Q-bus I/O project

John Wilson wilson at dbit.com
Mon Oct 12 11:12:03 CDT 2015


On Mon, Oct 12, 2015 at 11:30:39AM -0400, Paul Koning wrote:
>> ... but I'd rather go RoHS.
>
>I would recommend against that.  Not unless you are trying to create a
>commercial product where you *must* be RoHS to conform to the requirements
>of the bureaucrats.  Use real solder -- the job will be much easier and
>the result more reliable.  Real solder is still available, including
>solder paste.

Sure, but I've been using the new solder (and paste) for years now, and I'm
resigned to it.  I'd sell any of my doodads if someone wanted to buy them,
so I want them to be legal, even though I've never understood who was being
poisoned by the lead in solder.  I absolutely hated it at first (and I
still hate paying for it -- my last 1 lb roll of Kester RoHS solder with
water-soluble flux was over $100) but I've gotten used to it and really,
it's fine.  Nice shiny strong joints, if I run the iron a little hotter.
And the SMT stuff is all reflow so it's automatic.

>I would think a pick & place robot would be a fairly straightforward
>derivative of a 3d printer.

I was thinking the same thing.  A vacu-sucker head and some way of feeding
(and unwrapping) the tape with the components, and possibly a camera, are
all it needs ... not that that's trivial, but it's nothing compared to the
X/Y/Z mechanics.  Also, 500 gallons of new firmware, but, whatever!

>But is that really needed?  Placing the parts isn't all that terrible.

I wholeheartedly disagree!  It took me over four hours to position the
parts on this board and my back was killing me by the end.  (I can't
*believe* I made it four hours w/o knocking the board off the table or
putting my elbow on it or something idiotic like that.)  I wouldn't even
mind a very sloppy robot that doesn't have machine vision and isn't precise
about orienting parts.  I always go over with a microscope and square them
up anyway, and that part doesn't take long (and gives a nice OCD buzz).

>I have read (and posted here in the past) a nice article (in German) on
>the use of a toaster oven with some clever temperature control as an IR
>reflow soldering machine.  I haven't tried this yet, but it sounds like a
>good scheme, and would allow the use of BGA parts at least in moderate
>sizes.

It works *fantastically*.  That's what I do, using an Arduino and a homemade
shield with a MAX6675 on it (um ... hand-soldered, but SOICs are easy so
there's no chicken/egg problem), and a couple of buttons and a socket for
an eBay LCD module.  The only thing is, the (bare) tip of the thermocouple
has to be touching one of the boards (if it's just measuring the air temp,
you scorch the PCBs), and no matter how I think I've bent it, it has a mind
of its own and tends to sweep components away when I put it in place and
let go.

>At least one of the moderate cost small volume PCB fab shops will deliver
>solder paste screens along with the finished boards if you ask for it.

Those are huge time-savers.  I've had great luck with both ohararp.com
(charged by the sheet) and oshstencils.com (by the sqinch) for Kapton
(low-volume), and bayareacircuits.com (by the design) for stainless steel
(super nice but more expensive).  A small Corian cutting board makes a
great backboard for taping up the jig to do some pasting.

For one-offs, using a syringe by eye is fine.  It doesn't have to be very
precise.  The flux melts at a much lower temp than the solder pellets, so
the beautiful square blocks from the stencil turn into a pool anyway, but
then the solder mask does its job and the pool separates out onto the pads
(except for TQFPs -- gotta use solder-wick afterwards on them) once the
tin melts too.  So for SOICs/TQFPs you might as well just lay a bead down
across a whole row of pads.  It works just as well as a tiny dab on each.

John Wilson
D Bit


More information about the cctalk mailing list