It has been a while since my last update for my Keithley 2450 SMU RoadTest, however, it seems that there have been some relatively-recently released improvements to the firmware and KickStart 2 software which are worth noting. While I was aware of these updates a little earlier, I've been quite busy (as has the SMU) so I didn't have an opportunity to apply the updates and evaluate them until now. As a result, I decided to write this "short" blog post to cover these changes.


Firmware Update - Version 1.7.5

The Keithley graphical SMUs received an update on 4th January 2021, version 1.7.5. The firmware can be downloaded for the 2450 from Tektronix's website. Included in the update package is a changelog document:

It seems that this update now fully addresses an issue that was identified during my RoadTest where using the Output Off: High Impedance mode resulted in measurements not recommencing once the output is turned on. This change is a big convenience improvement especially when using the SMU via the front-panel for battery testing (as the High Impedance mode is necessary to avoid unintentional battery discharge).


Updating was straightforward - unzip the files to the root directory of a USB memory device, plug it into the SMU, navigate to the System Info screen and hit "Upgrade to Newer". The process taking several minutes and concludes with a dialogue requesting you reboot the SMU to apply the update.

Once completed, the firmware version will state 1.7.5b.


I can confirm that the issue with High Impedance mode has been fixed - measurement now starts automatically when the output is turned on. However, there is still a legacy that remains ...

... namely that when the output is turned off, the SMU will throw an Error 5078. This "error" merely indicates that the SMU cannot measure with output off in the High Impedance / 4-Wire Modes, but I don't consider it necessarily an error when the output is turned off by operator option. As it is denoted as an error rather than a warning, it results in a popup by default - the only way to eliminate this is to disable pop-ups for all errors and warnings as whole, which is not an optimal solution. This is a design quirk, but it's one that I encounter quite frequently because of the way I use the SMU, so I've disabled error pop-ups entirely and manually navigate to the error log when I need to read error messages.


Keithley Kickstart 2.5.0

A new version of Keithley KickStart 2 has been released since the review - the current version is now Version 2.5.0 which can also be downloaded from Tektronix's website.

This version, released 5th April 2021, has a number of changes. The changelog is excerpted below:

The key change for 2450 owners is the availability of a pulse mode in the I-V Characteriser app. This is selected with a radio button. The use of pulsed measurements can be important especially in semiconductor testing to avoid accumulation of self-heating during testing and to be able to characterise "short term" capability.

It should be noted that the 2450 SMU is not a "pulsing" SMU as such, so the pulsing capabilities are more limited. Pulsed testing can be quite demanding on equipment as it requires relatively high slew rates, rapid regulation feedback and high-speed measurement capabilities to obtain accurate results.

I checked the various limitations - it seems the shortest allowed pulse/off-time is 2ms, with the maximum being 10,000s. The maximum number of pulses is 100,000. In Voltage Pulse Sweep mode, the Voltage Range is set to Best Fixed with bias limit and current limit range dependent on the selected range. The corresponding limitations occur for the Current Pulse Sweep Mode as well. In addition, there are Voltage/Current Pulse Train and List Sweep modes as well. The corresponding measurement accuracy is limited by the pulse times - at 2ms, the NPLC is just 0.025, while at 10ms, the NPLC is 0.25. In these cases, mains-borne interference with measurements are quite likely and limitations with conversion accuracy are likely.

I was able to run a voltage pulse sweep test with a green LED connected to the output. The results are as expected - linear on the left, logarithmic on the right. Measurements below 1uA are noisy, likely because of mains borne interference causing the "cyclic" noise observed below about 1.5V. Regardless, having this pulse capability, even though somewhat limited, can be quite valuable.

I hooked up my Rohde & Schwarz RTM3004 oscilloscope to observe the output from the SMU with the LED connected (on the left) and with nothing connected (open circuit, on the right) running the default configuration 0-5V 21 pulses 10ms at 10mA as per the first screenshot. The pulse train is visible to the right of the trace, but there is also several seconds of about 1.0-1.5V AC peak-to-peak on the output during configuration process. I suspect this may be noise from the environment being picked up while the SMU is not driving the output at all.

The pulses themselves are very well timed - at 10ms on the left, it is very clearly right-on time. Even at 2ms, on the right, the pulses are very much on time with only the slightest curving of the edges during transitions due to slew-rate limitations. This is very encouraging.

A close-up of the AC in the beginning of the test shows that it has a 50Hz frequency which is my local line frequency - I noted that when the cable is disconnected from the SMU, the signal is still seen on the oscilloscope so it is likely to be pick-up from the relatively long test leads.



The latest firmware update from Keithley resolves issues with High Impedance Output Off mode and 4-Wire Sensing mode not returning to continuous measurement when the output is re-enabled, which is a big convenience for front-panel usage under those scenarios (e.g. for battery testing). When the output is disabled, an error pop-up is still presented (even though I think it should really be a warning), but that can be suppressed as long as all error and warning pop-ups are suppressed. The latest update to KickStart 2 brings new pulsed-testing capabilities to the 2450 which is an important functional improvement for testing of semiconductors where self-heating may be an issue or for characterising short-term overload capabilities. While the 2450 is not a "pulsed" SMU, the software allows for pulses as short as 2ms, as long as 10,000s, with the total number of pulses up to 100,000. This mode invokes the use of Best Fixed Range and automatically reduces NPLC to appropriate values which can be as low as 0.025PLC. As a result, pulsed testing will come with measurement accuracy trade-offs with the potential for mains borne noise to influence readings.