Hello, a question for those of you with manufacturing experience:
At what point do the added cost and efforts for automated assembly make
sense?
Assuming the board is mostly SMT, properly designed and not terribly
complicated or large.
I realize this is a very vague question. I am just trying to get an idea of
what kind of quantities people consider worthy of automated assembly.
Thanx,
Tony
Tony Rolando schrieb:
Hello, a question for those of you with manufacturing experience:
At what point do the added cost and efforts for automated assembly make
sense?
What /exactly/ do you mean with "automated assembly"?
Assuming the board is mostly SMT, properly designed and not terribly
complicated or large.
I realize this is a very vague question. I am just trying to get an idea of
what kind of quantities people consider worthy of automated assembly.
The answer is ... well, it depends. 
If the board has only (say) 10 components, we surely build up samples
(single boards) manually. However, we already assembled a single board
with several hundred components by machine.
Generally, using a pick & place machine is common for SMT at rather
small quantities already. OTOH, THT is commonly manually assembled up to
rather large volumes.
And if you don't have your own pick&place machine at hand, you'll
probably have a higher treshold for machine assembly.
Tilmann
"Tilmann Reh" <tilmannreh@despammed.com> wrote in message
news:g8mnt6$eb$1@cheetah.cadsoft.de...
Tony Rolando schrieb:
Hello, a question for those of you with manufacturing experience:
At what point do the added cost and efforts for automated assembly make
sense?
What /exactly/ do you mean with "automated assembly"?
Assuming the board is mostly SMT, properly designed and not terribly
complicated or large.
I realize this is a very vague question. I am just trying to get an idea
of
what kind of quantities people consider worthy of automated assembly.
The answer is ... well, it depends.
If the board has only (say) 10 components, we surely build up samples
(single boards) manually. However, we already assembled a single board
with several hundred components by machine.
Generally, using a pick & place machine is common for SMT at rather
small quantities already. OTOH, THT is commonly manually assembled up to
rather large volumes.
And if you don't have your own pick&place machine at hand, you'll
probably have a higher treshold for machine assembly.
Tilmann
Thanx for your comments. By automated assembly I was meaning pick & place. I
realize this was a vague question; I was hoping to get some various comments
regarding the topic.
Your last line really provided some insight for me...
And if you don't have your own pick&place machine at hand, you'll
probably have a higher treshold for machine assembly.
Where I work we contract out most assembly. We do somewhat large volume on
most products. We have planned some smaller products that will probably NOT
sell high volumes. I am wondering (aloud I suppose) what route we will take
for assembly of those boards.. I do not make these descisions, just
speculate I suppose.
Thanx,
Tony
Tony Rolando schrieb:
Your last line really provided some insight for me...
Where I work we contract out most assembly. We do somewhat large volume on
most products. We have planned some smaller products that will probably NOT
sell high volumes. I am wondering (aloud I suppose) what route we will take
for assembly of those boards.. I do not make these descisions, just
speculate I suppose.
Well, the answer is ... it still depends.
We do custom (industrial) electronics development with subsequent series
production, sometimes we also do contract manufacturing. Production lots
vary from 10 to >500, but not thousands. I assume this is what is
generally called "not high volumes".
For really low volumes (say, lots of 10 or 20 boards), we often use pure
THT and manual assembly, since this is much more efficient then.
However, as soon as you need some parts in SMT (for example, if you
don't get them in THT packages, or if it must be small), it generally
makes sense to do as much as possible in SMT then - combined with
machine assembly. However, we also have two products with one single SMT
chip on them (not available THT) and the rest in THT. This single chip
is then assembled manually. Or, talking about single quantities
(prototypes), we will assemble a single prototype board with (amongst
other parts) >400 0805 resistors, several tens of ICs (many in
fine-pitch)- of course we will use the machine for that, even if it's
only one board. Time for setting up the machine is much less than the
time we'd need to assemble this amount manually.
You can't give a general statement without caring about (all) details.
Chosing the optimum technology depends on very many details...
Tilmann
Tilmann Reh wrote:
Tony Rolando schrieb:
Your last line really provided some insight for me...
You can't give a general statement without caring about (all) details.
Chosing the optimum technology depends on very many details...
Tilmann
Agreed. It's all in the details. We do low-volume runs (10-100 per). Some
key details:
Setup charges vs Manual charges - there's usually a setup charge to load the
machine, but the run costs less. There's no setup charge for manual
assembly, but the run costs more. There's also typically a one-time
programming charge for the first run. Do the math.
Tape & Reel - machine runs require most parts on tape and reel. Do you have
tape and reel inventory?
Consistency - machine jobs are like ISO 9001 - if the setup is right, it's
always right; if it's wrong, all the boards are wrong. Once the programming
is done correctly machine jobs are usually more consistent both in the
current run and also for any later runs. How many future runs might this
board have?
Part density and complexity - 10 boards with 10 low-density SMT parts are
quicker to build manually. 10 boards with 500 SMT parts, of which a dozen
are very fine-pitch or BGA monsters, are far quicker and more precisely
done by machine. There is, unfortunately, no set quantity of parts, pitch
or whatever that you can plug into a formula to calculate where the
dividing line is.
We typically have prototype runs (3-10 boards) hand-stuffed unless they're
very complicated, because the programming and setup charges aren't
justified on a small-quantity run, when the board will almost certainly be
modified and the programming /setup will have to be redone anyway. Once a
board is in production, even small runs of 25 or so are typically
machine-assembled because it's quicker, higher-quality, and usually less
costly because the setup and programming remain consistent. We also have
most of our inventory on tape and reel.
Just some more detail data points to work with...
Robert you have make a very good explantion!
Juste for low volum, you can buy cheap stensil ( set more than one project
too) or use a seringue and set manuely solder cream, hand assembly SMD and
run to the oven.
With a stensil, juste assembly on a board, I have assembly 0,5mm pitch IC
( ARM7, sopp ++ )It is quicher than hand soldering.
Thierry
"Robert Johnson" <bob@renegadelabs.com> skrev i melding
news:g8ralo$qj1$1@cheetah.cadsoft.de...
Tilmann Reh wrote:
Tony Rolando schrieb:
Your last line really provided some insight for me...
You can't give a general statement without caring about (all) details.
Chosing the optimum technology depends on very many details...
Tilmann
Agreed. It's all in the details. We do low-volume runs (10-100 per). Some
key details:
Setup charges vs Manual charges - there's usually a setup charge to load
the
machine, but the run costs less. There's no setup charge for manual
assembly, but the run costs more. There's also typically a one-time
programming charge for the first run. Do the math.
Tape & Reel - machine runs require most parts on tape and reel. Do you
have
tape and reel inventory?
Consistency - machine jobs are like ISO 9001 - if the setup is right, it's
always right; if it's wrong, all the boards are wrong. Once the
programming
is done correctly machine jobs are usually more consistent both in the
current run and also for any later runs. How many future runs might this
board have?
Part density and complexity - 10 boards with 10 low-density SMT parts are
quicker to build manually. 10 boards with 500 SMT parts, of which a dozen
are very fine-pitch or BGA monsters, are far quicker and more precisely
done by machine. There is, unfortunately, no set quantity of parts, pitch
or whatever that you can plug into a formula to calculate where the
dividing line is.
We typically have prototype runs (3-10 boards) hand-stuffed unless they're
very complicated, because the programming and setup charges aren't
justified on a small-quantity run, when the board will almost certainly be
modified and the programming /setup will have to be redone anyway. Once a
board is in production, even small runs of 25 or so are typically
machine-assembled because it's quicker, higher-quality, and usually less
costly because the setup and programming remain consistent. We also have
most of our inventory on tape and reel.
Just some more detail data points to work with...
