USB 2.0 differential impedance - different results in different calculators

[Pak00]

Hi all,

I would like to ask about the correct differential impedance calculation in the USB 2.0 interface. I tried to calculate differential impedance in three independent calculators and calculate manually from the formula. I used a calculator embedded in Saturn PCB and an online calculator on JLCPCB and PCBWay web side.

The calculations are based on the selected option on the JLCPCB website:

For the same parameters, the calculator from the PCBWay website:


And Saturn PCB:


Summary:
PCBWay 107.60Ohm
Saturn PCB 85.48Ohm
Manual calculation 84.83Ohm

Why is the PCBWay result so different? What am I doing wrong? Currently, I do not know where the PCBs will be ordered, so I do not want to rely on only one calculator.
Maybe some of you can specify what parameters of the USB 2.0 differential pair he used for his stack-up of the 4-layer board.

Many thanks for the help.

[binayak]

Hi,
Try looking at these things:
- Include the effect of etch factor in trace design in all simulators (W2<W1). It will increase the impedance wrt. to the case where you provide single trace value of 6.03 mils.
- Include the effect of solder mask in all simulators. It will decrease the impedance wrt. to the case where you don't include SM.
- Try doing the same task in a software with 2D field solver. It will give more accurate result than softwares based on simple impedance equations.
- Try making all the modeling parameters same in all the softwares.

Hope my above 2 cents help!

[robertferanec]

PCB manufacturing process is more complex than a theory. PCB materials are for example squeezed and the final thickness may be different from the bare material thickness. There should not be so big difference (107 vs 85), but there always will be difference between numbers from the PCB manufacturer comparing to numbers from calculations based on bare materials.

PS: The PCB manufacturer should always measure the impedance after PCB is manufactured, so their suggested numbers should be correct. Maybe they have some correction coefficients included in their calculators? Or they use different materials .. or I maybe they have something wrong showing up on the website ... I am not sure.