Hi

I'm measuring a MEMS Switch on CPW line with a slot of  47/90/47 using Karl Suss Probe pitch 200 um., TRL calibration and SOLT calibration but in both cases the Insertion Loss is characterized by oscillation.
Otherwise the width of the Ground Planes are 793 um  

Someone have any suggestion?

Hi pancho_hideboo

90 um is the width of the central conductor and 47 um is the distance between the central conductor and the ground plane (board to board)

For the SOLT calibration, of course, i used a commercial SOLT kit calibration of Karl SUSS.

I report also the width of the ground plane ("Otherwise the width of the Ground Planes are 793 um ") because i think the oscillations are due to the finite ground dimension of the line respect the pitch of the probe so RF signal can be reflected by the external boarder of the Ground plane.

What do you thinck about it?

Ok in the picture you can see the dimensions of the CPW.

My question is if the oscillations in the measurement are due to the finite ground plane configuration of the CPW. I mean, in the case of the commercial standards used for the SOLT, the THRU has, usually, absorbers on the edges, to minimize the edge reflections and to simulate the ideal case of a CPW with infinite grounds, which is not my case. Moreover, I guess that the oscillation is enhanced around the 35 GHz frequency because the  pitch I am using is wider than the CPW geometry and I am not far from the edges, with possible enhancement of the higher frequencies reflections.

Which is your opinion ?

Attached S21 is insertion loss of MEMS Switch ?

Yes

You seemed to use commercial calibration fixture for SOLT. But what calibration fixture did you use for TRL ? Did you make it by yourself ?

Yes i make it and tihis are the dimensions of the standards

TRL (47/90/47) εeff ~ 5.6

1. THRUE  Lenght=500 um
Delay Time=3.94 psec
 
2. Line1  3-16 GHz  Phisical Lenght=3385um+500um  DelayTime=30.64psec

( Lenght=3385um ---> 28.94° at 3 GHz --- 154.4° at 16 GHz )  
   
3. Line2  10-40 GHz    Phisical Lenght=1068 um+500um    
DelayTime=12.37psec

( Lenght=1068um ---> 30.44° at 10 GHz --- 121.73° at 40 GHz )
 
And like reflect standard i use a short standard    

For SOLT and TRL calibration fixtures you used, sizes are same as "ScreenHunter_01_Mar__10_10_04.gif" ?  

For the TRL calibration the size is the same.
For SOLT calibration i use the KIT Karl Suss CS 3 (150 to 250 pitch)

Is it true short ? If not, how to estimate reflection coefficients ?

Yes is true short

giorgilloz wrote on Today at 4:57am:
For SOLT calibration i use the KIT Karl Suss CS 3 (150 to 250 pitch) What do you mean by 150 to 250 pitch ?
You mean you used different probe from TRL calibration ?

No i use the same probe for TRL calibration and SOLT calibration. 150 to 250 means that kit is valid for probe with pitch from 150 um to 250 um

Anyway sizes of CPW in DUT are same as TRL fixtures not SOLT fixtures. Is this correct ?

Yes it's correct

Using SOLT calibration, show me following two S21.
(1) S21 of THRU pattern of SOLT
(2) S21 of THRU pattern of TRL

I haven't this measure but after the several calibration i measured the thrue and i checked the overlap of S21 and S12 and of the S11 and S22 and i checked if the medium value of the S21 and S12 was 0 dB and if the S11 and S22 are under -20dB and all was Ok!
like in the case attached.

In actual CPW, ground plane width have to be finite never infinite.

Sorry the following might not be correct. You have to compare width of ground plane with effective wave length.

And this finite width of ground plane is less effective rather for high frequency as far as primary transmission mode.

If i understand the width of the Ground plane at  28 GHz (resonace) is a Lamda/4 (wavelenght 4000 um) so this can be the problem?