I am finding discrepancies in the ADS Momentum results with Modelithics capacitor models and S- Parameters.
, I have a copper structure built on RO4350.
Simulated this in ADS 2102 Momentum and brought the em model to schematics for co-simulation.
Have four configurations in the schematics (See the attached)
1. Using Modelithics ATC 600 S 10 pF cap connected to the copper and the ground plane which is connected through Via. Terminal 1 is used and S11 shows the result.
2. Used ATC S-parameter in Horizontal configuration for 600S 10 pF. The S parameter file is connected to the ground plane through Via. Terminal 2 is connected and S22 show the result.
3. Using Modelithics ATC 600 S 10 pF cap connected to the copper and the other end to the ground . Terminal 3 is used and S33 shows the result.
4. Using S-Parameter of ATC 600 S 10 pF cap connected to the copper and the other end to the ground . Terminal 4 is used and S44 shows the result.
Results from 1 and 2 are very different. 3 and 4 matches very well(Please see the attached). Not sure which is the correct answer here.
The Modelithics model with one terminal connected to the copper and other terminal to ground plane (Connected to real ground through Via) shows a type resonance as evident from S11 plot. But when this models one terminal is connected to ground rather than the ground plane Via combination, the resonance is not there.
Appreciate your help.
, I have a copper structure built on RO4350.
Simulated this in ADS 2102 Momentum and brought the em model to schematics for co-simulation.
Have four configurations in the schematics (See the attached)
1. Using Modelithics ATC 600 S 10 pF cap connected to the copper and the ground plane which is connected through Via. Terminal 1 is used and S11 shows the result.
2. Used ATC S-parameter in Horizontal configuration for 600S 10 pF. The S parameter file is connected to the ground plane through Via. Terminal 2 is connected and S22 show the result.
3. Using Modelithics ATC 600 S 10 pF cap connected to the copper and the other end to the ground . Terminal 3 is used and S33 shows the result.
4. Using S-Parameter of ATC 600 S 10 pF cap connected to the copper and the other end to the ground . Terminal 4 is used and S44 shows the result.
Results from 1 and 2 are very different. 3 and 4 matches very well(Please see the attached). Not sure which is the correct answer here.
The Modelithics model with one terminal connected to the copper and other terminal to ground plane (Connected to real ground through Via) shows a type resonance as evident from S11 plot. But when this models one terminal is connected to ground rather than the ground plane Via combination, the resonance is not there.
Appreciate your help.
Attachments
a) The difference between 1/2 and 3/4 is that 1/2 model the physical ground path from the cap to the PCB backside, whereas 3/4 use an ideal ground and exclude that physical return path. So models 1/2 are more realistic.
b) We need to look closer at your Momentum port definitions. All Momentum ports in this model need global ground reference (PCB backside). Don't use differential ports for the SMD, as Modelithics recently suggested in their Momentum appnote. That appnote is incomplete and misleading. Use regular ports with global ground reference.
c) From what I understand, models 1 and 2 are identical, except for the electrical model for the cap. Correct?
We need to look at the Modelithics model details. You have included the pads in the EM data, so pads must be switched OFF in the Modelithics data. That shunt path capacitance from pads is only a small effect here, but I wanted to mention it anyway.
d) If the large discrepancy between the Modelithics model and the s2p file is still seen, my next step would be to look at these files (just the 2-port without Momentum model). Extract capacitance, extract SRF, extract shunt path elements to see what is different.
I have attached the the ADS workspace where I did this analysis. Please have a look.
The difference between 1 and 2 is that 1 uses the Modelithics model for ATC 600s 10pF cap and 2 uses the S parameter data for the ATC 600s 10pF cap from ATC site. They should be close, but I see big difference in my result as posted.
As you noted, the difference between 1 and 3, 1 uses the real return path through the vias to ground on bottom side of the PWB. So, here the number of vias used to connect the ground plan at the top of the PCB and the bottom ground plane matters, right.?
For Modelithics model, I am using Sim_mode =2 and pad_mode =1. This as per Modelithics notes, will exclude the pads as pads are included in the Momentum.
My substrate definition is attached. I think this has the infinite ground plane and all of the ports are using this.
Looking forward to hear your valuable comments.
Attachments
> I have attached the the ADS workspace where I did this analysis. Please have a look.
> My substrate definition is attached. I think this has the infinite ground plane and all of the ports are using this.
Correct ... I checked the Momentum port setup, and this looks ok.
I could not check anything else because I don't have the Modelithics library.
The ATC S-parameter file (600S100.S2P) look good, and using that, I get the same results that you have.
> The difference between 1 and 2 is that 1 uses the Modelithics model for ATC 600s 10pF cap and 2 uses the S parameter data for the ATC 600s 10pF cap from ATC site. They should be close, but I see big difference in my result as posted.
Understood and agreed.
> As you noted, the difference between 1 and 3, 1 uses the real return path through the vias to ground on bottom side of the PWB. So, here the number of vias used to connect the ground plan at the top of the PCB and the bottom ground plane matters, right.?
Yes, correct.
> For Modelithics model, I am using Sim_mode =2 and pad_mode =1. This as per Modelithics notes, will exclude the pads as pads are included in the Momentum.
Sounds correct.
Not sure where the problem is, but the setup of the EM block looks fine. The difference must be in the cap-model itself. I suggest to contact Modelithics support.
I am trying to find out more on this with Modelithics help.
One more question on this.
As you can see, I am using the internal point port for connecting the capacitors and ADS 2012 doesn't do the calibration. I am wondering which port to use... the EDGE port with TML calibration or the internal point port I used?
Looking forward for your comments.
> As you can see, I am using the internal point port for connecting the capacitors and ADS 2012 doesn't do the calibration. I am wondering which port to use... the EDGE port with TML calibration or the internal point port I used?
I think the internal port (uncalibrated) is ok here. Edge port would require some tricks here, to make sure that the port is at the correct edge with the desired width. So lets keep it simple.
In the meantime, I did some experiment to check how sensitive simulation is to series inductance in the cap. That is VERY sensitive! By adding some extra length of line (MLIN length=0.3mm width=1mm) to the AVX data, I get almost the same results as the Modelithics data. So very small length of line = small inductance matters. Which means that we need to care about S2P data port reference location and these minor details, and possibly Momentum port is more critical than I expected.
> Which means that we need to care about S2P data port reference location and these minor details, and possibly Momentum port is more critical than I expected.
I looked into that. Comparing the internal port and the edge port (with no ref shift), we can get similar results if we add/remove some length.
This makes sense:
- According to the document that comes with the AVX data, their S2P are measured with reference plane at the *outer* edge of the pad.
- If we place the Momentum port on the edge of the pad, that is the *inner* edge.
- To move the reference plane to correct location, we need add reference shift in the EM setup, with shift = pad size in direction of the current flow.
Attachments
I am little confused about the "PAD". ..
Attached is the capacitor dimensions for a 0603 capacitor.The S -Parameter data is at the points marked blue.
In the momentum layout, the gap between the layout pads on which the capacitor will be soldered is 0.6mm. The capacitor length is 1.6mm. So , I have to use an edge port with reference shift of 0.5mm, is this right?
Attachments
The difference in results is caused by a small extra length (approx 0.3mm) built into the Modelithics data, compared to the AVX data. That's small, but it shows in results. Which means that we have to be very careful with modelling here!
> {quote:title=Leyo wrote:}{quote}
> In the momentum layout, the gap between the layout pads on which the capacitor will be soldered is 0.6mm. The capacitor length is 1.6mm. So , I have to use an edge port with reference shift of 0.5mm, is this right?
That logic would make sense. However, I found some inconsistent scaling of length when trying that ref shift into the metal area. Tthe reason might be that current isn't just flowing straight away from the pad.
I have used another approach, using "area pins". Area pin is more accurate (more realistic) than the single point for the internal port, which has undefined small size.
Draw a rectangle where the SMD connects to the pad (where current flows into the SMD) and then assign it as an area pin: "Edit > Edge/Area Pin". This places the port size and ref plane at the correct location in the metal plane (near the outer edge of the SMD -> Is this what Modelithics needs? -> Check this!).
Attachments
Thanks a lot for your help.
I have some questions about this thread. Could you please reach me at ali.kheirdoost@gmail.com ?
Thanks
Edited by: Alii on Jul 7, 2014 2:25 AM