[Practical Tips] How to choose the best probe for online signal testing?

Different from the ICT (In-Circuit Test System) test system, the signal online test here is relative to many serial bus interface tests such as PCIE, USB3/4, HDMI and even many 802.3xx standards, etc., which usually use fixtures and cables to directly connect to the oscilloscope input port for testing. These high-speed standard tests usually require transmitter configuration or switching to send out various test patterns, which are actually Open-Circuit or Broken-Link during testing.

Signal online testing is when the device or circuit under test is in the Active-Link state. For example, the MIPI D-PHY standard, whether DSI or CSI, is usually tested when the system is working normally. Similarly, the system-level test of the DDR storage interface is also performed when the storage controller and particles are on the board. At this time, the entire probe connection system including circuit components such as the front end and amplifier will be inserted into the signal circuit system. Therefore, the load effect of the probe becomes an important factor affecting the measurement accuracy .

Figure 1 DDR and MIPI-DPHY signal online test connection diagram

When it comes to probes, the most commonly used ones are 10:1 and 1:1 passive probes that match the 1MΩ impedance of digital real-time oscilloscopes. The following figure is a typical 10:1 passive probe circuit model:

Figure 2 10:1 passive probe circuit model

10:1 passive probes are usually standard probes for mid- and low-end oscilloscopes, suitable for general signal testing and are widely used, such as N2873A. However, due to its high capacitive reactance, even with a 4pF indicator, there is still a non-negligible load effect.

Table 1 Passive probe comparison table

1:1 passive probes such as N2870A do not attenuate the signal due to the characteristics of no attenuation ratio. Of course, there is no need to amplify in the oscilloscope, that is, the instrument background noise will not be amplified. At the same time, the minimum vertical sensitivity setting of the oscilloscope can be guaranteed, so it can be used for typical small signal tests such as power supply ripple and noise testing. However, due to its passive characteristics, it cannot bias the signal to remove the DC component on the signal, so it also has certain limitations. Another limitation is that it does not attenuate the signal, so it cannot be connected to an input voltage greater than 30Vrms.

Since this type of passive probe must match the 1MΩ oscilloscope impedance, an adapter such as the N5449A must be used when used with high-end oscilloscopes such as the Keysight V/Z/UXR series:

Figure 3. High impedance adapter connecting a 10:1/1:1 passive probe to a high-end oscilloscope with 50Ω input impedance.

The passive probes introduced in the previous section usually have a capacitive reactance of several pF to several tens of pF. Therefore, when probing some low drive capability signals or circuits, the circuits often cannot work properly due to their excessively high capacitive load. For example, when testing the clock signal output by the crystal, the circuit often fails to oscillate. Active probes, as high bandwidth and low load products, effectively solve this problem.

For example , Keysight N2795/6A provides 1/2GHz bandwidth, 1pf capacitance, and 1MΩ high input impedance. 1pF capacitance is low enough for many circuits.

Figure 4 N2795/6A single-ended active probe

High-bandwidth, low-voltage differential probes are the most important products for measuring weak and micro-electric differential signals. Keysight offers products with bandwidths ranging from 1.5 GHz to 30 GHz.

The InfiniiMax series currently has three generations of products:

If you are careful, you must have noticed why Keysight provides two product models for the 25GHz bandwidth product? What is the difference?

This involves the difference between the two circuit structure models of high-bandwidth differential probes:

Figure 5: RC and RCRC high-bandwidth differential probe circuit model structures

And its impedance-frequency response characteristics:

Figure 6 Differences in impedance-frequency response characteristics between the RC model and the RCRC model

As can be seen from the figure above, although the DC impedance of the RC circuit structure model probe is only half of that of the RCRC structure, it can maintain high impedance and high bandwidth. Keysight 113X/116X and MX0023A probes both use the RC model, showing a wide-band high impedance characteristic. RCRC probes, such as the N7000A series and N280X series, can support higher bandwidth and have high DC impedance, but the mid-band impedance is significantly lower than that of the RC probe, at the KΩ level.

Therefore, for DDR standard bus testing, considering that the DDR bus is in the High Z state when idle, dynamic ODT allows DRAM to switch between high or low termination impedance. When the termination impedance becomes high in the High Z state, the probe impedance needs to be high enough to reduce the probe loading effect. The RCRC probe has a low KΩ impedance and cannot form a high enough resistance to the circuit to generate false signals. In comparison, the RC probe has a significantly lighter loading effect. The following two figures compare the waveforms when using the RCRC probe and the RC probe for testing.

Figure 7 Comparison of DDR waveform test between RCRC probe and RC probe

In addition to the DDR bus, there are similar problems in the MIPI D-PHY bus and eMMC signal testing. However, in general, the D-PHY and eMMC rates are relatively low, and the 113xB or 116xB RC probes are usually selected by default, and the RCRC probes are not deliberately selected, so this comparison will not be revealed.

RCRC probes such as the N2803A are more suitable for eye diagram and jitter testing of low source impedance and ultra-high-speed Serdes because they can achieve higher bandwidth and higher DC impedance up to 100KΩ .

In addition to providing ultra-high bandwidth probe amplifiers, Keysight also provides a very rich and flexible front-end detection solution, such as the handheld differential point measurement N5445A with up to 30Ghz bandwidth:

Figure 8 Handheld point-to-point front-ends with various bandwidths, up to 30 GHz

In addition, the Mirco Probe Head MX0100A, which was launched in the past two years and is specifically designed for DDR bus-type small space welding testing, and the 23/26Ghz probe head MX0106A/MX0109A, which can be directly used for high and low temperature testing (-55°C to +150°C):

Figure 9 MX0100A (top) and MX0106A/MX0109A (bottom)

There are not many high-bandwidth RCRC probes available in the industry. Keysight N2803A is the leader in high-impedance probes with 30GHz bandwidth performance .

There are some products such as P7633 that claim to have a bandwidth of 33GHz, but they are essentially Z0 probes. As shown in the figure below, the CA-xxx used by P76xx is a 50 Ω connection front end such as various SMA connections. This SMA connection front end cannot complete the task of online signal testing. P76TA is a soldered front end, but its DC impedance is only 450Ω differential and 225Ω single-ended. This low impedance has a great load effect on the differential 100Ω circuit:

Figure 10 P76xx 33GHz probe in the industry

Therefore, in order to compensate for the signal distortion caused by the probe loading effect, the oscilloscope will " boost the measured signal gain ", and this algorithm compensation must also " assuming a 25 ohm signal source impedance ". Regardless of whether this assumption is universally applicable, this simple one-size-fits-all "boosting" of the signal will inevitably amplify the oscilloscope's background noise while amplifying the signal, swallowing up the poor margin of high-speed and low-voltage signals, and thus causing obvious errors and measurement uncertainties.

Figure 11 P76TA signal processing instructions (Source: P76xx User Manual)

The high-bandwidth, low-voltage differential probe mentioned above has another special case - InfiniiMode probe , which supports differential signal testing . It can complete the testing of four signals, differential, single-ended A, single-ended B, and common mode, in one welding .

Figure 12 MX0109A InfiniiMode front-end detection diagram

Figure 13 InfiniiMode probe signal detection results

for example:

The above is a brief introduction to the types and differences of probes commonly used in current circuit testing and some application scenarios. The following table is a full compatibility table of current Keysight's main probe products and oscilloscope models, please refer to it:

Table 3 Keysight oscilloscope and probe compatibility quick reference table

Figure 16 The weakest link in the signal transmission chain determines the overall performance of the measurement system

The above article provides a brief introduction to the types, characteristics and applicable scenarios of probes.

Due to space limitations, in actual use, in addition to the important indicators such as bandwidth, impedance, and capacitance that must be considered when selecting, there are also other indicators such as the probe's attenuation factor, dynamic range, bias capability, etc. These indicators and factors will also affect the measurement accuracy of the signal. We will continue to discuss this in the future, so stay tuned!

• [Dry Goods] High-bandwidth probe potential and efficiency improvement and front-end selection and welding guidelines

• The curtain of the optical era has opened, and the photoelectric probe fills the last shortcoming of the real-time oscilloscope!