Nice! I forgot to mention that you will have to deal with some flex in the system...
Seriously though, in this case most of the flex is caused by the rubber piston and your improvised motor mount. Those wings do flex on both ends, the motor and the syringe.
Yes, the syringe wings were a major cause of the flex, as was the rubber piston. In addition, there was a lot of hystersis between the rubber piston and the syringe piston.
Made the camera mounting bracket. Also intended to machine the camera side of the bracket, however it didn't quite go as planned. Need to redo the cam code for it.
I wasn't really happy with the mechanism for the tape index pin. That's the servo that just pushes a rod down so the gantry can pull the tape, indexing the next component. Despite the simplicity of the design, it was being difficult.
So I've changed the design to use a silicon tube as a flexible mating linkage, which as worked out superb! Bit more effort to machine, but it's more adjustable and flexible to get the smooth range of movement I want.
It pretty darn small. The bolts are actually 2mm bolts. The tubing used has an inside diameter of 2mm. I used a tiny length of brass tube for the joint to rotate about, very nice fit, and another very small larger diameter tube to offset set it's position from the cam plate.
New approach needs a new mounting bracket, which also mean a change in mounting points to suit the new position (doh). I like this mechanism far more than the rod system though. Apart from the fluid movement, it's easy to adjust the length.
Think twice about using a glue dispenser. One of the saving graces of reflow soldering is the surface tension of molten solder tends to pull misaligned parts into near perfect position. Gluing the parts "locks in" the parts to whatever your positioning error may be.
Secondly, glued parts are almost impossible to adjust or replace using a hot-air rework station. Glue is normally used only if you are placing parts on both sides of a board.
Mariss
I have no plans to have a glue dispenser at all. Only solder paste dispension. I find I can reflow a second side without using glue for small components. They don't fall off or move, and I'm using the same melting temp paste for both sides.
The solder paste dispension is for prototype work. If I'm doing a few boards then I prefer to stencil the paste.
Mariss... I must add, I appreciate the performance from the Gecko G540 unit. Nice bit of kit!
Eclipze,
Thanks for the nice comment; I really appreciate it. For what it's worth, I hand assemble the first prototype of everything we manufacture. It's the quickest way; I build the first one, machines build the next 10,000.:-)
What I use is a boom-mounted stereo zoom microscope, tweezers and a hot-air rework hand-piece. I first spread no-clean paste flux over the board and use a pencil soldering iron and 0.020" dia rosin core solder to "bump" all SMT pads. I then use solvent to clean the board.
Next, I spread some no-clean paste flux on a scrap piece of metal like a G201 cover. I fire up the hot-air hand-piece and use tweezers to dip an SMT component in the paste. Then, using tweezers, I hold the part in position on the board while I hit it with the hot-air under the microscope. This takes about 3 seconds per 0603 part and about 10 seconds for an SOIC-14. A 48-pin QFN package may take 25 seconds.
Most boards have 150 or so SMT parts so it takes about an hour and a half to hand assemble a prototype. This time takes into account head scratching, coffee sipping and admiring my own handiwork.:-)
It takes about four hours to load the parts trolley and program our pick-and-place machines. Since I don't trust any of my prototypes at all (I know who designed them), this setup could be wasted time. Once the miserable thing gets tortured into showing signs of life, the pick-and-place gets programmed. It then happily runs off copies every 50 seconds.
Mariss
I'm stuck in the realm of niche market small runs. Some items I only make 50 units, others only a few hundred. But in a lot of those cases it's not until you've sold the first 100 I consider getting the boards loaded. Too expensive to take the punt early for small run boards, and I don't always have all the parts on reels to be machine loaded. So I end up at least loaded 50 boards myself.
I have tried hot air, but after a few boards the paste start to smell a bit. Why they can't make a different flavour is beyond me. For prototypes the hot air is good, but for any production I get concerned about how hot and for how long the PCB gets with one component at a time. The SMT paste dipping process however does sound like a neat trick! Now that I can machine my own stencils, small runs are less tedious. It makes it a lot quicker and more accurate, and importantly the reflow is done under the exhaust fan. However loading the components is still time consuming. Is coffee a tax deductible consumable?
I could justify the cost of a proper surface mount loader. Don't have the volume to pay for it. But I only have a limited music library to keep me going while tweezer loading. Hence decided to spend a big chuck of time making this fun project - a little more stimulating. There is no real proper business justification into putting so much time into such a project. But I'm having a lot of fun in the process, so have ignored common sense. Hopefully with what I learn I'II become more proficient with manufacturing production aids.
I try not to have more than one prototype every month or so. The hand placing is so onerous it takes that long to get over the revulsion. Dental root-canal work seems like a pleasant alternative by comparison.
I use a hot-air rework station that has a presetable temperature and I always use 63Pb37Sn (leaded) solder for prototypes. 178C is much easier to work with than lead free at 228C. Things do begin to smell funny above 250C.:-)
By solder "bumping" I prep the board beforehand with solder on each SMT pad to leave a convex shaped solder meniscus. That way there sufficient solder already there when I flux, place and hot-air each part. I remove the tweezer about a second before removing the hot air. Surface tension then pulls each part perfectly into place.
Mariss
I've drilled and tapped eight points on the Z-Axis carriage, so I can now reduce the free play in the bearings with set screws. Actually run out of darn set screws, but just with 4 fitted it makes a big improvement. Adjustment will be difficult, as it's barely an 1/8th of a turn between free play --> ideal --> binding. I'II be thread locking them in place so they stay put. The Y-Axis is pretty good already, but there is a little movement in it. I might install set screws on the sides to remove the small rotation free play... all which contributes to the accumulated placement error. The long X-Axis carriage already has adjustment mechanisms.
A request for Mariss
Dear Eclipze: Please forgive me - I'm not hijacking your very interesting thread,
which I'm following closely, but in face of posts #66 & 68, by the
great Mariss, I think this is the proper place to post this request:
Dear Mariss: Your two posts above were very interesting, since I'd like to try my
hand at mounting SMD prototypes by hand, and nothing better than following
the methods used by "the master" himself. However, I'm left with dozens
of doubts and questions, so I'd like to ask you a big favor:
Would you be willing to start a new thread, describing your hand-assembly
methods, perhaps with a few photos and may be even a small video?
I know a succesful businessman like you must have very little time to
spare, but you've been so generous in your contributions to these forums
that I got the courage to ask...
Among my questions, that were raised by your posts, let me ask:
- Could you disclose maker, models and suppliers of your tools and supplies (I don't
think this would violate forum rules, right). In particular:
. "hot-air rework hand-piece" - I only found very expensive devices, costing more
than a thousand dollars; is there a model affordable for an amateur? What is the
size of the nozzle you use?
. "pencil soldering iron" - wattage? shape and size of the tip?
. "no-clean paste flux" - specs? type of dispenser: is it in a syringe? I understand
this is only flux, there is no solder paste in it, right?
. "boom-mounted stereo zoom microscope" - is there a model affordable for the
amateur?
. "tweezers" - I understand these have to be high quality, specially made for this
application, and probably you need more than one model/size, for different SMD
packages; could you elaborate?
Finally, what kind of PCB's are you hand-mounting - probably, you use only professionally
made PCB's, with some kind of surface treatment, rather than bare cooper, right? could
one use homemade PCB's, CNC-routed or made using tone-transfer?
By the way, I believe you meant 63Sn37Pb, instead of 63Pb37Sn - right?
Thanks a lot for your patience.
Nelson
Last edited by Arquibaldo; 01-29-2010 at 06:08 PM.
Arquibaldo,
Good suggestion! I have opened a new thread in the general electronics forum called "SMT the old fashioned way".
Mariss
P.S. I like to think of myself less as a "succesful businessman" and more as a practicing hands-on EE, floor sweeper and bottle washer.:-)
One wonders whether there isn't some clever way to at least CNC the solder bumping. The board is a known quantity from a CNC standpoint. There's no need to load reals of components. You still have to deal with manually placing the parts and applying the hot air, but perhaps a small increment of the work prototyping would be saved.Originally Posted by Mariss Freimanis
Cheers,
BW
Try G-Wizard Machinist's Calculator for free:
http://www.cnccookbook.com/CCGWizard.html
There are semi-automated placement machines available. The component wheel spins to reveal the component bucket to pick from. You pickup the component using ganty that you push by hand. Pickup with suction, take to the position as shown on the screen, rotate with the finger wheel and drop in place. They can have camera for easier vision, helps steady your hand, supports trays and even tube loading and also paste dispension. Quite expensive, but along the theme you are thinking.
Machined a new platform for the Y-Axis, as well as the servo mounting brackets. Both to support the new index pin servo design and also because I had some refinements in the design. The index servo bracket is also done. Assembled and all good!
Pictures below are the pulley brackets for the X-Axis.
I've done a fair bit of work re-designing the pickup head arrangement. Looking to improve accuracy and also to support head changing.
The solder paste needles are not manufactured with precision and it would be a problem to get them seated on the nozzle so they were straight... so they don't draw circles in the air when rotated for component placement. So I'm only going to use the metal needle tube from those now, and embed them in another part.
I've been waiting on some ring magnets, which have arrived. I'm going to use these to support changing the heads. It also doubles as a safety mechanism, whereby if the head collides with something at high speed it will simply be pulled off. Rather than break. The rings magnets provide the holes for the vaccum channel, and the head will have a tapered aspect so the head is pulled into accurate alignment.
I won't have an automatic head changer to start with... that's a future upgrade. But at least the system could be setup to pause and wait for the head to be changed manually, which will be quite easy given it's just a magnet holding it in place.
The new design also has two bearings to support the mechanism now. May be overkill, but it will avoid the tethered vacuum tube from effecting the alignment with a single bearing. The bearings came in a pack of two, so might as well use both :-)
Wow, you're really making progress there!
If the machine works well enough, you could even make some business out of them. I think lowest cost P&P machines cost over $10k today. Designing something incredibly price efficient is a talent and definitely appealing.
I've been working at the new pick up head design. I'm trying to incorporate the ability to have an automatic tool changer (ATC), by having a nozzle that can be easily removed. I have also liked the idea of having the nozzle "pop off" on collision, to avoid damage.
I initially started with a ring magnet on the adaptor and nozzle, with an o-ring seal. Using ring magnets as they have a hole in the centre, and use two (instead of one side have a metal washer) because I found they automatically snapped together with a higher initial degree of alignment. Using the o-ring seal required vertical or horizontal compression, and wasn't too confident. So I when with tapered mating surfaces.
The diagram below shows the approach I took. I machined a couple of parts just to get an idea on how well they would mate. Unfortunately if feels like more of a ball joint than a locked position. It also leaks too much air.
I'm planning to increase the slope of the taper from 45 degrees to 15 degrees, which will hopefully achieve a positive alignment mating lock. Then having and o-ring to seal on a vertical edge between the adaptor and nozzle. I might loose the pop off feature I wanted with a long taper.
I'm wondering if anyone had any ideas or comments on how to approach this. Keep in mind, I really need to machine the nozzle from the same face, to ensure accuracy between the hole milled for the needle through to the mating surfaces. If I have to flip the part to machine again, then it's likely to introduce an offset error.
Eclipze-
First, great engineering & machining effort - the design is coming along well!
Regarding the taper, a true "self-locking" taper doesn’t occur until you are under about 4° or 5°, so 15° should not self-lock too much. Self-locking Morse tapers are about 2.5°.
Looking at the cross-section view and your magnet placement, I would suggest getting the magnets as close together without touching as possible. The force greatly diminishes with distance. Also, adding properly shaped steel behind the magnet rings will greatly increase the attractive force. A "U" shaped cup like this one:
http://www.rockler.com/product.cfm?page=10713
or an assembly like this one:
http://www.kjmagnetics.com/products.asp?cat=110
The steel cup effectively redirects the magnetic field from the rear of the magnet back around to the front, adjacent to the other face and concentrates both the north and south poles to one face. If you make matching pairs of these magnets that will attract, this should greatly improve performance.
Also, if you are applying a positive pressure thru this joint, you should do everything you can to reduce the diameter of the o-ring. This will reduce the effective "piston" diameter and decrease the force that the magnet needs to overcome and seal.
You may want to look into "Quad" o-rings:
http://www.mcmaster.com/#o-rings/=5qaynd
which have double sealing surfaces and require far less force to create a seal.
Or, a piston cup (not the one from the movie CARS) could be used so that it expands when pressure is applied and helps it seal better:
http://www.mcmaster.com/#u-cup-seals/=5qaz9r
Keep up the great work!
Keith
Last edited by NEATman; 02-08-2010 at 12:48 PM. Reason: Typo's...
Thank you for your comments NEATman :-) I actually found the magnet force from the ring magnets I have to be too high. I can hover one 30mm above another and it will pick it up. I currently have a 2mm gap to reduce the force, as otherwise it would make it too difficult to remove the tip, particularly for auto tool changing. But otherwise, forcing the nozzle attachment through an o-ring shouldn't be an issue if the Z-Axis drives it through.
A self locking morse taper is getting close to a vertical wall at 4 to 5 degrees. I will have to pay more attention to this aspect. This information is quite helpful! I was somewhat hoping the taper would be good enough that a film of vacuum grease would be all that was needed to seal... rather than needing an o-ring. But I'II most likely introduce an o-ring into the next design.
I would not bother with o-rings or grease to seal the vacuum leaks and insteaf focus on the alignment and centering part. Since you producing vacuum on demand, it is okay to leak some.
On this ATC nozzle system that I'm using (from Madell), there is no O-rings of any kind and the fit and finish is not leak free. Still works quite well. Only one ring is a magnet, the one on the motor shaft. The nozzles themselve have a regular bushings (white metal on brass body) which looks like SS 440 or something similar. Imagine what would happe when you put several nozzles with magnets on the same nozzle rack. Will they attract/move in their sockets? Probably yes. It think this is why Madell is is only using one magnet on the shaft.
"http://www.youtube.com/watch?v=bYfD0UUKyek"]YouTube- Homemade SMT pick and place machine #3 Nozzle Changer and plastic chips
Posting Permissions
Reply with Quote