50 ohm 30GHz termination in microstrip ring coupler question

[yefj]

Hello,Shown below a ring quadrature coupler i have built.
However one critical problem is that from the diagram bellow i need the isolation port to be matched .
In theory i need to put 50Ohm termination to this port .
However i dont trust is 30GHz to put resistors because they can cause problems when i will solder them.
Is there a trick in transmission line theory that i can use to make a 50Ohm termination for my isolation?
Thanks.

 

[yefj]

Hello, i have the footprint of the device.
i need an s2p , model or RLC model at 35GHz.
using the plot you posted for 0201,how can i know the RLC model of the resistor?
is it a series or parralel model?
Thanks.

 

[volker@muehlhaus]

Don't make things too complicated, use R=50 Ohm.

 

[yefj]

Yes R=50Ohn C=0.0206pF L=1.7e-5nH
are they in series or in parralel?
Thanks.

 

[D.A.(Tony)Stewart]

Are R,L, C all in parallel or series? Yes both.

L=1.7e-5nH is rather low considering a common thin trace is ~ 1 nH/mm and this is a ~ 0.5 mm 3D body flip-chip.

The package drawing is not to scale, but the dimensions indicate a 201 package are a 5 mil cube between terminals (20-10 mi= 10 mil), which is also the same as the substrate thickness.

The terminals will overlap the PCB pads. Deviations will occur due to the pad length and area (conductor overlap), the bump diameter, and the width of the coplanar-waveguide transmission-line launch. Simulations are required to ensure the method of attachment in the assy. has no singularity on the PCB in the case< 35 GHz.

The flip-chip probably uses NiCr sputtered alumina & Au plate, which is likely what your board requires. It will be just a tad more sensitive than your 10 kHz current buffer.

I don't know how one would calibrate your antenna feeds with shunt to load the resonant line lengths.

Due to surface roughness, current distribution will be non-unform at 35 GHz.

It may need tapered lines with a special calibration LRM or TRL. (Line-reflect-match or thru-reflect-line) for the test jig.
I have only done production fixtures up to 6 GHz for Iridium Satphone LNA/Tx chips.

 

[volker@muehlhaus]

Yes R=50Ohn C=0.0206pF L=1.7e-5nH
are they in series or in parralel?
Thanks.

You really want to include parasitics of 0.017 pH? Get serious!

 

[D.A.(Tony)Stewart]

@volker@muehlhaus Is that L=1.7e-5nH a real or accurate value? That is an unrealistic Zo at 35 GHz and does not match the graph

I would expect more from a 0.25 mm length.
Unless it's a wrap-around, then C is higher, even if it was air.

 

[volker@muehlhaus]

I would expect more from a 0.25 mm length.

If the 0201 width is on the order of feedline with, you can "hide" the physical device length as a part of the feed line. I think that is why Vishay mounted it flip chip.

My advice for yefi is to go ahead and built a PCB prototype of that termination (feedline + 0201 resistor + radial stub) and measure it. The resistor with tiny parasitics as shown here is certainly not an issue.

 

[dick_freebird]

In my observation the "done thing" is to represent what lies
within the physical extent of the device in the model (or
subcircuit), and leave modeling the interconnect outside those
extents to the responsible engineer. Then the "guts" get a
marker layer to prevent double-counting interconnect, or you
jump to a second set of post-layout models that don't show
(say) the inductance of contact to PCell-boundary.

Where / whether to assign features like the solder lands, I guess
you'd have to determine whether they are already embodied or
not. But I wouldn't impose any "rough distance to X" R, L, C
within the model proper.

 

[yefj]

Hello, when i try pure 5Ohm i get good result but when i define L=1.73e-14 C=2.06e-14.
when i increase the capacitance of the resistor i get a very good results, i dont know if the simualation is good.
is there some S2P file i could use for a simulation?
Thanks.