I’m trying to copy a switch which has contacts on both sides. I’d like to get it fabricated as a double-sided PCB. I have drawn the contact layout as a DXF and successfully created a footprint containing edge cuts and front and back copper layers, with pads linked to the contact shapes.
The original switch piece uses contacts along the edge to link one side to the other. I have no experience of KiCAD or PCB manufacture so thought this was probably an unusual one-off, but have since discovered that edge plating exists. See the attached photographs for details of the original switch contacts. (I’ve made more than one post as it seems new users can’t have more than one image).
I can use blind vias for most of the contacts as they are opposite each other on the board, but there are several where the edge bypasses the neighbouring contact to link with one offset on the other side.
Can I replicate this in KiCAD and create a Gerber file to get this made, or is it an unusual and therefore expensive addition? Would I be better off using vias for the contacts that need to be linked and are opposite each other, and manually soldering some material to the offset contacts when I receive the board?
You may not get very many replies to your post. I looked into edge plating about a year ago and decided it was not worth all the extra effort/costs. It may be do-able at this point time, but I don’t know how to tell you how to do it.
Edge plating is chemical (the same process to get copper into vias) and will at first plate all holes and edges of the panel.
The normal PCB process is:
drill holes, cut slots (those will become plated),
chemical copper deposition over the entire panel,
drill non plated holes,
cut out boards from panel.
The typical way to do castellations is to put normal THT holes at the edge of your board and then cut them in half when cutting out the boards from the larger panel.
For your plan to work, the board house will have to make small cuts into the board to create the “pads” on the side. Or you could place non-plated holes which will be cut in half.
Because all of the steps are done anyway, there is no extra cost at volume, but your typical shops like pcbway will add an extra charge to cover the cost of communicating with you to get it right because they need to put your small order onto their large multi-customer panel and get the edges right. If you produce a lot and do your own panel, it shouldn’t cost extra.
In any case, I would be more worried about the longevity since with a standard process the copper on the side is only 20µm (plus 5µm nickel, 0.1µm gold for ENIG) and your example looks like stamped metal. There is a process of adding 20µm+ of gold but you are not going to like the prices, it is usually called “hard gold” or “gold edge connector”.
Thank you for the thoughts. Yes, it does appear as though the original has been bent around the plate. I don’t know if it was laser cutting, as the switch piece is from the late 1970s. It was likely a special and I don’t know a way to make it apart from manually cutting the contacts and somehow adhering them to the board. It doesn’t sound like edge-plating is the way to make it happen.
I’ve sent an e-mail to PCBWay as I read they have a good UK support service, so perhaps they’ll tell me how it’s prohibitively expensive!
I’ve looked at castellations and they won’t work as they will go straight through the board to the opposite contact. This would be fine for the majority of the contacts that simply wrap right over the top of the board, but the ones that are offset have to have a contact that in some way travels along the edge.
For another switch of the same decade, I manually etched two pieces of copper-clad board, but that one didn’t have contacts that were electrically connected between each side of the board. Attached is a picture of the same failure mode on this other switch that necessitated the repair. The hole at the end holds a crossbar that prevents the switch spring from pulling the contact piece out of the front panel, so it’s under strain whenever the switch is disengaged. It appears this strain causes failure at this point.
The design with the edge plating is subsequent to this nylon example, longer, and appears to be more robust - interestingly, it’s made from two layers of PCB material, like my repair! They must have thought it a good idea too.
If it was from the '70ies then etching is more likely.
Etching is a regular process to be used on sheet metal. Lots of nameplates for example are deep etched (or cast instead), then filled with enamel powder and baked in an oven. Etching can also be used for very fine details such as 1000 slots in an encoder disk.
It does not look like it can be made with a “Standard” PCB process. This probably means you have to find a PCB manufacturer that does special jobs, or modify the design so most can be done in a standard process and then modify it yourself to get to a finished product.
A possible way is to design a PCB which is edge plated over the whole side, and then cut slots in them with either a dremel, x-acto knife, a saw or similar to separate the pads, and possibly fill the cuts again with epoxy.
My suggestion would be to read carefully manufacturer’s requirements about how your drawings should be.
After some changes that they will probably apply into your uploaded(proper drawed) files, and maybe also some explaining emails with images like the above, you should be able to confirm the result into their 3D viewer.
I recently handled a board that was completely plated around the edges except for the breakaways where it was attached to runners during the SMT placement / wave solder phases (RF shielding).
The edges were nowhere near as smooth as would be expected for switch contacts.
I would be inclined to say that the easiest method to make these would be to have the PC boards made with the edge bare, but with vias connecting the sides as needed. Then for the edge itself, have some thin sheet copper laser-cut then bent over a form to get a series of ‘fingers’ and which would then be lap-soldered in place.
This looks to me to be a better solution. The copper alloy/thickness of the material could be selected for the best balance of wear/conductivity/corrosion; something that could not likely be done with edge PCB electroplating.
Non-ferrous metal sheets such as copper and brass up to around 0.2mm are relatively easy to cut with a normal pair of scissors.
If you’re curious how well this work, then start with a normal soda can (Aluminimum) which are easily available in supermarkets or just lying around in the street.
Every time my stock runs low I start looking around a bit when cycling into town to grab a few lying around. Avoid the “red bull” cans, they stink enormously, and the smell does not go away easily.
If you do not have fixed sliders on the narrow edge then your design with vias would be totally OK.
IMHO, you should place vias at the place wheres free of fixed contacts. Gold plating as a must.
Caution with this. Many laser cutting shops will refuse to cut copper. One vendor that I was in contact with told me that the reason why is their lasers use copper for mirrors (copper is an excellent IR reflector). You would need to find a laser cutting shop that has the correct equipment and processes for cutting copper.
I’m not saying that it is impossible, but the lower cost laser cutting houses might not have the specialized equipment available. Check this link for a white-paper-like explanation of the challenges and techniques to overcome laser cutting copper:
I guess Sprig’s prophecy was innacurate based on the discussion this has caused!
I’ve just had a reply from PCBWay. It was somewhat lacking in English (who uses txt speak when replying to an inquiry?!) but the gist was that they do edge plating and it must be included in the Gerber. So there must be a way to define it.
I’ve looked harder at the switch, and considered the interesting and thoughtful replies in the thread. Yes, it’s likely the original is bent metal rather than plated. Presumably the bending over the top of the switch adds strength.
There are no mechanical contacts with the upper thin surface of the switch - these folds are purely to provide electrical continuity to the other side. However, there are only two that are offset and need special consideration. The rest connect to the contact opposite so could be done with vias, if it’s not too much to place a via on every pad. Not having done a PCB before with one of these services, the whole process of production appears rather magical and crazy.
So that leaves two that could be edge plated (each plating perhaps 5mm long) or linked manually subsequently. My reservation with this is the bulk of the extra material interfering with the sliding of the switch past the contacts.
I’m not too concerned with wear and tear. I can’t imagine a bit of sliding wearing through the contacts in a short time. It’s not like the record switch is in constant use. If it works, the PCB switch version could be replicated easily, and perhaps be made with thicker copper.
I don’t think the cutting out approach will work, simply because of the effort involved in accurately cutting and positioning the conductors. While possible with scissors and time, I am unsure about the adhesive working against the sideways sliding pressure, and about burnishing the edges down so that it slides properly. I’m sure I could do it, but I was hoping for a less labour-intensive method which the PCB houses seemed poised to provide. I have a house to build and about 6" of desk space to do repairs on when I push my keyboard back in my temporary office, and that space is being competed for by about a dozen projects!
However, thank you for all the thoughts. These are really sparking my imagination!
Think about how PCB’s are made.
(Some Youtube vid’s may help, Euro Circuits has some nice video’s)
How double sided PCB’s are made:
Start with FR4, which is fully plated with thin, copper (typically 17.5um) on both sides.
All plated holes (and side plating routing is done.
Inside of the holes are made slightly conductive with some “carbon substance”.
The whole panel with PCB’s is put into a bath, and with electrolysis copper is added on all conductive surfaces. Top & Bottom is thickended to 35um, all (conductive) holes get approx 17.5um copper.
The places where copper must stay is covered with some acid resistant “paint”. (This includes the holes and the side plated area’s.)
The PCB is etched in a bath.
Etch mask is removed, Solder mask added, etc.
(I have omitted a few steps here, youtube explains it better.
The problem is with your intermittent edge plating. The step of making the sides conductive and adding copper with electrolysis are not selective. They work over all exposed area’s. Your PCB will also be made from a big panel with lot’s of PCB’s in it, and your side plating will be made in a narrow routed slot.
This slot is only accessible from the top side, and a router bit that creates the slot is round. That bit can not reach into the corners between your intermittent edge cut parts.
End result is that even manufacturers who do side-plating will probably not make this PCB for you.
This image comes close. It also has milled out gaps after the side plating. Maybe it is good enough for you.
There is something I would like to have cut from sheet copper as soon as I can get a proper CAD drawing made, so when I get ‘a round tuit’, I’ll look at having it cut via water jet instead.
