Alioth's Journal: Success!

My second take at the W5100 breakout board has been a success! I have it connected to my ALIAC-2 Z80 based computer, and I've managed to ping it.

Here's a pic of the board with that tiny and rather awkward to solder chip (seriously, the chip alone took longer getting the solder joints right than all the passives on the PCB put together)

http://www.alioth.net/tmp/w5100-online.jpg

Hopefully, by tomorrow, ALIAC-2 will be on the internet!

There's still a little trouble - even with this new breakout board, I'm having random trouble with the lower half of the LCD. This I now suspect to be being caused by a glitch in the improvised chip select logic on the breadboard. I will probably need the logic analyser to find out for sure though...

However, the board working does validate my basic PCB layout especially with regard to the analogue parts and placement of components; I couldn't get the crystal as close to the chip as I would prefer because...well, it's just impossible to get it any closer, and the short run of the 1.8v analogue supply having no choice but to run alongside the xtal signal seems to not be having any deleterious effects (I managed to keep that run down to about 3mm though).

Even with all the pain of making this board, it's been a very worthwhile learning experience: it's caused me to hugely improve my PCB making process (finding better toner transfer paper to make finer features, and a more reliable process) as well as greatly speed up my method of soldering tiny surface mount passives. I can now do 1206 (120 thousandths of an inch by 60 thou) passives faster than pin through hole components, and 0603 (60 thou by 30 thou - in metric terms, about 2mm by 1mm, don't breathe on the board too hard!) sized ones almost as quickly. I've found the best way of soldering them by hand is to melt a little blob of solder on one pad, manoevre the component in with a small screwdriver, then re-melt the solder with the iron tip and just nudge the component's end into the blob. Then conventionally solder the other end to its pad.

However, I think my next take at a W5100 based board (i.e. when I assemble the first ethernet card based on this chip for the Sinclair Spectrum), I'll use solder paste and hotplate reflow. I think it will just be much more reliable than hand soldering this insanely fine pitch package.

Congrats

[anominous]

I'm very impressed by your soldering skills and perseverance - good work sir!

Re:

[Alioth]

It's more preserverence than soldering skills :-) If you can neatly solder through hole parts, you can also do fine pitch SMD - the technique is fairly easy, tack a pin down at each corner with a small amount of solder, then after checking the IC is still lined up on the pads properly, melt a small blob of solder on the tip of the iron and just drag it down the pins. Then clear up the solder bridges with solder braid. A magnifying glass helps :-)

The patience comes in getting the chip properly lined up, then