Introduction

Probing signals with a bandwidth below 10 MHz and voltage sensitivity above 100 mV is a no-brainer. Regardless of the type of signal or the source impedance, the venerable 10x passive probe is the answer. However, at bandwidths >10 MHz and with voltage sensitivity <100 mV, the 10x passive probe may not be the best option.

In this Application Note, we introduce a simple, easy-to-implement, and low-cost alternative to the 10x passive probe specifically for power-rail measurements.

An Alternate Power-Rail Probing Method

Six conditions specific to power rails convey to them a unique set of challenges when compared with signal lines:

  • A switching power supply emits large near-field RF noise
  • The output impedance of the rail can be less than 1 Ω
  • There can be a large DC offset
  • The signal of interest may be as little as 10 mV
  • For low-current rails we want to avoid a low DC impedance load
  • There may be fast transients with bandwidths >100 MHz

An alternative method for probing a low-impedance, fast-switching source is the source series termination method, comprising a 50-Ω resistor in series between the DUT and the coaxial-cable connection. The coaxial cable is then connected to the oscilloscope’s analog input set for 1 MΩ termination. An equivalent circuit model and a simple implementation appears in Figure 1.

When measuring most high-bandwidth signals, 50 Ω is the recommended input impedance to the oscilloscope. This does not hold true, however, for power rails. There are two important problems created by a 50-Ω input impedance at the oscilloscope when measuring power rails.

For one, with 50-Ω input impedance to the oscilloscope, the maximum voltage one can probe is typically around 5 V. Higher voltages will dissipate too much power in the 50-Ω resistor and the oscilloscope may suffer damage.

For another, using 50 Ω for the oscilloscope’s input impedance will introduce a DC load to the DUT. If the DUT is a 3-V source, the 50-Ω load will draw 60 mA. If the DUT can source 100 A, a 60-mA draw from the probe is negligible. But if the DUT is a 100-mA low drop-out (LDO) power source, a 60-mA draw from the probe will dramatically affect the LDO power source’s performance.

Thus, we must use the oscilloscope’s 1-MΩ input termination. On one hand, this enables measurement of a voltage range of ±40 V with a negligible DC current draw. On the other, when probing a low-impedance source with a direct coaxial-cable connection, the oscilloscope’s 1-MΩ input termination will generate large reflections from fast transient edges. The equivalent circuit is shown in Figure 2.

After being launched into the coaxial cable from the low-impedance source, an initial transient signal travels down the coax cable to the oscilloscope’s 1-MΩ input termination. From there, it reflects back to the source. When it again reaches the source’s low impedance, it reflects with a sign change. This reflection makes it to the oscilloscope input and pulls the input signal low, then reflects and repeats. The end result is a large ringing signal at the scope.

The answer to this problem: Add a 50-Ω source series termination to the DUT. As a result, half of the source voltage is initially launched into the coax cable, which reaches the oscilloscope’s 1-MΩ input termination and reflects back. The initial voltage measured by the oscilloscope is 2x the launched voltage, which is exactly the voltage of the source.

When the reflected signal makes its way to the source, it sees the 50-Ω resistor in series with the low impedance of the source. As long as the source impedance is less than 5 Ω, there is virtually no reflection and the reflections are terminated.

In the example shown in Figure 3, a 5-V switch mode power supply (SMPS) with a 0.1-Ω source impedance switches on with a rise time of a few nanoseconds. It is measured with the oscilloscope set to a 1-MΩ input termination. At left in Figure 3, there is a direct coaxial-cable connection with multiple reflections are clearly evident. At right, the addition of a 50-Ω source series resistor to the DUT source-terminates the coax cable and prevents multiple reflections.

This simple approach of adding a 50-Ω source series resistance at the end of the coaxial cable is a low-cost alternative to probing power rails at high bandwidth. It is an alternative to using a 10x probe, with the advantage of being a 1x probe that does not attenuate the signal.

Which is better: 10x passive probe or 1x, 50-Ω-termination probe?

These two probing methods can give very similar results when probing low-impedance power rails. Figure 4 shows the measured voltage transient noise on an 18-V SMPS using both probing methods on the same scale. These measurements are DC coupled, with offset applied by the oscilloscope, on a scale of 50 mV/div. This is a very sensitive scale of about 0.25 %/div.

The advantage of a 10x probe is a measurable voltage range of ±400 V compared to the ±40-V range of the 1x probe method.

The disadvantage of the 10x passive probe is its 10:1 signal attenuation. This means that when measuring small voltages, the signal-to-noise ratio (SNR) may be reduced.

The typical peak-to-peak noise of the oscilloscope’s front-end amplifier is about 1 mV. With a 10x probe, this is a tip voltage noise of 10 mV. If the goal is to achieve a SNR of 20 dB (10:1) and the noise level is limited to 10 mV by the oscilloscope amplifier, do not use a 10x probe if voltage changes of less than 10 mV x 10 = 100 mV are of concern. This is the lowest voltage change measurable by a 10x probe while still managing a 20-dB SNR.

If 100 mV is 1% of the DC signal level, then the signal level would be 10 V. For rail voltages below 10 V, a 10x passive probe would not deliver an acceptable SNR.

In the measurement of the 18-V DC level SMPS described above, the slightly better SNR of the 1x probe is evident. In Figure 5, a 3.3-V SMPS was measured with the same two probes. Now, we clearly see the better SNR of the series terminated 1x probe.

This distinction defines the condition to decide which probing method to use:

If your application is for a voltage level of 10 V DC or higher, the 10x passive probe will be a better choice. It offers higher voltage range with acceptable SNR and is quick, simple, and easy to use. And, everyone has a 10x passive probe.

If your application is for a voltage level of 10 V DC or lower, a 1x passive probe with source series termination will be a better choice. It will provide a better SNR.

But, if your application has a bandwidth above about 500 MHz, neither approach will be acceptable. Even in the best case, the 10x probe is fundamentally limited to less than 500 MHz by its attenuating coaxial cable and the equalization circuit built into the probe. The source series terminated probe method is limited by the tip loop inductance and the 50-Ω resistor. If the tip loop inductance is about 15 nH, the resulting low-pass frequency cut-off is about 500 MHz. Thus, in high-bandwidth applications, consider using an active rail probe such as Teledyne LeCroy’s RP4030 active voltage-rail probe.

Conclusion

While the 10x passive probe is suitable for many power-rail probing applications, the 50-Ω source series terminated probe is a low-cost, easily homebrewed alternative to the 10x probe, providing improved SNR in low-voltage rail applications. However, for high-bandwidth scenarios, Teledyne LeCroy’s RP4030 voltage-rail probe is the best option.