Connecting two pins on a 32 pin Atmega328

[dan3460]

Working on a prototype that uses a atmega328 and I'm using the 32pin version (LQFP-32). This is the second run as I made changes including the chip, the cost for 5 boards was about $30 including the freight. I received the boards and installed the MCU with the crystal and the programing connector to make sure that I that least I can talk to the MCU. In version 1 I used a DIP version of the chip and I made the grave mistake of relaying on the schematics and footprint from KICad and I missed two pins that needed to be connected to Vcc, pin 6 and pin 18 (AVCC).
Last night I tried to connect with a thin wire the two pins after several hours of trying I gave up, everything that could go wrong went wrong. The wire would get stuck on the tip of the soldering iron, a bridge will form between some pins and went cleaning the bridge will pull the wire, etc etc etc.
Is there any way to this?
I will spun a new board eventually, there are things that I'm changing anyway but I wanted to test the different parts of the board before committing to a new version. What do you think?

[metrologist]

so you have pads and no traces from the chip on those pins? If you do maybe you can pick up another node.

Or I go under a desktop microscope now adays for all my soldering work. My most prestigious accomplishment was swapping a 148 pin FPGA and it worked.

[Ian.M]

0.8mm pitch patch-wiring rework is a PITA but not impossible.  No coffee (or other caffeine) for 24H,  fine magnet wire with solder-through enamel and the chip end pre-tinned,  good flux, and a blob of bluetack to hold the wire in the right place and it should be doable.  As you haven't connected them anywhere sensible on the board you can lift the pins to get them clear of the adjacent ones if you have to - then you could probably get away with Kynar wirewrap wire.

However if you are getting on a bit, you may have no choice but to find someone with younger eyes and steadier hands.

[NivagSwerdna]

Working on a prototype that uses a atmega328 and I'm using the 32pin version (LQFP-32). This is the second run as I made changes including the chip, the cost for 5 boards was about $30 including the freight. I received the boards and installed the MCU with the crystal and the programing connector to make sure that I that least I can talk to the MCU. In version 1 I used a DIP version of the chip and I made the grave mistake of relaying on the schematics and footprint from KICad and I missed two pins that needed to be connected to Vcc, pin 6 and pin 18 (AVCC).
Last night I tried to connect with a thin wire the two pins after several hours of trying I gave up, everything that could go wrong went wrong. The wire would get stuck on the tip of the soldering iron, a bridge will form between some pins and went cleaning the bridge will pull the wire, etc etc etc.
Is there any way to this?
I will spun a new board eventually, there are things that I'm changing anyway but I wanted to test the different parts of the board before committing to a new version. What do you think?

Sounds like a normal day to me.   

Bridged pins is not unusual.  Apply flux. Solder wick off excess and draw iron across and away and the bridge should go.

Attaching wires to pins should be OK..  just use thin wirewrap wire and apply flux first and line the end of the wire up and along pin.

On my recent v0.00 boards I have defects... keep attaching outputs to pins that are input only. 

And... if you are aging like me... you need serious magnification.  I love my widefield microscope... makes things 1000x easier.

[dan3460]

Thanks for the answers and indeed old fart here. I bought a cheap usb microscope and is surprisingly good, I have been using it for all kind of repairs but this is the first time that I'm doing this fine work.(fine as in small, I tend to butcher things) 

metrologist, yes I did not run a connection to the pad, the two pads that I need to connect are isolated right now.

[Ian.M]

USB microscopes can be rather 'laggy' which can make fine work harder than under an optical micriscope or a video microscope with a CRT or fast responding flat panel screen.

[Nusa]

For prototype purposes, you can probably skip the pin 6 repair if pin 4 is properly connected to VCC. The other 328p package types only have a single VCC pin, so I'd bet you can get away with it for now. Fix it for the final product, of course.

[NivagSwerdna]

If you don't want to spend the $$$ on a microscope just yet then... previously I used a magnifying head band visor (<$20 on EBAY) which will really help but the depth of field is quite shallow and you need to get very up close and personal.

I found some soldering jobs complely impossible without magnification but with the visor do-able and with a microscope a joy.

Another trick is to wear multiple pairs of glasses at the same time. 

[metrologist]

My mag visor has additional swing down magnifiers for even more magnification and shallower DOF 

[Siwastaja]

0.8mm pitch patch-wiring rework is a PITA but not impossible.

Utterly trivial once you get the hang of it.

With 0.5mm pitch, it's kinda PITA, but well doable. (see the attachment)

No coffee (or other caffeine) for 24H

My rule of thumb is: when my hands are shaking too much and the rework doesn't seem to go right, it's either due to too little or too much coffee. In either case, the problem can be fixed with more coffee. 

Once I bought thinnest insulated, pre tinned standard wire that was available, a sort that solders well and is thin enough so that it can be soldered to 0.5mm pitch QFP leg, my issues were gone forever. Kynar wire wrap wire sometimes doesn't solder too well (since it's not designed to be soldered), and magnet wire is PITA to strip.

The key is not to fear, and not to give up. Most often, everything can be fixed back together and rescued, no matter how desperate it may look; it will take time, but you'll gain invaluable experience while doing that.

[Buriedcode]

Rather than start a pissing contest about how fine a pitch people can work at or how much equipment they have, to the task at hand.

As some have said, TQFP32 ain't too bad.  Here's how I do it.  Use either solderable enamel wire or relatively thin insulated single core wire - like kynar.  Worst case, 1/0.6mm single core.

Cut off more than enough, use tweezers to roughly hold one end where it should go, fingers hold the other end to gauge the length required. 
Cut and strip insulation ~2-3mm of both ends. 
Tin both ends. 
Dab of flux on the pin you wish to solder too,
Holding the wire with tweezers, place the tip of the wire on the top of the pin you wish to solder to. (end of wire butted up against the plastic package)
Get a small blob of solder on the tip of your iron, about 1mm across.
dab the blob onto the wire (which is sitting on the pin), let flux do its job.
Keep holding the wire there for a couple of seconds.
Do other pin.
For convenience, tack down the to the PCB with a tiny spot of superglue, or preferably Kapton tape.

What really helps:
Solderable magnet wire, or more popular, Kynar wire (insulation doesn't shrink/burn when the iron touches it).
Jelly flux, applied with a needle.

I have never used a microscope - but then I only do rework every couple of months so it isn't worth the investment.

Example of the solderable enamel work:
http://elm-chan.org/docs/wire/wiring_e.html

Example of using kynar wire:
http://jeelabs.org/wp-content/uploads/2015/04/DSC_50201.jpg