I'm desinging a very simple 4 layer board with a USB connection. I defined an impedance of 90 Ohms (differential) in the layer stackup. This gave me a track width of 0.3 mm with a gap of 0.334 mm, which roughly agrees with my hand calculation. I used the differential pair routing option, matching the suggested measures as close as possible. I do need one pair of vias to get from a SMD connector to layer 3 where the routing is done, but apart from that the trace is a straight line.

However, the signal integrity calculates an impedance of 239 Ohms. This impedance doesn't change, no matter what I do. For test purposes, I changed the trace width, the dimension of the board, the thickness of the substrate and added a via fence around the differential pair traces. None of these measures caused any change to the impedance in Signal Integrity.

Am I doing something wrong? I mean I wouldn't expect the Signal Integrity impedance to be exactly the same, but this is off by a factor of more than 2.5... Has anyone experienced this problem before and found a solution to it? Any help would be highly appreciated.

Edit: I'm using Altium 19.1.8.

This is my layer stackup This is my signal integrity analysis output


1 Answer 1


What version of Altium are you using? Before 19, trace impedance analysis was an option, but Altium themselves freely admitted it didn't work right. For this reason, I'd be skeptical of their SI offerings. Best bet is to run your trace geometry past a couple of the many online impedance calculators to make sure you're in the ballpark.

  • \$\begingroup\$ Ah sorry, I forgot to mention that I'm using Altium 19.1.8. I actually tested my hand calculation with this one: www.eeweb.com/tools/edge-coupled-stripline-impedance and it gives me 96 Ohms, which is within the tolerance I would expect, considering that this online calculater doesn't account for my slightly asymmetrical stackup. \$\endgroup\$
    – Andy P.
    Mar 10, 2020 at 14:15
  • \$\begingroup\$ I actually tried some other calculators and got similar results. But it's good to know that Altium's SI is not very reliable. Thanks for your answer! \$\endgroup\$
    – Andy P.
    Mar 10, 2020 at 14:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.