[SI-LIST] : RE: 2.5 GHz in FR4

About this list Date view Thread view Subject view Author view

From: Larry Miller (ldmiller@rhapsodynetworks.com)
Date: Wed Apr 04 2001 - 08:46:50 PDT


Broadside coupled, keep goodly distance apart for crosstalk.

That is about all that my mgmt is comfortable with....

One thing:

When you get into this business you are not the driver on the board
parameters.

You need to go to a back/mid plane specialist fab house and let them tweak
the board thicknesses for their Er and processes. The UltraCad and Polar
calculators we all use are more than adequate for rough estimate of board
layer parameters (I use 4.0 - 4.2 for Er) for design/layout purposes.

The fab houses will come in and give proposals for board parameters based
upon solid experience (including whether or not you need to go to Getek or
better!). I am assuming that you work for a company/project big enough that
these guys will visit-- hey, this stuff is not for the weak of heart and
light of wallet!

You order by impedance and include TDR test coupons and reports (not usually
a problem since you only get one board per panel!)

Our results were extremely satisfactory.

"Welcome to the Age of Digital Microwave!", as Michael King has said.

Larry Miller

-----Original Message-----
From: Arun Vaidyanath [mailto:arun@sisilk.com]
Sent: Tuesday, April 03, 2001 1:24 PM
To: Larry Miller
Subject: 2.5 GHz in FR4

Larry,
That's great to hear that FR4 works well at 2.5g. Could you share info on
whether you use loose coupling - broadside or edge?
And what is short length on plug in board? What is the total length of the
gig nets?
We are designing a similar system and I am not sure what spacing to use
between pairs? I could play it safe and keep really low Kb levels but I need
the board routing density.
The other problem I see is what to use for tan delta and er of the FR4.
Different mfgrs. have different values - not to mention the er
characterization at 1 mhz!

By the way, the new ZD from Tyco simulated real well - if you use SLM; go
for multiline models and they probably still simulate well; but take a week
to run!

Thanks in advance for anything you share.

-arun

-----Original Message-----
From: owner-si-list@silab.eng.sun.com
[mailto:owner-si-list@silab.eng.sun.com]On Behalf Of Larry Miller
Sent: Tuesday, April 03, 2001 12:04 PM
To: 'Michael Nudelman'; si-list@silab.eng.sun.com
Subject: RE: [SI-LIST] : RE: 2.5 GHz in FR4

Not at all masochistic.

This was an extremely easy board to route. AMP, Erni and Teledyne (in no
particular order) have expended great amounts of research for high speed
back/mid planes using differential pairs for the telecomm and networking
industries and have reduced it down to a turn-the-crank operation. (The app
notes are readily available on their websites.)

Plus, the quotes we have regularly gotten from backplane/midplane vendors
show the cost for Getek to be about 1.8X that of FR-4, and the more exotic
ones like the Rodgers materials go up exponentially.

Thirdly, the other materials (Getek and up) have comparative availability
problems when you are using slabs that are 17" or 21" wide by 24" or more
high.

Lastly, for serial backplanes, the SERDES manufacturers have fitted their
chips up with pre-emphasis logic to hugely reduce the data-dependent jitter
(DDJ) caused by the roll-off vs. frequency of FR-4.

If anyone from Tyco or other such fabricators is listening, please chime in.

Larry Miller

-----Original Message-----
From: Michael Nudelman [mailto:mnudelman@tellium.com]
Sent: Tuesday, April 03, 2001 10:20 AM
To: si-list@silab.eng.sun.com
Subject: RE: [SI-LIST] : RE: 2.5 GHz in FR4

This is quite masochistic. Do you have to do it to yourselves and use FR4,
unless you are running couple of inches of hi-speed wires?

-----Original Message-----
From: Larry Miller [mailto:ldmiller@rhapsodynetworks.com]
Sent: Tuesday, April 03, 2001 12:57
To: 'Gaboian, Jerry'; 'Zabinski, Patrick J.'; si-list@silab.eng.sun.com
Subject: RE: [SI-LIST] : RE: 2.5 GHz in FR4

I, too, must take issue with practically every statement made by Mr.
Zabinski.

We are using stripline. We do not have vias except the connector pins (AMP
HS-3).

We are using 10-mil stripline traces on the backplane, 5-mil stripline on
the plug-in boards (have to, the boards get too thick otherwise). Yes, the
5-mil sections are lossy because of skin effect; luckily, these are short.

We have backplane traces that are over 12" long (can't help it).

In spite of all this, our system happily percolates along at 3.125 GHz (2.5
GHz with 8B/10B coding).

Larry Miller

-----Original Message-----
From: Gaboian, Jerry [mailto:j-gaboian@ti.com]
Sent: Tuesday, April 03, 2001 5:00 AM
To: 'Zabinski, Patrick J.'; si-list@silab.eng.sun.com
Subject: RE: [SI-LIST] : RE: 2.5 GHz in FR4

Check out this site from AMP.....

It will surprise you on how long of a trace one can use.....
They also compare other materials, widths act.....

http://www.amp.com/simulation/files/papers/dc99cmdh.pdf

Regards,
Jerry

-----Original Message-----
From: Zabinski, Patrick J. [mailto:zabinski.patrick@mayo.edu]
Sent: Tuesday, April 03, 2001 6:24 AM
To: si-list@silab.eng.sun.com
Subject: [SI-LIST] : RE: 2.5 GHz in FR4

Sreejith,

It is possible to use common FR4 in 2.5 GHz systems, but you
need to follow a few design constraints:

* Do not use stripline. The discontinuities presented by
the vias are most often killers. Avoid all vias in the high
speed nets and use microstrip (i.e., top-surface only).

* Use wide traces. FR4 is lossy at 2.5 GHz, so try to compensate
for the high dielectric loss by decreasing the metal loss. Use
wider traces which will reduce the resistive losses in the metal.

* Use short traces. If you need to route 2.5 GHz signals
on nets that are 10 inches (25 cm) long, I doubt if your
system will function. Keep the traces *short* (say, less
than 1 inch [2.5 cm]).

Hope this helps,
Pat

> HI All,
>
> i am doing a 2.5Ghz PCB design. i was wondering whether FR4 material
> is sufficient for such high speed signals? or i should look at some
> other material for these desings? and if FR4 is sufficient is
> there any
> special care i have to take (other than impedance matching
> and differential
> routings)
>
> thanks in advance,
> Sreejith
> Paxonet Comm. (I) Ltd.
> Blore, India

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Jun 21 2001 - 10:11:26 PDT