linc2 pro

Hello everybody,

i have calibrated the VNA 360 and measured diffrent TRL cal-kit, transistor with line. I want to konw how can i extract (de-embedding) the real transistor S-parameters.
The band of frequency that i want to get S-parametres transistor is from 1 - 30Ghz.

Is there any availlable application to use it ?

I measured thru 'T', Reflect 'R', line 'L' and Transistor with tow line on in the left and the other in the right pad of transistor...........

pls help me .......

thx in advance

nassim

You can search the paper of U.L.Rohde " General Noise de-embedding procedure for Packaged two port linear active device".

Go to below link download the file
http://www.synergymwave.com/Articles/default.asp

A New Efficient Method of Designing Low Noise Microwave Oscillators

Read the section 4.

Thanks nccu for your contribution, but i think you have not understood what i'm looking for ..........

I want to extrtract real S-parameters of transistor!

thx again

M

Hi,

With your VNA, you can perform a TRL calibration, and then you will obtain S parameter data at device accesses.

It's very easy.

Measuring the cal kit it means nothing...

Hi jayce,

Thx for your contribution.

i have realized TRL cal-kit and i have measured their S-parameters.

the band is from 1 to 30 GHz or from 6 to 30 GHz since the line of TRL calibration is limited from 20° to 160° so with VNA alone we can not perform your idea.....

Maybe only for narrow band.......

Have you understood what i mean ? In ADS there de-embeed2 component but i do no how can i proceed ............

thx again

N

Yes I know what you mean, you have a large line for small frequencies and a small line for high frequencies.
But in many case you can just use one, I tried , that will be more precise than using deembedding in ADS , because to use ADS for deembedding process you need TRHU measurement of each cal kit launches , but it not easy or impossible to get it with some test fixture.

is it a packaged device ar on wafer measurement?

thx again jayce,

It's a package device.......When i tried to calibrate with therealized cal-kit i did not have a good calibration compared to standar cal-kits given by manufacture of Wilstron VNA, te reason why i tried to calibrate VNA with that standards and measured each cal-kit realized (TRL) for de-embedding.........

I found in litterature that Agilent developped De-embedding using TRL calibration in ADS for 4-port.

I do know how can i proceedddd..

Mouloud

You can use SOLT calibration and get the whole band calibrated. LINC2 RF simulator has very good de-embedding calculator. All what you need to do is to measure your device S-parameters with test board and then measure input part of the test board and output part. You will have three S-parameter files: one for device plus test board, one for the input part of the board and one for the output part of the board. Then insert all three files into LINC2 calculator and get the de-embedded S-parameters file. If you have no access to LINC2 you may send me your files and I will de-embed them for you.

Dear RF-OM,

Many thanks for you.. Un fortunately, i do not have LINC2 RF and did not find it as free software. Also, i do not have sample substrate to realize the cal-kit standard so i have performed only :

TRL calibration
- Open circuit ( 1,5cm) : Reflect (R)
- 02 Lines 50 Ohms (i.e, 2x1,5cm) : Thru (T)
- Line 50 Ohms ( 2x1.5cm + L/4 à la fréquence de travail) = 3cm + 1.85 mm : Line (L)
operating frequency : 24,125GHz,
substrat parameters : - epsilon r = 3.63
- H= 0.254 mm
- T=35 micro m .
Could you perform the calibratin with the software LINC2 ? i yes, i will send to you the measured files......
I looked for internet i have found that ADS also perform calibration but i do know how! i tried to use de-emebed2 component avalaiblein ADS and it works but i'm not sure that gives correct result!

Thx again

Added after 8 minutes:

and of course Transistor with two lines, one (1.5 cm) on the left pad of transistor and ther other on the right pad of transitor with (1.5cm)

_____1.5 cm line ______Transistor_______1.5 cm line _______

thx

Yes, LINC2 is not free, it is almost free (I paid for my PRO copy about S400). I did not get your answer about calibration. Probably you mean de-embedding. I include the picture of LINC2 de-embedding calculator here in order to give you better understanding of what is required and what will be in result. If it is what you have, send me the input file S1, output file S2 and Cascaded file CAS. I will calculate DUT file for you. I think if you measure your files on the same substrate that you will use, the result should be good. 1.5cm line for 30GHz looks to be long. Are you sure that it is necessary to have so long lines? They will multiply reactance and severely affect performance.

Thanks you again RF-OM,

Yes, i have measured cal-kits in the same substrate and i want to make de-embedding..Is S1 and S2 are open, i.e, line without transistor in the right and in the left? In attached file, you will find:

- S1
- S2
- the complete DUT with S1 and S2.

About line 1.5 cm, is used just to make well the standard and to make transistor a way from test fixture.

All files are in mdm extension. the DUT has many biasing so you can take that vds=2V, vgs=-0.52V and Vds=2V, Vgs=-0.56 V

Thank you in advance
N

Added after 41 minutes:

S-para,

sorry i have chaned the extention since i could not sen them as it was so now are in doc but you can chage the extention of course.

Extension is not a problem, I may change it. The problem is that S-parameters are defined for certain characteristic impedance. I believe for your case it must be 50 Ohm. If you measured open lines without termination you got wrong files. Think about your input and output lines as two port devices. You measure the whole set of 4 S-parameters for two port device. There must be S11, S12, S21 and S22 and all must be measured with 50 Ohm terminations. You need to remeasure these files with good 50 Ohm terminations. Good means good for the whole frequency band. The best way is to use short (about 10 - 20 mm) end-launch (say 47 mils semiridgit coax cable) to connect VNA to the end of the line. Very important is to keep exposed central conductor as short as it possible. 1mm of wire is about 1nH which is roughly 6.3 Ohm at 1 GHz. So 0.5mm of exposed central conductor will provide about 3 Ohm per GHz of frequency! This is not easy but possible to do accurate measurement. Do not forget to use Port Extension to move calibration planes to the end of the end-launch after calibration. The best way to do this is to calculate delay through known dielectric constant of the cable dielectric and its length. Do not use electrical delay feature of VNA because it works only for selected S-parameter, not for all 4 of them. Port extension will work fine. Then you may check Port Extension values with Smith chart by looking on short and open points and correct slightly if necessary.

Hello RF-OM,

Exactlly, S-para defined for 50 ohms impedance charactereistic. an Open is a line without terminatio so why we have to teminate it with 50 ohm ? i have s-parameters in data sheet, i you try to get s-parameters of DUT we can verify if we are closer to the manufacture measurement or not. So could you try by extraction just to verify what can we get.!

thanks again

M

Okay, I will. There is small problem. I may download only files below 500kB. You may see that I am over the 400MB limit of downloading. Please send this big file to my pm and I will download two small files.

Added after 8 minutes:

I download these two .doc files, then renamed .doc to .s2p. Something is wrong, files are corrupted. Probably you used some converter to save them, or Word did it. Never use word processor for S-parameter files, only Notepad. Could you please check your original files. If they are good change the extension to .dat . This should go through e-mail well. And please check how many test point in the file. Often simulators have a limit of 1000 points for S-parameter files.

Hello everybody,
I have seen all of you write,but I have a question as follows. S parameters of transistor with DC biasing is only useful isn't it?
If the anwer is yes, the above-mentioned is also gotten with DC biasing isn't it?
This is my question that S21 maybe amplified and how you can extract precisely S parameters of transistor .

Yes, S-parameters file is measured for certain bias conditions and you need to chose one that close or match your needs.

To extract transistor S-parameters de-embedding technique is used. This is the way how all manufacturers measure S-parameters for their products.