Does it usually happen that the simulated result and the measurement are not the same

[torbai]

I recently designed a S band LNA. But the measured result is far more different from the simulated one. Luckily, I can change the amplifier into a L band amplifier by replacing some of the lumped inductors and capacitors(all from murata). But after I do the same process in simulation, I still cannot obtain the result from measurement.

So I have to ask, does it usually happen that the simulated result and the measurement are not the same?

PS: I'm using ADS, the transistor is Avago ATF541M4, and I'm using Rogers RO4350 board.

 

[curiouscase]

Simulation and real world results are always in interest. You can see in many research that they compare the simulated result with fabricated ones.

As simulation result are based on theoretical, which as far tries to govern all the physical variables with define equations(relations). More accurate result can be defined with more advance programs which tries to relate the real variables like parasitic. Having said this, there are number of things which makes your simulated result vary from real ..calibration, real element being used etc.

 

[volker@muehlhaus]

So I have to ask, does it usually happen that the simulated result and the measurement are not the same?

It really depends on the user's simulation methodology and experience with the tool. It is possible to get good agreement, but it is not trivial at all.

Typical error sources:
- Real components have parasitics. Circuit simulation with ideal R,L,C isn't approriate. Using manufacturer S2P libraries is more accurate than using ideal elements, and using substrate scalable models (Modelithics) is more accurate than using manufacturer S2P libraries.
- EM simulation will include coupling between elements, but that coupling is missing if you use schematic level microstrip models. Also, EM simulation is useful where layout models don't exist, or their valid range is exceeded.
- From many years in EM support, my experience is that most users make mistakes in their EM models: not enough attention payed to mesh density, port grounding concepts not understood.

Besides these basics, there are many other things that can go wrong. Some designers simulate Rogers substrate with the correct substrate data (great) but then order solder resist in manufacturing that wasn't included in simulation.

 

[torbai]

curiouscase and volker, thanks for your replies.

I have used Momentum for EM simulation, and murata design kit is used.

Maybe it is because my lumped element are too close to each other. (My work almost only consists of lumped elements and devices in order to make it easier for adjustment or debugging)

And how to simulate solder resist in ADS?

 

[BigBoss]

curiouscase and volker, thanks for your replies.

I have used Momentum for EM simulation, and murata design kit is used.

Maybe it is because my lumped element are too close to each other. (My work almost only consists of lumped elements and devices in order to make it easier for adjustment or debugging)

And how to simulate solder resist in ADS?

There may be some reasons for discrepancies between simulations and results..
-Poor models
-Wrong configured EM/Circuit simulations
-Unconsidered side effects ( closures,connectors,unwanted-especially-inductive couplings,vias etc. )
-Measurement methodology/instrument errors..

 

[volker@muehlhaus]

I have used Momentum for EM simulation, and murata design kit is used.
Maybe it is because my lumped element are too close to each other.

Possible error sources:
1) reference planes between SMD data and Momentum ports for the SMD do not agree.
2) SMD simulated with differential ports between the pads, which effectively floats any capacitance to ground. https://muehlhaus.com/support/ads-application-notes/momentum-port-global-ground-or-differential

There was an application note by Modelithics on Momentum port configuration where they made that mistake (2). The appnote was updated after I showed the Modelithics guys that this can lead to serious simulation error.

And how to simulate solder resist in ADS?

You can add it as an additional dieelectric. However, that is only relevant if you have microstrip coupler etc where the capacitance across a narrow gap is changed by the extra dielectric.