The main thing is that she suggested I make a harness: I plug in connectors from the power supply into one end, and the other end plugs into the board. I attempted to do without it and while everything did fit fine on the board with the 24-pin ATX header and EPS12 header, the position of the two 12v pins on the 24-pin ATX header made it too awkward to utilise. I didn't have much of a choice but to take her advice and engineered the harness, which utilises the 24-pin ATX plug (which is very necessary), the EPS12, and a PCIe power connector for a total of 8 wires (the "middle" 12-volt pin of the PCIe isn't used, since by specification it's "not connected").
Anyway, I arranged it to where 3.3v, 5v, +5VSB, and PS-ON get one pin each (since none of them pull much current in general. 12v gets 8 pins (which I think I mentioned not long ago, and 10 GND pins. I had planned for 9 GND pins, but since the closest connector was 22 pins I'm not using 20 pins), I added an extra ground.
I found a 22-pin header with a flange, and while it is a bit large, it'll prevent anyone from attempting to plug in a 24-pin connector. I"m just hoping that it doesn't accept a 20-pin connector... I'll have to see later. It saved quite a bit of space compared with the 24-pin connector and 8-pin connector.
Since the maximum amperage is 5 (or was it 4?) amps for the power output to the fan/light board, I utilised a more sane trace width, which allowed me to fit the traces fairly well,. The lights I think I did at 3 amps, since the draw probably won't be more than 2.
Arrangement probably took the most time, since i wanted to keep the layout as compact as possible while allowing the traces to be run. I did a rough arrangement of the parts and checked it with the design rule thingy.
I ran the air-wires, routed them, ran more, etc., and was able to successfully fulfil the design rule requirements without much rework.
At some point I forgot about the mounting holes, but I remembered and inserted them.
I think the most annoying thing is the square relay, since it requires 5mm of spacing from the board edge.... as stupid as that is.
Anyway, it looks fairly messy, but still worked out fine.
Before and after ratsnest tool.
Bottom plane is GND, top plane is 12v. Again, it's a rough drawing, but I'm fairly surprised of the turnout. I'll eventually clean it all up so I can make the mounting plate.
It's good that I don't have to buy the "full" version of EAGLE to be able to create a larger board for this part of the project. One of the options I was considering (but didn't want to employ) was to split the main board into two. Since it would've made the project that much more of a headache, I'm glad it didn't go into that direction.
I did the maths for the light pipes and standoffs, finding 19/32in would be the best choice with the 2.0mm thick aluminium alloy plate, because I wouldn't have much thread on the top nut of the switch with 5/8in. I then did the maths for the light pipes and found a large gap between the SMD LED and the light pipe, so I went back to the maths and calculated for 1.6mm thick instead.
With the 1.6mm thick plate, I found that the 5/8in standoffs were much closer to the calculated distance and that it puts the light pipe closer to the LED, though still further than the light pipe's manufacturer's suggested maximum distance of 0.02in.
I tinkered around at some point, seeing that there's dead space on the other side of the board where the LEDs and resistors are, trying to use 3mm and 5mm (T 1 and T 1-3/4 cases respectively) along with light pipes, but the distance from the LED and the light pipe was still further than the SMD option.
I realised just now that there's SMD LEDs with taller casings that I didn't explore, and while I might explore that option, I think what I have is the cheaper option, but whenever I tinker around more with that, I'll see if it puts it closer to the light pipe's specified distance.
I really don't like mixing SAE and metric measurements, but there's not much I can do about it besides utilise maths to convert when necessary.
I also placed the parts down for the fan/light power distribution board and ran some airwires, but not much more than that.
Somewhat sloppy, incomplete "rough draft".
I wasn't ever worried about this board, since I figured that it would easily fit within the constraints of the 100mm by 80mm board size limitation. I'll eventually clean it up and such, but it'll be much easier than the main board, since there isn't a lot of parts with different sizes.
Oh, right, with the main board, I removed some of the power relays since the switches themselves are able to handle the current of the function (the lights and blower fans), which saved me a lot of space. I also had to add some smaller relays for the fuse indication lights, since I didn't look around for the part because I had forgotten that the LEDs would need them for power regardless if I place the relay before or after the fuse.
The fuses will more than likely be difficult to replace, but they should only blow if there's a problem somewhere - they're there to protect the traces and board connectors from melting which could lead to a costly fix (the main board itself will be about 62 USD for three boards).
The difficult part is now over, and though the remainder is tedious, it's still part of that "downhill slope."
Even though the project isn't at a state where I can purchase what's needed and build it, I've still learned quite a bit.
- Do the maths and utilise sane trace widths instead of trying to utilise a trace width that is equivalent to the connector's current rating.
- Custom harnesses may be much more beneficial than direct connections
- Utilising different connector styles may seem more costly, but may be a better investment than utilising the same connector style.
- Draw mounting holes (plated/unplated) before arranging parts.
- Open up options by consulting others for ideas.
No comments:
Post a Comment