The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Usefulness of s-parameters on chip
https://designers-guide.org/forum/YaBB.pl?num=1032754377

Message started by Ken Kundert on Sep 22^nd, 2002, 9:12pm

Title: Usefulness of s-parameters on chip
Post by Ken Kundert on Sep 22^nd, 2002, 9:12pm

The previous post (Modeling differential/common-mode input impedance) brings up an issue I’ve been wondering about for a while. Just how useful are s-parameters now that RF circuits have moved on chip. Consider a receiver front end, with an on-chip LNA, mixer, etc. I can see where s11 of the LNA would be useful to know so that one can match its input. But the output of the LNA does not see a controlled impedance. How does one even measure s21 without a well defined load impedance. One can always combine the mixer with the LNA, but that only moves the problem because the output of the mixer often does not see a controlled impedance either. You could place a controlled load impedance at the output of the LNA to facilitate the measurement of its s-parameters, but then what do you do with them?

Are s-parameters starting to outlive their usefulness? Will they be confined to the world of boxes and pipes?

-Ken

Title: Re: Usefulness of s-parameters on chip
Post by RFICDUDE on Sep 29^th, 2002, 9:03pm

Hi Ken,
I am surprised no one replied sooner, but here is my two cents.
S-parameters for RFIC design is useful mainly for circuits that need to interface with external fixed impedance sources and/or loads.

They can also be extremely useful for LNA chip design for finding matching networks that need to achieve specific gain, noise, and linearity specifications.

Other uses are for checking potential instabilities on supplies or other I/O pads. You can put a port DC source on a supply and check the impedance to see if the real part goes negative. This is an indication that there may be potential instability issues.

We rarely need to use s-parameter analysis on chip because most circuits are deliberately impedance mismatched.

Still, s-parameters are very useful for characterizing the impedance of active and passive components. Almost every analog/RF component model is fitted from an array of measured s-parameter and DC I-V data over many different conditions. Therefore, in a sense, high frequency circuit simulation could not be possible with out s-parameters.

RFICDUDE

Title: Re: Usefulness of s-parameters on chip
Post by Ken Kundert on Oct 1^st, 2002, 9:05am

When using s-parameters to design an LNA, would you only use s₁₁, or do you use all four s-parameters? If you need all four, what do you do for port 2 (assuming it is driving an integrated mixer and so does not see 50 Ohms at its output)?

Title: Re: Usefulness of s-parameters on chip
Post by zinmel on Feb 20^th, 2003, 2:26pm

I think if the circuit is highly integrated RFIC chip, then s-parameter won't be that important as conventional microwave circuit design. Therefore, for a LNA to drive mixer on-chip I didn't even look at the S21 because of load termination is only a capacitive load. Actually, I believe the information of voltage gain should be more useful than S11.
Regarding to the issue pointed out by "RFICGUIDE", I think he is right, if the signal in the chip needs to be drawn out, then s-parameter is getting more and more important. Actually, I used to be a RF board designer. I used s-parameter a lot, especially for designing the matching circuit of differential I/O SAW filter. The SAW filter has the frequency dependent termination because it is in the position after mixer and before IF amplifier. Both of them are the signals from and into the chip, therefore, package will make its source and load termination very frequency dependent. Due to isolation between SAW's input and output is very bad in the passband, without the helps of circuit optimization it will be very difficult to design its input and output matching circuit.
But like I just said, for a highly integrated RFIC, signal in the chip is less preferable to be drawn out of chip, although I don't know so far, there is any circuit can replace IF high-Q SAW filter such as the application of GSM, probably Gm-C filter. That is the only puzzel in my head?? probably someone else can answer this question. If there are some counter circuit on-chip to replace the SAW filter or some high-Q devices, then the signal in chip won't be necessary to come out anymore. If that is the case, I believe the s-parameter will only be necessary like the point connecting the antenna to chip or filter to chip. And also if that is the case, then as Ken put at the very beggining, the usefullness of s-parameter will become less and less. ???
-Zinmel

Title: Re: Usefulness of s-parameters on chip
Post by Frank Wiedmann on Feb 24^th, 2003, 2:19am

If you want to eliminate the SAW filter, the best way is probably to choose a different system architecture. A Google search for "low-if receiver" will show you some examples.

I agree with everyone else that s-parameters do not make a lot of sense as long as you stay on chip and do not have matched input and output impedances for your circuits. I prefer to look at voltages and currents in this case.

On the other hand, s-parameters are important at the input and output nodes of the chip in order to achieve proper matching to the system impedance (and possibly to check stability, as mentioned by RFICDUDE).