ENIG frequency response 1

[zappedagain]

I vaguely recall that the ENIG (Electroless Nickel/Immersion Gold) finish on a PCB starts to have a lot of attenuation at high frequencies (from the skin loss in nickel or the thin gold layer?). Can anyone point me to documentation about that effect, or tell me what frequency range it affects? I'm supporting a 10 Gbps PCB design and unfortunately our experienced RF engineers are long gone.

Thanks,

Z

 

[MacGyverS2000]

I Googled "RF attenuation ENIG" and found some useful entries, though this one probably has some immediately useful data points to consider...

https://ece.uwaterloo.ca/~oramahi/IEEE-TADVP-Surface-Feb2008.pdf

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Comcokid]

Makes sense. Skin effect is an issue many engineers overlook and PCB designers don't know of. Even surface roughness on both the top and bottom of a PCB trace can create trouble for high frequency/high performance RF circuits which is why some RF PCBs are specified to use only rolled copper rather than electro deposited copper, and plating-up the PCB to a specific thickness for these is a no no.

Reminds me of a story (from an article?) of a troublesome problem a engineer once diagnosed. Low cost coax for cable TV has a center conductor of copper plated steel to save cost. The engineer found that some lots of coax spools had high attenuation that varied with frequency. The issue was that the copper plating was too thin over the steel, and how the skin effect over frequency responded to the thin copper.

 

[zappedagain]

Thanks Dan. That covers the frequency range I'm interested in. I believe the issue at higher frequencies is that this loss continues to increase.

It is interesting, that paper from 2008 shows about 1.5 dB/in at 5 GHz for ENIG on FR-4. The other Google finds from Taconic in 2005 and Rogers in 2009 shows 0.15-0.3 dB/in at 5 GHz for ENIG on higher performance materials (Teflon, Ceramics). YMMV!

Z

 

[MacGyverS2000]

The loss appears pretty linear, just a different slope for each finish. For microstrip, the increase in loss is probably negligible for the vast majority of designers. For differential pairs, though, it's a 4-5x increase in loss. Over even a 6" board, that's huge!.

My one experience with OSP is mixed, and my last experience with silver was a complete bust (heavy oxidation before the bag was even opened). I don't believe I've designed for above 5Ghz, and most is 3Ghz and below, so I can afford a bit o' loss and will stick with ENIG whenever possible.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[zappedagain]

This is timely. PCB Design magazine just published an article on surface finishes:

http://iconnect007.uberflip.com/i/586473-pcbd-oct2015/52

Z