The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> S-parameters used in circuits
https://designers-guide.org/forum/YaBB.pl?num=1367887535

Message started by aaron_do on May 6^th, 2013, 5:45pm

Title: S-parameters used in circuits
Post by aaron_do on May 6^th, 2013, 5:45pm

Hi all,

I am trying to use s-parameter data in an amplifier design. However, I have encountered a problem. I have attached an image below. The circuit that i'm trying to model with s-parameters is inside the red box. Now suppose I have two different circuits as shown. Both circuits will have the same s-parameters. However, the absolute voltages V1 and V2 wrt ground are different for the two circuits.

I have noticed that in ADS, the circuit is always interpreted as in the second figure, while in Cadence V1 and V2 can be floating (more like the first figure). However, neither interpretation is necessarily correct. So how can I resolve this problem?

thanks,
Aaron

Title: Re: S-parameters used in circuits
Post by RFICDUDE on May 6^th, 2013, 6:25pm

Hi Aaron,

I am a little confused.
Is your question about the characterization simulation (where you are trying to generate an s-parameter model) or is it about the simulation result when you use the s-parameter block?

Title: Re: S-parameters used in circuits
Post by raja.cedt on May 6^th, 2013, 11:00pm

Hi aarron,
i guess you have S-parameters data set, which you are trying to re simulate as a black-box. If so you should use grounded ports means two ports should have 50ohm wrt ground. I didn't understand I have noticed that in ADS, the circuit is always interpreted as in the second figure, while in Cadence V1 and V2 can be floating. According to me both are wrong as you don't have gnd on both sides.

As matter of fact, if you don't have 50ohm wrt gnd then there will some capacitor between node v2 and gnd which alters the results.

Title: Re: S-parameters used in circuits
Post by aaron_do on May 6^th, 2013, 11:00pm

Hi,

its about the simulation result. I don't think it matters how I generate the s-parameter data.

I have generated the s-parameter data, but there doesn't seem to be any way for the schematic simulator to know which figure is correct.

thanks,
Aaron

Title: Re: S-parameters used in circuits
Post by aaron_do on May 6^th, 2013, 11:06pm

Hi Raja.Cedt,

why do you say I "must" have grounded ports. So you are saying all of my s-parameter data must be referenced to the same point (ground)?

In ADS, the s-parameter element only has one reference port, while in cadence, each port has its own reference.

thanks,
Aaron

Title: Re: S-parameters used in circuits
Post by Ken Kundert on May 7^th, 2013, 7:42am

N-ports are specified only in terms of the port voltages and currents. On a two port there are 4 terminals, but only two voltages are considered. Thus, they are innately under specified. Many people implicitly assume that two of the terminals are grounded, but this is an unnecessary restriction. If you give it up and allow your ports to float, then you must accept that the common mode voltage on the port is completely ignored. If that is not acceptable to you, then what you have is not a two port, it is a 3-port or a 4-port, and you have only included 2 of the 3 or 4 ports.

-Ken

Title: Re: S-parameters used in circuits
Post by RFICDUDE on May 7^th, 2013, 6:24pm

Interesting ...

The original question is very general:

The two simulators express the ports on an n-port device differently (Cadence nport has +/- terminals per port and ADS only has one terminal per port plus a global reference), so what is the difference (if any) when each simulator is using the same s-parameter file description?

As you mention, two-ports are general so there is no need to define a global ground to reference the ports to. In the old days, s-parameters were explicitly ground referenced because most circuits and network analyzer measurements were single ended. But now days there are many differential devices where the differential mode and common mode responses are generally different.

So, my question is what is the appropriate way to accurately represent both the common mode and differential mode responses of a multi-port network?

Title: Re: S-parameters used in circuits
Post by aaron_do on May 7^th, 2013, 6:34pm

Hi Ken,

thanks for the reply. You have confirmed what I was thinking.

regards,
Aaron

Title: Re: S-parameters used in circuits
Post by aaron_do on May 7^th, 2013, 7:03pm

Hi RFICDUDE,

I think for differential circuits, a differential port needs to be represented by two ports, one for the positive terminal and one for the negative, and each one referenced to ground.

The question is then, where is the ground reference in your EM model? In the schematic simulation, all reference ports would be tied to the same point, so ideally in your EM model, you also need to make sure all ports are referenced to the same point. Since this doesn't seem to be possible, the next best thing is to have all ports referenced to an ideal ground plane (low resistance and inductance). However, my problem is that if your circuit doesn't have an ideal ground plane (i.e. its not real), then there might be problems. For example, some of the return current might flow through your ideal ground plane instead of the real return path.

So basically I'm still wondering what is the most accurate approach to EM modeling...

OK I realise it might be difficult to follow my thinking, so if its unintelligible, please let me know and I'll try and draw some figures.

thanks,
Aaron

Title: Re: S-parameters used in circuits
Post by RFICDUDE on May 13^th, 2013, 7:37pm

Hi Aaron,

In my (somewhat limited) experience of EM simulating passive structures, I too find that I need to use a single port for each terminal of interest in a simulation (two ports for each differential signal). The, so-to-say, global ground reference is important. In the days when a die was attached to a lead frame or paddle that is grounded, the global ground reference was easily defined as the backside of the die substrate. Now days the backside of the die is not attached to a grounded piece of metal, so we are much more reliant on the substrate tie downs to define the substrate ground. This greatly complicates (I think) the ground reference and the dependency on using the substrate ties to define the ground reference.

I hope others add their observations.