Skip to content
/ JISA Public

Standardised instrument control, data handling and simple GUI creation library for making experiment control programs.

License

Notifications You must be signed in to change notification settings

OE-FET/JISA

Repository files navigation

JISA - "LabVIEW? Never heard of her."

JISA is a library that I created, here in the Cavendish Laboratory, because I really (really really really) do not like LabVIEW. If you're ever in the mood to be forcibly bored into an early grave, ask me why. You can read all about it, as well as how to use it, on the JISA Wiki.

In essence then, the purpose of JISA is to act as an alternative (and standardised) means of creating experimental control systems. Written in Java, it can be used in a variety of different languages, but works particularly well in:

  • Java
  • Kotlin (Used in examples below)
  • Python (See PyJISA)

It comprises, largely, of three sections:

  1. Standardised Instrument Control
  2. Data Handling
  3. GUI Building Blocks

Example Usage

Using JISA, one can create measurement programs like this:

with minimal fuss like so:

fun measurement(source: ISource, meter: VMeter, tm: TMeter?) {

    // Define data table
    val I       = Column.ofDoubles("Current", "A")
    val V       = Column.ofDoubles("Voltage", "V")
    val T       = Column.ofDoubles("Temperature", "K")
    val results = ResultList(I, V, T)

    // Create user-input form
    val form     = Form("Parameters")
    val currents = form.addDoubleRange("Currents [A]", Range.linear(-1e-3, +1e-3, 11))
    val iTime    = form.addDoubleField("Integration Time [s]", 20e-3)
    val dTime    = form.addTimeField("Delay Time", 50)

    // Create table and plot, put all together in a grid
    val table = Table("Table of Results", results)
    val plot  = Plot("Plot of Results", results, I, V)
    val grid  = Grid("Measurement", form, table, plot)
    
    // Add a toolbar button to run the measurement
    grid.addToolbarButton("Start") {
        
        // Configure instruments using values from form
        meter.useAutoVoltageRange()
        meter.setIntegrationTime(iTime.value)
        source.setCurrent(currents.value.first())
        meter.turnOn()
        source.turnOn()
        
        // Set each current, wait, and measure in turn
        for (current in currents.value) {
            
            source.setCurrent(current)
            
            Util.sleep(dTime.value)
            
            results.addRow { row ->
                row[I] = current
                row[V] = meter.getVoltage()
                row[T] = tm?.getTemperature() ?: NaN
            }
            
        }
        
        // Turn it all off at the end
        smu.turnOff()
        meter.turnOff()
        
    }
    
    // Show the grid in a window, exit program when closed
    grid.setExitOnClose(true)
    grid.showThenExit()
    
}

Running this by passing it the required instruments, and clicking the "Start" button, results in:

1. Standardised Instrument Control

JISA implements standard interfaces for each "type" of instrument, meaning that instruments are easily interchangeable. For instance:

// Routine that expects a DCPower object
fun powerUp(power: DCPower) {

    power.setCurrentLimit(1.0)
    power.setVoltage(12.0)
    power.turnOn()

}

// Both AgilentE3644A and K2200 implement the DCPower interface
val power1 = AgilentE3644A(GPIBAddress(12))
val power2 = K2200(SerialAddress("COM9"))

// Thus they can both be used without altering the powerUp routine
powerUp(power1)
powerUp(power2)

That is, JISA ensures that no matter how the instrument functions "under the hood", its controls are always presented to you through a standard API. Furthermore, any instrument that can be thought of as being a collection of sub-instruments is treated as such, allowing further interchangeability. For instance, if we connect to a Keithley K2612B series dual-channel SMU, an Agilent SPA, and a Keithley 2450 single-channel SMU like so:

// Connect to instruments
val keithley = K2612B(TCPIPAddress("192.168.0.5"))
val agilent  = Agilent4155X(GPIBAddress(20))
val k2450    = K2450(USBAddress(0x05E6, 0x2450))

then their individual sub-instruments can be extracted as stand-alone instruments. That is, for the Keithley K2612B, its two SMU channels can be extracted as stand-alone SMU objects, whereas the Agilent SPA can be treated as a collection of SMU, VMeter, and VSource objects (representing its SMU, VMU, and VSU channels, respectively), and the Keithley 2450 is simply represented as an SMU object itself.

// Extract lists of sub-instruments
val list1 = keithley[SMU::class]    // All channels with SMU capabilities
val list2 = agilent[SMU::class]     // All channels with SMU capabilities
val list3 = agilent[VMeter::class]  // All channels with voltmeter capabilities
val list4 = agilent[VSource::class] // All channels with voltage-sourcing capabilities

// To extract just the first (i.e., index 0) SMU sub-instruments in each
val smu1 = keithley[SMU::class, 0]
val smu2 = agilent[SMU::class, 0]

// Keithley 2450 is just an SMU object (no sub-instruments)
val smu3 = k2450

Therefore, if we have a measurement routine that expects a single SMU, then it can be run on any of these instruments without changing the routine itself:

data class IVPoint(val V: Double, val I: Double)

// Write a method to be given an SMU channel without having to specify what make/model
fun voltageSweep(smu: SMU): List<IVPoint> {

    // Create list to hold results
    val results = ArrayList<IVPoint>()

    smu.setIntegrationTime(0.1) // Set the integration time
    smu.useAutoRanges()         // Use auto ranging on current and voltage
    smu.setVoltage(0.0)         // Set to source 0.0 V
    smu.turnOn()                // Enable output of channel
    
    // Sweep voltage from 0V to 50V, recording measured currents in list
    for (voltage in Range.linear(0, 50)) {
    
        smu.setVoltage(voltage)
        
        val current  = smu.getCurrent()
        results     += IVPoint(voltage, current)
        
    }
    
    return results

}

// Can pass any SMU to it and it will run without needing to be changed
val results1 = voltageSweep(smu1)
val results2 = voltageSweep(smu2)
val results3 = voltageSweep(smu3)

2. Data Handling

JISA provides a simple means of creating tables of data which can then be directly output as CSV files:

// Create results storage
val V     = Column.ofDecimals("Voltage", "V")
val I     = Column.ofDecimals("Current", "A")
val T     = Column.ofDecimals("Temperature", "K")
val table = ResultList(V, I, T)

// Take 10 readings
repeat(10) {

    // Add data by specifying columns
    table.addRow { row ->
        row[I] = smu.getCurrent()
        row[T] = tc.getTemperature()
        row[V] = smu.getVoltage()
    }
    
}

// Take another 10 readings
repeat(10) {

    // Add data by providing it in column order
    table.addData(
        smu.getVoltage(), 
        smu.getCurrent(), 
        tc.getTemperature()
    )
    
}

// Easy output as CSV file
table.output("data.csv")  

3. GUI Building Blocks

val smu = K236(SerialAddress("COM3"))

// Create user-input panel
val params = Fields("Parameters")

// Add input fields to it
val minV = params.addDoubleField("Min V [V]", 0.0)
val maxV = params.addDoubleField("Max V [V]", 60.0)
val numV = params.addIntegerField("No. Steps", 61)

val V       = Column.ofDecimals("Voltage", "V")
val I       = Column.ofDecimals("Current", "A")
val results = ResultList(V, I)

// Create plot
val plot = Plot("Results", results)

// Add panel and plot to a grid
val grid = Grid("Main Window", params, plot)

// Add start button to toolbar
grid.addToolbarButton("Start Sweep") { // This code will run when clicked

    // Get whatever the current values in the input fields are
    val start = minV.get()
    val stop  = maxV.get()
    val steps = numV.get()

    // Loop over the range they define
    for (voltage in Range.linear(start, stop, steps)) {
        /* do measurements etc */
    }
    
}

// Show the grid in a window
grid.show()

JISA the Polyglot

JISA is written in Java, but because Java is all about magic beans, you can actually use it in almost any language. More specifically, any language that can either be compiled to java byte-code or interpreted by a Java program (so basically anything) can use the library.

For example, here's the same program written in Java, Kotlin, Python and even gosh-darn MATLAB:

Java - Classic style, robust but verbose, like a northern grandparent.

import jisa.devices.smu.K2450;
import jisa.results.*;
import jisa.addresses.*;
import jisa.maths.Range;
import jisa.Util;

public class Main {

    public static void main(String[] args) throws Exception {

        SMU            smu     = new K2450(new GPIBAddress(0, 20));
        Column<Double> V       = Column.ofDoubles("Voltage", "V");
        Column<Double> I       = Column.ofDoubles("Current", "A");
        ResultTable    results = new ResultList(V, I);

        smu.setVoltage(0.0);
        smu.turnOn();

        for (double v : Range.linear(0, 60, 61)) {

            smu.setVoltage(v);
            Util.sleep(500);

            results.addRow(row -> {
                row.set(V, smu.getVoltage());
                row.set(I, smu.getCurrent());
            });

        }

        smu.turnOff();
        results.output("data.csv");

    }

}

Kotlin - Slick, simplified and concise without rocking the boat

import jisa.devices.smu.K2450
import jisa.results.*
import jisa.addresses.*
import jisa.maths.Range
import jisa.Util

fun main() {

    val smu     = K2450(GPIBAddress(0,20))
    val V       = Column.ofDoubles("Voltage", "V")
    val I       = Column.ofDoubles("Current", "A")
    val results = ResultList(V, I)

    smu.voltage = 0.0
    smu.turnOn()

    for (v in Range.linear(0.0, 60.0, 61)) {
    
        smu.voltage = v
        Util.sleep(500)

        results.mapRow(
            V to smu.voltage,
            I to smu.current
        )
        
    }
    
    smu.turnOff()
    results.output("data.csv")

}

Python - "Screw your traditions, I'm a snake from the early 1990s"

To use in regular Python, take a look at PyJISA here. Otherwise, take a look at GraalPy here.

# Import and start PyJISA
import pyjisa.autoload

# Import JISA classes as if they're Python classes
from jisa.devices.smu import K2450
from jisa.addresses import GPIBAddress
from jisa.results import Column, ResultList
from jisa.maths import Range
from jisa import Util

def main():
    
    smu     = K2450(GPIBAddress(0,20))
    V       = Column.ofDoubles("Voltage", "V")
    I       = Column.ofDoubles("Current", "A")
    results = ResultList(V, I)
    
    smu.setVoltage(0.0)
    smu.turnOn()
    
    for v in Range.linear(0.0, 60.0, 61):
        
        smu.setVoltage(v)
        Util.sleep(500)
        
        results.mapRow({
            V: smu.getVoltage(),
            I: smu.getCurrent()
        })


    smu.turnOff()
    results.output("data.csv")


main()

MATLAB - Why?

function main()
    
    smu     = jisa.devices.K2450(JISA.Addresses.GPIBAddress(0,20));
    V       = jisa.results.Column.ofDoubles("Voltage", "V");
    I       = jisa.results.Column.ofDoubles("Current", "I");
    results = jisa.results.ResultList({V, I});

    smu.setVoltage(0.0);
    smu.turnOn();
    
    for v=jisa.maths.Range.linear(0.0, 60.0, 61)
    
        smu.setVoltage(v);
        jisa.Util.sleep(500);
        results.addData([smu.getVoltage(), smu.getCurrent()]);
        
    end
    
    smu.turnOff();
    results.output('data.csv');
    
end

We can then extend this program easily, with only two lines, to display a plot of the results as they come in. Taking the example in Kotlin:

fun main() {

    val smu     = K2450(GPIBAddress(0,20))
    val V       = Column.ofDoubles("Voltage", "V")
    val I       = Column.ofDoubles("Current", "A")
    val results = ResultList(V, I)

    // Make a plot that watches our results
    val plot = Plot("Results", results)
    plot.show()

    smu.setVoltage(0.0)
    smu.turnOn()

    for (v in Range.linear(0.0, 60.0, 61)) {
    
        smu.voltage = v
        Util.sleep(500)

        results.mapRow(
            V to smu.voltage,
            I to smu.current
        )
        
    }
    
    smu.turnOff()
    results.output("data.csv")

}

Resulting in:

Supported Instruments

Some of our currently implemented devices (the list is a bit too long to put here):

Spectrometers have relatively limited support to trigger scans as part of wider routines, we plan to improve this in the future.

Model Type Class
Keithley 236 SMU (Single-Channel) K236
2400 Series SMU (Single-Channel) K2400
2450 SMU (Single-Channel) K2450
2600B Series SMU (Multi-Channel) K2600B
6430 SMU (Single-Channel) K6430
2200 DC Power Supply K2200
2182 Voltmeter K2182
Agilent / Keysight 4155B/C SPA / SMU (Multi-Channel) Agilent4155X
4156B/C SPA / SMU (Multi-Channel) Agilent4156X
B1500A SPA / SMU (Multi-Channel) AgilentB1500A
E3644A DC Power Supply AgilentE3644A
Cary6000i Spectrometer AgilentCary6000i
AIM-TTI TSX3510P DC Power Supply TSX3510P
OI / Mercury ITC-503 Temperature Controller ITC503
IPS-120 Magnet Controller IPS120
ILM-200 He Level Meter ILM200
MercuryITC Temperature Controller MercuryITC
Lake Shore 336 Temperature Controller LS336
Lake Shore 331 Temperature Controller LS331
CryoCon 22C Temperature Controller CryoCon22C
Stanford Research SR830 Dual-Phase Lock-In Amplifier SR830
SR560 Voltage Pre-Amp SR560
Eurotherm 2408 Temperature Controller ET2408
Pico Technology USB-TC08 Thermometer (Multi-Channel) USBTC08
Arroyo TEC Temperature Controller ArroyoTEC
Bruker 70v Spectrometer Bruker70v
Pegasus Pegasus Translation Stage / Prober Pegasus

Prerequisites

Before being able to use JISA, you will need the Java Development Kit (JDK) 11 or newer installed.

Linux

Either you'll already have OpenJDK installed or you simply need to run something like:

sudo apt install openjdk-11-jdk

Windows and MacOS X

You can download pre-built OpenJDK packages (with installers) from the Adoptium (previously known as "Adopt OpenJDK") website:

https://adoptium.net/en-GB/temurin/releases/?version=11&package=jdk&os=any&arch=x64

Using JISA

You can use JISA in your project simply by including the JISA.jar file as a library. This will work so-long as your project uses Java 11 or newer.

If using IntelliJ IDEA, this means adding JISA.jar to your project directory and, inside IDEA, right-clicking on it then selecting "Add as Library..."