> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Dave McGuire
> Sent: 06 January 2006 02:12
> To: geda-user@seul.org
> Subject: Re: gEDA-user: trace calculation[Scanned]
>
> On Jan 5, 2006, at 4:27 PM, DJ Delorie wrote:
> -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Dan McMahill
> Sent: 05 January 2006 22:07
> To: geda-user@seul.org
> Subject: Re: gEDA-user: trace calculation
>
> Robert Thorpe wrote:
> > Microvias in particular
> I suggest it to be W and not W/2. Then the width "bump" would be 0.
I wasn't describing a change. I was describing the way it already
happens to work.
On Fri, Jan 06, 2006 at 09:12:09AM -0500, DJ Delorie wrote:
>
> > > As an FYI, pcb always uses round flashes for traces. A trace of width
> > > W has a corner with radius W/2 on the outside, not a sharp 90 degree
> >
> > Why W/2 and not W?
>
> The diameter of the rounded ends of traces is W, to
> > As an FYI, pcb always uses round flashes for traces. A trace of width
> > W has a corner with radius W/2 on the outside, not a sharp 90 degree
>
> Why W/2 and not W?
The diameter of the rounded ends of traces is W, to match the
thickness. Therefor the radius is W/2. That's just the way it
On Thu, Jan 05, 2006 at 09:03:57PM -0500, DJ Delorie wrote:
>
> As an FYI, pcb always uses round flashes for traces. A trace of width
> W has a corner with radius W/2 on the outside, not a sharp 90 degree
Why W/2 and not W?
CL<
> angle like the PDF's reports. That probably eliminates even the
On Thu, Jan 05, 2006 at 01:39:11PM -0500, DJ Delorie wrote:
>
> > Why? Via doesn't have much more inductance than a piece of trace,
> > does it?
>
> At high enough frequencies, traces are waveguides, not just
> conductors. Vias have nontrivial geometry relative to the signal.
>
> That reminds m
On Thu, Jan 05, 2006 at 02:17:36PM -0500, DJ Delorie wrote:
>
> > is that it is enough to just miter the corners of a 90degree bend,
> > or (better) use two 45 degree bends to transition between a 90
> > degree bend for all but the very highest frequencies.
>
> Would it be useful to add an optimi
On Thursday 05 January 2006 05:21 am, Karel Kulhavy wrote:
> Would it be possible to enter material constants and layer
> ordering into the PCB and then make a function that would
> display resistance, inductance and capacitance of a track
> (alone or with reference to infinite groundplane at a
> d
> On Thu, Jan 05, 2006 at 08:13:23AM -0800, Larry Doolittle
wrote:
> > I don't even need to handle vias -- the critical nets
> > that this would be used on most shouldn't use vias. :-p
On Thursday 05 January 2006 01:16 pm, Karel Kulhavy wrote:
> Why? Via doesn't have much more inductance than a
Vias also have problems with continuity of the ground plane. There was a white paper a couple years ago that showed good results with four ground vias arranged around the signal via (I forget what they did for differential pairs). Unfortunately, I don't have a copy any more :-(
---
On Jan 5, 2006, at 4:27 PM, DJ Delorie wrote:
Isn't tuning trace _lengths_ more important on pcbs than what types
of corners you're using?
On my AMD motherboard, the bus traces have s-shaped wiggles in them to
keep the lengths all the same. Rounded corners, too.
I have several boards from
As an FYI, pcb always uses round flashes for traces. A trace of width
W has a corner with radius W/2 on the outside, not a sharp 90 degree
angle like the PDF's reports. That probably eliminates even the
trivial differences between corners.
An arc corner would get rid of the 90 degree inside tra
round corners are slightly better.
-- Original message -- From: DJ Delorie <[EMAIL PROTECTED]> > > > My experience at ~1GHz (2Gbps FibreChannel) is that the difference > > between round corners and 45 degree pairs is small, but if you're > > pushing the limits like we wer
Robert Thorpe wrote:
Microvias in particular are thin and have significant inductive
reactance at high frequencies. Normal vias are not so bad.
you can minimize the effect of a via by nominally matching it to the
trace. In other words by varying the via diameter you have a handle on
its impe
On Jan 5, 2006, at 2:17 PM, Phil Taylor wrote:
Or another way to consider this: Isn't it possible to make a trace
that a
certain frequency cannot pass through due to nothing other than its
length?
Nope. One of the simplest solutions to Maxwell's equations is a wave
attached to an infi
> My experience at ~1GHz (2Gbps FibreChannel) is that the difference
> between round corners and 45 degree pairs is small, but if you're
> pushing the limits like we were, it can be significant.
Which way was better?
My experience at ~1GHz (2Gbps FibreChannel) is that the difference between round corners and 45 degree pairs is small, but if you're pushing the limits like we were, it can be significant. The biggest problem we had with round corners was a CAD package that generated gerbers with line segments fo
> The rounded corners are probably overkill.
I guessed that. Maybe they've found it improves their quality or
yield. Maybe they just think it looks pretty.
On Thu, Jan 05, 2006 at 04:27:20PM -0500, DJ Delorie wrote:
>
> On my AMD motherboard, the bus traces have s-shaped wiggles in them to
> keep the lengths all the same. Rounded corners, too.
The rounded corners are probably overkill. Those traces have
to carry 3.2 GB/s, IIRC. So response up to
> Isn't tuning trace _lengths_ more important on pcbs than what types
> of corners you're using?
On my AMD motherboard, the bus traces have s-shaped wiggles in them to
keep the lengths all the same. Rounded corners, too.
Phil et al. -
On Thu, Jan 05, 2006 at 04:17:44PM -0500, Phil Taylor wrote:
>
> 50 ohm hard coax makes sharp bends all the time, but it's still 50 ohm.
> Reflections do exist but they're way down. This is no doubt because the parts
> are designed very carefully.
Right. Someone tuned the excess
DJ Delorie <[EMAIL PROTECTED]> wrote:
> makes the board look pretty, but I wonder if gentle curves provide
> better waveguide performance than corners?
>
50 ohm hard coax makes sharp bends all the time, but it's still 50 ohm.
Reflections do exist but they're way down. This is no doubt because
> > is that it is enough to just miter the corners of a 90degree bend,
> > or (better) use two 45 degree bends to transition between a 90
> > degree bend for all but the very highest frequencies.
>
> Would it be useful to add an optimization that puts rounded corners on
> 90s instead of miters? W
On Thursday 05 January 2006 13:39, DJ Delorie wrote:
>> Why? Via doesn't have much more inductance than a piece of trace,
>> does it?
>
>At high enough frequencies, traces are waveguides, not just
>conductors. Vias have nontrivial geometry relative to the signal.
>
>That reminds me of another opti
I've detailed this in the past, but to summarize:
> -You would need to have some way for the program to in advance how much
> angle (arc) you wanted to cover.
It's automatic. The straight parts are always tangents to the curves
around obstacles. See the png I posted.
> -You can probably pac
> is that it is enough to just miter the corners of a 90degree bend,
> or (better) use two 45 degree bends to transition between a 90
> degree bend for all but the very highest frequencies.
Would it be useful to add an optimization that puts rounded corners on
90s instead of miters? We already h
DJ Delorie wrote:
Why? Via doesn't have much more inductance than a piece of trace,
does it?
At high enough frequencies, traces are waveguides, not just
conductors. Vias have nontrivial geometry relative to the signal.
That reminds me of another optimization I'd like to implement in the
> That reminds me of another optimization I'd like to implement in the
> future.
It would look like this: http://www.delorie.com/pcb/pullit-sample.png
> > Why? Via doesn't have much more inductance than a piece of trace,
> > does it?
>
> At high enough frequencies, traces are waveguides, not just
> conductors. Vias have nontrivial geometry relative to the signal.
Indeed. Folks doing > 1GHz design (to name a rough cut-off point)
spend a lot of
l.org
> Subject: Re: gEDA-user: trace calculation[Scanned]
>
>
> > Why? Via doesn't have much more inductance than a piece of
> trace, does
> > it?
>
> At high enough frequencies, traces are waveguides, not just
> conductors. Vias have nontrivial geo
> Why? Via doesn't have much more inductance than a piece of trace,
> does it?
At high enough frequencies, traces are waveguides, not just
conductors. Vias have nontrivial geometry relative to the signal.
That reminds me of another optimization I'd like to implement in the
future. Instead of c
> Why? Via doesn't have much more inductance than a piece of trace, does
> it?
It does. Impedance changes of any kind are bad for high speed signals.
Matt
On Thu, Jan 05, 2006 at 08:13:23AM -0800, Larry Doolittle wrote:
> Friends -
>
> On Thu, Jan 05, 2006 at 11:21:40AM +0100, Karel Kulhavy wrote:
> > Would it be possible to enter material constants and layer ordering into
> > the PCB and then make a function that would display resistance,
> > induc
Friends -
On Thu, Jan 05, 2006 at 11:21:40AM +0100, Karel Kulhavy wrote:
> Would it be possible to enter material constants and layer ordering into
> the PCB and then make a function that would display resistance,
> inductance and capacitance of a track (alone or with reference to
> infinite groun
Would it be possible to enter material constants and layer ordering into
the PCB and then make a function that would display resistance,
inductance and capacitance of a track (alone or with reference to
infinite groundplane at a different given layer)? You would define
starting and ending point and
36 matches
Mail list logo
