Re: [CAD_CAM_EDM_DRO] taig pick and placing
Posted by
Jon Elson
on 2007-06-06 10:31:35 UTC
Graham Stabler wrote:
think speed was important. But, this machine is SLOOOOW! Of
course, with good servo drives, it could go a LOT faster. The
centering pit is BRILLIANT! But, I decided to go with a
commercial machine, and got a GREAT deal on an older Philips
CSM84, which will hold over 50 mechanical feeders plus the
vibratory stick feeder and the mechanical alignment fixture for
large chips. It is an old-style machine, with a huge amount of
stuff done by pneumatics or even mechanical linkages. All the
small feeders are operated by a plunger on the head that pumps
the feeder ratchet N times to move the component tape in 4 mm
steps. (Most small parts are on 4mm intervals on the tape.)
This thing has no vision, and the Z axis is pneumatic. It does
have 3 heads with different nozzles on them, and I took off one
set of centering arms so I could use one nozzle for the big
chips, like 144-pin FPGAs. Each nozzle has a 3-level vacuum
sensor on it, and it can be set to require any of those vacuum
levels to know the chip has been properly picked up. It has a
bunch of air restrictors so you can set the rate of lift and
drop for the nozzle. I had to slow down the lift rate for the
big chip nozzle so it wouldn't drop the heavy chips when lifting
them.
It also has a "beam sensor", this is a fiber-coupled lens
arrangement that projects a spot on the board, and it uses the
head X-Y motion to measure the cneter of holes or fiducial dots
on the board. I don't have fiducials on my boards, so I use a
pair of mounting holes as fiducials, and it locates them and
then references the rest of the board off those points, both to
translate and rotate the coordinates. It has a CCD camera that
is used only to project an image on a video screen. This can be
used to teach component locations, or to teach the location of
chips in the feeders. You only have to teach one standard
feeder on each feeder bar, but it is good for the vibratory
feeder, where all the chips glide up to a stop rail, so their
centers depend on the number of pins. I made my own add-on
feeder lane to add wide-so size chips to the lanes already built
into the vibratory feeder.
The software is pretty primitive, basically a custom 80186 board
that looks a lot like an original PC/AT. It even appears to
have the ISA bus strung out along a 50-pin cable, but no
standard slots. It has serial ports, but I haven't gotten
XON/XOFF handshaking to work right. It might be a problem with
my comm program. It works if I download the placement data at
about 10 chars/sec, not that big a problem as the files are
small. It has no disk drive, not even a flopy, that was an option.
I've got some pictures at http://jelinux.pico-systems.com/CSM84.html
I have done a run of my servo amp boards, and a couple small
runs of my Universal Stepper Controller boards on it. I do the
back sides (mostly capacitors) first, reflow, then do the front
sides and reflow again. I do the reflowing in a toaster oven
with a thermocouple ramp-and-soak controller. If I put the
thermocouple in the air, the boards fry. If I put the
thermocouple in a hole in the board, then they come out real
good. I use my homemade laser photoplotter to make artwork to
etch .005" brass sheet to make solder stencils. I make
mirror-image films for both sides of the brass, and laminate the
brass with Riston dry-film photoresist. I put a spacer of
cardboard through so the machine develops enough pressure, as
the rollers are set for .062" boards. I think you have to scrub
the brass real well to remove some kind of surface protectant or
it doesn't etch well. I made one great set of stencils, the
next set didn't etch evenly. Next time I will scrub the brass
very well first, and I think the stencil will come out OK.
The P&P machine, of course, is way faster than I can stencil the
solder paste or run the boards through the reflow oven. I can
do 4 servo amp boards, but only 2 controller boards at a time.
Jon
> Someone posted this video a while back but I don't think the websitePretty cool! When I wanted to build a P&P machine, I didn't
> existed at that time, just saw it on make magazine's blog.
>
> http://www.ciciora.com/picknplace.html
>
> Its very slow BUT steady. There is also a bit on the website where he
> uses a pneumatic syringe system on the Taig for the solder paste
> dispensing, obviously good enough for work at this level.
think speed was important. But, this machine is SLOOOOW! Of
course, with good servo drives, it could go a LOT faster. The
centering pit is BRILLIANT! But, I decided to go with a
commercial machine, and got a GREAT deal on an older Philips
CSM84, which will hold over 50 mechanical feeders plus the
vibratory stick feeder and the mechanical alignment fixture for
large chips. It is an old-style machine, with a huge amount of
stuff done by pneumatics or even mechanical linkages. All the
small feeders are operated by a plunger on the head that pumps
the feeder ratchet N times to move the component tape in 4 mm
steps. (Most small parts are on 4mm intervals on the tape.)
This thing has no vision, and the Z axis is pneumatic. It does
have 3 heads with different nozzles on them, and I took off one
set of centering arms so I could use one nozzle for the big
chips, like 144-pin FPGAs. Each nozzle has a 3-level vacuum
sensor on it, and it can be set to require any of those vacuum
levels to know the chip has been properly picked up. It has a
bunch of air restrictors so you can set the rate of lift and
drop for the nozzle. I had to slow down the lift rate for the
big chip nozzle so it wouldn't drop the heavy chips when lifting
them.
It also has a "beam sensor", this is a fiber-coupled lens
arrangement that projects a spot on the board, and it uses the
head X-Y motion to measure the cneter of holes or fiducial dots
on the board. I don't have fiducials on my boards, so I use a
pair of mounting holes as fiducials, and it locates them and
then references the rest of the board off those points, both to
translate and rotate the coordinates. It has a CCD camera that
is used only to project an image on a video screen. This can be
used to teach component locations, or to teach the location of
chips in the feeders. You only have to teach one standard
feeder on each feeder bar, but it is good for the vibratory
feeder, where all the chips glide up to a stop rail, so their
centers depend on the number of pins. I made my own add-on
feeder lane to add wide-so size chips to the lanes already built
into the vibratory feeder.
The software is pretty primitive, basically a custom 80186 board
that looks a lot like an original PC/AT. It even appears to
have the ISA bus strung out along a 50-pin cable, but no
standard slots. It has serial ports, but I haven't gotten
XON/XOFF handshaking to work right. It might be a problem with
my comm program. It works if I download the placement data at
about 10 chars/sec, not that big a problem as the files are
small. It has no disk drive, not even a flopy, that was an option.
I've got some pictures at http://jelinux.pico-systems.com/CSM84.html
I have done a run of my servo amp boards, and a couple small
runs of my Universal Stepper Controller boards on it. I do the
back sides (mostly capacitors) first, reflow, then do the front
sides and reflow again. I do the reflowing in a toaster oven
with a thermocouple ramp-and-soak controller. If I put the
thermocouple in the air, the boards fry. If I put the
thermocouple in a hole in the board, then they come out real
good. I use my homemade laser photoplotter to make artwork to
etch .005" brass sheet to make solder stencils. I make
mirror-image films for both sides of the brass, and laminate the
brass with Riston dry-film photoresist. I put a spacer of
cardboard through so the machine develops enough pressure, as
the rollers are set for .062" boards. I think you have to scrub
the brass real well to remove some kind of surface protectant or
it doesn't etch well. I made one great set of stencils, the
next set didn't etch evenly. Next time I will scrub the brass
very well first, and I think the stencil will come out OK.
The P&P machine, of course, is way faster than I can stencil the
solder paste or run the boards through the reflow oven. I can
do 4 servo amp boards, but only 2 controller boards at a time.
Jon
Discussion Thread
Graham Stabler
2007-06-06 06:22:28 UTC
taig pick and placing
Jon Elson
2007-06-06 10:31:35 UTC
Re: [CAD_CAM_EDM_DRO] taig pick and placing
Graham Stabler
2007-06-07 02:05:14 UTC
Re: taig pick and placing
Jon Elson
2007-06-07 09:34:46 UTC
Re: [CAD_CAM_EDM_DRO] Re: taig pick and placing
Graham Stabler
2007-06-07 15:26:15 UTC
Re: taig pick and placing
Jon Elson
2007-06-07 18:38:52 UTC
Re: [CAD_CAM_EDM_DRO] Re: taig pick and placing
Peter Homann
2007-06-07 19:42:22 UTC
Re: [CAD_CAM_EDM_DRO] Re: taig pick and placing