Understanding Source Measure Unit Specifications

Before purchasing any type of test equipment, it is important to fully understand the specifications and their capabilities relative to the application you intend to use it for. In the case of a Source Measure Unit (SMU), you may already be familiar with basic specifications such as channel count, while other specifications may be less familiar.

This article walks through the key specifications for an SMU and examines several important features, with worked examples along the way, using the Yokogawa GS200 DC voltage/current source and measurement unit as the reference instrument throughout.

You should expect a source measure unit to be well-characterized in the bulletin or specification literature provided with the instrument. Let's discuss the most important SMU specifications, as well as some features that matter during the selection process.

Accuracy

Accuracy is one of the first specifications to consider when selecting an SMU. Without accuracy, the SMU could not reliably produce a given output (or measurement) on a repeatable basis. On the source side of the SMU, for each voltage or current range, accuracy is listed as a plus/minus percentage of the setting and will often include an additional fixed amount of uncertainty as well.

For example, when the Yokogawa GS200 voltage source is set to the 1V range, it exhibits a one-year accuracy of ±0.016% of setting + 120μV. Ninety-day accuracy, in many cases, may be listed as well, stating a better accuracy for the first ninety days of a one-year calibration period.

So for this example, if the SMU is adjusted for 1.000 volts, it can be expected to always fall between 0.99972 VDC and 1.00028 VDC for the entire one-year calibration period of the instrument.

Stability

Stability is another specification to consider. In the voltage source example, stability describes any short-term or long-term deviation of the voltage source from its set point, over a stated period and at a stated temperature range.

For the GS200, 24-hour stability is listed as ±0.001% of setting + 10μV at 23°C ± 1°C. While accuracy for a 1VDC setting over one year is 0.99972 VDC to 1.00028 VDC, the stability specification predicts that the same 1VDC source setting will remain between 0.99998 VDC and 1.00002 VDC over any 24-hour period, again assuming a 23°C ± 1°C environment.

But what if the environment is not strictly controlled to 23°C ± 1°C as listed in the specification? Most SMU specifications will also list a stability spec for a wider temperature range, such as 23°C ± 5°C, so the example above can be recalculated using that specification. In addition, most SMU specifications will list an additional temperature coefficient that can be added to the accuracy specification.

In the case of the GS200 set to 1VDC, add ±0.0009% of setting + 7μV per each 1°C change in temperature. This additional accuracy uncertainty is added to both the 90-day and the 1-year accuracy specifications for the ranges outside the controlled band, typically 5°C to 18°C and 28°C to 40°C. Again, expect an SMU to be well-characterized for this in the bulletin or specification literature provided with the instrument.

Resolution

Resolution is the smallest step that the source can be adjusted to. Using the GS200 as an example, on the 1V range one can select steps of 10μV. For instance, the source could be set to 0.999990 VDC. At the same time, the GS200 would be able to measure the voltage source current to the nearest 10μA.

Range

Range is important too. The overall maximum and minimum range for the SMU output is usually best specified using a graph, since the SMU can be used not only as a high-accuracy voltage or current source, but also as a high-accuracy programmable electronic load. The operation range for the GS200 is the same during sink operation as it is during source operation.

Figure 1. GS200 four-quadrant range.

The instrument divides its total range into a number of sub-ranges. While the GS200 can go up to 30 VDC and 200 mA, the voltage range is further divided into 10mV, 100mV, 1V, 10V, and 30V ranges. Each range has a unique resolution associated with it. For example, the 10mV range can be adjusted in 100nV steps, while the 30V range can be adjusted in 1mV steps.

Source range can also be described in terms of polarity and over/under allowances. The GS200 can go from -32VDC to +32VDC and, in fact, can perform four-quadrant operation by acting as a current source or a current sink across ±30 V and ±200 mA. It is not hard to imagine how important this can be for fine-tuning the SMU output to suit the application.

Figure 2. Source operation.

Figure 3. Sink operation.

Output Resistance

Output resistance is usually specified for each of the sub-ranges: simply the internal resistance of the source: a series resistor for a voltage source, and a parallel resistor for a current source. The GS200 provides ≤2mΩ for a voltage source on the 1V range, and ≥10MΩ output resistance for a current source on the 200mA range.

Output Noise

Output noise is the noise sourced by the source unit in the SMU, characterized and specified separately for the voltage and current sources across one or more frequency bands. For the 1VDC voltage source example on the GS200, noise is specified as 10μVpp total noise within a DC to 10Hz band, and 60μVpp total noise within a DC to 10kHz band.

An SMU will naturally provide its very best (minimum) output noise performance when the measure unit is off, which internally disconnects it from the source. Current source noise is specified similarly, typically in units of current such as μApp.

Common Mode Rejection Ratio (CMRR)

CMRR characterizes the ability of the source to reduce, or minimize, any signal present on the Device Under Test (DUT) to a smaller signal. For example, if a DUT had noise on it, or was drawing current at a rate of 50Hz, can the SMU reduce that noise by very much?

The answer is yes, and CMRR is specified in decibels referenced to the test waveform, or the noise/other waveform present on the DUT, at a stated frequency or frequency range, typically 50Hz/60Hz. For the GS200, the voltage source has ≥120dB on the 1VDC source, meaning the SMU can reduce a noise source equal-to-or-less-than -120dB. The current source CMRR is specified similarly, not in dB, but rather in amps per volt.

Both the current source and the voltage source will often have these specifications provided in easy-to-use formats, such as a table correlating range, resolution, stability, accuracy, temperature coefficient, output resistance, output noise, CMRR, and more.

Other Specifications

Limiter Functions

To protect the device under test, an SMU almost always has a built-in limiter function. When the SMU is sourcing voltage, a user-defined current-limiter adjustment lets the engineer set an absolute value beyond which a limiter circuit will activate. So if the limiter is set to 10mA, the SMU will not permit current to exceed 10mA under any condition.

Similarly, when the SMU is sourcing current, a user-defined voltage-limiter adjustment is also available. The limiter is not an alarm condition: the SMU quietly limits the output and otherwise resumes normal operation. The GS200 limiter, for example, ranges from 1mA to 200mA in 1mA steps, and 1VDC to 30VDC in 1V steps.

Response Time

Response time is typically specified in units of time, usually milliseconds, and under very specific conditions. It is the amount of time the SMU takes to change from any current output setting to within 0.1% of the final value, at maximum output, maximum resistive load, and with no limiter operation. For the GS200, that time is 10 milliseconds or less for all voltage source and current source ranges.

Figure 4. GS200 response time, transitioning cleanly from -32V to +32V.

Because an SMU's response time is typically much faster than that of a power supply, it can be prone to overshoot or oscillation under the right conditions. An SMU must strike a balance between transient response and stability under all conditions, so most SMUs will publish a maximum capacitive and inductive load specification. For the GS200, the largest capacitive load permitted is 10μF, and the largest inductive load is 1mH. Stability is guaranteed if the reactive portion of the load lies between these values.

An SMU will typically have a four-terminal connection. When an SMU is measuring voltage in current source mode and the current becomes large, the voltage drop in the lead wire can become significant. In such a case, the SMU can measure the voltage while eliminating the voltage drop due to lead-wire resistance by using a four-terminal connection (4W) and SENSE terminals near the DUT. The effects of lead wires also appear in voltage source mode; using the four-terminal connection lets the GS200 apply the specified voltage to the DUT while reducing the effect of lead-wire resistance. The GS200 features this four-terminal connection on its front panel.

Input Voltage

The input voltage specification is an important consideration too. While one tends to think of the SMU as a source, it can also sink current and therefore be considered an electronic load. As such, it has limitations. For the GS200, the input voltage specification is 32V_PEAK between the Hi and Lo terminals, and 42V_PEAK between the Lo terminal and ground. Furthermore, a maximum of 250V_PEAK is permitted between the ground terminal and the case.

Source Guard

Since output noise is one of the most important characteristics of a high-performance SMU, caution must be taken when connecting the SMU to the DUT so as not to degrade noise performance. The internal source guard exists for that purpose. While the SMU features excellent (minimum) noise performance, common-mode noise can still be an issue: it can affect the source circuitry internal to the SMU and worsen the noise delivered to the DUT.

Figure 5. Common-mode noise conducting into the SMU source circuit.

The guard feature is simply a switch between the front-panel ‘G’ terminal and the internal guard, a floating internal shield around the source circuit. This noise-abatement technique keeps common-mode noise current out of the Lo lead and, therefore, out of the source itself, and can be used in two ways:

• Guard set to ON diverts any ground-referenced Lo-side noise current through the guard and back to the common-mode noise source.
• Guard set to OFF permits a common third-wire, added as a preferred path for common-mode noise to follow.

Figure 6. Common-mode noise conducting through Lo and guard (guard on).

Figure 7. Common-mode noise conducting through ground and guard (guard off).

When used properly, the guard feature assures the best noise performance the SMU is capable of in the presence of common-mode noise sources.

Conclusion

Referring back to the question of what a source measure unit actually is, the differences between an SMU and, say, a power supply with a meter on it become apparent while studying the specifications. That can be a number of things: accuracy, stability, resolution, noise, common-mode rejection, and others. In one word, it is performance. In addition, the more capable SMUs will, frankly, often have longer bulletins or spec sheets, and a lot more can be expected of them.

SMU specification bulletins will also discuss the instrument in terms of functionality and features. Functions provided by an SMU should include programmability, sweep functionality, synchronization and trigger capabilities, and software and drivers. On the feature side, expect specifications on connectivity (Ethernet, USB, and GPIB), on-board servers such as a web server and an FTP server, and on-board media for storing measurements, programs, and setup files.