I don't really remember too much about how it was sparked, but I did remember that I had a PCI cover plate thing that has nine holes drilled 1cm apart. I actually have two of these, and I think one of them is being used for the Intel build at the moment, but that's not important for the moment. The plates came from a couple PCI capture cards that was used to record video and sound for a surveillance system (it's what the Dell originally was for before the HDD(s) corrupted to where the restore disks were of no use.
Anyway, I figured out that the very best way to switch between the soundcards was to have an internally-mounted board that has holds the latching relays to have the all the audio jacks as close as possible to each other, and to reduce the lot of external cables and the box. I opted for some pushbutton switches that has an LED in them instead of a toggle switch to make it easier to find, for durability, and for look. I had recalled seeing something that fit the description quite a bit before then when I was poking around for cheaper part alternatives for work.
I found some PCB-mounted audio jacks that fit the profile I needed (since 1cm spacing between the audio jack centres is fairly close), but was offset with the needed plated slot that was indicated in the datasheet. I drew a couple of variations for the part to accommodate with what OSH Park's DRU would "allow" me to do, though it was more of just ignoring the errors and tweaking the DRU a bit for some inconsistency I found between OSH Park's capabilities and their DRU.
Anyway, I drew a board to test OSH Park's capabilities seeing that they can do plated slots, but (like castellations) isn't fully supported.
I dunno what's floating around that's causing it to auto-zoom like this, but it should be good enough.
I didn't change all of the of them to be the same, and it doesn't show the actual cutouts with the holes, but you can see the lines with the holes where the cutouts are supposed to be. I think I stuck the third one in to satisfy the minimum size requirement, but I don't remember. Eventually I ordered the board along with something else, since it was fairly small (it also got automatically upgraded to the swift service thing as well).
I did draw another test board because I was having troubles uploading it to OSH Park... Rather it was that OSH Park's server was having trouble processing them, though it seemed to be fine the second time around... It also might've just been a hiccup in the transmission as well.
I was fairly surprised when I received the boards, since I was fairly edgy with their "we can do it, but we can't guarantee it'll turn out well" sort of attitude (for lack of better word). The slots weren't parallel with the board edge, but with the thickness of the pins and the width of the cutout, it's not going to matter much (the cutouts were parallel to each other, however). The large hole turned out fairly well, but there was the minimum distance missing from the hole plating from the board edge (0.015"). I'm glad the plated slots turned out decently, since I really wanted to avoid using the large hole for trace routing issues and having to utilise a lot of extra solder to fill the remainder of the hole.
I took pictures of the board, but lost it when my phone locked me out of the encryption after the November Android patch. I'll take a couple pictures when I'm at home and whatnot, but I probably won't upload them until the next post for this project, which will be in March when I get the audio jacks.
With the ability to use the plated slots confirmed, I then did the baseline design for the board, which was just lining up the audio jacks properly. I first decided to use the 4-pin Molex power connector to get power to the board, since the stand-alone SATA power isn't readily available (though if it would, it'd probably be more expensive anyway), and drew and tossed it onto the layout just to have it there for later placement.
I then decided how I wanted to connect the buttons to the board, and I had originally thought of using a 6-position connector and having a 6-conductor cable (22 AWG wire) route it to the front to a breakout board that would hold the SMD resistors for the LED and some 3-position connectors for the switches. I eventually decided that it would be much easier to solder a wire to each terminal than to awkwardly tie the positive to one of the switch terminals and the anode of the LED (and also that it would make the resistor have to be on the ground line instead of the positive line... though it doesn't matter).
I think the project sat for a while before I came back to it and realised that it would be much more efficient to run the wire from the switch directly to the board, since it would be less parts, less cost, and less work. Luckily I hadn't done any work with the design of the board, so it's not like I had to fix anything. I also decided that I would use the free circuit of one of the five DPDT latching relays to switch which light would be lit up, which would allow me to use just one resistor and to reduce the power consumption (by a measyly 20mA).
I decided to utilise the relays (and layout of the relays) from the first design for the board of the soundcard switchbox version 3, since it would be awkward to attempt to implement something that uses the single-coil relays that the soundcard switchbox actually uses. I also considered utilising only one pushbutton, but that would require me to utilise another relay (along with the fact that it might be difficult to properly select the side since the pushbutton would constantly switch between the soundcards until released).
I arranged the relays so that they would span the same amount of space as the audio jacks and then took some time to decide how I wanted to arrange the jack order to make routing the traces easier, I eventually opted for a similar arrangement to the soundcard switchbox, where all the greeen jacks are adjacent to each other, and so on. Basically it's where the output of a colour (green, to say) is flanked by the inputs of the same colour (I'll have a screenshot up in a moment).
With the traces done I placed the Molex power connector and the 4-position connectors before getting some measurements to find that I'm hitting the PCI express header housing, but is easily remedied. I can't remember if I had done the traces for the connectors or not, but it's probably not that much of a concern.
Anyway, I have the LED set up so that the LED of the inactive card is lit. For example, if I'm in Linux, the button to switch to the Windows sound would be lit, making it easy to find in the dark. With the silk screen, I marked which side is the motherboard, the pinouts for the connectors, the ordering of the audio jacks, and "Soundcard Switcher 3.0" on the flip-side (which would be visible through the window in the day).
Before and after ratsnest tool.
I really wanted to keep as much of the audio signals on the "top" side of the board (red traces), but didn't have a whole lot of options. Obviously, both sides have ground planes to somewhat shield any EMI, though the audio traces are relatively short anyway that it probably won't pick up anything. I remember having to redo the traces for the two 4-position connectors, but don't remember why - it might've been the movement, but I can't be certain. All the connectors and components are mounted on the top of the board (which becomes the bottom when mounted in the case), and I might do a bit of sleeving or something to make the cables look nice for the little they appear before disappearing into the power supply cover thing of the case.
It was a bit tough to figure out how I wanted to mark the ordering of the audio jacks, but eventually I decided to go with a sort of "type" and colour scheme (so WG is Windows Green, OG is Output Green, and so on). Also for once, I actually used the proper colour for the center and subwoofer channel which is orange and not peach.
I honestly really wished I spent more time thinking about this and coming up with this, but from the time I was developing the third version of the switchbox, I hadn't the crimper that I do now (and that I'd thought would never be so versatile). Such is life though.
The pushbuttons will be mounted in the front on a external 5.25 inch to 3.5 inch adapter that holds the card reader, and should look fairly nice, if the holes are drilled properly.
I think I've exhausted all I have to say about the project thus far, and I'll have to figure out how to nicely drill the holes so it looks fairly professional into the adaptor.
