USB trace length differences...

On Sunday, November 6, 2022 at 4:58:38 PM UTC-8, Don Y wrote:
Back of napkin suggests I can support about half-an-inch trace
length [difference] (if we reduce it to purely an issue of length) for USB2.
Before I start the layout, I have to sort out how to attach
\"discrete wires\" to the USB signals (to an offboard connector).

What\'s the \"safest\" way to allow for differences before
nailing down the mechanicals?

It\'s ugly, but you could make multiple wire-attach points on the
printed wiring board, and select the cable-to-board landing according to
test, with zero-ohm resistors installed to stretch (or spot-mill the copper to shrink)
trace length.
 
On Monday, 7 November 2022 at 04:28:34 UTC, whit3rd wrote:
On Sunday, November 6, 2022 at 4:58:38 PM UTC-8, Don Y wrote:
Back of napkin suggests I can support about half-an-inch trace
length [difference] (if we reduce it to purely an issue of length) for USB2.
Before I start the layout, I have to sort out how to attach
\"discrete wires\" to the USB signals (to an offboard connector).
What\'s the \"safest\" way to allow for differences before
nailing down the mechanicals?
It\'s ugly, but you could make multiple wire-attach points on the
printed wiring board, and select the cable-to-board landing according to
test, with zero-ohm resistors installed to stretch (or spot-mill the copper to shrink)
trace length.

What speed do you want to run at? That will make a big difference to the answer.

John
 
On 11/6/2022 9:28 PM, whit3rd wrote:
On Sunday, November 6, 2022 at 4:58:38 PM UTC-8, Don Y wrote:
Back of napkin suggests I can support about half-an-inch trace
length [difference] (if we reduce it to purely an issue of length) for USB2.
Before I start the layout, I have to sort out how to attach
\"discrete wires\" to the USB signals (to an offboard connector).


What\'s the \"safest\" way to allow for differences before
nailing down the mechanicals?

It\'s ugly, but you could make multiple wire-attach points on the
printed wiring board, and select the cable-to-board landing according to
test, with zero-ohm resistors installed to stretch (or spot-mill the copper to shrink)
trace length.

Well, ALL of the solutions we\'ve looked at are \"ugly\", to some
degree. Small size is really a bitch to work with (unless
you have an equally small design! :< )

And, when you\'ve already used the back side of the PCB, you
can\'t even rely on THAT for an \"out\"!

The idea was offered to put a tiny MCU *at* the USB connector
end of the pigtail -- and, in doing so, change the requirements
for that pigtail. The added software would be a cakewalk so it\'s
just a packaging issue.

Better to find these problems before layout than after!
(something about 10 pounds in a 5 pound sack...)
 
Don Y <blockedofcourse@foo.invalid> Wrote in message:r
> Back of napkin suggests I can support about half-an-inch tracelength (if we reduce it to purely an issue of length) for USB2.Before I start the layout, I have to sort out how to attach\"discrete wires\" to the USB signals (to an offboard connector).[No, cannot be a flex cable, etc. Must be individual wireswith NOTHING on the PCB end (not even a connector).]I can make assembly a bit easier by spacing out the attachmentpads for the individual conductors.Again, BEFORE STARTING THE LAYOUT, I can hope that the placeschosen for these attachment points lead to trace lengths thatare \"close enough\" to make the assembly work, electrically.Or, I can plan on adjusting the trace lengths so that theassembled device (including pigtail) fits the specification(which would allow the discrepancies of the pigtail andtraces to be adjusted in concert -- subject to themechanical constraints).I realize most folks just mount a USB connector on their PCB.Or, a pigtail that is terminated in a PCB-mountable connectorthat ensures the conductors in the pigtail are matched, inlength.What\'s the \"safest\" way to allow for differences beforenailing down the mechanicals?

Solder pads?
Isnt the spec for usb 100ohm, some max pf and a max length ( trip
time)?


Cheers
--


----Android NewsGroup Reader----
https://piaohong.s3-us-west-2.amazonaws.com/usenet/index.html
 
On 11/7/2022 2:06 PM, Martin Rid wrote:
> Solder pads?

Yes, but even finding *accessible* space for four decent-sized pads
is a challenge (it\'s a really tight layout; I\'ve \"integrated\" as
much as possible and still struggling to fit in the space available)

Isnt the spec for usb 100ohm, some max pf and a max length ( trip
time)?

There are, also, constraints as to the amount of skew that can exist
between signals (~100ps for USB2, IIRC). (And I think *90* ohms).

Most folks put a connector on a board. I\'ve got a connector on
a pigtail. But, the lengths of the wires in the BULK of the
cable are matched.

OTOH, there\'s nothing that says I have to keep them at these
lengths. If, for example, I need a bit more wire to \"reach\"
an attachment point for one signal, that lengthens that
signal line. If the trace length to the pad is correspondingly
shorter, then it\'s a wash (largely).
 

Welcome to EDABoard.com

Sponsor

Back
Top