USE DISCOUNT CODEEXPERT30TO SAVE $30 USD
On the conflict between signal and return thickness vs PDN thickness.
miner_tom , 04-05-2022, 09:17 AM
Robert,
I took one of your courses on schematic capture and layout and loved it! I heartily recommend them.
I see a lot of information, especially from experts like Eric Bogatin, concerning the height above copper for getting good "returns" for high speed signals. This enhances signal integrity. There is also a good reason for keeping the height above copper, thin, with respect to keeping crosstalk down. I believe that Robert even did a video with Eric claiming that the closer the return layer is to the signal layer, the less chance of electric field radiation has on creating crosstalk.
However, when it comes to the Power Distribution Network design, it is believed that the thinner the dielectric layer is between PWR and GND, the higher the capacitance and therefore the lower the AC impedance is in the PDN, which is important if the impedance target of the PDN is to be obtained.
So, it seems to me that there is a conflict between keeping the top, for example, layer, close to its return layer, for signal integrity purposes, vs keeping that return layer (if it is GND) close to its associated PWR layer for achieving PDN targets.
Can you comment.
Regards
Tom
I took one of your courses on schematic capture and layout and loved it! I heartily recommend them.
I see a lot of information, especially from experts like Eric Bogatin, concerning the height above copper for getting good "returns" for high speed signals. This enhances signal integrity. There is also a good reason for keeping the height above copper, thin, with respect to keeping crosstalk down. I believe that Robert even did a video with Eric claiming that the closer the return layer is to the signal layer, the less chance of electric field radiation has on creating crosstalk.
However, when it comes to the Power Distribution Network design, it is believed that the thinner the dielectric layer is between PWR and GND, the higher the capacitance and therefore the lower the AC impedance is in the PDN, which is important if the impedance target of the PDN is to be obtained.
So, it seems to me that there is a conflict between keeping the top, for example, layer, close to its return layer, for signal integrity purposes, vs keeping that return layer (if it is GND) close to its associated PWR layer for achieving PDN targets.
Can you comment.
Regards
Tom
qdrives , 04-05-2022, 10:36 AM
Now I am not a expert on PDN, but if I look at this quote from the article below "Engineering the peak impedances below the target impedance may make the PDN very expensive. This would require many MLCC capacitors, and thin dielectric between the power and ground planes, for example." I would say that also for PDN it is beneficial to have thin dielectrics between power and Gnd.
miner_tom , 04-05-2022, 10:56 AM
qdrives,
I don't disagree that there are other considerations besides "target impedance" to be considered. I will read the article that you referred to . And, we are saying the same thing when you say "that also for PDN it is beneficial to have thin dielectrics between power and Gnd".
t does not necessarily take many MLCC caps to achieve proper target impedance as shown by some of the PDN analyzers or optimizers from, say Ansoft. What I am saying is that it can't be one way or the other in terms of minimizing crosstalk and maximizing signal integrity vs achieving target impedance in the PDN. There has to be a compromise and I don't see a lot of literature addressing the compromise. It all seems to be one way or the other.
Tom
I don't disagree that there are other considerations besides "target impedance" to be considered. I will read the article that you referred to . And, we are saying the same thing when you say "that also for PDN it is beneficial to have thin dielectrics between power and Gnd".
t does not necessarily take many MLCC caps to achieve proper target impedance as shown by some of the PDN analyzers or optimizers from, say Ansoft. What I am saying is that it can't be one way or the other in terms of minimizing crosstalk and maximizing signal integrity vs achieving target impedance in the PDN. There has to be a compromise and I don't see a lot of literature addressing the compromise. It all seems to be one way or the other.
Tom
robertferanec , 04-05-2022, 11:29 PM
thank you @miner_tom - this could be maybe a good topic to discuss with some experts. I am putting this on my todo list of future videos.
PS: Usually I do not have many options how to place powers in the stackup so I have not really thought about that much.
PS: Usually I do not have many options how to place powers in the stackup so I have not really thought about that much.
qdrives , 04-06-2022, 07:41 AM
@miner_tom I do not fully understand what you think the problem is with thin dielectrics to reach you target impedance in the PDN. Is it that you need a higher impedance?
Perhaps one of the bigger problems with a low plane impedance is that the anti-resonance between the planes and MLCC would be to much.
The other question is: how thin are you or do you want it to be?
Perhaps one of the bigger problems with a low plane impedance is that the anti-resonance between the planes and MLCC would be to much.
The other question is: how thin are you or do you want it to be?
miner_tom , 04-06-2022, 08:43 AM
qdrives,
I believe that there is a language issue between us. What I am saying is that one can tune for the best PDN (anti resonance and added MLCC included) or tune for the smallest dielectric possible between the signal layers and their associated returns. Where to put the compromise, is the question.
Robert implied that this would be a good topic to discuss with some experts. I agree and would love to see that discussion.
Tom
I believe that there is a language issue between us. What I am saying is that one can tune for the best PDN (anti resonance and added MLCC included) or tune for the smallest dielectric possible between the signal layers and their associated returns. Where to put the compromise, is the question.
Robert implied that this would be a good topic to discuss with some experts. I agree and would love to see that discussion.
Tom
Use our interactive Discord forum to reply or ask new questions.