Hi Steve,

 

It's my understanding that Nickel plating is very lossy to high freq RF... you might try not plating it at all.. leave it bare copper. Use OSP to keep it clean until assembly. High freq RF does not seem to be affected by copper oxide on the surface of the trace.  Also the smaller the trace is and the closer it is to the substrate, the more signal is affected by deviations from the nominal physical configuration of the microstrip. . A small change in the line width will be a large percentage change in the overall impedance of the trace. This particular 4 mil line with a 2 mil dielectric configuration would see more changes and inconsistencies in impedance matching along the circuit due to changes in process variations... just because of the physical geometry and the difficulty in maintaining uniformity and consistency in the width of the trace and its percentage change with etching and cross section of the dielectric under the trace and its distance from the ground plane. A larger width trace with thicker dielectric would be less sensitive to changes in geometries because the percentage overall change would be less.

It also occurs to me that the stripline model might be more successful in this sort of design... you might want to simulate a model like that and see what you come up with, based on the application.

 

Wish I could help you more... but I think you need to get an RF Microwave Engineer into this one....

 

 

 

-----Original Message-----
From: Steve kelly [mailto:[log in to unmask]]
Sent: Friday, January 24, 2003 11:07 AM
To: [log in to unmask]
Subject: [TN] FW: HI - SPEED NOISE PROBLEM

 

 

 

-----Original Message-----
From: Steve Kelly [mailto:[log in to unmask]]
Sent: Friday, January 24, 2003 12:09 PM
To: IPC ([log in to unmask])
Subject: HI - SPEED NOISE PROBLEM

 

Dear Technetters:

 

I have a question about noise levels at high speed on a flex circuit. The circuit is:

1) surface microstrip design  (.004 tracks on one side - x- hatched ground plane on other)

2) impedance is approx. 50 ohms

3) construction is:   .0005 kapton

                             .0005 adhesive

                             .0007 RA copper )

                             .002 kapton        ) DuPONT AP

                             .0007 RA copper )

                             .0005 adhesive

                             .0005 kapton

4) part is pad plated/selective plated

5) finish is immersion gold over electroless nickel

6) circuit is less than .5 inches long

 

When we measure the S parameters of the 50 ohm trace, the bandwidth starts dropping at 6 GHZ , peaks at 13 GHZ and then drops -  I have a pdf. file which I can send off-line or post it to somewhere?

 

Any thoughts?

 

Thanks

Steve Kelly

 

PFC Flexible Circuits Limited