Subtleties of scope probes can lead to garbled measurements when examining power circuits.

KEN JOHNSON, DAVID MALINIAK, TELEDYNE LECROY CORP.
When an oscilloscope is used for debugging, validation, or device characterization, measurements generally take place with the help of a scope probe. There are several types of scope probes because manufacturers optimize different types for specific applications.

In some cases, though, a given probe may have an Achilles’ heel. For example, high-voltage, active differential probes are excellent for all-purpose uses. But they may not be the best option for a specialized task such as measuring conduction loss in a power-semiconductor device.

One such specialized task is the measurement of small signals riding on a high-voltage bus. Examples of such signals include upper-side gate-drive signals, floating control signals, and sensor signals.

Two key probe specifications determine why a probe is, or isn’t, useful in power-electronics applications: high-voltage isolation and common-mode rejection ratio (CMRR).

High-voltage isolation is the maximum common-mode voltage an attenuating probe can safely handle. In the power-electronics realm, the maximum common-mode voltage is the dc bus voltage. One must use an isolated probe to properly measure signals floating on the dc bus. That small 3-to-24-V gate-drive signal might be riding atop a 500-V bus voltage, so the probe must be able to safely handle the sum of the two.

Common dc bus voltages include: 500-V dc for 120/240-V ac line inputs; 1,000-V dc for 600-V ac-class line inputs; 1,500-V dc for grid-tied solar photovoltaic inverters and UPS systems; and 6,000-V dc for 4,160-V ac inputs.

Conventional high-voltage-rated probes carry a safety agency rating from bodies such as Underwriters’ Laboratory (UL). The rating indicates a maximum safe common-mode voltage to ensure that neither the oscilloscope, device under test (DUT), or operator can sustain harm.

CMRR is the ability of the differential amplifier to ignore the component of a signal that is common to both inputs. For starters, the perfect differential amplifier does not exist: In the real world, a diff amp cannot remove all the common-mode signal. Then, differential probe/lead pairs must be perfectly matched for frequency response, which is difficult to realize with attenuating probes.

The measure of effectiveness of common-mode rejection is CMRR expressed either in decibels or as a ratio of rejected voltage. You might see CMRR reported as 100,000:1 or as 100 dB, but they mean the same thing:
20log₁₀ (V_SIGNAL/V_MEASURED) = CMRR_dB

A conventional high-voltage (high attenuation) probe topology has a tough time realizing a high CMRR at high frequencies, but oscilloscope manufacturers do the best they can by binning, sorting, testing, and calibration. It’s not difficult to get a handle on a given probe’s CMRR in a field measurement.

A DIFFERENT PROBE TOPOLOGY

One probe technology — high-voltage, fiber optically-isolated (HVFO) probes — optimizes both CMRR performance and high-voltage isolation. To understand why, it is useful to compare the topology of a conventional high-voltage differential probe or amplifier with that of the HVFO probe.

Often, users of conventional probes aren’t aware that they employ high levels of attenuation, or that the probe is grounded. Because these probes are grounded, they use two leads, and current flows in both. On the negative side, current flows because of the influx of common-mode voltage into the probe. On the positive side, it flows due to the common-mode voltage plus signal swing into the high side of the differential amplifier. Thus, the probe is measuring small signals plus the common-mode voltage across the lead capacitance, which results in more probe loading on the device under test, especially at high common-mode voltages.

In addition, the probe pair must be precisely matched both in impedance and in frequency response to maintain CMRR across the probe’s rated frequency range. As noted earlier, this matching is difficult to realize in practice.

In contrast, the HVFO probe topology possesses inherent advantages for maintaining a high CMRR and for minimizing probe loading. The single-ended HVFO topology acts differentially in a way because the whole battery-powered amplifier is floating. As a result, it measures only the small-signal swing. This, in turn, results in only a small load current drawn from

the circuit. The signal lead, being coaxial, does not require matching to realize high CMRR, while the fiber-optic isolation between transmitter and receiver also contributes to the CMRR specification.

One significant difference between the HVFO and conventional high-voltage differential probes is its three-lead configuration: The blue lead is coaxial; its center conductor carries the signal current. The return path for signal current is through the coax cable outer conductor. The green lead connects to measurement reference and to the blue coaxial cable’s outer signal conductor. The black lead also connects to the measurement reference.

Connecting the green and black leads to the same point ensures that the signal and return currents are equal and opposite at the tip of the common-mode choke located near the amplifier jack. As a result, the current in the black reference wire (shield current) drives the reference voltage for the single-ended amplifier, and it accounts for and minimizes any parasitic capacitance effects. Note there is no connection to earth ground whatsoever.

Its topology gives the HVFO optically-isolated probe a significantly higher CMRR than that of a conventional high-voltage differential probe/amplifier. An example of the latter, Teledyne LeCroy’s HVD3106A, is an exceptionally good probe for its class; its CMRR plot-vs.-frequency shows CMRR topping out at 85 dB at 60 Hz and dropping to 65 dB at 1 MHz and to 40 dB at 5 MHz.

The nearby CMRR plot for the HVFO reflects use of the probe’s 1X tip and shows a CMRR of 140 dB at 60 Hz, 120 dB at 1 MHz, and 85 dB at 10 MHz. This level of performance is what makes the HVFO a candidate for upper-side gate-drive measurements and sensor-voltage measurements, floating in-circuit in either case. Similarly, the probe works well for EMI/RFI measurements in situations where a sensor signal from a board gets bombarded with EMI within a test chamber as it travels to the oscilloscope outside of the chamber.

Estimating probe CMRR

A simple test will give a reasonable estimation of a probe’s CMRR. Though not highly accurate, it will correlate well enough with the manufacturer’s specifications to serve as a reality check.

The yellow trace is an upper-side gate-drive (V_G-E) signal acquired with a Teledyne LeCroy high-voltage fiber-optic probe (HVFO). The signal has an amplitude of about 15 V dc as it swings from low to high. It floats on a dc bus voltage of about 465 V.

The blue trace is the same signal acquired with a Teledyne LeCroy HVD3106A high-voltage differential probe, which delivers good CMRR for its class. But here the positive and negative leads connect at the measurement reference location, which, in this case, is the emitter or source location of the upper-side gate-drive device.

With both the positive and negative leads connected at the emitter, the HVD3106A probe should not pick up any signal. The fact that it does demonstrates the probe’s inability to separate interference from the true signal of interest (V_G-E). A probe with near-perfect CMRR would give a flat line; instead, we get a transient spike with an amplitude of about 1 V. That spike represents measured common-mode interference.

Meanwhile, the magenta trace is acquired using a Teledyne LeCroy high-voltage fiber-optic (HVFO) probe with all three of its leads (signal, ground, and shield) connected at the emitter. Thus, the HVD3016A is giving us a CMRR of about 15:1. From the equation 20log₁₀ (V_SIGNAL/V_MEASURED) = CMRR_dB, the CMRR is about 24 dB. An eyeball estimate of the rise time for the upper-side gate-drive signal is about 40 nsec. Using the 0.35/T_RISE rule of thumb gives a bandwidth of about 9 MHz.

Compare these estimates to the data published in the HVD3106 data sheet: The data sheet CMRR plot shows about 30 dB at 10 MHz (using the 500x attenuation path required for the high common-mode voltage). Note that characterization of probes for data-sheet plots takes place in highly controlled and optimized conditions. These optimized conditions probably account for the difference between the 24 dB and the data sheet’s 30 dB.