Maker Pro
Maker Pro

differential stripline question

OK, I lied, it's probably questionS.

I am using a 1508 pin 1mm pitch BGA package. I wish to route out all
the differential high speed signals. I will need to use thin traces,
3.5 mils with 4 mils spacing, on Dk 3.3 Melco 4000-13 material 10 mils
apart, to route with edge-coupled 100 ohm routes out of the via array.
However, once out of the via forest, I wish to change to thicker
traces to minimize losses, however just with 5 mil traces I already
need something like 24 mils air gap to keep 100 ohms differential
impedance on that configuration.

1) How do I route the transition from one set of rules to another?
Just sort of eyeball it, make the traces thicker as they spread apart?
With 100ps edge rates, how long or short should this region be? What
exactly the hell is happening when two different set of rules give
close to 100 ohms impedance, is the configuration of fields different
somehow? I'm assuming the skinnier traces kill off higher frequency
signals quicker, yes?

2) Do thicker traces help reduce losses in a give maetrial, or is it
better to switch to a more exotic material like Rogers 4350 and keep
the original rules and so be able to route better?

3) Is there something I can read that's been translated to PCB-monkey
speak, that can give me the basics of this very complex subject? I
realize this subject is probably as fraught with dangers and staunch
opinions as power supply decoupling debates (array of different values
or lots of the same value caps?)

4) Yeah I'm cracking open the Black Magic books and a Bogatin book,
but any "voice of experience" remarks appreciated.

TIA to all.
 
OK, I lied, it's probably questionS.

I am using a 1508 pin 1mm pitch BGA package. I wish to route out all
the differential high speed signals. I will need to use thin traces,
3.5 mils with 4 mils spacing, on Dk 3.3 Melco 4000-13 material 10 mils
apart, to route with edge-coupled 100 ohm routes out of the via array.
However, once out of the via forest, I wish to change to thicker
traces to minimize losses, however just with 5 mil traces I already
need something like 24 mils air gap to keep 100 ohms differential
impedance on that configuration.

1) How do I route the transition from one set of rules to another?
Just sort of eyeball it, make the traces thicker as they spread apart?
With 100ps edge rates, how long or short should this region be? What
exactly the hell is happening when two different set of rules give
close to 100 ohms impedance, is the configuration of fields different
somehow? I'm assuming the skinnier traces kill off higher frequency
signals quicker, yes?

2) Do thicker traces help reduce losses in a give maetrial, or is it
better to switch to a more exotic material like Rogers 4350 and keep
the original rules and so be able to route better?

3) Is there something I can read that's been translated to PCB-monkey
speak, that can give me the basics of this very complex subject? I
realize this subject is probably as fraught with dangers and staunch
opinions as power supply decoupling debates (array of different values
or lots of the same value caps?)

4) Yeah I'm cracking open the Black Magic books and a Bogatin book,
but any "voice of experience" remarks appreciated.

TIA to all.

Oh crap, it's "Nelco" not Melco...
 
D

David L. Jones

Jan 1, 1970
0
OK, I lied, it's probably questionS.

I am using a 1508 pin 1mm pitch BGA package. I wish to route out all
the differential high speed signals. I will need to use thin traces,
3.5 mils with 4 mils spacing, on Dk 3.3 Melco 4000-13 material 10 mils
apart, to route with edge-coupled 100 ohm routes out of the via array.
However, once out of the via forest, I wish to change to thicker
traces to minimize losses, however just with 5 mil traces I already
need something like 24 mils air gap to keep 100 ohms differential
impedance on that configuration.

1) How do I route the transition from one set of rules to another?
Just sort of eyeball it, make the traces thicker as they spread apart?
With 100ps edge rates, how long or short should this region be? What
exactly the hell is happening when two different set of rules give
close to 100 ohms impedance, is the configuration of fields different
somehow? I'm assuming the skinnier traces kill off higher frequency
signals quicker, yes?

The 100ohm differential impedance is usually not important at all, it's the
50ohm SE that matters:
http://www.speedingedge.com/PDF-Files/diffsig.pdf
Although it ultimately depends upon your actual circumstances.

Not having to fight the 50ohm SE/100ohm DIFF non-linear trade-off balance
can give you a lot extra routing space.

Dave.
 
Top