I roughly understand what a differential pair is (in cables, at least, which take the form of pair(s) of twisted insulated conductors). How does this apply to the PCB, beside the fact that the two signal is as-close-to-one-another-as-possible? What are the “squiggly” sections used for? Is it for simply making sure that the two sections’ length is equal? Are there any limitations on via placements?
Given those, can I make my own diff pairs?
The tracks of a twisted pair are not as closely together as possible. The idea is to create a controlled impedance. Via’s can be problematic in differential pairs, but it depends a lot on what the differential pair is used for. For USB3, it can be GHz signals, while for most other uses the frequencies are lower, and signal flanks are less steep and that allows for more impedance mismatches in the pair.
The “Squiggly sections” are indeed for length matching (KiCad has built-in functions for that) Lengt matching is important for a high speed bus, for example for DDR IC’s a fast ADC or for LVDS for monitors.
For how to create them, read the manual:
The same document is very likely also available via: PCB Editor / Help / Help
1 Like
Well my intention is to be … yeah I dont think it’s going to work. Maybe it will
I have 2 USB connectors and a chip. One is a male and the other is a female. I currently intend on routing the two USB to some pin headers and allow you to swap the connections’ (for the chip) to either connector to either be a host or a device.
What is the speed for your USB signal? USB can be anywhere between 1.5Mbps and 10Gbps.
It’s for USB 2.0 (a single pair). so 480KHz, I believe.
Obviously, it would be best to have the best signal quality since I have considered to use it as some sort of extender (thereby bypassing the chip). Maybe that will be exclusively for power, in which case the signal pairs are irrelevant.
I think being able to switch between the two (male and female) connector to be quite important, even though I worry about signal integrity. That said, I have cases with very flimsy solders&pin headers where non-twisted different-length random wires can carry enough signal integrity for the computer to recognize the device (a USB keyboard I think).
If, however, that it indeed will not work (e.g. the pin header being too noisy), then I will just solder bridge the pads and it shouldn’t be very important. (to have the pins left out)
USB 2 is not always 480Mbps. If you connect a keyboard to an USB 2 hub then it will still be 1.5Mbps between the keyboard and the hub. I do think that the hub re-interprets the message and uses a 480Mbps uplink to the PC. With 480Mbps signal integrity is an important issue, but it’s far less critical then at 10Gbps. With 480Mbps you can probably afford some via’s or small stubs created by connector pins. USB is a quite complex standard and I do not have much experience with it. Others can probably give better advise here.
1 Like
I know the USB-OTG specifically require the use of either Micro-A/B or Mini-A/B ports, which have the ID pin. However, both of them are complete garbage (in terms of durability/mechanical strength), and thus I have chose USB-B for downstream and maybe USB-A for upstream (since I have a few receptacles lying around).
Maybe I should have just tied the data pins all together and just never plug 2 things into it at the same time. However, currently I plan on drawing power over the downstream port, so there need to be a way to cut the data connection (to whatever that the power is being drawn from) to avoid inteference.
In a finished product (that is not a “development board”) I would imagine either the Micro-A/B connector to be used, or the chip tied into a specific mode of operation and use a single connector.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.