Any thoughts on attaching wire leads to very fine pitch IC's such as .5mm TSSOP's or even some of the leadless packages?
Yes, the right way to do prototypes is to make a board, or buy a breakout board with the right footprint.
But what about the wrong way to do it?
Wire wrap wire is just plain too large. Single strands of bare wire pulled out of standed cables can approach working, though soldering is tricky and they aren't insulated so have to be routed carefully.
Thought about using extremely fine heat-strippable magnet wire... tried unwinding a 32 ohm dollar-store headset but the wire kept breaking before I got much unwound.
Apparently the metallurgy of an actual wire bonding system is incompatible with tin-lead prepared leads.... (hmm, so how does it work inside the package - are those ends untinned? could one chemically strip the plating somehow?)
Honestly, believe me, I've tried it. Make a prototype board or buy an adapter, if you value your time and temper at all. By the time you've got the thing hung together you could have had a couple of generations of prototypes.
It helps if the chip is glued upside down and alternate leads are bent up and down. I've heard of people doing it successfully. Someone has managed to get a Xilinx Virtex (BGA) chip working by soldering wires to the balls with the chip upside down.
I know this technik as "dead buck prototyping" described in a nice to read appnote from linear. I use the thin isolated wires from the primary side of a old small transformator. Easy to remove the isolation with temp > 400 °C. Therefor I use 2 different solder irons.
A few times I have needed to read out TSSOP mask-ROMs with unique pinouts. I've epoxied the chip down to a flat insulating surface and bent alternate pins up. Once configured that way it is fairly easy to solder wire-wrap wires to the pins.
As far as leadless packages go, it's best to get a prototype board but while I was waiting for my first run of PCBs for a recent project, I did glue down a MEMS accelerometer (ADXL322JCP) and solder wires to the pads. Wouldn't want to do this with a BGA though.
If you have a nice stereo microscope or very good eyesight, and a Metcal or similar soldering iron, then it is quite feasible to use the leadless packages such as LFCSP upside down with enamelled 'magnet wire' and 0402 components for decoupling etc. I usually start with a piece of blank copper-clad fibreglass as a groundplane and put the chip upside down on that. Then if there is an exposed paddle in the middle of the package, I make several links from the paddle over each corner of the IC to the ground plane. After that I solder 0402 decoupling caps tombstone-style to the ground plane around the perimeter of the IC whereever there is a supply pin. Then I wire it all up using enamelled wire which I unwind from an old
48 Volt relay. I have wired up an ADF4001 like this and it worked great. I've used various other RF parts like this too.
Of course if you can afford a multilayer PCB then that is better but ordinary single layer and double sided PCBs might not be better than the dead bug method for RF performance, since the via inductance of a 1.6mm thick board would probably be a nanohenry or more.
Have access to the scope at work, and the metcal setup... though I notice our stock tips are a bit big for the finest pitches, so I might have to pick up one of my own to use for "research" projects. Any suggestions which specific tip would be best? One of the .016" variety? Straight or bent?
A relay! Didn't think of that as a source of wire. I'd tried dollar store headphones, and was about to try a cellphone charger from the same source/price.
Reminds me a somewhat anxious coworker that was programming big ACTEL antifuse/fine pitch TQFP FPGAs in the $1K ballpark too.
In a attempt to do it well he used that little suction cup tool to grasp and handle them and put one after another into the programmator: you sure don't want to bend one pin. Take the tray, pick the bugger, put the tray down then insert it in the prog. ... Take the tray, pick the bugger, then suddenly the suction cup let the FPGA go. Made him jump... Oh shit all the tray went flying...
Dissasembled the $1 charger, pried up a core lamination and twisted it around my needlenose to pull it, knocked out the rest, and I have a bunch of fine enamled wire.
Tried soldering to it. I think this will be possible but it's not trivial. Piece knotted around an end or wire wrap wire (simulated TSSOP lead) and simply soldered did not make contact.
Wire does not like to pierce the solder on the end of the iron, but if I get it in the puddle, and add some fresh solder (presumably for the rosing core flux) it will then tin if I hold it in there for a while. It's interesting pulling the wire through the puddle without tinning it
- this stuff does not tin as easily as the wire a former employer used for speaker inductor crossovers, which took only a quick dip in a miniature solder pot.
But I think it will work, and may actually be small enough to tie to the leads, which should help with the accidentally ondoing one connection while working on another problem. Of course that means it's probably too inductive for RF...
I may also tying bare wire (either this stuff chemically stripped, or something else) in situations w`here the very short routing and the circuit topology can keep things seperate, then glue it down (thinking of a TSSOP to a .100 header to an FPGA development board, where the pin order happens to be regular but arbitrary could be fixed in software anyway)
Oh, about the charger... the rectifier PCB has spots for an electrolytic cap, transistor, and resistor, unpopulated next to the four diodes - looks like it might be a simple current source topology. I had actually looked at the output waveform compared to the nokia original and decided it was no worse... but now I realize I should have looked at it under load. Will do so before using the non-sacrificed twin on the phone again.
I use STTC-022 for the actual soldering to the IC pins and STTC-125 for tinning the ends of the enamelled wire - I think it's a bit hotter which helps. The STTC-125 is also good for anything which involves soldering to the ground plane.
The advantage of the relay is that they don't seem to dunk these in any varnish after they've done the winding, so they are easy to dismantle. Also the wire is much thinner than anything that you can buy easily as enamelled wire. Beware the fumes from soldering the enamelled wire, I believe they are quite bad for you and you shouldn't breathe them.
For stuff which doesn't need to be insulated (e.g. short wiring) I use silver-plated strands out of some flexible teflon-insulated wire (RS365-688 in the UK). I have also used strands from the braid of teflon coax in the past - just strip and cut off one inch from the end of a piece of teflon coax and unravel the braid. I find that wire-wrap wire is much too thick for soldering to the IC itself, because when I bend the wire, it applies too much stress to the solder joints.
I use STTC-025 for almost everything. heavy jobs get the STTC-138 (happily solders onto 1.6mm Cu sheet), and occasionally I use the STTC-145, but with a bit of flux the -025 works fine for 0.5mm leadpitch. And I have a 40x binocular microscope too :)
bending legs of smt ICs can be a very bad idea. especially if the resultant fault can destroy a 5kW block of power electronics.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.