2
\$\begingroup\$

I am currently working on a 4-layer PCB that uses a Cypress PSOC4200 with a USB Full Speed connection. However, I am completely overextended with calculating the impedance and the resulting requirements of the traces.

I am working on a printed circuit board with the following layers:

Layer 1: Top (USB Signals) Layer 2: GND Layer 3: VCC Layer 4: Bottom

I want to layout the D+ / D- Signal Lines in CADSOFT Eagle, however this program does not offer calculation. Can anyone point me in the right direction? I have seen the PCB Saturn Software, but I am not sure what to fill in. I am planning to order the PCB from this manufacturer:

http://www.pcb-specification.com (Tag Layer Build ML4)

\$\endgroup\$

2 Answers 2

Reset to default
1
\$\begingroup\$

For FS mode of operation you don't need to keep precise impedance of traces. The trace impedance control is only important for HS modes (480 Mbps) and above.

It is also strange that you didn't mention the "stack-up" of your particular PCB design, which is critically important when selecting trace width for impedance-controlled design.

Just for reference: ExpressPCB (quick turnaround for 4-layer PCB) offers 0.305 mm dielectric spacing between inner (ground) layer and signal layers. If you want a 50-Ohm trace, the trace width must be around 0.51 mm wide. There are plenty of PCB trace impedance calculators that allow pretty accurate trace selection, Google for "pcb trace impedance calculator".

\$\endgroup\$
0
\$\begingroup\$

Is there a specific reason why you want to calculate exact impedances? Does the soc specify you to do so? The data sheet would usually have a reference schematic that you can follow. For most of my USB interfaced peripherals I use 0.254mm traces that are not more than 25mm from the device pins they are connected to. I also put in 24R resistors on each line.

\$\endgroup\$
5
  • \$\begingroup\$ Agree completely! Unless the length of the line is greater than 1/8 wavelenth for your highest frequency of interest, the use of transmission line theory and controlled impedance traces is a bit of a luxury. Anyone who says otherwise is probably trying to sell you something you don't need or doesn't know transmission line theory too well. If you get stuck, there's a great book on transmission lines by Waddell books.google.co.uk/books/about/… \$\endgroup\$
    – N.G. near
    Commented Jul 31, 2016 at 13:39
  • \$\begingroup\$ Transmission line effects are always present, I'd use a 1/10 or 1/20 wavelength to decide how much emphasis should be given to it. P.S the web is full of simple controlled impedance calculators. There are also free 2d field solvers for this. \$\endgroup\$
    – Mike
    Commented Jul 31, 2016 at 14:40
  • \$\begingroup\$ Yes, you are right, trying to sell transmission line theory for USB FS mode is exactly the OP does not need. The Nyquist frequency of FS signal is 6MHz, so your 1/8 wavelength rule comes down to about 140 inches of trace, or abut 3.5 m. \$\endgroup\$
    – Ale..chenski
    Commented Jul 31, 2016 at 19:26
  • 1
    \$\begingroup\$ That's misleading, because the rise/fall times usually have the high-frequency content. No matter is 6MHz or 50... \$\endgroup\$
    – Junius
    Commented Jul 31, 2016 at 20:05
  • \$\begingroup\$ Not really, because the high-frequency content in signal communication is actually a parasitic content. Physical layer designers actually tend to limit the high-frequency content by limiting edge rates to minimum, especially for high-speed busses. That's why USB specifies the faster limit for fall/rise time in HS mode, but has no slow limit, which is defined only by ability of signal to meet the eye opening (signal mask) at destination end. \$\endgroup\$
    – Ale..chenski
    Commented Jul 31, 2016 at 20:48

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.