Electronics > Projects, Designs, and Technical Stuff

USB differential routing

(1/3) > >>

shadewind:
I'm using the FTDI FT232RQ (the QFN version of the FT232R) for a little USB-to-serial design I'm working on. I have some questions about the routing of the differential data lines:


* The characteristic impedance of each line is supposed to be 45 ? ohms if I understand correctly. My board is a two layer 1.6 mm FR4 board and using a line calculator (TXline) and assuming the dielectric constant for FR4 is 4.7, I calculate the micro strip width to 134.711 mil which must mean I'm doing something wrong. Or am I?
* For some reason, the pins on the FT232RQ are flipped with regards to the USB connector which means I must pass at least one of the traces down through a via and back up. According to most data sheets, this is very bad but according to most datahsheets but according to Lee W. Ritchie (http://www.speedingedge.com/PDF-Files/diffsig.pdf), it is not and he seems to know what he's talking about. What to do?

mikeselectricstuff:
This isn't an issue at USB low or full speed as data rate is only 12mbits/sec - it's only high-speed (480mbits) where you start needing to be more careful.

amspire:
Mike is right.

The FTDI FT232RQ is a USB 2.0 Full Speed device which means the USB runs at 12MHz as apposed to a High Speed device running at 480MHz.

So basically as long as the FTDI FT232RQ is close to the USB socket the tracks will be too short to behave like transmission lines.

Richard

shadewind:
I figured as much but it would still be interesting to know what I would have done at high speed. Are my micro strip calculations wrong or is it just that my ground plane is too far away (1.6 mm)?

joelby:
Pretty much. As a rule of thumb, a 50 ohm microstrip trace width should be twice the thickness of the dielectric, or the same width as the thickness for 75 ohms.

Your options could include using a 0.8 mm board, or a four layer board, where you could have a top-GND thickness of, say, 0.15mm.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version