A very inexpensive way is to buy some scrapped hard drives and/or circuit
boards. Most people find it easier to get new ones rather than fix bad
surface-mount boards. Examining these things lets you get familiar with
individual parts, component marking, and even soldering/desoldering.
Hobbyist-level SMT doesn't seem to take much in the way of expensive
equipment. A couple of months ago I made my first foray into SMT
work. It went very easily. I'll describe it here on the off chance
it's useful to you (or someone else).
I wanted to build a circuit using no through-hole components, so that
the board could be mounted flush against something (the board contained
a temperature sensor, and I figured this would help keep it at the
same temperature as the thing it's measuring). The small size was a bonus.
I used toner-transfer sheets and an ancient Radio Shack PCB etching kit
that's been gathering dust for the last ten years. Components were
mostly 0805 discretes, one 8-SOIC, and a couple of other components
of similar size and pad-spacing. (PCB design was done using xpcb on Linux.)
I had a butane hot-air torch lying around, so I decided to try to
use it for hot-air soldering. It's a "MASTER Ultratorch", and claims to
have a temperature of 1300 deg C. IIRC, it cost me about US$100 new.
Hot air technique: I put solder paste on the pads with a toothpick.
Too much can cause a solder bridge hidden under the component (had to
fix a couple of these afterwards). Place all the solder, then place all the
components using tweezers, being careful not to smear the paste.
Then preheat the board. I did this by holding the board over a candle
flame for a few minutes. It should be fairly hot to the touch during this
time. If you don't do this, then when you try to melt the solder, the
components will be blown off the board by the steam from the solder paste.
Then, apply the hot air. Wave the torch over the board a bit to continue
the "preheat" then focus it on one component or one lead of a component.
The solder will melt, then a fraction of a second later some of it will
get sucked under the component lead as the lead comes up to temperature.
When that happens move on to the next component. You can do a lot
of components quickly this way.
The non-hot-air technique I used was just the obvious technique of putting
down some solder paste, putting down the component, and touching each
lead with the tip of a small (12W) soldering pencil. I used that for
the 8-SOIC (the only component I would have had to order more of if I had
ruined one) and a relatively bulky RJ-11 (easier to deal with this way).
If I were making more I'd try using the hot-air technique on the IC also.
I was pleasantly surprised with how quickly it went --- including etch
time, this was as fast as wiring up the circuit on perfboard, and
the result was much nicer-looking. I may start prototyping things this
My inspiration for the "preheat" period was the data sheet for one
of the more temperature-sensitive components, an electrolytic
cap from CDE. It had a nice temperature-vs-time graph of a
typical soldering operation, with warnings about which limits
were critical. I probably exceeded the limits here and there but
not by a whole lot. At any rate, the circuit works, and as a hobbyist,
that's what I care about.
As for expense, compared with building the same device using through
hole components: the discretes didn't cost noticably more as SMT
than they would have with leads. I had to buy the solder paste, of
course, but it wasn't very expensive, and I'd have had to buy solder
anyway. If I hadn't already had the hot-air tool, it would have
been somewhat expensive, but it's a one-time cost. A good magnifier
might have made the parts placement easier, but all in all, not too
expensive or difficult.
Wim Lewis <wi...@hhhh.org>, Seattle, WA, USA. PGP keyID 27F772C1
My ambition is to have all [my] arguments seem annoyingly plausible -Joe Slater