Maker Pro
Maker Pro

SMT R-chain layout choices

D

D Yuniskis

Jan 1, 1970
0
Hi,

I have an SMT design that must support three R choices
at final assembly. R1 < R2 < R3

Differential stuffing is not an option. Boards need
to be identical up to final assembly/test.

One option is to install three R's in a series chain:
R1, R2 - R1, R3 - (R2 + R1) and short none, the last, or
the last *two* R's depending on the configuration desired.

Another approach is to arrange three paralleled resistors such
that R1 is formed by R3 || Ra || Rb and R2 formed by R3 || Ra.
Then, removing Rb and possibly Ra, as required.

The first approach could be accomplished with solder switches
on the board -- but that eats up real estate (this is a *tiny*
board!). Bridging R's with shunts seems like it could lead
to components becoming "unsoldered" in the process.

The second approach could possibly be accomplished by *cutting*
the R's in question (gasp!).

Are there other approaches that I should evaluate?

Thx,
--don
 
D

D Yuniskis

Jan 1, 1970
0
D said:
One option is to install three R's in a series chain:
R1, R2 - R1, R3 - (R2 + R1) and short none, the last, or
the last *two* R's depending on the configuration desired.

Grrrr... s/+ R1//
 
D

David L. Jones

Jan 1, 1970
0
D said:
Hi,

I have an SMT design that must support three R choices
at final assembly. R1 < R2 < R3

Differential stuffing is not an option. Boards need
to be identical up to final assembly/test.

One option is to install three R's in a series chain:
R1, R2 - R1, R3 - (R2 + R1) and short none, the last, or
the last *two* R's depending on the configuration desired.

Another approach is to arrange three paralleled resistors such
that R1 is formed by R3 || Ra || Rb and R2 formed by R3 || Ra.
Then, removing Rb and possibly Ra, as required.

The first approach could be accomplished with solder switches
on the board -- but that eats up real estate (this is a *tiny*
board!). Bridging R's with shunts seems like it could lead
to components becoming "unsoldered" in the process.

The second approach could possibly be accomplished by *cutting*
the R's in question (gasp!).

Are there other approaches that I should evaluate?

You can get zero ohm resistors, but that would require removing and then
reinstalling parts. So the 2nd option of removing parallel resistors sounds
the least effort.

Depending upon what your requirements, you might be able to use an EEPOT
(digital programmable resistor) and have a small header on the board (or
simply probed as part of the bead of nails tester) so you program the
resistor value after it's assembled.

Dave.
 
N

Nemo

Jan 1, 1970
0
D Yuniskis mulls
I have an SMT design that must support three R choices
at final assembly. R1 < R2 < R3

Differential stuffing is not an option. Boards need
to be identical up to final assembly/test.

One option is to install three R's in a series chain:
R1, R2 - R1, R3 - (R2 + R1) and short none, the last, or
the last *two* R's depending on the configuration desired.

Another approach is to arrange three paralleled resistors such
that R1 is formed by R3 || Ra || Rb and R2 formed by R3 || Ra.
Then, removing Rb and possibly Ra, as required.

The first approach could be accomplished with solder switches
on the board -- but that eats up real estate (this is a *tiny*
board!). Bridging R's with shunts seems like it could lead
to components becoming "unsoldered" in the process.

The second approach could possibly be accomplished by *cutting*
the R's in question (gasp!).

Are there other approaches that I should evaluate?

Is a trimpot too unstable for the application?

I have seen select-on-test resistors (not SMT) fitted into a pair of
single-pin sockets, but that may take up too much room. Of course the
resistor can then fit above SMT components, so you win some real estate
back.

Snap-off sections of the board.
 
K

krw

Jan 1, 1970
0
Hi,

I have an SMT design that must support three R choices
at final assembly. R1 < R2 < R3

Differential stuffing is not an option. Boards need
to be identical up to final assembly/test.

One option is to install three R's in a series chain:
R1, R2 - R1, R3 - (R2 + R1) and short none, the last, or
the last *two* R's depending on the configuration desired.

Another approach is to arrange three paralleled resistors such
that R1 is formed by R3 || Ra || Rb and R2 formed by R3 || Ra.
Then, removing Rb and possibly Ra, as required.

The first approach could be accomplished with solder switches
on the board -- but that eats up real estate (this is a *tiny*
board!). Bridging R's with shunts seems like it could lead
to components becoming "unsoldered" in the process.

The second approach could possibly be accomplished by *cutting*
the R's in question (gasp!).

Are there other approaches that I should evaluate?

Cut lands to the Rs. Route the lands on the surface, between pads big
enough to jumper, and forget the solder mask. A quick cut (one at
each end of the trace is better) and it's customized.
 
D

D Yuniskis

Jan 1, 1970
0
Hi Dave,
You can get zero ohm resistors, but that would require removing and then
reinstalling parts.

Correct. Or, adding spare sites in which to install those (as shunts).
Two of them wouldn't take up *much* space, but...
So the 2nd option of removing parallel resistors sounds
the least effort.

Yes. Unsoldering one or two components has to be no more
effort than *soldering* one or two shunts! The downside
of the parallel approach is it makes testing harder -- i.e.,
you can only test the R1 condition (all R's installed).
Depending upon what your requirements, you might be able to use an EEPOT
(digital programmable resistor) and have a small header on the board (or
simply probed as part of the bead of nails tester) so you program the
resistor value after it's assembled.

Yeah, I thought of that but I'm not sure how well that would work
in this circuit topology. And, these are three *specific* R
values so I would have to evaluate the consequences of the
pot being set "wrong".
 
D

D Yuniskis

Jan 1, 1970
0
Hi,
Stressing the resistors is a bad idea (can cause debris); there are

Yes. I was only offering it up as an "inspiration" for unusual
approaches to this problem. :>
jumper blocks intended to be cut-once for trim, you only need
two jumpers for selecting three values.

Yes. Getting them *small* enough is the issue and, as I mention
in another reply, it makes testing harder as you can only test
the one "none cut" condition (i.e., if the two shunted R's
are incorrect values, you will never know at final test)
For a dollar, you can get two position DIP switches.

Two much real estate and two much money for the value that
it provides (you can cut/solder components cheaper than that
dollar spent)
If boards are swapped for field maintenance, the remanufacture
of old boards will be easier if you use switches not clipped jumpers.

Understood. I am hoping to keep costs low enough and reliability
high enough that this is not a factor.
 
D

D Yuniskis

Jan 1, 1970
0
Hi Nemo,
D Yuniskis mulls


Is a trimpot too unstable for the application?

Yeah, I think so. As I mention in another reply, there are
a set of three or four *fixed* values that are acceptable
here. I don't want to risk dealing with a pot that was
set to some *other* value midway between two "legitimate"
values.
I have seen select-on-test resistors (not SMT) fitted into a pair of
single-pin sockets, but that may take up too much room. Of course the
resistor can then fit above SMT components, so you win some real estate
back.

Now *that* may be an idea! I.e., use a regular 1/8W resistor
with pigtails and install it above the other components.
This "costs" roughly (labor) the same as soldering/unsoldering
components that are already populated on the board. Only needs
two pads -- which can be made "far enough apart" just by straddling
some larger component, etc.

It also lets the circuit be tested with all of the possible R values
(the other approaches have limitations on this) easily.

I will have to stew on that...
Snap-off sections of the board.

<frown> Board is already *tiny*. I think it would be hard
making a *tinier* portion that you could nick off.
 
D

D Yuniskis

Jan 1, 1970
0
krw said:
On Tue, 12 Jan 2010 00:42:00 -0700, D Yuniskis


Cut lands to the Rs. Route the lands on the surface, between pads big
enough to jumper, and forget the solder mask. A quick cut (one at
each end of the trace is better) and it's customized.

I think that would be hard to do at this scale (the board is
tiny intentionally!). Putting in big pads would use up more
space than the SMT R's in the first place.
 
K

krw

Jan 1, 1970
0
I think that would be hard to do at this scale (the board is
tiny intentionally!). Putting in big pads would use up more
space than the SMT R's in the first place.

The lands only have to be as large as you want them to be. A wire
between 0603 pads would be plenty big enough to cut without special
tools. The pads would allow a 0-ohm resistor to be attached in case
you change your mind. Just be careful not to run any lands underneath
the cut area.
 
Top