[flippersplace]

Hi, its fascinating reading. I don't mind it here but I think it actually deserves its own thread.

I was trying to buy a couple of second hand feeders out of the States this week to see how they work, what they would tell me (having never seen any first hand) and if I could use them

If I could have maybe at least 4 but preferably up to 8 automated 8mm tape feeders I could accomplish most PCB's I want to with a lot less effort. Manually forwarding, peeling and lining up tapes is getting to be a chore.

[flippersplace]

Wow that SiPlace machine is insane. Coincidentally I am planning at two heads/gantries for my next machine

[Eclipze]

I'm amazed that the Siplace machine can hold the components on a wheel that spins and accelerates so fast!

[RomanLini]

...
First, you are not mentioning any theta axis. You'll have to layout resistors in 0 and 90 degree and SO IC's likely in all 4 different orientations.

No, my machine has nozzle rotation within 180 degrees (actually about 200 degrees).

...
For blind pick and place this is actually a better way to go since the rotation of a component will double the offset error.

I'm not sure of your math but it shouldn't increase the offset error from the component in the pocket, provided the nozzle is accurately centred in the centre of the pickup pocket. Any parts offset will just be rotated, not increased.

...
Second, when you advance the tape with a pin, I would recommend you go in opposite feeding direction for 0.1mm or so (and make the pin dia only 1.3 or 1.4 mm). Else, you pull the pin, while it is reaming against a perforation hole and all components will go flying.

Hmm, if I understand you right you are worried that pulling the pin out of the tape will snag (because the pin is tight in the feed pin hole) and pull the tape up? My tape guides have side lips so the tape can't move vertically. It's a good point though, to check for th epin being tight and any chance of snags.

So you know quite a bit about the Siplace Multistar machines? Did you do design work on them or have something to do with operating them?

[stewi]

You may look for older type Siemens feeders such as 141100, 141102, 141104, 141105, 141107
All these feeders have two 8 mm tracks per feeder, one motor which advance either left or right side.
AdoptSMT GmbH - Used Machines and Spare Parts for the Automatic PCB Assembly Process - Spare Parts
All these feeders are quite heavy to ship. Cost of shipping is likely more than the value of the feeder.
Adopt is in Austria and Germany.
I'll get back to you with more info on Mon.

[stewi]

I used to manage final assembly and technical service in Germany and US. I also refurbished used machines when establishing my own business in 1995.
Recently, I went back to the roots of it all by designing a low cost CNC machine and signed up for this forum.

There are several things, I’ve learned and would recommend on new designs.
First, from the very beginning we were using venturi vacuum generators. This is a very simple design and the vacuum is generated directly at the placement head.
JV Series Mini Compressed Air Venturi Vacuum Generators
Or build your own.
The 12 nozzle revolver head contains three separate vacuum circuits. One for pick up and placement, one for rejecting a component (vision/vacuum error) and one for the rest 10 nozzles. A garden hose size airline is required to supply enough air for the venturis.
The valves for switching from vacuum to air pressure are custom made pistons, activated by very fast stepper motors.
http://www.ahs-antriebstechnik.de/pd...-Portescap.pdf
These little motors are indeed the turbo engines among the steppers. A 90 degree turn is faster done than a solenoid piston activation.
A larger disc motor with incremental encoder feedback rotates the head.
Camera vision system for the components is located at 12 o’clock.
Linear motors operate X and Y-axis with the introduction of the X series machines, replacing timing belt drives we used for two centuries.
There is nothing cheap on this machine, a true Mercedes under the pick and place machines.

You can use “vice versa” placement in order to determine offset errors from pick up. (See pic below) Placing parts in an array, one at 0 degree and next in 180 degree. The correction value is half as much as the measured offset.

Yes, the pin transport can create “bad” vibrations on the tape. Consider using a spring loaded 3 mm, 30-45 degree cone shaped pin, which doesn’t insert all the way into the hole.

[RomanLini]

Thanks for the info on the Siplace machines. I went and looked at a few Siemens Siplace videos on youtube and they are VERY nice. Siemens have a really nice design philosophy compared to many electronic manufacturers, they make stuff that is really reliable and works really well, quite a difference from many modern electronic products!

...You can use “vice versa” placement in order to determine offset errors from pick up. (See pic below) Placing parts in an array, one at 0 degree and next in 180 degree. The correction value is half as much as the measured offset.

Good measuring technique! Thank you.

...Yes, the pin transport can create “bad” vibrations on the tape. Consider using a spring loaded 3 mm, 30-45 degree cone shaped pin, which doesn’t insert all the way into the hole.

Hmm, hopefully my guides that don't allow vertical tape movement should save that problem. But based on your expert advice I will polish my stainless feed pin to make it very slippery and see how that goes.

That is an excellent idea too of a sprung feed pin inside a larger pin! The larger pin would hold the tape down when the feed pin retracts, then the large pin retracts last which won't lift the tape. I'll keep your idea for a last resort if my simple pin causes problems, thanks again.

Nice low-cost build thread too Stewi!

[flippersplace]

this thread doesn't have much attention anyway

[flippersplace]

The sprocket wheel and the cover tape are driven by two separate motors. The sprocket wheel advances the tape forward, than turns slightly backwards in order to engage the advancing pins with the rear edge of the perforation hole, this way the peel off force of the cover tape are not moving the tape out of position.
An additional solenoid activated slider covers the component to prevent rejection out of the pockets during fast motion.

Valuable info thanks. I'm constantly surprised how little it takes to jump parts out of a peeled tape

[stewi]

I used to participate in the smtnet.com forum, which may be more of your interest.

[rgee]

hi guys
just a few comments on feeding
i did not have enough time to make a feed pin so i decided on only one nozzle and i used that to feed the tape. as most of the board was 805 this was not a problem. rather than use the sprocket hole i just plunged into the last vacant component hole so i was pulling the tape from the center. i have a dialog box in my Gcode converter software for the size of nozzle and pocket to calculate the movement necessary to accurately feed tape.
for the parts to big for the small nozzle, i just paused machine after placing all small parts and slipped a small piece of silicone tubing over the end of nozzle. one update to machine will be a whole new head with larger nozzle
rather than making changeable nozzles. all large parts were tray picked.
for the most part a small and large nozzle would place most parts.

regarding surface tension for part placement correction, even the best placement by a laser corrected machine will be useless if the board design
is not thermally balanced. example if you have a 805 with the track at the top entering from the side and the track at the bottom entering from the outer side, the part will twist on it's land no matter how well it was placed.
place a 805 cap with one end on the ground fill and a 10thou track on the other end, it is a safe bet it will tombstone. pad design and stencil aperture sizes play a large role here, and some solder paste's are just better than others. that all said there is no substitute for a good placement.
Rgee

[stewi]

Hi rgee,
imagine the component isn't picked up by the nozzle and falls back on to the tape. Upon next pick, you can either damage the nozzle, or the tape gets lodged with the component sticking out of the pocket. I guess, you have a very robust Z-axis, which doesn't bend if the tape doesn't advance.

Tombstone can have many other causes. Since subcontractors don't have any impact on the board design, they often use adhesive dots after screen print, which either cures in the preheat zone, or UV cure before re-flow.
With your home-made pick & place you could use a pin instead of the nozzle, dip the pin in an open container of SMT adhesive (Amicon, Loctite)
and place the pin between the pads. The UV curing glue is best placed, that it is half exposed on both edges of the component.

BTW, guys, are you doing only single sided boards? Don't you require an adhesive applying machine anyway?
What about screen printing? Have you considered applying solder paste dots out of a syringe, using your CNC pick & place machine?
Have you considered using silver epoxy dispensed with your machine? I know silver epoxy is ridiculous expensive, but stencils, solder paste and re-flow ovens are expensive as well.
Curing the epoxy versus re-flow any oven will do and no tombstone effect.

[RomanLini]

Stewi- You should check out Eclipze's build thread;
http://www.cnczone.com/forums/rc_rob...ce_build.html?
he has a lot of clever development there for solder paste dispensing by machine.

Personally I will be using stencils, as I have a precise small CNC mill easily capable of milling the holes in plastic brass or aluminium stencils. Stencils also offer much better throughput as you can be applying paste by stencil while the last PCB is doing the PNP.

Rgee- It sounds like you are talking about centreing issues with home made PCBs? On commercially produced PCBs you generally use the solder stop mask to define the pad areas and careful use of pad sizes and shapes and the correct pad corner rounding gives very nice centreing of the part on reflow. Also the size of the stencil hole and stencil thickness (amount of solder per pad) makes a difference.

[flippersplace]

On Eclipses thread:

I tend to agree with the majority, that vision system may be an overkill.
Most of the commercial machines use board conveyors, which either require mechanical board alignment (troublesome) or optical (fiducial) board recognition.
If we insert the board manually in the machine, we’ll likely use pins in the board mounting holes and get the boards pretty well pre-aligned.
Larger components, except fine pitch, can be aligned with mechanical jaws at the head or in an external alignment station.
Smaller than nozzle components could use a special shaped nozzle, assuming that we get component pick up fairly accurate.
What I would recommend:
Mount a USB microscope to the placement head.
Determine camera to nozzle offset by placing a small blob of silly putty in the placement area.
Lower z axis with nozzle into the silly putty to leave a print.
Zero out X and Y.
Jog the camera over to the nozzle print in the silly putty.
Write down X and Y offset.
If your entire placement program can run with an offset, than you can watch on the monitor pick up and placement. Else, you can modify your program in Excel, adding or subtracting the offset value in pick up and placement.
Make corrections, as required for pick and placements and run your batch of boards.
For inspection after placement, you can run your board program with the camera offset again.

Yeah done all that. The following is simply what I've found in going through all the above (sensible) suggestions..

.. we’ll likely use pins in the board mounting holes and get the boards pretty well pre-aligned..

I used pins for multiple boards but currently only do a single board at a time and align with the USB camera per below as is more accurate (mostly due to my inability to drill peg board holes consistently )

.. Larger components, except fine pitch, can be aligned with mechanical jaws at the head or in an external alignment station...

includes the pain of bump centering

Note again, in my experience, lead free solder does not pull parts into position like leaded solder, (and you will go lead free one day even for your own use) so up your expectations of the no of parts that need centering eventually. Maybe it improve will one day as the solder & paste formulations improve..

.. What I would recommend:
Mount a USB microscope to the placement head.
Determine camera to nozzle offset by placing a small blob of silly putty in the placement area.
Lower z axis with nozzle into the silly putty to leave a print.
Zero out X and Y.
Jog the camera over to the nozzle print in the silly putty. ...

I have the downward USB microscope on my machine. I did the silly putty / clay thing. a bit painful / clumsy / annoying. So now I have an upward looking USB camera as well. The head origin is at zero when the top camera is aligned with the bottom camera.

Calibrate/check a head offset:
- Position the head over the upward camera
Calibrate/check a feeder offset:
- Position the downward camera over the first component in the feeder row
Calibrate a PCB before placement (currently per rotation):
- Position the downward camera over the center of optionally:
- - the PCB cad 0,0 origin of the PCB (more or less the guillotined corner for a coarse setting)
- - a nominated PAD on the PCB for fine adjustment (after previous discussion with RomanL on these threads)

.. If your entire placement program can run with an offset, than you can watch on the monitor pick up and placement....

IMHO no you can't, because the camera will be offset to the head pickup point. The camera needs to be vertical to be accurate unless you guarantee that all pickups & placements are the same height

.. Write down X and Y offset.
If your entire placement program can run with an offset, than you can watch on the monitor pick up and placement. Else, you can modify your program in Excel, adding or subtracting the offset value in pick up and placement.
Make corrections, as required for pick and placements and run your batch of boards. ....

I used to put the head / feeder offsets back into Excel and regenerate the GCode from Excel with fixed values for the head / feeder offsets.

However, if an offset value changes through tweaking / removing / reattaching a head, or even just reloading a feeder row for fussy components, it is annoying to have to edit the offset reference table in Excel & regenerate the GCode etc.

Also, re checking an offset, entering manual coordinates gets old fast.

So now I store each head / feeder X & Y offset in nominated Mach 3 variables. My Excel generated GCode instead of referencing the head X Y offset directly, uses the values stored in the relevant Mach 3 variables instead. This means I can quickly adjust a head / feeder and have it immediately take effect.

Of course I have Mach 3 VB dialog buttons for moving to and setting the offset for each head / feeder / bump pit edge / PCB calibration point so checking/setting alignments is almost as quick and easy as it could be now.

Finally because Mach3 variables are volatile (to my knowledge):

- Before powering the machine down, I can enter any altered head / feeder X Y offsets into the reference tables in my spreadsheet.

- On power up, Generated GCode for each PCB includes GCode macros that only if called (again, by a Mach 3 VB Macro Dialog button, repopulate the X & Y offset Mach 3 variables for each Head / Feeder. (But I'm fundamentally lazy, so when I'm dong a run of the same PCB across days, I just leave it running)

My current Mach3 variable mappings:

Mach3 Variable, Offset & Macro allocations

A working generated GCODE:

[flippersplace]

Stewi, a question.

For components supplied in reels:

If I get a new component I add it to my schematic library and map that to its PCB footprint.

The PCB footprint has a rotation as defined in PCB Cad package which is is by default the 0 rotation for any subsequent Pick Place export

Is there an accepted standard for the position of each part at the 0 rotation based on its original rotation in the component tape.

I.e. Is is 0 rotation per the feeder coming from the left for example?

Thanks.

[stewi]

As far as I remember, it has been defined as a standard that a component picked up at 0 degree and placed at 0 degree is not rotated by the head if the board uses the standard math coordinate system with the datum lower left corner, X pointing to the right and Y to the top.
So, I would expect that a chip is placed parallel to the X axis at 0 degree placement angle.

[flippersplace]

$400 bucks! Ouch.

http://www.cotinc.com/smt_site/downl...er%20Flyer.pdf

[stewi]

On Eclipses thread:
IMHO no you can't, because the camera will be offset to the head pickup point. The camera needs to be vertical to be accurate unless you guarantee that all pickups & placements are the same height

Translate IMHO for me, I'm only a semi US citizen.
Siemens used to have a machine with board load at the front and feeders in the rear. If you apply the same offset to the feeder bank, than you would see the pick up.
Also, the very first Siemens machine had the capability to look through the hollow shaft Z-axis with a fiber optic camera. We could insert a glass with scale marks in the jaws and used this feature to teach PCB positions. In this case, you would not have to deal with any offset, regardless from which feeder bank you would pick.

However, Eclipse was looking for a method to move the board and align it, not using pegs. I'm used to an aluminum hollow shaft Z-axis and would be scared to death moving x and y while z is down.
Also, I do assume you do double sided boards as well, so, instead of mounting the PCB to a flat plate, you would rather clamp the board on the edges.

I got somehow fixed on the idea using the nozzle vacuum sensor to find the board position and correct all placement data on the fly (I would have to do that anyway if I would use a vision system).
In this case the regular nozzle would pick up an adapter slightly larger than the board holes, 4 mm diameter or so. There is a small hole of 1mm diameter in the center. Since the adapter is larger, the nozzle adapter will not insert in the board holes. Now, you move X and Y over the position of the first board hole and lower Z. You shouldn't get any change in vacuum, while air flows through the board hole. Move x step by step until the edge of the hole will block the nozzle hole. Do the other edge. Do the same in Y to get the hole center and a second board hole. I think it works. Haven't done the math yet.

[stewi]

$400 bucks! Ouch.

http://www.cotinc.com/smt_site/downl...er%20Flyer.pdf

I'll let them know.

[Eclipze]

I'd use the SMD tape index pin for moving the board around, rather than the placement nozzle. But I had another thought, that uses standoffs for the board. This would facilitate double sided boards. One corner would have a standoff post with a magnetic base. This would be aligned first, then the opposing corner aligned second... whereby it would rotate about the magnetically held first post.

This would probably need to be expanded to be a bit more universal... such as the standoffs having a 90 shelf to reference from the corner of a PCB, given the PCB doesn't have a suitable hole etc...

Perhaps with multiple posts using a magnetic base, boards could be successively loaded (gently) without need for re-alignment.