Electronics > Beginners
Is oxide on surface of RF conductor problem?
<< < (4/4)
TheUnnamedNewbie:


--- Quote from: DC1MC on July 26, 2018, 06:29:54 am ---There is no resistivity on hi freqs, just impedance :), if you have your metal so thin that the normal surface oxide will matter, than is too thin. For hi freqs, usually the bigest problem is not the DC resistance but change of the impedance and q factor of the circuits, not to mention the resonance frequency. this why there a attempts to protect the conduits.


--- End quote ---

What?

At high frequencies resistance still very much matters, as this resistance is what causes the loss. Surface oxide can increase the loss of the traveling wave for a number of reasons - A less perfect surface finish of the conductive layer, which increases losses (this is why rolled copper is less lossy than plated copper), a conductive 'alloy' of oxide and non-oxidized material that carries a lot of current due to skin-effect, again increasing loss. I think for many applications at 'low' (low being a few GHz) the oxide layer acting as dielectric will have negligible impact on impedance since it is so tiny compared to the wavelength. What does impact it is it's effect on the resistance component of the conductors and that can increase the impedance. (Remember that a basic transmission line model starts with series R and L, with parallel G and C - in most cases we ignore R and G, but if R becomes significant, it will start impacting the impedance).


--- Quote from: DC1MC on July 26, 2018, 06:29:54 am ---Of course as you've mentioned the thicker the metal, the less influence has the oxidation, but for many applications this solution is not possible (strip antennas and friends). Better put a 4 microns layer of gold on it and be done  8).


--- End quote ---

I'm not sure this is true either - at a certain point, most of your energy will be at the surface anyways due to skin effect. Making things much thicker past this point doesn't really help, and so the impact of oxidation will be almost the same.




--- Quote from: DC1MC on July 26, 2018, 11:24:51 am ---Well, there IS a reason why the strip line and PCB antennas from 2,4GHz up are covered with gold, mainly for the fact that the gold will conduct better, and stopping the oxidation will not hurt either.
And sadly I'm far removed from a microwave wizard  :-DD.

 Cheers,
 DC1MC

--- End quote ---

This is also not true.

The reason starts out with soldermask. I copy this from a stackexchange answer I wrote a while back:

--- Quote ---Soldermask is applied as a liquid. As such, its thickness may not be as well controlled and predicable as the thickness of the substrate and conductor layers. In addition, it may have an unpredictable profile - how does it "flow" in between the traces? All of this means that you cannot accurately model the impact of the solder mask on your line, and cannot predict the impedance of the trace.

This is even important on any distributed element filter or microwave component such as a directional coupler, resonator, power combiner, etc. In these cases, a very small shift in the ϵeff of the system will potentially shift the center frequency out of the band of interest.

With high-performance RF substrates we can get very accurate models, provided we know very precisely the etch profile of the process. The unpredictable nature of solder mask ruins this.
--- End quote ---

So, we don't want solder mask on top of the transmission lines. But copper oxidizes a lot. As a result we want to add a layer to stop this from happening. On normal boards there are a number of ways of doing this - I myself am not an expert. Most of them I think are called "HASL" (hot air solder leveling). I believe this involves some tin alloy being plated onto the board. But the problem is that the thickness of the layer is unpredictable - not a problem for normal PCB work, but unacceptable for the precise tolerances we have on our transmission lines. So instead, we tend to use much thinner platings on those PCBs, usually Hard gold (electroplated gold over electroless nickel), ENIG (Electroless Nickel, Immersion Gold) or ENIPIG/ENEPIG (electroless nickel, immersion/electroless Palladium, immersion gold) - the difference between these has to do with wirebond capabilities. These platings are very, very thin - a few micrometers at best, and tend to have much better surface finish.

However, all of these finished has higher loss than pure copper, as can be seen in this paper: https://ece.uwaterloo.ca/~oramahi/IEEE-TADVP-Surface-Feb2008.pdf or this magazine: http://iconnect007.uberflip.com/i/586473-pcbd-oct2015/53?m4=. The argument that they use these finishes for lower loss is a myth spread by people who assume that gold being a better conductor than copper must mean putting it on top of copper gives us lower losses - in most cases the layer of gold is tiny (less than a micrometer thick) and the nickel needed to support it is much thicker (a few micrometers), and nickel is a much worse conductor. In addition, surface finish is hurt. The reason they use gold is because gold doesn't form an oxide layer, unlike most other plating, which form a thin, passivating oxide (=oxide layer that is not penetrable by oxygen, and as a result protects the underlying material from oxidizing further).



To put some stuff into context, I work with very high frequencies (tens to hundreds of GHz), so I don't know how some of this stuff translates to more 'common' frequencies in the <6 GHz applications.







DC1MC:
Hi TUN, this is really fascinating stuff, thanks for such a nice organized answer with an overview of the current technologies and research, just two extra bits:

- when I telling the resistance/impedance thing, it was just to point that above some freqs there is no more pure resistance.

- you seem to be an actual microwave expert, so just a personal curiosity, what about those pated ceramic antennas, I seem to find them lately in phones and 5GHz capable routers, are they directly plated (gold over ceramic) or they use the same multimetal technology (Cu + Ni + Au) ?

 Cheers,
 DC1MC
fonograph:
My guess is the ceramic antenas are either due to higher dielectric constant of ceramic so it radiates more easily or due to thermals becose it have higher thermal conductivity.
TheUnnamedNewbie:

--- Quote ---- you seem to be an actual microwave expert, so just a personal curiosity, what about those pated ceramic antennas, I seem to find them lately in phones and 5GHz capable routers, are they directly plated (gold over ceramic) or they use the same multimetal technology (Cu + Ni + Au) ?

--- End quote ---

I'm not really an expert on all things microwave (I wouldn't call me much of an expert to begin with really), and antenna design for those bands is not something I have experience with. I think they are also based on copper, but I could be wrong. I know it is possible to do them with just gold, but that stuff quickly gets really, and I mean really expensive (some guys at the research lab I work at have done ceramic substrate PCBs, and their order of 10 pieces, cost more than my grandmother's car). Keep in mind though these are PCBs that have 10 um holes and stuff, and the manufacturing is more similar to IC manufacturing than the process of making your 2 dollar prototype board in China.

The reason they use those ceramics in antenna modules is likely because the high dielectric constant makes an antenna shorter for a certain frequency and/or the low loss tangent of the dielectric makes them efficient. They might also have very predictable dielectric behaviour that does not change much with respect to frequency and processing.

The thing about those antennas (I believe, again, no actual experience here) is that they allow people to get very high performance antennas with some form of directivity. Antenna design is an art with quite a bit of handwaving, more than anything else in microwave engineering, and it takes a lot of expertise and expensive software to do it. Using module antennas gives people the ability to get a very predictable antenna pattern, without having to go to quite omni-directional antennas such as half-wavelength dipoles (which in addition to not being as directive must generally be mounted outside of the enclosure).
Navigation
Message Index
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod