Well, I drew out one adapter design, to see where we were
in the ballpark.  I designed a simple 28-pin SOIC to DIP
adapter board. Basically, it is a board with regular
plated-through holes for a 28-pin DIP (0.32" drill),
with SMD pads for a SOIC-28 inside the outline for the
DIP-28. I routed traces on a 0.05" grid on two sides, and
managed to do it with 10 vias. (I went through a little
bit of hell with Eagle trying to get the pad layout for
the SOIC chip commensurate with that for the DIP so that
it would pass the DRC or even route; their part design was
a little flakey -- it was somewhat assymetetrical. But I
straigtened it out, and I'll report it to Cadsoft).

There is a PostScript printout of it in 2x scale at
ftp://ftp.drzyzgula.org/pub/electronics/adap28.ps.
(Ghostview seems to do fine with it and will show it in
color. If you print it in B&W, the top and bottom traces
won't be distinguishable, but if you pay attention to the
vias you can figure it out. When I get the Eagle board
file packaged up I'll put that on my Adapter page.)

My idea is that you could simply solder some Mil-Max
pins into the DIP holes; e.g. Digi-key #ED5055-ND (at
$6.30 per hundred) might be made to work, although there
might be a better choice; suggestions are welcome.  Also,
after plating I'm not sure the DIP holes in my design
are a good size; they should be carefully matched to
the choice of pin.

This board works out to 1.4"x0.7". I plugged some guesses
as to parameters into the Advanced Circuits quote engine,
and came up with the following for a production run
of 1000 with a 4-week turn: 62 cents each. Now, for an
adapter like this, there might be better/cheaper choices
for materials... all y'all with more experience making
boards are welcome to make suggestions.

But the bottom line seems to be that at least this design
could be made, as a kit with pins that need to be soldered
in by hand, for a little less than a dollar. Clearly,
at a minimum, there would have to be a little markup for
cost recovery in handling and such, so maybe the total
price would have to be more in the range of $2.50 to $5.00
for sales on a non-profit basis; in keeping with Paul's
comments below, ten of them might cost $25. This is still
far less than one pays for commercial adapters.

Thoughts? Comments?

--Bob

Quote Type: Production
Quantity: 1000
Unit Price: $0.62
Board Subtotal (Qty * unit price): $620.00
NRE Tooling: $125.00
Lead Time: 15-day
Material type: FR4
Material thickness: 0.062"
Layer count: 2
First dimension: 1.4"
Second dimension: 0.7"
Finish Plating: SMOBC
Copper weight: 1 oz
Trace width/space: 0.008"
Smallest hole size: 0.024"
Gold Fingers None
Qty SMD pads - top: 28
Qty SMD pads - bottom: 0
SMD - pitch: 0.05
Solder mask sides: 2
Solder Mask type: LPI
Legend silkscreen: 0
CNC route points: 4
Tab Route?  No
Scoring?  No

On Mon, Aug 23, 1999 at 07:37:28AM +1000, Paul B. Webster VK2BZC wrote:
> Tom Handley wrote:
>
> >  Wagner, if we could do adapters for $1 in single qty, I would sell my
> > house to raise startup capital ;-)
>
>   Whilst understanding that Tom is pooh-pooing the practicality of
> getting down to $1 a board, it seems to me that if it were to be
> practical at all, there is *no way* you would *ever* offer them as
> "single qty".
>
>   I note the local electronics shop now sells resistors only in packs
> of five, or twos for the large ones (as Tandy = Radio Shack always did).
> This seems to me the only sensible approach.  I can't think of a time
> when I ever needed to buy one resistor; for whatever reason I wanted the
> one, it would always have been sensible to have the remainder in stock
> for next time - to avoid a trip where possible.
>
>   If you were making these adapter boards (and I think they should be
> made), you would *never* want to ship less than ten to anyone.  At that
> rate, they should still be cheap to do.
> --
>   Cheers,
>         Paul B.

--
============================================================
Bob Drzyzgula                             It's not a problem
bob@drzyzgula.org                until something bad happens
============================================================