".

Browse and select a file. Enable "First Column is Independent Data" if the first column of the data is independent data (swept). Name the variable in the Variable Name field.

The data should be formatted as a list or matrix of numbers. Semicolons (" ; ") and spaces (" ") are used to indicate breaks between values. Other characters are treated as zeroes. Begin the data with !Units unitindep unitdep to define a unit of measure for the data. Other rows that begin with a ! are ignored as comments.
 

Example (Choose Real, check First column as independent) :

Importing Complex Variables:

Complex data can be imported in several formats. A typical usage is shown below, where the independent vector is frequency (MHz) and the dependent is S21 in DB and ANG format. The same conventions apply here as for reals; spaces, tabs, and semicolons define breaks between entries.

Example using rectangular coordinates (Re + Im):

Notes

1. Complex data should come in pairs of columns; two elements are needed to specify a point in the 1D complex space. A warning is given if there is an odd number of columns (excluding the independent vector).
2. To use the dB scale for complex numbers, the unit should be specified as dB; otherwise the absolute scale is used based on whatever unit is defined. For example, input impedance should have a unit of "Ohm" which can also potentially have a phase; thus it cannot be in Ohms and dB simultaneously.
3. Typical units: dB, dBm, dB10, dB20, Abs, Ohms, V, A, mil, pF, nH
4. The independent variable must be real ( this will typically correspond to time or frequency, both of which are real quantities).