Differential pair impedance calculations

F

Fibo

Guest
I'm about to lay out a USB 90 ohm diff pair, and I go to the eeweb impedance online calculator

https://www.eeweb.com/tools/edge-coupled-stripline-impedance

I punch in my numbers from the fab house

Trace Thickness = 1.4 mil
Substrate Height = 14.4 mil
Trace Width = 6 mil
Trace Spacing = 8 mil
Dielectric = 4.5

and I get 89.9 Ohms

Then I try to verify against another impedance calculator, I pull up Saturn PCB Toolkit and plug in the same info

Conductor Width = 6 mil
Conductor Spacing = 8 mil
Conductor Height = 14.4 mil
Base Copper Weight 1oz
Dielectric 4.5

and I get 77 Ohms


Does anyone have a tool they like for these type of impedance calculations? I don't know why the solutions are are off, I suspect user error

These are for inner layers with a ground plane...

Top
GND
Sig1
Sig2 <--- USB Diff Pair here
GND
Bot

Much Thanks!
 
On Tue, 18 Jun 2019 19:20:02 -0700 (PDT), Fibo <panfilero@gmail.com>
wrote:

I'm about to lay out a USB 90 ohm diff pair, and I go to the eeweb impedance online calculator

https://www.eeweb.com/tools/edge-coupled-stripline-impedance

I punch in my numbers from the fab house

Trace Thickness = 1.4 mil
Substrate Height = 14.4 mil
Trace Width = 6 mil
Trace Spacing = 8 mil
Dielectric = 4.5

and I get 89.9 Ohms

Then I try to verify against another impedance calculator, I pull up Saturn PCB Toolkit and plug in the same info

Conductor Width = 6 mil
Conductor Spacing = 8 mil
Conductor Height = 14.4 mil
Base Copper Weight 1oz
Dielectric 4.5

and I get 77 Ohms


Does anyone have a tool they like for these type of impedance calculations? I don't know why the solutions are are off, I suspect user error

These are for inner layers with a ground plane...

Top
GND
Sig1
Sig2 <--- USB Diff Pair here
GND
Bot

Much Thanks!

That's tricky, asymmetric buried diff pair. Looks like the eeweb calc
assumes symmetric. Saturn does asymmetric as an option.

For tough cases, we use an EM simulator, like ATLC or ATLC2.

It is fun to run various impedance calculators and compare the
numbers. Some use the old Motorola formula, where wide traces have
negative impedances.

For short runs and modest data rates, a little error doesn't matter.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics
 

Welcome to EDABoard.com

Sponsor

Back
Top