Reflow of 'solder blob' Jumpers

[Christoph Loew]

Hi everyone,

i was wondering :
has anyone used different solder paste masks for a reflow-soldered
PCB to configure solder blob jumpers as open/closed ?

I was thinking of using this technique to set I2C-bus addresses for
a bunch of otherwise identical PCB's that will be manufactured on one
panel; it would save using 0-Ohm-Resistors for the configuration.

Can this be done reliably ?

Any hints appreciated,

Chris

 

[Robert Baer]

Hi everyone,

i was wondering :
has anyone used different solder paste masks for a reflow-soldered
PCB to configure solder blob jumpers as open/closed ?

I was thinking of using this technique to set I2C-bus addresses for
a bunch of otherwise identical PCB's that will be manufactured on one
panel; it would save using 0-Ohm-Resistors for the configuration.

Can this be done reliably ?

Any hints appreciated,

Chris

NO.
If the gap is small enough (say 5 mil or less), it might be possible
to reflow an excess of solder to bridge the gap.
But then, the gaps you want open might short by of a solder bridge
evenwithout an excess of solder.
The wider the gap, the harder it would be to bridge, making a desired
short less possible.
The more narrow the gap, the more likely an accidental bridge, where
one is not wanted.
Layout one unit with back-to-back (flat parts) "D" shapes and a 8-10
mil gap between, with a 5-mil trace connecting them in the center. This
makes them all a short.
Then during step and repeat during panelization on the final film,
photographically break traces, one unit at a time, in a binary pattern
(position N has binary pattern N).
The panels made then will give a set of units programmed from zero to
M, where M+1 is the count of the number of units on the panel; 1000
panels made fromthat film will give 1000 sets for assembly.

 

[Christoph Loew]

NO.
If the gap is small enough (say 5 mil or less), it might be possible
to reflow an excess of solder to bridge the gap.
But then, the gaps you want open might short by of a solder bridge
evenwithout an excess of solder.
The wider the gap, the harder it would be to bridge, making a desired
short less possible.
The more narrow the gap, the more likely an accidental bridge, where
one is not wanted.
Layout one unit with back-to-back (flat parts) "D" shapes and a 8-10
mil gap between, with a 5-mil trace connecting them in the center. This
makes them all a short.
Then during step and repeat during panelization on the final film,
photographically break traces, one unit at a time, in a binary pattern
(position N has binary pattern N).
The panels made then will give a set of units programmed from zero to
M, where M+1 is the count of the number of units on the panel; 1000
panels made fromthat film will give 1000 sets for assembly.

Thanks for the quick answer.
The process you describe sound very reasonable - changes are still
possible but reliability is not compromised. I'll probably patch the
Gerber data during panelization to implement your idea.

Meanwhile, a colleage told me he had once seen the 'reflowed blob'
technique used with pads shaped like this :

__ ____
/ \\ \
/ \\ \
| \\ |
---- \\ ----
---- // ----
| // |
\ // /
\__//___/

(crappy ASCII art of a circular pad with the separating gap in a
'V'-shape)

which allegedly improved the success quote sufficiently if the gap was
the minimal gapsize allowed for exposed copper and the complete circle
was covered in solder paste for the pads which are supposed to be
shortened while the pads which are to remain open were kept free of
paste.

After your reply I'm much more sceptical of this - does that sound
possible ?
Did you try this once or have you seen this used anywhere ?

(I can recall the Commodore 1551 Floppy drive's ID jumpers - pre-bridged
solder blobs as you described, but I haven't seen anything similar
since)


Chris

 

[sPoNiX]

Hi everyone,

i was wondering :
has anyone used different solder paste masks for a reflow-soldered
PCB to configure solder blob jumpers as open/closed ?

I don't think you could do it with a solder paste mask. However, such
links are widely used on panel meters and in car alarm keyfobs etc and
are perfectly reliable when make with a soldering iron.

sPoNiX

 

[legg]

I don't think you could do it with a solder paste mask. However, such
links are widely used on panel meters and in car alarm keyfobs etc and
are perfectly reliable when make with a soldering iron.

sPoNiX

He wants to save the cost of a single smd placement, not the
exhorbitant costs implied in 'manually' soldering or trimming
anything...........

RL

 

[Rich Grise]

__ ____
/ \\ \
/ \\ \
| \\ |
---- - ----
---- - ----
| // |
\ // /
\__//___/

(crappy ASCII art of a circular pad with the separating gap in a
'V'-shape)

I've modified your diagram, to show the jumper that Robert Baer was
talking about. In this case, the cut doesn't need to be a V, just
back-to-back D pads will do it.

I wouldn't trust a solder bridge to make a contact.

which allegedly improved the success quote sufficiently if the gap was
the minimal gapsize allowed for exposed copper and the complete circle
was covered in solder paste for the pads which are supposed to be
shortened while the pads which are to remain open were kept free of
paste.

After your reply I'm much more sceptical of this - does that sound
possible ?
Did you try this once or have> you seen this used anywhere ?

It's like he said. Where you want a solder bridge, it won't, and were you
don't want one, it does. ;-) Put it in copper. Zero ohm jumper resistors
are available too, so you could program it at the pick&place level.

(I can recall the Commodore 1551 Floppy drive's ID jumpers - pre-bridged
solder blobs as you described, but I haven't seen anything similar
since)

Cheers!
Rich

 

[Robert Baer]

Thanks for the quick answer.
The process you describe sound very reasonable - changes are still
possible but reliability is not compromised. I'll probably patch the
Gerber data during panelization to implement your idea.

Meanwhile, a colleage told me he had once seen the 'reflowed blob'
technique used with pads shaped like this :

__ ____
/ \\ \
/ \\ \
| \\ |
---- \\ ----
---- // ----
| // |
\ // /
\__//___/

(crappy ASCII art of a circular pad with the separating gap in a
'V'-shape)

which allegedly improved the success quote sufficiently if the gap was
the minimal gapsize allowed for exposed copper and the complete circle
was covered in solder paste for the pads which are supposed to be
shortened while the pads which are to remain open were kept free of
paste.

After your reply I'm much more sceptical of this - does that sound
possible ?
Did you try this once or have you seen this used anywhere ?

(I can recall the Commodore 1551 Floppy drive's ID jumpers - pre-bridged
solder blobs as you described, but I haven't seen anything similar
since)

Chris

It does not matter much in using "creative" shapes; it is the gap that
can help/hinder bridging/nonbridging.
All of what i mentioned is assuming the use of solder paste and
heating for reflow.
A solder wave can give completely different results.
I have seen extremes: (1) the whole back side of a PCB covered with a
minimum of 1/16 thick solder, and lots of "icicles"; (2) visually
identical to solder paste / heat tunnel coverage (uniform).
So, if one could reliably and repeatibly run a solder wave system, an
"arrow" shaped gap could reliably be bridged.
But what would be the point; there still would be no programming.
*Something* must be altered at a given gap/nogap in order to reliably
have a programed result.

BTW, that double D method was popular in a number of boards in the
1980s (i saw it done mostly on floppy boards).

 

[Al]

It does not matter much in using "creative" shapes; it is the gap that
can help/hinder bridging/nonbridging.
All of what i mentioned is assuming the use of solder paste and
heating for reflow.
A solder wave can give completely different results.
I have seen extremes: (1) the whole back side of a PCB covered with a
minimum of 1/16 thick solder, and lots of "icicles"; (2) visually
identical to solder paste / heat tunnel coverage (uniform).
So, if one could reliably and repeatibly run a solder wave system, an
"arrow" shaped gap could reliably be bridged.
But what would be the point; there still would be no programming.
*Something* must be altered at a given gap/nogap in order to reliably
have a programed result.

BTW, that double D method was popular in a number of boards in the
1980s (i saw it done mostly on floppy boards).

Sorta off topic, but... Many years ago I did some analytical work on
shorts in a feedthru connector. It enclosed a pi filter. Two of the
elements were circular, ring shaped, ceramic capacitors which were
soldered on the outside edge to the case and the inside edge to the
feedthru conductor. This were soldered blind using solder preforms and
the whole shebang would reflow nicely when it was heated in an oven. On
occasion, the flowing solder would bridge betwe"flowing solder blobs." I
coined the word "flobs" to describe the event. Saved some ink in the
reports ;-)

At any rate, since we had hundreds of these and could not afford to
scrap and rebuild due to schedule/cost constraints, we designed a
temperature cycling, monitored shorts test which weeded out the "almost
shorted" bunch. And a shorts test weeded out the shorted ones prior to
the temperature cycling. No problems in the field later as I recall.

Al

 

[John Woodgate]

Two of the
elements were circular, ring shaped, ceramic capacitors

Phenomenally good capacitors, with vanishing inductance. Many don't
resonate at any frequency; they turn into lossy transmission lines at
GHz frequencies. Can be tricky to embody, as you found.

 

[ranmyaku]

Couldn't you just play like the computer manufacturers and put jumper
pins on the board you could enable or disable with that clever little
jumper. Seems a lot easier that way.