Pass current to "burn" stray vias?

[f95toli]

A while back I designed a fairly expensive 3-layer PCB (gold plated RF laminate, not cheap).
Unfortunately, we have now discovered that I've accidentally put a stiching via, i.e. a vias for connecting top and internal ground planes in the centre of an important RF line which is now permanently shorted. There are 5 PCBs that we would ideally like to repair.

The via diameter is small (0.127mm) and in the centre of of 0.65mm line meaning removing it using a drill will be challenging..

However, I spoke to one of our engineers and he mentioned that he had heard of people "burning away" vias by simply passing a high current through them. The ampacity of this vias should only be ~1A so the amount of current needed might not be very large (a few A at a ~5V or so)

I was wondering if someone has actually tried this? My worry is that the pcb will simply de-laminate as it heats up.

The alternative would be to find a specialist company that can use laser ablation to remove the via, this might still be cheaper and faster than new ordering a new PCB

 

[anorlunda]

I think the word is fusible link.

 

[Dullard]

You might be able to do what you describe (I've done it). Unlike drilling, you will have no control of how clean, how well, or how permanently separated the remaining conductive material might be. If it's for prototyping, you don't really have much to lose. If it has to be right (and you can't get new boards), it is sometimes possible to 'excavate' to the problem, fix it, and 'rebuild' your way back out (with epoxy). 3-layer is a good candidate for this. If it sounds like I've screwed up PCBs before, it's because I have. Good Luck.

 

[f95toli]

Thanks. Drilling would probably be better, but would(?) require the use of a 0.1-0.15mm drillbit which I suspect would be very difficult to use without specialised equipment (CNC or a PCB router) which I don't have access to.
If the lines were wider I would certainly go for drilling.

 

[berkeman]

I've accidentally put a stiching via, i.e. a vias for connecting top and internal ground planes in the centre of an important RF line which is now permanently shorted

Even if you are able to unshort it and not leave other damage, will the $Z_0$ discontinuity not be an issue?

find a specialist company that can use laser ablation to remove the via

I'd be inclined to pursue this option instead of mechanical drilling. I've blown shorts before on PCBs with moderate-current power supplies, but that was generally on first articles of regular PCBs (not super-expensive like your new ones), and generally because I couldn't figure out where the short was, and needed the heating to figure out what I'd done wrong in the layout.

 

[hutchphd]

I too have quite successfully drilled and plugged boards with epoxy. It depends on how much space you have to play with and the wavelength of the rf. Hopefully you use a "large" drill and can rebuild the trace as required.
Of course I was fixing someone else's mistake.

 

[f95toli]

Even if you are able to unshort it and not leave other damage, will the $Z_0$ discontinuity not be an issue?

Possibly, although one good thing about coplanar waveguide is that the metal near the gap is more important than the middle of the centre conductor. It would not be perfect but might work. Ideally the via/short should be removed from opposite side (drilling/ablating) which -in theory- should leave the centre conductor mostly intact. There is nothing on the opposite side that I need to worry about.

I'd be inclined to pursue this option instead of mechanical drilling. I've blown shorts before on PCBs with moderate-current power supplies, but that was generally on first articles of regular PCBs (not super-expensive like your new ones), and generally because I couldn't figure out where the short was, and needed the heating to figure out what I'd done wrong in the layout.

Yes, you are probably right. The PCBs are worth a about £400 each which is not a huge amount of money, but since there are 5 of them it adds up.
One issue is that I need to solve this in the next few weeks and currently everything takes a long time because of COVID19 (many companies are operating at a very reduced capacity or not at all) so I was trying to come up with an "easy" solution that only requires the equipment I have in the lab.

 

[Joshy]

I agree with the others: I think the drilling is the better option. Trying to pass high current sounds like high risk. I haven't seen mentions of possibly adding another stub or if there's a matching network to change those components instead... not sure if it's a narrow band design and acknowledge that double stubs might have forbidden regions.

 

[Tom.G]

Another vote against trying to burn it out.

A safer way than either drilling or burning is use a hobby knife (in the States, X-Acto is the major one) with a #11 blade, that's the triangular shaped blade with an acute angle at the tip.

Clamp down the board and use a stand-alone magnifying glass or magnifying eyeglasses. Working carefully, you can cut off the corner/shoulder of the via where it connects to the planes. Then insert the blade tip into the via hole and twist it to get a little more Copper out.

Have a good stock of blades, they are extremely sharp but brittle, you will both dull and break several of them.

Let us know how it all works out!

Cheers,
Tom

p.s. At least you got an "easy" one in that it's a 3-layer board. (It'a real head-scratcher on 8-layer boards.)

 

[Joshy]

I was thinking more about this today. I've never seen what happens to vias when the current density is high. If it "burns" does it vanish or are there stubs left behind? I'm curious if anyone has pictures or can show if they've come across it because I could not immediately find it online. Unintended stubs obviously wouldn't work very well for RF.

Edit:

I moved into spoilers my earlier concern. I was curious and able to investigate it further and I was wrong about its impact (the hole). I didn't look into anything with really really high frequency like > 10 GHz.

Spoiler

Note I comment before the spoiler that this concern has been mitigated.

Something else I'm not sure if it's worth considering... would be a fun simulation and maybe just overly thinking it from the microwave stuff... is to see what the hole will do to the impedance. I'm imagining that for a moment it'll look like a T-junction because of the hole it'll briefly separate into two transmission line that acts as a splitter followed by a combiner. The T-junction is a classic textbook problem just before Wilkinson to show that it's inferior and mismatched (Pozar has it in Chapter 7); so: Maybe even though the "via burning" thing sounds like high risk... putting a hole in the transmission line might not be great too.

You mention the size of the hole it sounds like it might be small enough length to get away with, but when someone is using more expensive material I'm thinking they might be working at higher frequency. It's hard to get away with things at higher frequency.

 

[dlgoff]

A safer way than either drilling or burning is use a hobby knife (in the States, X-Acto is the major one) with a #11 blade, that's the triangular shaped blade with an acute angle at the tip.

agreed,

they are extremely sharp but brittle,

@Tom.G Safety glasses are a must.

 

[tech99]

When we had whiskers shorting a Ni-Cad battery the method was to blow them away using a splat from a large capacitor. This would avoid general heating of the PCB and as the energy is known, it could be built up in stages for each shot.

 

[Tom.G]

Just after my Internet went down (infrastructure problem extended by borderline competence), I realized the simplest approach is have the board house that made the boards drill out the vias. After all, they do have the equipment and know EXACTLY where they are!

And I wholeheartedly concur with @dlgoff; if you're doing it yourself...

Safety glasses are a must.

And a last note: If you tackle it, ordinary reading glasses around 1.75 diopter strength make decent magnifying spectacles. At least here in the States they are readily available without a doctors prescription.

Cheers,
Tom

 

[eq1]

You might find the following link helpful.

https://www.ultracad.com/articles/viacurrents.pdf

Someone built a handy JavaScript calculator based off it.

http://circuitcalculator.com/wordpress/2006/03/12/pcb-via-calculator/

The comments in the calculator link are pretty useful too.

 

[f95toli]

We had a meeting on Friday where we decided to use the PCBs as are they are for now; the two lines that are shorted won't be needed for a few months (22 out of 24 lines are working) and by then things will hopefully be back to normal (or whatever the "new normal" will be) which will make it easier to deal with. There is also always the possibility that we will discover other issues while using them so we might as well wait.

Anyway, thanks for all your help and suggestions .