Momentum Differential ports DC voltages

[rkarnaty]

I have noticed this recently, whenever I EM a structure with differential ports somewhere in the layout and do DC/ HB simulations. The DC voltages shown at the differential port seem incorrect. Am I looking at it incorrectly? I have attached two pictures. In the first image, i have a resistor of 5 ohms across differential port from momentum . In the second image I have a schematic equivalent of the momentum model. i get different absolute node voltages across the resistor(The voltage difference across the resistor is same in both cases though). i am trying to understand if I am using differential ports incorrectly

 

[BigBoss]

Why differential Ports ?? There is no reason to use them here.
Differential Ports are used in very special applications, not here.

 

[volker@muehlhaus]

The DC voltages shown at the differential port seem incorrect.

This is what you get with differential ports if there is no connection to global circuit ground in schematic. In that case, the Momentum port is floating and circuit simulation assumes symmetric drive.

If you want to ground one side of the Momentum port, to pull that to 0V, you can do that in schematic by placing the circuit ground symbol.

Differential Ports are used in very special applications, not here.

Yes and no. If the attached element is a pure series element, one differential port in Momentum can be used here. One port terminal at one pad, the other port terminal at the other pad, that is perfectly valid in that case.

If the attached element has both series and shunt elements (e.g. parasitics to ground) then two ports with reference node at the ground plane should be used.

 

[rkarnaty]

If you want to ground one side of the Momentum port, to pull that to 0V, you can do that in schematic by placing the circuit ground symbol.

I am planning to use differential ports(SMD or auto) for series elements, but neither of the two pins in the differential port are at 0V. So is it better to skip the differential port to avoid the floating voltages? Interestingly the voltages outside the momentum block in the schematic do not seem to be affected by this, making me wonder if its fine overall

Yes and no. If the attached element is a pure series element, one differential port in Momentum can be used here. One port terminal at one pad, the other port terminal at the other pad, that is perfectly valid in that case.

I have gone through your application notes regarding SMD differential ports. Can a SMD cap be considered a series element even when it is used as a shunt component where the second port is connected to a global ground as shown below?

 

[volker@muehlhaus]

So is it better to skip the differential port to avoid the floating voltages?

No, it all depends on how the model is wired in schematic. I use differential ports very often.

Can a SMD cap be considered a series element even when it is used as a shunt component where the second port is connected to a global ground as shown below?

Yes, this also is a single impedance where the input current into one pad is the same as the output current at the other pad. That can be modelled accurately by a single port between the pads.

But looking at your screenshot, I notice something unexpected. Your Momentum ground is connected to schematic global ground, and still the ground voltage is not zero. It seems that your DC sources are not wired properly, and the ground potential of your source is floating, with no DC path to schematic global ground.

If all your Momentum ports are differential, and there is no connection of Momentum "ground" polygons to schematic global ground anywhere, results are not technically wrong but your reference potential is shifted. Zero volt is whatever you choose as the reference potential, by connecting it to schematic global ground. So if there is no such ground path yet, you can connect one of the source grounds to schematic ground.

If that doesn't work, can you upload your ADS workspace or show the entire schematic?

 

[rkarnaty]

Here is a full schematic where i showing the discrepancy. I verified that the global ground is wired properly to the DC sources

The transmission lines have a apparent voltage drop. even though the right transmission line is grounded, the differential port is showing -500 mV

 

[volker@muehlhaus]

I see ... indeed the series port in the middle only knows the voltage drop and current across the port, if it is defined as one differential port between the two terminals. The EM results don't provide information about node voltage relative to other ports, just the impedance across the port.

As I wrote above, that is perfectly valid as long as you connect a series element there. But if you want to connect something to other nodes, like global ground, or probe relative to other nodes, that information is missing and you get this "floating" result.

In this case, you need to use ports with reference all at the same ground. Sorry for the confusion!