About the influence of a ground plane on the differential pair impedance
mulfycrowh , 07-17-2023, 12:49 PM
Hi everyone,
I am looking for articles about this topic.
Thanks for help.
I am looking for articles about this topic.
Thanks for help.
qdrives , 07-17-2023, 05:18 PM
Something like this: https://www.signalintegrityjournal.c...o-return-plane
mulfycrowh , 07-18-2023, 03:59 PM
@qdrivesI don’t think that it is what I was waiting for.
I didn’t explain very well.
I meant the gnd plane being on the same layer than the differential pair. Please refer to screenshot.
I didn’t explain very well.
I meant the gnd plane being on the same layer than the differential pair. Please refer to screenshot.
mulfycrowh , 07-19-2023, 07:05 AM
I didn't find any article about the topic.
Now the idea could be to set up a new rule for polygon clearance around differential pairs in order to enlarge it, something between 1 and 2mm instead 0.2mm.
What do you think?
Now the idea could be to set up a new rule for polygon clearance around differential pairs in order to enlarge it, something between 1 and 2mm instead 0.2mm.
What do you think?
mulfycrowh , 07-19-2023, 10:47 AM
And the result with a clearance of 1mm is as shown on the screenshot.
qdrives , 07-19-2023, 02:53 PM
Seeing I got the first part wrong, let me ask what it is you are thinking of.
1) Do you have a Gnd return (plane/polygon) below/above the tracks?
2) I do not think you would design this on a single layer board, what is the distance from this signal layer to the next?
3) Are you asking because of the other discussion on co-planar waveguides? The one where I mentioned that the 'Gnd' on the same layer still had an influence > 2mm according to most calculators.
4) Are you thinking here as to have the two signals couple into each other and not a return plane? Thus making them more true differential pairs.
If it is point 3, then I found it very interesting and frightening at the same time. However, I think most calculators are wrong as otherwise you could never calculate impedance correctly in a 'normal' situation.
1) Do you have a Gnd return (plane/polygon) below/above the tracks?
2) I do not think you would design this on a single layer board, what is the distance from this signal layer to the next?
3) Are you asking because of the other discussion on co-planar waveguides? The one where I mentioned that the 'Gnd' on the same layer still had an influence > 2mm according to most calculators.
4) Are you thinking here as to have the two signals couple into each other and not a return plane? Thus making them more true differential pairs.
If it is point 3, then I found it very interesting and frightening at the same time. However, I think most calculators are wrong as otherwise you could never calculate impedance correctly in a 'normal' situation.
mulfycrowh , 07-19-2023, 03:20 PM
@qdrives 1/ Yes, we have
2/ This is a 16 layers stackup
3/ No
So, for you, the best option is to remove all gnd polygon pours on all signal layers?
Is there any calculator for differential pairs guarded by gnd polygon pour?
2/ This is a 16 layers stackup
3/ No
So, for you, the best option is to remove all gnd polygon pours on all signal layers?
Is there any calculator for differential pairs guarded by gnd polygon pour?
qdrives , 07-19-2023, 04:22 PM
So, for you, the best option is to remove all gnd polygon pours on all signal layers?
https://www.broadcom.com/info/wireless/appcad
https://www.cadence.com/en_US/home/tools/system-analysis/rf-microwave-design/awr-tx-line.html
mulfycrowh , 07-20-2023, 12:18 AM
@qdrivesThank you so much for all this information.
I checked everything with different calculators, there is a slight difference of impedance between with or without coplanar waveguide for a same width and gap.
The tolerance about the impedance is generally 10% and 20% for LAN.
The calculation shows we are always within this range if we consider "differential" in AD instead "differential coplanar".
I checked everything with different calculators, there is a slight difference of impedance between with or without coplanar waveguide for a same width and gap.
The tolerance about the impedance is generally 10% and 20% for LAN.
The calculation shows we are always within this range if we consider "differential" in AD instead "differential coplanar".
Use our interactive Discord forum to reply or ask new questions.