“The ever-faster switching transistors on modern wide-bandgap power devices (SiC, GaN) make measurement and characterization a considerable challenge, and in some cases almost impossible. This has changed with the advent of isolated probing technology, which finally allows designers to confidently measure previously shunned half-bridge and gate driver waveforms. By understanding the challenges in detail and using appropriate probing techniques, power supply engineers can more quickly and efficiently characterize and optimize their designs.
The ever-faster switching transistors on modern wide-bandgap power devices (SiC, GaN) make measurement and characterization a considerable challenge, and in some cases almost impossible. This has changed with the advent of isolated probing technology, which finally allows designers to confidently measure previously shunned half-bridge and gate driver waveforms. By understanding the challenges in detail and using appropriate probing techniques, power supply engineers can more quickly and efficiently characterize and optimize their designs.
Half-bridge circuits (Figure 1) are widely used in a variety of applications in the field of power electronics and are fundamental circuits used in modern designs to efficiently convert electrical energy. However, the benefits of this circuit can only be realized if the half-bridge, gate drivers, and wiring are properly and optimally designed. When measurements do not match expected results, it can be difficult to extract meaningful details about the device under test. To make matters worse, based on probe position and other factors, the waveforms can vary significantly, ultimately costing the designer more than the cost.
Figure 1. Half-bridge circuits are widely used to efficiently convert electrical energy in modern designs.
Efficiency and power density requirements often change with application design requirements, such as whether to optimize price/performance. The requirement to improve energy efficiency in power density determines the topology of the design, which in turn affects the measurement equipment and techniques to be considered. Table 1 summarizes the most important metrics and measurements for half-bridge and gate drivers.
Table 1. The most important measurements for gate driver and half-bridge configurations.
Accurate power measurements are inseparable from the performance of the measurement system in several areas, including Voltage handling, common-mode rejection, connectivity, temperature handling, and the ability to measure very small currents. Despite the ever-evolving power design requirements, the actual development of test and measurement technology has been lagging a bit. In some cases, designers are forced to develop custom measurement solutions or only approximate partial measurements, ignoring possible optimizations.
At the most basic level, these measurements are performed using an oscilloscope and a corresponding set of probes. Oscilloscopes are rarely a problem when it comes to making accurate and reliable power measurements. And the biggest challenge is getting the signal from the test point to the oscilloscope. Therefore, choosing the right probe for the job is critical, whether passive or single-ended, traditional high-voltage differential probes, current probes, or isolated probes.
Single-ended probe C Low side measurement
Most oscilloscopes come with a set of passive or single-ended probes. These probes can only accurately measure signals referenced to the oscilloscope ground level and are limited to making low-side measurements. By isolating the oscilloscope, or using a pair of probes to make pseudo-differential measurements (see discussion later), you can perform high-side measurements with passive probes, but this method is generally not recommended.
When considering how high probe performance is required for a measurement task, people typically focus on bandwidth. Conventional thinking holds that the higher the bandwidth, the higher the performance. Indeed, bandwidth is an important metric that determines the highest frequency at which the peak-to-peak amplitude of a sine wave can be measured. But in reality, you’re not measuring a sine wave in the frequency domain, you want to Display and measure the signal over time, i.e. measure the signal in the time domain.
Therefore, the performance metric of greatest interest in half-bridge and gate driver measurements is rise time. The rise time can be calculated from the bandwidth, but the only way to reliably know the rise time and full time response of your measurement system is to actually measure the rise time using a step signal that is much faster than you measured signal.
A measurement system with insufficient rise time performance can experience step response distortion, as shown in Figure 2, including nonlinearity, rounding, and dip. It can be difficult to confirm whether these distortions actually come from the measurement system or the device under test, and the real answer can only be found by characterizing the measurement system. To avoid these measurement errors, the rise time of the probe selected must be faster than the rise time of the device under test.
Figure 2. Rise time specifications are more important than bandwidth specifications in power device measurement accuracy.
Figure 3 shows the significance of a fast probe, where a 1 GHz passive probe is used to measure the low side of a high speed FET driver whose rise time is shown in the data sheet
Figure 3. The 1 GHz Tektronix TPP1000 passive probe can accurately measure high-speed FETs due to the 450 ps rise time specification.
Measuring gate driver current
When measuring gate driver current, many designers use an imposed current shunt instead of a current probe for the simple reason that using a current probe to measure the inductance of the loop affects the circuit. Typically, the design will already have a resistor in series between the gate driver and the gate. To minimize insertion impedance, the resistance of the current shunt is kept very low, so the voltage drop across the current shunt is also very low. The current is obtained by measuring the voltage drop across the current shunt and dividing by the known resistance of the resistor.
Connecting the current shunt to the low side usually means that one terminal is grounded. The main difference between placing it on the low-voltage side and placing it on the high-voltage side is that placing it on the low-voltage side will reduce or effectively eliminate the common-mode voltage, which will appear in phase on either side of the current shunt at the same time. Therefore, it is generally recommended to place current shunts on the low voltage side, especially at high voltages. In high current applications, ground bounce can appear as a common mode signal.
One of the techniques for breaking ground loops is to “isolate the oscilloscope” or isolate the circuit under test. A floating ground would break the connection to ground, and in theory a differential measurement could be made between two test points because the oscilloscope ground had been broken. This method is inherently dangerous as it defeats the electric shock protection and may damage the measuring equipment.
Floating testing may be suitable for some measurements, especially at very low frequencies, but be aware that without a low impedance ground connection, radiated and conducted emissions from the oscilloscope may interfere with the measurement as noise. Also note that when ground is interrupted at higher frequencies, the ground loop may not be interrupted because the “floating” circuit will remain coupled to ground through large parasitic capacitances, causing ringing and waveform distortion. Figure 4 shows the float measurement on the high side gate driver. Ringing and distortion are evident, with up to 28 V overshoot.
Figure 4. Ringing, distortion, and 28 V overshoot are evident in this high-side gate driver floating measurement.
Pseudo-differential measurements (instead of passive probes) can also be used, possibly satisfying some low-frequency signal measurements. The measurement is done by taking two ground-referenced signal measurements and using the oscilloscope to subtract the two oscilloscope channels. In Figure 5, the oscilloscope subtracts the CH2 waveform from the CH1 waveform, resulting in a red waveform. Both inputs must be set to the same scale, and the probes must be identical and closely matched. The common-mode rejection ratio (CMRR) in this technique is poor, as shown in Figure 5, especially at higher frequencies, which can exceed the oscilloscope input range. CMRR refers to the ability of an oscilloscope to reject the common-mode voltage of two test points during differential testing.
Figure 5. Pseudo-differential measurements have limited performance but are adequate for very low frequency signals with low common-mode signals.
For most GaN and SiC applications, differential probes are a good choice for accurate low-side measurements and some high-side measurements. But for higher performance devices, the most likely scenario is that traditional high-voltage differential probes are not the best choice because of their insufficient common-mode rejection at higher frequencies. This becomes an obvious problem when performing high-side voltage measurements, as small differential voltages are measured in the presence of large common-mode voltages during fast switching transitions.
A common misconception is that differential probes are floating. In fact, traditional differential probes are based on differential amplifiers, which are connected to ground. Unfortunately, this connection limits the common-mode voltage range, degrades the frequency rating, creates ground bounce, and limits the common-mode rejection ratio beyond a certain MHz of bandwidth.
These limitations are especially apparent when testing powered GaN or SiC devices, which have ultra-fast switching rates and even nominal common-mode voltages. For example, a 100 MHz bandwidth differential probe provides -70 dB CMRR at DC, -50 dB CMRR at 1 MHz, and drops to -27 dB CMRR at 100 MHz, which is approximately 22:1.
It is difficult to see such a poor index in the technical data of the probe product, because the decrease of the rated value with the frequency cannot become the index promoted by the manufacturer. You’ll need to look through the user manual to find graphs like Figure 6, but we can easily calculate the impact of poor CMRR. For example, for a common-mode voltage of 600 V, the resulting error is 27 V (600 divided by 22). This behavior is eye-catching because it is impossible to accurately measure a high frequency 15 V differential signal with a probe with such a large error in the presence of a 600 V common-mode voltage.
Figure 6. A 100 MHz bandwidth differential probe drops the CMRR rating to -27 dB as frequency increases.
Another consideration when calculating common-mode rejection is the connection between the probe and the DUT. Most common mode rejection specifications only include the probe and don’t take into account additional connection options such as large hook clips.
Due to the lack of adequate probing accessories, many power supply designers have turned to some alternative techniques to perform high-side device measurements, such as measuring the low-side first, deriving high-side results using full simulation, examining thermal characteristics, EMI proximity probing, if these If none of the methods work, it’s just trial and error.
High-Performance Isolated Probe
The limitations of single-ended and differential probes become more apparent when testing SiC and GaN power devices with their ultra-fast switching rates and high nominal common-mode voltages. Since these signal capture problems stem from the need for grounding, viable solutions employ probe techniques that cannot rely on grounding to be more or less immune to common-mode voltages. Operating entirely over fiber, this isolated probe offers a number of advantages, including bandwidth up to 1 GHz, a large differential voltage range, and perfect common-mode rejection at all frequencies.
When performing high-side VGS measurements, engineers need to see enough waveform detail to confirm the simulation and evaluate signal characteristics such as ringing compared to the ideal state represented in Figure 7. The high-side VGS is turned on, the first area represents the CGS gate-source charging time, followed by the Miller platform. After the channel conducts, the gate will charge to its final value.
Figure 7. This is a schematic diagram of the ideal state of the high-side VGS.
Figure 8 compares high-side VGS measurements using traditional high-voltage differential probes with high-performance isolated probes, and it is evident that it is difficult to extract meaningful information and make design decisions based on the measurements provided by traditional probes.
Figure 8. Isolated high-voltage differential probes provide the confidence needed to optimize device performance.
In contrast, isolated high-voltage measurement systems provide the resolution and repeatability needed to measure, characterize, and optimize design performance. The correlation of the Miller platform and the switch-to-node transition can be clearly seen. This waveform clearly shows previously hidden resonance and signal details, providing the confidence needed to optimize performance and develop designs without being too conservative.
High pressure side, low pressure side interaction
For GaN devices with tight tolerances, the parasitic coupling between the low-side switch and the high-side gate in the switching node is one of the more difficult problems to diagnose. Figure 9 shows how overshoot or ringing from the high side is transferred to the low side. This situation is unknowable without an accurate high-side measurement being performed, and can create numerous problems, at least leading to switching and efficiency losses and degradation, and at worst, simultaneous opening of both low-side and high-side switches resulting in catastrophic Fault.
Figure 9. The ability to view actual waveforms makes it possible to diagnose and resolve issues such as parasitic coupling between switching nodes.
It is safe to say that half-bridge and gate driver measurements present a number of challenges that must be overcome in order to take full advantage of the latest wide-bandgap devices. This requires the right measurement technology and robust measurement solutions. Often, the source of the problem is not the oscilloscope, but the choice of probe. High-side gate measurements are particularly difficult, but many of the associated challenges can be overcome by understanding the common-mode rejection ratio and how isolated high-voltage differential probes can make accurate and reliable measurements in the presence of high common-mode voltages.