This blog focuses on using the Keithley KickStart 2 software with the 2450 SMU, which provides the I-V Characterizer function for PC-connected I-V tracing.


*PLEASE READ* - This Review is with Caveats!

Unfortunately, the SMU I received appears to have a hardware fault causing problems with its 200V range and also produces some occasional unexpected readings even at lower voltages. I have been in contact with representatives from both the Singapore and US branches of Tektronix since 22nd May 2020 reporting my findings and attempting to work through the issues towards a resolution. However, in part due to COVID-19 disruptions, it did not appear that a full resolution of the issues would be forthcoming prior to the deadline for review delivery.


While reading this review, please keep in mind that the unit I received does have hardware faults which are likely to have directly contributed to the issues observed noted at the end of this chapter.


Installing Keithley KickStart 2

The KickStart 2 software is intended as a start-up software to let users get up and running with their instruments immediately without any programming, supporting a range of Keithley and Tektronix products including SMUs, oscilloscopes, DAQs, DMMs and PSUs. It consists of a number of apps, most of which are hardware specific. The software is licensed with the KICKSTART-FL-BASE license, but a free 60-day full-featured evaluation is provided and used to perform this review.


The latest version of KickStart 2 can be downloaded from Tektronix’s website. At the time of this review, this was version 2.3.0.

The installation process was straightforward, however, towards the end of the installation, permission is asked for the collection of anonymous usage data. This seems to be more of a “thing” with free programs rather than licensed software, so it seemed a bit out of place to me. The installation is completed with restarting the computer.

KickStart 2 launches with its splash screen. As I am using the free trial, I am reminded about this at every launch and with a count-down displayed in the title-bar of the program in red.

The home screen of the app shows a list of auto-detected instruments which can be changed to a list of apps. The licenses are configured through the settings dialogue. Apps can be launched either from the apps list or by double-clicking on the instrument which brings up the app selector.

This chapter will focus on the I-V Characterizer App. For I-V Tracer details, please see the next chapter of this review.


I-V Characterizer App

The I-V Characterizer App is one of two available apps for the 2450 SMU within KickStart. This app is the main attraction for SMUs, allowing for up to four SMUs to be controlled by the app to perform PC-connected I-V sweeps of devices.

On first launch, if your device is in SCPI language mode, KickStart 2 will ask for permission to change the command set to TSP. This is necessary to run the I-V Characterizer.

The main I-V Characterizer screen allows the user to configure the sweep on the SMUs. The configurations are rather flexible, allowing for voltage/current sweep/bias/list, dual sweeps, linear/log sweeps with a variable number of steps, delay, repetition and various limits. Which measurement features are active can also be configured to improve speed or accuracy, select which input terminals are used, compensate for high-capacitance loads, configure sensing modes and output off modes.

The data is tabulated and graphed on the computer and can be exported to CSV, Excel and graph images at various resolutions. In the example above, the graph illustrates the behaviour of a 45VDC-rated MOV being swept with a 20mA current limit.

Above are examples of directly exported graphs of a white LED in reverse bias, in forward bias, a 9V relay coil being swept up to 20V and a 1Mohm resistor.

Exported data can easily be plotted and analysed in other software, such as Microsoft Excel. This is an example of an NE-2 neon bulb being swept in the positive direction. Initially it is struck and lit, extinguishes, before re-lighting as the strike voltage is hit and then reducing to run voltage almost immediately. The negative characteristic of such gas discharge tubes is easily demonstrated.


Experiment: I-V Curves of Diodes

As I-V testing is typically associated with semiconductor devices, I decided to do some testing of commodity parts that I had to hand. I start with diodes, since they are the simplest – the venerable 1N4004 power and 1N4148 signal diodes.

As I was not afraid of destroying the diodes, I swept both of them up to a current of 1A (the maximum the 2450 is capable of) even though the 1N4148 is rated at 300mA. The effects of exceeding the current limit significantly can be seen, especially above 600mA of forward current where voltage and power dissipation would have increased dramatically. By comparison, the 1N4004 wasn’t breaking a sweat – this is no big surprise.


When plotted on a log scale, the 2450’s impressive accuracy starts to become clear – the diode characteristic begins even at very low voltages, but is hard to see on a linear scale. Currents below 1nA seem to be smoothly captured despite the “crude” banana-clip and glass coffee-jar setup (for protection against high voltage and exploding components).

With the ability to generate -200V, I thought it would be good to try and see what the reverse breakdown characteristic of the diodes looked like. As the 1N4004 is rated for 400V, it would survive this very easily, passing no more than half a microamp of leakage current. Again, the graph shows smooth currents down well below this. The 1N4148 is rated for around 75V, but this unit may have been damaged by ESD and only withstood about 40V of reverse bias before its current shot up avalanche-style, rapidly hitting into the current limiter (off graph scale).


This shows how convenient the data export feature is from KickStart 2’s I-V Characterizer. True to its mission, I did not need to code anything to run these tests and perform these analyses. Let’s push on a little further and try something more interesting.


Experiment: I-V Curves of LEDs in Forward & Reverse

This section might as well be titled “I have some 5mm LEDs and I’m not afraid to destroy them,” as this section looks at such LEDs almost exclusively. It’s common knowledge that different coloured LEDs generally use different semiconductor materials to produce their characteristic wavelengths (with the exception of phosphor-based LEDs). The band-gap of each of these semiconductors results in a different forward voltage, which can be easily measured.

Eight different LEDs were curve traced, resulting in the following traces. There will be unit-to-unit variances, but as expected, all showed the expected I-V characteristic, although the forward voltages and slope of the curve differed slightly. Between the high-intensity and low-intensity LED variants, amber seemed very similar and red was not entirely dissimilar either.


Because the LEDs typically do not begin conducting until higher voltages compared to diodes, this allows a chance to see the low-current measurement abilities of the 2450. Keep in mind that this was done using the front panel banana jacks on a crude setup, but it seems extremely impressive that the background noise is around tens of picoamps without any special precautions. The closest I’ve been able to get with conventional power supplies and digital multimeters is only within the hundreds of nanoamps, so this is a big “leap” for low-level measurements.


But one big question remained – what happens to LEDs in reverse bias? I’ve been taught to avoid reverse biasing LEDs and to not use them as diodes because they would be damaged by as little as 5V of reverse bias. I’m not so sure about that. So I decided to sweep down to -200V with a current limit of 20mA to see what would happen to these set of eight LEDs.

The answer seems to depend on the type of LED. The red high intensity was first to give out at about 30V of reverse bias, followed by the blue and green high intensity variants at about 45v. Amber high-intensity managed to make it to 65V, while white managed almost 90V. The low intensity LEDs generally fared much better – the red LED failed by 190V while the green tolerated 200V while having some current leak by. The amber LED appeared to tolerate 200V with no measurable current flow. After failure, the current flow was achieved at voltages near zero, so the failure was destructive to the device.


What sort of destruction happened to the device? I decided to try and find out using a high-intensity amber LED, sweeping it forward, then back, then forward repeatedly to see how the I-V curves evolved.

The first run is the perfect I-V curve with very low current until the forward voltage is hit. After successive reverse polarity damage, the first time results in a sudden “peak” in current consumption early on in the sweep before settling back down. Later sweeps also had some of these discontinuities. The LED still seemed to be visually functioning and creating light, but perhaps not as efficiently as regions may have become shunted and then “burned away”.


The difference is much starker when plotted in the log scale – the LED basically has become “leaky” at lower voltages. This is perhaps a good way to detect LEDs that have been damaged.

After the initial destructive reverse bias run, successive runs showed that reverse voltage tolerance was much diminished – in essence, the diode was less of a diode and more like a resistor, at least at first. This indicates damage to the junction and possibly formations of shunts. The moral of the story? Don’t exceed the reverse breakdown voltage – and if you do, respect any current/power limits to avoid permanent damage. This experiment was done with the expectation and intention of damage … so I count this to be a success!


Issues of Note

While testing, a number of interesting issues and limitations seemed to crop up. Some of these are almost certainly related to the unit I received, while others are perhaps a little more curious.

The first thing to notice is that the provided CD in the package contains the original KickStart rather than the later KickStart 2. This version did not require a license to my knowledge, so I attempted to install it anyway on the understanding it would be missing the functionality to bring newer features (e.g. I-V Tracer App Installation) to the SMU. Unfortunately, I couldn’t get it to work properly on my Windows 10-based desktop running NI-VISA 19.5. It would install and run, but failed to detect the instrument which was connected by LAN. Perhaps with some persistence, I might be able to get it work but its inclusion in the bundle seemed a bit strange given that it seems to be six-year-old software.

During high-voltage (200V range) tests, when the voltage swept near 60-80V positive, the unit would buzz, hiss and the relays would click out as long as there was some load on the output. Overheating messages, loss of instrument connection, corrupted instrument displays and unusual voltages would result. While investigations are ongoing, it appears that this is most likely due to a hardware fault within the 2450 SMU I received.

This resulted in strange-looking I-V curves with data in the tables that did not make sense. Also, in some cases, even operation at low voltages resulted in “overflow” values being recorded in the data table which resulted in non-sensical I-V plots. Again, this seems highly likely to be related with the other hardware fault, although the investigation is ongoing. For the purposes of the above results, some tests were repeated a number of times to obtain overflow-free results, or invalid data points were removed from the test data.

KickStart 2 also occasionally failed to trace when commanded and spurious errors appeared on longer testing runs. It seems the software may not be entirely free of bugs, especially when used for longer periods or after starting, aborting and restarting traces in quick succession.


It seems that aside from I-V Characterizer and I-V Tracer, there are no other modes for the SMU within the software. Thus, the software can’t offer features such as data logging which you might be able to do with the DMM-related modules which could easily have extended to the SMUs, which is unfortunate.



The KickStart 2 software is intended as a start-up software to let users get up and running with their instruments immediately without any programming, supporting a range of Keithley and Tektronix products including SMUs, oscilloscopes, DAQs, DMMs and PSUs. It consists of a number of standard apps and add-on apps. It is licensed as KICKSTART-FL-BASE option, with a 60-day fully-featured evaluation period which is used for this RoadTest.


For the SMU, KickStart 2 offers a competent I-V Characterizer app that has support for multiple SMUs with most of the necessary settings allowing for customisation of the sweep type and parameters. It offers a table of values and graph which is updated in real-time as the test progresses, along with a progress bar that offers estimated times. It allows data to be exported in CSV, Excel and PNG screenshot image formats at various resolutions.


Using KickStart 2 and the 2450 SMU, I was able to characterize a variety of devices, including diodes, LEDs, metal-oxide varistors (MOVs) and NE-2 neon bulbs in forward and reverse bias. The exported data was used to build the graphs which show the differences in behaviour between different devices. The data illustrated the superior low-level capabilities of an SMU compared to power supplies and digital multimeters I had previously used, with the background current noise in the tens of picoamps even when using banana leads on the front panel. Previously, I was only able to measure down into the hundreds of nanoamps.


The software generally performed well, with some anomalies experienced likely due to hardware faults in the 2450 SMU that I received. Despite this, I noted the occasional spurious bug that required a restart of the program. KickStart 2 is also a little less flexible than I imagined, as other apps are not available for SMUs. I would imagine something like data logging, similar to what might be available for DMMs, could be easily adaptable to the SMU but is not offered. It was also discovered that the CD included contained the original version of KickStart from 2014 which had issues detecting the instrument when installed on my Windows 10 desktop machine.


Users will have to evaluate whether KickStart 2 provides the functionality and value they expect, as the price of a KICKSTART-FL-BASE license is not inconsiderable and its capabilities are still relatively limited. Of course, if one is comfortable with programming their own solutions either using SCPI or TSP remote control, such options are fully available as an alternative.



This blog is part of the Keithley 2450 SMU with I-V Tracer RoadTest.