The same via has different impedances?

Public number: Mr. Express

Author: Huang Gang

On the processed PCB board, two high-speed team members conducted a via impedance test on the board. As a result, the impedance tested by the two is very different. Can you believe it?

This is a test board with SMA coaxial connector, which is used for internal testing of various PCB structures. Among them, the performance of the 2xthru de-embedding routing we made is particularly important, and it determines our de-embedding accuracy. As for the specific principle of testing de-embedding, Mr. Gao has already described in an article before, so I won’t repeat it here.

After the PCB board came back, the master Chris asked his apprentices Lei Bao and Tang Niu to test the board. Not long after, Chris went to the laboratory to check the testing of Lei Bao and Tang Niu. As a result, he heard a very “fierce dispute” between Lei Bao and Tang Niu as soon as he walked into the laboratory.

Leibao: The 50 ohm SMA via impedance is not bad, reaching 45 ohms, which meets the 10% requirement;

Tang Niu: How could it be so good! I clearly measured it was less than 42 ohms, which obviously did not meet the requirements.

Then they did show Chris the results of their test. One test reached 45 ohms, and the other was less than 42 ohms!

What is going on? It is obviously the same via and the same network analyzer is used for testing. Why is there a significant difference between the test results of two people?

Just when Lei Bao, Tang Niu and some fans of Mr. Express were puzzled, the well-informed Chris found a possible point. He immediately sat in front of the network analyzer to operate the 2xthru cable, and then Chris found the reason for the difference between Lei Bao and Tang Niu.

What exactly did Chris operate on the network analyzer? I saw him quickly switch the test stop frequency, and then test and compare the same 2xthru trace at different test frequencies, and set the stop frequency to 10GHz, 20GHz, and 30GHz. , 40GHz, and then conduct impedance test on the same signal!

As a result, something incredible happened! We looked at the impedance of this via under different test frequencies and found that there is really a significant difference!

The impedance of the via at 10GHz is 47 ohms, at 20GHz it is 45 ohms, at 30GHz it is 43 ohms, and when tested at 40GHz, the impedance is only 41 ohms.

I also asked Lei Bao and Tang Niu later, and found that Lei Bao was tested with the 20GHz frequency, and Tang Niu was tested with a higher frequency of 40GHz, so there was a “dispute” between the two of them.

But the impedance change is not a problem with the S-parameters they tested. Whether you use 20GHz or 40GHz for frequency domain testing, the measured S-parameters are the same, only the difference between 20GHz and 40GHz!

Chris looked back at Lei Bao and Tang Niu, and found that the two of them still seem to understand the phenomenon of different impedance values ​​measured. Then, fans of Mr. High-speed, do you understand?