Several measurement considerations when selecting a modular source measure unit (SMU)

1. IV range

Selecting a source measure unit (SMU) with the appropriate voltage and current range for your device is critical to the success of your application. The IV range is typically represented by the quadrant diagram in Figure 1, which refers to the voltage and current values ​​that the SMU can source or sink. The terms source and sink describe the flow of power into and out of a device. A device that sources current provides current to a load, while a device that sinks current acts like a load, passively absorbing the incoming current and providing a return path for the current.

Figure 1. The four quadrants represent the current or voltage that a device can source or sink.

In the quadrant diagram above, quadrants I and III represent devices sourcing current, while quadrants II and IV represent devices sinking current. Devices that can source current in both quadrants I and III are sometimes referred to as bipolar because they can source both positive and negative voltages and currents. The term "four-quadrant SMU" is often used to describe bipolar SMUs that can source and sink current.

For example, the NI PXI-4132  four-quadrant SMU has a maximum voltage output of 100 V and a maximum current output of 100 mA; however, it cannot output 100 V and 100 mA simultaneously. In this case, the quadrant diagram provides the information needed to easily determine the maximum voltage and current combination that the SMU can source or sink. Simply listing the maximum voltage and current for an SMU with multiple ranges does not provide enough information to determine if the instrument meets the IV requirements of the device.

Figure 2. NI PXI-4132 IV Range

Table 1 summarizes the input and output capabilities per channel of each NI power supply and SMU device.

Table 1. Input and output capabilities of each channel of NI instrument products

2. Accuracy

The measurement resolution of a power supply or SMU is the smallest change in voltage or current measurement that can be detected by the hardware. The output resolution of a power supply or SMU output channel is the smallest possible change in output voltage or current level. These measurements are usually expressed in absolute units, such as nV or pA. Resolution is usually determined by the analog-to-digital converter (ADC) used for the measurement, but high-precision SMUs are often limited by other factors such as noise.

Sensitivity is the smallest unit of a given parameter that an instrument can detect and meaningfully measure under specified conditions. This unit is usually equal to the measurement resolution within the smallest range of a power supply or SMU.

In general, you should use the smallest range of the SMU to achieve the best accuracy. This information can be found in the instrument's specification manual. Here is an example:

Table 2.  NI PXIe-4139  Current Programming and Measurement Accuracy/Resolution

3. Source measurement accuracy

The measurement or output level of a power supply or SMU may differ from the actual or required value. Accuracy expresses the uncertainty at a given measurement or output level and can also refer to the deviation from an ideal transfer function as shown below:

y = mx + b

Where m is the ideal gain of the system

x is the input to the system

b is the bias of the system

y is the output of the system

When this formula is used for power supply or SMU signal measurements, y is the output reading of the device, x is the input, and b is the offset error, which can be zeroed before measurement. If m is 1 and b is 0, the output measurement is equal to the input value. If m is 1.0001, the measurement result deviates from the ideal value by 0.01%.

For most high-resolution, high-accuracy power supplies and SMUs, accuracy is a combination of offset error and gain error. These two errors are added together to determine the overall accuracy of a particular measurement. NI power supplies and SMUs typically express offset error in absolute units (such as mV or μA), while gain error is typically a percentage of the reading or requested value.

The typical source-measure accuracy of an SMU is at or below 0.1% of the programmed output. This information is provided in the specifications manual for each NI SMU instrument.

Table 3.  NI PXIe-4139  Voltage Programming and Measurement Accuracy/Resolution

4. Measuring speed

The measurement acquisition window or aperture time directly affects the speed and accuracy of your measurements. Some SMUs can modify the instrument's aperture time, giving you the flexibility to expand the acquisition window for high-precision measurements or reduce the window for high-speed acquisitions. Expanding the measurement aperture allows the instrument more time to sample and average, which reduces measurement noise and improves resolution. The following graph shows the function of measurement noise as a function of aperture time at different current ranges.

Figure 3. Measurement noise as a function of aperture time.

To achieve high-precision measurements, use an aperture time that provides the appropriate resolution while still minimizing the overall test time. Conversely, for less precise measurements or when digitizing signals such as line or load transients, a smaller aperture time should be used. For example,  the NI PXIe-4139  can sample at rates up to 1.8 MS/s, which can help you study the transient characteristics of the SMU output in detail. Depending on the current range, measurements can reach 1.8 MS/s when the noise is 1 nA – 10mA.

5. Source update rate

The update rate of an SMU determines the rate at which the SMU output voltage or current changes. For example, an SMU with an update rate of 100 kS/s can source current to the next point every 10 us. An SMU with a fast update rate can perform lengthy IV sweeps much faster than a traditional SMU. In addition, an SMU with a fast update rate can source current for non-traditional sequences such as sine waves.

Figure 4. Control the update rate of an SMU by varying the source delay, or the time difference between the start of a voltage step and the start of a measurement.

6. Transient response

Transient response is the response of a power supply to a sudden change in voltage or current, which is usually caused by an external event such as a load change or an internal event such as an output voltage step.

External load changes

Changes in external load current can cause a sudden change in voltage, causing the voltage to temporarily drop below the expected voltage output. Transient response is the time it takes for the power supply voltage to recover to a certain voltage value (ΔV) after a change in load current (ΔI). Fast transient response is critical for powering mobile devices. A large, instantaneous change in the load current consumed by the device under test (DUT) causes the output voltage to drop sharply, and then the control circuitry of the power supply restores the output voltage to its original value. For a typical programmable power supply, this can take hundreds of microseconds. The 20µs transient response of the NI PXIe-4154 (when set to "Fast" mode) enables analog circuits to quickly respond to changes in load current during testing. This short recovery time makes it an optimal choice for many wireless communication devices that use pulsed communication protocols.

Figure 5. Typical definition of transient response.

Changing SMU Output

When the SMU output changes, the transient settings of the instrument define the rise time of the output and the time it takes to reach the expected output and settle into a stable state. An ideal transient response has a fast rise time without any overshoot or ringing. Under many loads, there is a trade-off between transient response and power supply stability. If you want the fastest transient response, the device should have a high gain-bandwidth product (GBW), but the higher the gain-bandwidth product, the higher the chance that the device will become unstable under a specific load. Therefore, most devices sacrifice performance in many cases for stability. Other devices can be customized to a small extent to optimize performance in different situations. For example, many traditional SMUs have a "high capacitance" mode that is designed for use with devices with up to 50 uF of capacitance.

Certain NI SMUs feature a digital control loop technology called NI SourceAdapt , which enables you to custom tune the transient response of the SMU to achieve the best response for any given load. This provides the best SMU response while also achieving the shortest settling time, which reduces wait times and test times. In addition, this technology not only eliminates overvoltages, protecting the device under test (DUT), but also eliminates oscillations, ensuring system stability. Because the adjustment of the SMU response is done through programming software, you can easily reconfigure a SMU designed for high-speed testing to one designed for high-stability testing—maximizing your return on test equipment investment and achieving better test results.

Figure 6. The  NI PXIe-4139  features configurable transient response settings to flexibly load compensate the output control loop.

The following table lists the NI SMU models that feature NI SourceAdapt technology:

Table 4. Quickly view NI products that use NI SourceAdapt technology.

7. Sequence or sweep

SMUs typically have two output modes: single point or sequence. In single point mode, the SMU outputs only one value, while in sequence mode, the SMU outputs a series of values ​​and measures IV data at each point.

Single point source mode

Single-point mode is typically used to capture IV data of a certain value, such as testing the forward voltage of a diode, or using the SMU to power a device under test (such as powering an integrated circuit at a constant voltage). Use cases for single-point mode include developing software-timed sequences that loop through a series of single-point SMU outputs in software. Software-timed sequences can be used in place of hardware-timed sequences when the SMU does not support changing a specific function without prior planning.

Sequence Mode

When operating in sequence mode, an SMU outputs a series of hardware-timed values, providing advantages such as faster and more deterministic outputs (as well as synchronization with other PXI instruments). This process involves the SMU sourcing a DC voltage or current, then measuring the voltage and current, and then looping back to the next point in the sequence. Depending on the SMU capabilities, you can change the output level, current or voltage range, aperture time, and transient response at each step in the sequence. For storing large sequences, SMUs offer two methods: dedicated onboard memory and support for low-latency data streaming from the host to the SMU. For example, the NI PXIe-4138 and NI PXIe-4139 PCs stream data from the host to the SMU over a high-bandwidth, low-latency PCI Express connection and allow you to transparently output sequences with millions of setpoints and properties.

Sequence mode is often used for IV characterization or aging testing, and is often essential for applications that require tight synchronization with other instruments, such as testing RF integrated circuits.

8. Pulse Generation

Most semiconductor test applications using SMUs involve some form of source-measure operation. In sequence mode, this process typically involves the SMU sourcing a DC voltage or current, then measuring the voltage and current, and then looping back to the next point in the sequence. A basic DC sweep gradually increases the output in increments until each point in the sequence is completed, as shown in the following figure, which shows a five-step sequence of current values.

Figure 7. Example of a five-step sequence for a basic DC sweep.

In some applications, especially high power applications, attempting to sweep a sequence without turning off the SMU output can result in incorrect behavior or complex test setups. For these applications, the pulsed output of the SMU is preferred because it allows you to source and measure at different set points while minimizing heat dissipation through the DUT. Pulsed sweeps are similar to DC sweeps in that both involve outputting a setpoint, waiting for the output to stabilize, and then taking a measurement. Pulsed testing differs in that the source returns to the bias level after a very short pulse duration. In most cases, the bias level is set to turn off the DUT (for example, 0V or 0A).

Figure 8. Pulsing the output allows the source to return to the bias level before switching to the next setpoint.

Under ideal conditions, the pulse and DC sequences in the previous two figures should return the same IV data. However, as mentioned earlier, the DC sequence dissipates more heat through the DUT, which can lead to erratic behavior and less than ideal test results. This is why pulse testing is preferred for these types of applications. When testing in pulse mode, the pulse width should be long enough to allow the device to reach a fully on state for stable measurements, but short enough to minimize self-heating effects on the device under test. A fast and clean SMU response is especially important when generating pulses, since the SMU always starts from a pulsed bias level rather than gradually increasing the output in small increments.

Specific SMUs allow you to generate pulses beyond the range of traditional DC power supplies for applications that require higher currents. For example, the NI PXIe-4139 can generate pulses up to 10 A at 50 V, providing up to 500 W of instantaneous power. Pulse widths can be as short as 50 microseconds, depending on the load and SourceAdapt control settings. Short pulse widths not only reduce test execution time, but also minimize heat dissipation in the device under test, facilitating test engineers who might otherwise need to add a heat sink or other thermal control mechanism.

Figure 9. NI PXIe-4139 IV Range

9. Channel Density

One of the main advantages of modular SMUs is their compact size. Traditional SMUs come with dedicated displays, processors, power supplies, fans, knobs, and other redundant components, which complicates the process of building high-channel-count systems. Because modular SMUs share components with the chassis and controller, they reduce redundant components and take up much less space than traditional instruments, ultimately reducing the size and power consumption of the test system.

The number of channels required for an application changes over time. Many applications no longer have the one or two channels of a traditional box SMU. This is especially true for parallel IV test systems in the semiconductor industry, which require a large number of SMU channels in a compact space. With NI modular SMUs, you can combine multiple instruments in a single PXI chassis to create a high-channel-count solution with up to 68 SMU channels in a 19-inch 4U rack space, compared to the four to eight channels available with traditional SMUs. The compact size and modularity of the PXI platform also enable you to combine SMUs with other PXI-based instruments, such as oscilloscopes, switches, and RF instruments, to build high-performance mixed-signal test systems.
 

Figure 10. Use high-density NI SMUs to build a system with up to 68 SMU channels in a single 4U chassis.

10. Timing and Synchronization

A trigger is a signal that starts an operation on a device. An event is a signal from a device that indicates that an operation has completed or a state has been reached. You can use triggers and events to synchronize multiple operations within a single NI power supply or SMU or with other PXI/PXI Express devices. Many applications involve multiple instruments, such as oscilloscopes, signal generators, digital waveform analyzers, digital waveform generators, and switches. For these applications, the inherent timing and synchronization capabilities of PXI and NI modular instruments allow you to synchronize all of these instruments without the use of external cables.

When using this trigger function, you can select from the following trigger types:

• Start: After receiving the trigger, the source unit and the measurement unit start to execute the operation.
• Source: After the device receives this trigger, the source unit starts to modify the source configuration.
• Measurement: The measurement unit receives the trigger and starts measuring. When the measurement unit is measuring, the trigger is ignored.
• Sequence advance: After completing one sequence iteration, the source unit waits to receive this trigger before starting the next iteration.
• Pulse: The source unit waits for the trigger to be received and starts to switch from “Pulse Bias” to “Pulse Level”.

An example of a PXI platform optimized for triggering is the NI PXIe-4138/4139 module. The module sends and receives triggers and events through the PXI chassis backplane, simplifying programming and system wiring. These modules can also be hardware-timed and have a high-speed sequence engine to synchronize handshakes between multiple SMUs.

Figure 11. Diagram of the sequence engine used for triggering and timing

NI PXIe-4138 and NI PXIe-4139 modules also take advantage of the high bandwidth and low latency of PXI and support direct DMA data streaming between the host and SMU. This enables you to transparently transfer large amounts of waveform and measurement data at the instrument's highest update rate (100 KS/s) and sampling rate (1.8 MS/s), eliminating the bandwidth and latency bottlenecks of traditional instrument buses.

11. Software, Analysis Functions and Customization

Determining the software and analysis capabilities is important when choosing a modular SMU for your application, as this factor can help you choose between the two instruments.

Standalone SMUs typically use basic register-level commands and vendor-defined functions, while modular SMUs are user-definable and flexible to address application needs. Box SMUs offer many standard functions that meet the common needs of many engineers. As you can imagine, these standard functions do not address all application needs, especially for automated test applications. If you need to define the measurements that the oscilloscope will make, you should choose a modular SMU instead of a standalone SMU with fixed functions. Modular SMUs can take advantage of the PC architecture while also allowing you to customize the application to your needs.
 

NI SMUs are fully programmable using the free NI-DCPower driver software. NI-DCPower is an IVI-compliant instrument driver that comes with your NI power supply or SMU and communicates with all NI programmable power supplies and SMUs. NI-DCPower has a set of operations and properties to enable the functionality of the power supply or SMU, and the software includes an interactive soft front panel.

Figure 12. Using the soft front panel with a modular SMU to quickly make measurements.

In addition to the soft front panel, you can program the modular SMU in NI-DCPower driver software using NI LabVIEW, NI LabWindows™/CVI, Visual Basic, and .NET to implement common and custom measurements for a variety of applications. The driver also supports configuration-based express vi in ​​LabVIEW.

Figure 13. Programming a modular SMU using LabVIEW software.

12. Connectivity for high-precision measurements

Measurements made using remote sensing, sometimes called four-wire sensing, require four wires to the device under test (and a four-wire switch if a switch system is used to expand the number of channels). Remote sensing enables more accurate voltage output and measurement when the output lead voltage is significantly lower. When remote sensing is used for the DC current output function, the voltage limits are measured at the sense lead ends, not at the output terminals. Using remote sensing to measure the voltage at the DUT terminals is more accurate than close-end sensing. Ideally, the sense leads should be as close to the DUT terminals as possible.

Another aspect to consider is guarding. Guarding is designed to eliminate the effects of leakage current and parasitic capacitance between the high input (HI) and the low output (LO). The guard terminal is driven by a unity gain buffer followed by the voltage at the HI terminal. In a typical test system using guarding, the Guard is located between the HI and LO terminals. With this connection, there is effectively a 0 V voltage drop between the HI and Guard, so there is no current leakage from the HI. There may be some leakage current between the Guard output and the LO, but the current is provided by the unity gain buffer, not the HI, so this does not affect the output or measurement of the SMU.

For example, the measurement circuitry of the NI PXIe-4138/4139 can simultaneously read the voltage and current at the output terminals (near-end sensing) or the sense terminals (remote sensing). These measurements are made by two integrated ADCs that are always synchronized.

In addition, as shown in Figure 10, the NI PXIe-4138/4139 output connector has two terminals, Guard and Sense. You can use the Guard terminal to isolate the cable and test fixture. If remote sensing is enabled, you can use the Sense terminal to compensate for the current-resistance loss voltage drop of the cable and switch.

Figure 14. The NI PXIe-4138/4139 output connector has Guard and Sense terminals.

13. Next Steps

Modular SMUs are able to perform the same or even better measurements as traditional instruments, while providing a platform to support modern technology with measurement and channel capabilities to meet changing needs. However, whether purchasing a traditional SMU or a modular SUM, the factors discussed above are very important. Considering application requirements, cost constraints, performance, and future expandability in advance can help you choose the instrument that best meets all your needs.