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.

re: crosstalk noise reduction

Status
Not open for further replies.

jaya sree

Member level 3
Joined
Nov 9, 2009
Messages
55
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,288
Location
india
Activity points
1,791
re: crosstalk noise reduction

hi all ,

One of the ways to prevent crosstalk delay is to upsize victim.
But in case of crosstalk noise , aggressor is switching and victim is not switching. So how come upsizing victim will reduce crosstalk noise.

thank you
jaya sree
 

Re: crosstalk noise reduction

Cross-talk noise is something which causes a glitch at the victim due to a transition appearing at the aggressor.

This is especially due to the capacitive coupling between nets/wires.

Answer to your question: By upsizing the victim i.e decreasing the resistance of the victim, it will help the victim net to maintain a strong static voltage thereby eliminating crosstalk noise.

Refer to the following link:

https://vlsisystemdesign.com/kunal58625/php/crosstalk.php

All the best.
 
Last edited:
  • Like
Reactions: matter

    matter

    Points: 2
    Helpful Answer Positive Rating
Re: crosstalk noise reduction

hi prashant ,

I did not understand this statement in the link

"Hence, the first solution to reduce crosstalk noise, is to increase the Resistance of Victim driver (RV).i.e. downsize the victim driver, so that, the high resistance of the victim driver restricts the supply of current and charging of victim net capacitance during the rise time (tr) of aggressor signal, which would in turn reduce the bump height.

The second solution to reduce crosstalk noise, is to increase the Capacitance of Victim load (CV).i.e. upsize the victim load, thus the resistance will reduce, which will in turn help the victim net to maintain a strong static voltage."

Generally , to reduce crosstalk noise , I upsize driver of victim net. But I understood from the link info that i need to downsize the driver to have high resistance. Please correct me , If i went wrong.

Thank you
jaya sree

- - - Updated - - -

Please note that I am talking about fixing crosstalk noise only . I got to know that to fix crosstalk delay , upsizing victims helps . But to fix crosstalk noise , upsizing victim wont help.Please tell if this is correct and if yes why
 

Re: crosstalk noise reduction

Yes. You are confused with the terms victim driver and victim load. Let me explain.

The first solution is downsizing the victim driver, it means that when victim net is maintained at a higher resistance it will restrict itself from aggressor's high voltage acting on it at that instant of time.

eg: Think victim is at constant logic '0'. When there is a sudden 0 to 1 logic transition in aggressor's net, then due to coupling capacitance b/w Aggressor and Victim the aggressor's sudden logic change tries to alter the victim's logic '0'. But we made the victim's net as high resistance. So it can resist the current flow through the coupling capacitor and maintain its logic '0' thereby eliminating the crosstalk noise.

The second solution (upsize the victim load) is trying to say that if we maintain a strong logic '0' at victim net the logic transition (0 to 1) at aggressor cannot affect the strong '0' at the victim. It talks about noise margin levels. This is an analogous approach to the previous one.

Hope its clear now.

Let me know if not understood. I will try to make it much clear for you.

All the best.
 

Re: crosstalk noise reduction

Hi Prashant,

I dont think that reducing the driver strength of victim will reduce the crostalk..
If the driver of victim net would be weakened further, the transition on victim net would reduce fruther and the aggressor will have more impact on the victim net.
I have reduced crosstalk in the designs, by increasing drive. strength of victim net.

In fact sometimes, when the net is very long we break the net and add buffer.

I dont get how you are improving crosstalk by reducing the drive strength of victim net. Have you ever tried it or its just concept ??

regards.
Shobhit
 

Re: crosstalk noise reduction

Hi Prashant,

I dont think that reducing the driver strength of victim will reduce the crostalk..
If the driver of victim net would be weakened further, the transition on victim net would reduce fruther and the aggressor will have more impact on the victim net.
I have reduced crosstalk in the designs, by increasing drive. strength of victim net.

In fact sometimes, when the net is very long we break the net and add buffer.

I dont get how you are improving crosstalk by reducing the drive strength of victim net. Have you ever tried it or its just concept ??

regards.
Shobhit
Its not about reducing the crosstalk noise. Its about saving the victim from aggessor's sudden transitions. Hope you are not clear with my explanation.
What i mentioned is true and is validated in a few text books. Refer to the link mentioned in above posts.
Let me know if you are clear with this.

- - - Updated - - -

Let the crosstalk noise exist. But dont make it affect the victim's logic value. Thats the idea of the above explanation.
 

Re: crosstalk noise reduction

I agree with shobit. The solutions to reduce noise in the link seems counter intuitive. I actually went ahead and tried the approaches and it made things worse. Then if I did the exact opposite of what is suggested in the article (which is what I usually do), my noise slack improved.

Prashant - "Let the crosstalk noise exist. But dont make it affect the victim's logic value. Thats the idea of the above explanation."
This might work in theory and it assumes a lot. One of the major flaws in this theory is that it assumes that the total crosstalk aggression is going to be a constant. In practical chip design, by taking the approaches in the link, the total aggression will NOT remain the same as before, instead you will be increasing the total aggression. What is the use of increasing the aggressive noise and then trying to resist it. This theory works only if your total aggression remains a constant.
 

Re: crosstalk noise reduction

The best solution to avoid crosstalk noise is to use a shield between the aggressor and victim which eliminates the cross coupled capacitance.

Above mentioned techniques focus on saving the victim from aggressor's transitions. Also maintaining a strong '0' at victim's net improves the noise margin of victim's net and thereby resisting the crosstalk noise.
 

Re: crosstalk noise reduction

Btw thanks for posting the link Prashanth. Although I don't agree with the solutions, it was an interesting take on xtalk. It got me curious to double check what I learned so far.
 

Re: crosstalk noise reduction

Btw thanks for posting the link Prashanth. Although I don't agree with the solutions, it was an interesting take on xtalk. It got me curious to double check what I learned so far.

I agree with your explanation too.

Always approach 1 (downsize the victim) enables us to reduce the crosstalk noise to some extent apart from saving the victim.

I agree that in approach 2 (upsizing the victim load) we are just trying to save the victim but not on fixing the crosstalk noise.

So i want to conclude by saying that in approach 2 we are not reducing any crosstalk noise.

Thanks to jayasree.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top