Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Question about crosstalk

Status
Not open for further replies.

EDA_hg81

Advanced Member level 2
Joined
Nov 25, 2005
Messages
507
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,298
Activity points
4,808
data bus spacing to avoid crosstalk

When checking the signal integrity, have to separate two traces routed out from two neighbor pins for preventing from crosstalk.
I am wondering how those two pins prevent from crosstalk? Since the distance between them is fixed. The thickness of dialectic is fixed too.

Thank you.
:D
 

hi sidra here....to prevent the cross talk one thing has to be changed..the structure of the system (as u hv talked about dielectric) or the transmitting technique of the signal....like after dividing the spectrum into channels in cellular phone system guard channels are left between the channels to avoid crosstalk or any kind of interference.....by the way i like yr glasses.....ha ha..sidra_maheen84@yahoo.com
 

    EDA_hg81

    Points: 2
    Helpful Answer Positive Rating
Thank you sidra.

But I am still confused.

For example on 256Mb SDRAM, the distance between address or data bus are so close.
If the traces routed out from those buses are easy to get crosstalk, how they can make sure no crosstalk is going to happen between address or data bus?
Thank you:D
 

there are many ways to prevent crosstalks intercoupled between high-speed switching address and data bus lines.

for sure, u cannot change the dielectric of the substrate or PCB because it is fixed by technology.

1. widen the space or gap between address lines. likewise, do the same for data lines. this reduces capacitive coupling between lines.

2. place ground lines between address lines. likewise, do the same for data lines. this reduces inductive coupling between lines. this is more commonly used in VLSI.

3. route all address lines close together but in parallel between microprocessor and memory devices. the intercoupling can become cross-coupling cancellation due to all lines are coupling one another. this is a statistical method proven and used. but this is recommended for low-speed bus, but high-speed bus. likewise, do the same for data bus.
 

    EDA_hg81

    Points: 2
    Helpful Answer Positive Rating
Thank you SkyHigh,

Would you explain #3 more clearly?

Thank you again.
 

In layman principle, this is how it works.

Do you know how UTP (unshielded twisted pair) works and why it has good noise and coupling cancellation in a bundle of wires that carry different signals? This is precisely the same principle.

The whole idea is about a random process, if you have heard of it.

Let's say there are 3 signal lines in the address bus, A, B and C, running in parallel from the microprocessor to the memory device. If A is changing from 0 to 1, thus couples to B with a small "hump". C could be changing from 1 to 0, thus couples to B with a small "dip". Therefore the net of hump and dip is zero, hence a cancellation of intercoupling effect.

I hope you understand what I meant.

I suggest to you these books if you are interested. I personally have a copy of [1] because it is a good handbook for practising engineer in VLSI and high-speed system design. It is a book written by Intel engineers.

[1] High-Speed Digital System Design: A Handbook of Interconnect Theory and Design Practices (Hardcover)
by Stephen H. Hall, Garrett W. Hall, James A. McCall

[2] High-Speed Digital Design: A Handbook of Black Magic by Howard Johnson

[3] Signal Integrity Issues and Printed Circuit Board Design by Douglas Brooks
 

    EDA_hg81

    Points: 2
    Helpful Answer Positive Rating
u can reduce the cross talk by adjusting the D/H ratio as
cross talk = 1/(1 + (D/H)^2)
D - distance btn the traces
H - height of the dielectric

as the maximum return current travel just below the conductor and spreads out a little on both the sides of the trace. when u adjust D/H ratio u can control cross talk and a spacing a spacing of 3W btn the traces will work.
W - width of the trace
but practically in a high density board sticking to a spacing of 3xW is impossible...so adjust D/H and space the traces atleat (1-2)xW
... i think it will eliminate cross talk to a good level.
 

    EDA_hg81

    Points: 2
    Helpful Answer Positive Rating
Hi all
Cross talk can also be reduced by providing good return path for the signals.In EDA_hg81 case 3 things can be done.
If u have space u can increase the space between the traces this will help to some extent in reducing crosstalk.
Other wise u can route address and data buses in two different layers.Route in X direction on one layer and in Y direction on the other by this was the effect will be cancelled.
U can try introducing a ground plane or VCC plane underneath the layer this is the excellent way to reduce the crosstalk as the GND or VCC layer acts a perfect return path for the signals and also as a shield.

regards
prasanna
 

to minimize the cross talk you have to chosse the right seperation and trace width for the lines.
the more seperation the better cross talk. it also will change characteristic impedance .you have to use some free impadeance/microstripline calculator like TXline from MWO,you can aquire it in the web site or from agilent,

mehdi
 

any body help me.
if u run gaurd trace between two coupled traces,how much it will usefull to avoid crosstalk.
 

spaceing is better option thn gaurd trace which cause extra problems and extra vias
 

cross talk will depend on this equation
H^2/(H^+D^2)
IF THE HEIGHT IS GOING TO INCRESING FROM REF PLANE THRN THE CROSS TALK WILL BE MORE

AND IF DISTANCE IS GOING OT INCRESING CROSS TALK WILL BE REDUCE
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top