How to generate the S parameter from DC ?

[jaya000]

s parameter from DC

I hv read about S parameter from DC. The author of the book hv generated S parameter from DC. The original S parameter r from .1Ghz. I want to know how he has done it? The document is attached herewith.

 

[Mazz]

s parameter from DC

The paper says how it is done. It is a simple math extension to 0 Hz done by a linear simulator.

If you have the simulator, simply add your s2p file and simulate it from DC! Accuracy is not guaranteed, of course.

I hope it can help.
Mazz

 

[xxargs]

Re: s parameter from DC

The paper says how it is done. It is a simple math extension to 0 Hz done by a linear simulator.

If you have the simulator, simply add your s2p file and simulate it from DC! Accuracy is not guaranteed, of course.

I hope it can help.
Mazz

But many simulators reject using frequency outside S-parmeters block, so I think using intellegent guessing and interpolate from curves in linear simulator and some assumtion (for example 180 degree phase shift in DC) to make good guess missing values.

if lucky enogh find spice parameters of transistor - you can take up measure in Spice simulator in 50 Ohm bridge, other way using transistors DC-parameters and AC-parameters as Hfe,gm,Cce,Cbe,Cbc etc. to make simplified transistor model in spice and if curve is close to S-parameters curve in 100MHz - 500 MHz range - i thinking simulators value is usable and accurate enough for frequency below 100 MHz to DC

but remember assumtion this way is newer fast and quick solution compare to real measured data, ie. You need spend (much) time to make model, verify if your assumtion is close enough to real life units.... and, remeber manufacturer S-parameter data is a 'typical' value and real unit can have sligthy difference and in critical near unstable coupling design can give unwanted effects as self oscillating or different gain and bandwith not seems in simulator - manuactorys 'improvement' (mostly after shifting silicon plant or process without notice) with no change type number of component can make big problems in production ie. know stable design now begin to oscillate etc. depend of manuafactorys 'improvement' higher band width product etc.


Is a not easy life to hold product going year after year depend of component manufacorys change of process whole time internaly without notice in 'standard component'...


---

question: - from which book take the text above ??

 

[jaya000]

Re: s parameter from DC

XXARGS
Thanks for ur reply!
The pages r from "J.F.White High Frequency Techniques - An Introduction to RF and Microwave" book.
Here i want to discuss more about measuring real s parameters.
I am saying real means like real car performance in the field & in ideal conditions, it differs. Then why not test S parameter in real conditions upto applications of 300Mhz or more.
Real condition testing means make the real sample good quality factory made PCB with actual biasing components & fit input output 50 or 75 ohm connectors as required by VNA. Then u can make different biasings as required and hv S parameters.
Can someone put some light into my thinking. I am actually as newbie to this field. hence looking for guidance from RF Gurus here.

 

[xxargs]

Re: s parameter from DC

XXARGS
Thanks for ur reply!
The pages r from "J.F.White High Frequency Techniques - An Introduction to RF and Microwave" book.
Here i want to discuss more about measuring real s parameters.
I am saying real means like real car performance in the field & in ideal conditions, it differs. Then why not test S parameter in real conditions upto applications of 300Mhz or more.

To make good simulations you want smallest piece you can get from manufactory - ie ideal world - chip only, after this model of bonding, package etc. - every item outside chip represent lumped/transsmisions component and have big influence of trasistors performents outside package ie. extra inductance, capasitance etc. - special in high frequency.

in real life you can only get S-parameter for whole transtor with package in special manuafactured test fixture, advanced wide band range DC-bias network and lot of calibration to deembedd all outside infuence best as possible and only measure transistor itself without so much error.

this work need highly skilled measure engineer and take lot of time - and if high power circurit, scary ie. very easy to burning costly VNA for slightiest small misstake - most RF-power designes make self testfixture and measure itself on samples - experieced designers never trust manuafuctures data and many time also miss lot of importent information from manuafactory (for example reference planes place in measurement of S-parameters) or S-Parameter is a poor quality (mostly case...) or not take up measurement in different currents or tempeperature range.

Real condition testing means make the real sample good quality factory made PCB with actual biasing components & fit input output 50 or 75 ohm connectors as required by VNA. Then u can make different biasings as required and hv S parameters.

Many time can you buy manuafctorys reference fixture (and in old time often describe in data sheet/book so you can make one itself) but needs now close contact with manuafctorys engineers and development process with feedback in cooperation - this support is not easy to get if you work for small company and low volume customer...

many of days manuafactcorys have not longer own skilled measure group to make good public engineering data for customer - now in most case maked from consult or external companys - often in close cooperation to big EDA-companys as agilent/MWO/ansoft etc. and for this companys purpose (ie. you needs pay for model now in this companys development platform)

Can someone put some light into my thinking. I am actually as newbie to this field. hence looking for guidance from RF Gurus here.

Im also newbie - after around eigth years of RF-working in different levels


In most case, get a VNA and spectrum analyzer with tracking generator and practice many many hours in exprimentation on labb togheter with simulators... difference between experienced RF-engineer and newbie is a how many thousand hours you spend in lab and make fixtures, make filters, make couples, make calibrations, make misstake, burning components and some case burning instrument for €4000 in repair cost... books, theory, S-parameters and simulators is only a guide for your work and learning...

 

[jaya000]

Re: s parameter from DC

Nice reply from XXARGS.
Thanks for nice reply.
But i accept ur kindness in telling that u r still newbie. This means i am just new born.
As mentioned by u that S parameter changes by different layout/conditions etc. When u r not going to use the device in ideal conditions than what is the use of ideal S parameters.
The ideal S parameter will be different than in the real conditions under which the device is going to be used. Then why not measure the S parameters under the actual conditions under which it is going to be used.
In car industry, they do all the simulations in ideal conditions, but for real test they test the vehicle in different actual conditions. Then they make the final product.
In RF it is not possible to get real conditions in ideal S parameter testing. The S parameters change under different test conditons. So i think it is better to measure them in actual conditions.
U hv here said that
"manuactorys 'improvement' (mostly after shifting silicon plant or process without notice) with no change type number of component can make big problems in production ie. know stable design now begin to oscillate etc. depend of manuafactorys 'improvement' higher band width product etc."
Due to this only i am more & more convinced that it is better to use actual conditions for testing S parameter. So there will be difference in testing S parameters & designing with them I might be wrong. Hence seeking advice from more expeienced people here.