Electronics > Projects, Designs, and Technical Stuff
USB differential pair on 2 layer board
(1/1)
Jester:
I realize a 2 layer board is less than ideal for this, however that's what were stuck with for this application.
I'm laying out a board that will make use of an open source Arduino board for the uC section, because the code is already written and tested on the MkrZero board. I imported the PCB files and when I perused the USB differential pair, first thoughts are this is not right. Trace width is 0.25mm (9.84 mils), and space is 0.2mm (7.87 mils). This is a 1.6mm board. Assuming FR-4
Viewed as an edge coupled external plane, Zdiff =148 \$\Omega\$
However, both the top an bottom are flooded with ground plane as shown.
Viewed as a coplanar waveguide, Zo is close to 90 \$\Omega\$, not sure about Zdiff?
Looks like I need a more versatile calculator.
Is this just a compromise, the total length is only about 37mm (1.5"), or does the top flooded plane significantly lower Zdiff?
PCB house tells me to send it and they will tweak as required, however for 90 \$\Omega\$ based on edge coupled, the trace width would need to be closer to 40 mils and that's neither a tweak or feasible.
martin1454:
If you can - Use a 0.8mm PCB instead of 1.6mm, that will make it easier to approximate the 90 Impedance - Use as thick traces as possible, and be aware that if you let the ground pour on the top layer get too close, it will also alter the impedance
Rerouter:
usb on a 1.6mm thick board always calls for crazy thick traces, instead focus on keeping the distance you need to route your USB lines short, this way you dont need to worry about impedance anywhere near as much,
you can reduce how much trace width you need a bit by routing a picketed ground along side the differential pair. (coplanar Wave differential pair, its not in any free tool I have seen, but you can use it to trim down your Zo)
If you want to follow the spec as close as possible, 0.18mm gap, 0.8mm trace, 0.8mm gap to ground, yes it is crazy sized, and near to the ratio limits, but if you want close to 90 diff, 50 Zo, then that should put you inside the limits.
A good tool for this is "Saturn PCB Toolkit"
martin1454:
Also a small note - You will probably use USB 2.0 low speed mode, and not high speed mode, which means the impedance is not that important (compared to high speed at least) but still aim for the best possible impedance.
Jester:
Any comments on the apparent 150 \$\Omega\$ Zdiff on the Arduino board? It obviously works.
Wild guess as to how much the top ground plane helps?
When you say short, how short is short? If less about 1" and no expectation beyond USB 2.0 speed is this a reasonable compromise?
Navigation
[0] Message Index
Go to full version