skimask wrote:So...with the 30ga wire...You're just stripping off 1/8" (or less) and laying the end of the wire into the solder of the joint of the IC sockets or whatever parts you're hooking up?
I'm I'm thinking right, once you get the pieces/parts soldered in, get some wire, solder one end down, bend the wire around a screwdriver or something to make nice round bends (vs sharp bends), run the wires in 'bundles' to where they need to go and finish them off.
I'm sure there's many techniques and not all techniques will work in all situations. In general, I tend to solder the two components to be connected in place. I'll then take a length of the wire (e.g. Oki KSW30B-1000):
strip a short length and solder it to one end by flowing the joint and literally sticking the wire into it. I'll route the wire such that I avoid overlaying other joints to avoid interfering with future joints/interconnects, keeping an eye on separations for insulation/safety and even crosstalk avoidance (rare). I'll cut the length as needed to the other component, strip its end and solder it there. Here's a project I'm working on right now, about 1/4 (if that) finished at this point. The wiring you see is between a PSoC and a graphical LCD and I'm doing testing and code development right now:
(NOTE: Look for the 0603 capacitors at the top-right, nestled between pads and see how well they fit...)
I prefer "bundling" for cleanliness and, for me, ease of debugging. I hate seeing rats-nest work but that's a personal thing. Cross-talk will likely never be an issue for most of this sort of thing because you're not likely going to be prototyping high-speed circuits using this method. At "normal" microcontroller speeds crosstalk will never be a problem. If you've got high-currents and/or high-speeds you're probably going to have to do an actual PWB with impedance matching and proper space/trace widths.
What's the easiest way to strip off that tiny bit of insulation? Just melt it off?
I use a tool similar to this:
for stripping that wire. Many manual wirewrap tools have built-in stripping devices. Any decent electronics store will have these things.
Is that method, for lack of a better word, reliable? I'm sure thousands of people are doing it that way. Just seems to me that a chunk of 30ga solid wire would break off fairly quickly. Does that sort of construction work well with high speed stuff, or just well enough?
Keeping in mind it's a prototype, not a finished, hardened product, reliability can be just fine. If lots of vibration or abuse is in the prototype's future, it'll break. You might be able to pot it in conformal coat to harden it but it won't even be as reliable as a proper PWB. Still, there are lots of very old projects that were built in the "old days" with wire wrap still working just fine.
Re high speed: It depends on your definition of "high speed." I've never bothered to try prototyping 400MHz SDRAM -- I'd just do a proper board for that -- but never in my years have I had issue with run of the mill micros and electronics. I recently built a quick proto of a PIC32 connected to a 128x128 OLED and using DMA to send data to the OLED. Signal pulses were on the order of <80nS with quick sub-microsecond edges and the proto seemed to work fine. I would simply use such a proto to validate a concept or firmware, not with an eye to selling or expecting too much out of it.
Reason I ask is I've got a project where I'm going to be 'networking' a dozen or so PIC18F26K20's at 64Mhz (16Mhz external), connected to WiFi, GPS, Camera, WiiNunchuck, LCDs, F-RAMs, SD cards, OBD2, touch-pads, etc. Not every signal will be required to run at that speed, in fact most of it will be inter-PIC comm's will be done using SPI or I2C, but the edges of those signals will obviously be up there in speed. I suppose I could just slow everything down until it works reliably.
A lot depends on the serial clock rate and how solid the board power is overall. If your clock rates are up in ranges where track length on a PWB would matter, forget it
This sounds pretty ambitious to build all as one prototype. Is this going to be all on one board or will it be separate boards? You might think about protyping segments of the design as proof of concept and then using something like Sparkfun's PCB connections and just get a board(s) made.
And I suppose this method wouldn't be any faster or slower than using the 3-hole boards that I have been using. Might even be easier. I'm thinking a person could make a short 'bus' by weaving a length of wire along a length of pads to interconnect them, for instance, ground at one end, power at the other end of the board. Ok...now I'm just babbling to myself...
"Busses" are kind of a pain but can be done by connecting more than one 30AWG wire to a node. This way you end up with two or perhaps three connections at one node max, which is really the upper limit of practicality as far as I'm concerned. You could run a traditional data or address bus this way.