Is someone interested in building a SMD pick and place ?
I mean a little, not-so-fast, not for very small package pick
and place.
Any idea about hardware and software is really appreciated !

I am interested.. I already started something but the results are really not very good.. I made a "bindomatic"... The space between the bearings blocks is too big and it just binds all the time. I will have quite a hard time to fix it without spending much more money than I already have..

Do you have any clean ideas in mind ? The software part will not be the hardest if your PCB program can export XY coordinates for the parts centering.]

Best regards,
Alexandre

I am interested.. I already started something but the results are really not very good.. I made a "bindomatic"... The space between the bearings blocks is too big and it just binds all the time. I will have quite a hard time to fix it without spending much more money than I already have..

Do you have any clean ideas in mind ? The software part will not be the hardest if your PCB program can export XY coordinates for the parts centering.]

Best regards,
Alexandre
Cute....a bindomatic... :banana: you can probably patten it...but seriously, I suspect that you have an alignment problem, or is your base is not flat & stable to the movement that will torque your rails out-of-alignment.

Are you trying to place electronic components onto pads for some vapor soldering? If so then it's more then just simple x-y...there is component size, type, orientation, alignment to the pcb's datum lines, etc.

Cute....a bindomatic... :banana: you can probably patten it...but seriously, I suspect that you have an alignment problem, or is your base is not flat & stable to the movement that will torque your rails out-of-alignment.

The "bindomatic" new word is something I read at one of John "cranky" ********** message somehwere.. It really defines my first go on the machine very well...

I used Igus bearings and the guy that made the mechanics for me ignored my warnings that the bearings should be at 1/2 the width of the space perpendicular to them as Igus clearly states for their bearings. It will just keep binding until that rule is covered or I get other kind of bearings.

Best regards,
Alexandre

Are you trying to place electronic components onto pads for some vapor soldering? If so then it's more then just simple x-y...there is component size, type, orientation, alignment to the pcb's datum lines, etc.

They will go into a reflow oven. Component type and orientation is also part of most BOM generated by PCB's programs. And for small productions that could even be entered by hand without much trouble. If someone can make me do a nice machine I can come with the software that will read the info from a text file and move the steppers around to make it all work.. My problems really are with the mechanics and specially the lack of good mechanical tools. I would love to see some simple and fast stuff that could be produced with just hand tools.. Or something that I could just submit to a machine shop and have the parts done easily. This machine is different from a router because it has to be much faster and do not have to cope with cutting forces.


Best regards,
Alexandre

I think we can cut a big problem in some more smaller problems, and so on till we find a good, working, inexpensive solution. Is this way OK for you ?
The machine is SOFTWARE + HARDWARE.
Software can be a problem for me, but probably not for Alexg, so first problem is solved
Hardware: we can cut in 3 parts: the X Y Z plus Rotation " plotter",
feeders, and the most complicated part, the "head" with all the pick and place goodies.
Alexg: can you tell us more about your experience?
Thanks to all, we are a good team.
Paolo

I think we can cut a big problem in some more smaller problems, and so on till we find a good, working, inexpensive solution. Is this way OK for you ?
The machine is SOFTWARE + HARDWARE.
Software can be a problem for me, but probably not for Alexg, so first problem is solved
Hardware: we can cut in 3 parts: the X Y Z plus Rotation " plotter",
feeders, and the most complicated part, the "head" with all the pick and place goodies.
Alexg: can you tell us more about your experience?
Thanks to all, we are a good team.
Paolo

Ok.. Let's forget the software by now... It is not trivial but more than easy enough to solve when the machine is operational..

The Z, rotation and vacuum pick and place is also quite easy. What I made was using 2 small steppers with screws attached to them that go on a Igus Drylin N "shuttle". The top one drives the "shuttle" up and down and the bottom one is connected to a small vacuum pickup tool and makes the rotation.

X and Y should be quite fast to be usefull..

In my humble thinking we can survive with about 0.3 mm of combined precision. That should not be a big problem..

Another problem to be dealt with afterwards is how to take the components from their original rools. But that is a hole new story...

Best regards,
Alexandre

I have just received from Distrelec a " P 830 vacuum pick up ":
it does NOT work, suction cups are not precise, and move easely.
What do you suggest ( and are using ) to pick up the SMD?
What vacuum pump do you use ( refrigerator or bigger pump) ?
The suction cup is my biggest problem, I do not think to have
problem with the " plotter", I will use FLEXLINK or BOSCH RETHROT
alluminium and accessories, plus SYNCHROPOWER belts, as used in
my Bungard PCB driller : simple, noiseless, fast, and , after all, inexpensive.
If needed, I can send you pictures and RS catalogue part numbers.
How are built suction cups in commercial pick & place ?
Any picture of simple " squaring station "?
Paolo

I have just received from Distrelec a " P 830 vacuum pick up ":
it does NOT work, suction cups are not precise, and move easely.
What do you suggest ( and are using ) to pick up the SMD?
What vacuum pump do you use ( refrigerator or bigger pump) ?
The suction cup is my biggest problem, I do not think to have
problem with the " plotter", I will use FLEXLINK or BOSCH RETHROT
alluminium and accessories, plus SYNCHROPOWER belts, as used in
my Bungard PCB driller : simple, noiseless, fast, and , after all, inexpensive.
If needed, I can send you pictures and RS catalogue part numbers.
How are built suction cups in commercial pick & place ?
Any picture of simple " squaring station "?
Paolo

I am planning on just using the table for small components... Not for the big ones that need the suction cups. I use a aquarium pump modified to make vaccum for the manual pick and place I use today and "flat" hypodermic needles for picking up.. It works very nicely for manual application and should be enough for the machine also. The area is very small, the vaccum pump does not need to be powerfull..

I think I can get Bosch stuff down here in Brazil also.. Not cheap but maybe be easier than other options...

Best regards,
Alexandre

Can you please tell me more? Did you simply "cut" the needle?
I will test as soon as you answer me !
A commercial "suction nozzle" is about 150 USD !!!!!!
Paolo

Thank you for the GREAT idea, Alexandre !
I could not wait, so I built the gear with a flattened needle and
an aquarium pump ( preiously used for the PCB bubble etcher )
and a piece of tube: it works great!!
I'm here in front of my pc, answering you with one hand, and with the other
shaking with decision the needle with attached a SMD 1206 resistor:
it stays attached to the needle without problems!
Now I'll find a small 200 step stepper, and I'm thinking the best way to
attach the needle to the stepper, then to the Z axis.
Is a good idea to use a solenoid for Z axis as in LPKF pcb drillers ?
Paolo

Thank you for the GREAT idea, Alexandre !
I could not wait, so I built the gear with a flattened needle and
an aquarium pump ( preiously used for the PCB bubble etcher )
and a piece of tube: it works great!!
I'm here in front of my pc, answering you with one hand, and with the other
shaking with decision the needle with attached a SMD 1206 resistor:
it stays attached to the needle without problems!
Now I'll find a small 200 step stepper, and I'm thinking the best way to
attach the needle to the stepper, then to the Z axis.
Is a good idea to use a solenoid for Z axis as in LPKF pcb drillers ?
Paolo

I cut the needle with a sharp "xacto" blade rollingundefined it back and forth slowly to get it as flat as possible and if necessary I use very fine grit sandpaper to finish it up. www.smallparts.com has some needles already flat. They are nice..

The Z axis motors do not need to have 200 steps at all !!! You do not need much precision for rotation and the up and down movement has much precision added by the driving screw. I am using a small motor that was originally used to position a tape head in a streamer backup unit.. A solenoid on the Z axis would be a bad idea... Even resistors have different heights depending on the value and maker. Capacitors are much worse. We need some precision there.

The attachment of the needle to the rotation head needs something done on a lathe... I can't imagine other way.. The format would be something like this:

-----------------------
| |___
\
\ /-----
\ /
------- -------
\ /
\ /


Bad ascii art.... :-)

If you can get me a easy and cheap way to do the XY movement I can solve most of the other problems ;-)

Best regards,
Alexandre

After some expensive experiments with CNC, I will use:

FRAME: Flexlink aluminium structural system, does not require milling,is a sort of " Lego" game, just cut and go, with lot of accessories,angle, brackets,fixings, etc.
MOVEMENTS:it has to be lightweight,fast, easy to move,precise, easy to build, inexpensive: 3 or 4 small V-grooved bearings running on 2 shafts, there is an alluminium accessory to fix 2 shafts on Flexlink lateral T slots, they can move with a breath of wind ( try this with 4 linear bearings... )
POSITION CONTROL : no expensive and noisy ballscrews, but easy to install, noiseless,inexpensive Synchropower timing belts with related timing pulleys
( a professional version of movements used in every ink-jet printer)
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Milling and lathingis are expensive, time consuming and sometimes not very precise; consider the use of some extruded structural aluminium similar to Flexlink, after all you can save money and time and have a very precise machine.
A timing belt is very precise and ( very important for us) allows some small misalignement ( users of leadscrews know what I mean ! )
and no backlash at all. You can see Flexlink also on www.rs-components.com
and there are many similar aluminium systems.

Do you use a stepper on Z axis ? No problem if feeders and PCB are not very well leveled?
About needle :your plastic tube rotates 360 degrees , or 180 degrees ?
Will you consider the use of an IC as pressure sensor ( to know if something goes wrong, and to gain some speed when picking the SMD component )?
How small are your steppers?
Paolo

Do you use a stepper on Z axis ? No problem if feeders and PCB are not very well leveled?
About needle :your plastic tube rotates 360 degrees , or 180 degrees ?
Will you consider the use of an IC as pressure sensor ( to know if something goes wrong, and to gain some speed when picking the SMD component )?
How small are your steppers?
Paolo
Yes, one stepper for up and down and another one for rotation. The neddle actually rotates 270 degrees. I have not got to the point of actually picking up the components :-( My idea is to have some flexibility on the needle coupling to cope with the small differences in the level of the board and feeders. In the first version I am planning on using just cut pieces of the components tapes. Nothing fancy...

Best regards,
Alexandre

Just arrived from the famous "my friend with a lathe".
We tryed to drill the axis of a small stepper ( we can !)
to have an axial output for the air-vacuum from the needle
( pratically: the needle is on the axis of the stepper, the axis is drilled,
the tube is connectet to the top of the axis, coaxial to the needle, soory, my
english...). We think to use a small stepper normally used as carriage
driver in ink-jet printers.They are the smallest steppers with bearings I can find. We want to try to use some elicoidal spring to push needle against PCB and feeder ( Z axis), to avoid problems with not-so-leveled PCB and feeders.
Z axis will have gear and pinion, to gain speed.
No idea for the squaring station, but it is not important now, till I use 1206 SMDs.
I hope to find ( for free ) a refrigerator pump, just to see how much void it can create.
SMD pick and place....not so easy, but lot of fun !
Paolo
PS Is there a way to exchange pictures ?

Just arrived from the famous "my friend with a lathe".
We tryed to drill the axis of a small stepper ( we can !)
to have an axial output for the air-vacuum from the needle
( pratically: the needle is on the axis of the stepper, the axis is drilled,
the tube is connectet to the top of the axis, coaxial to the needle, soory, my
english...). We think to use a small stepper normally used as carriage
driver in ink-jet printers.They are the smallest steppers with bearings I can find. We want to try to use some elicoidal spring to push needle against PCB and feeder ( Z axis), to avoid problems with not-so-leveled PCB and feeders.
Z axis will have gear and pinion, to gain speed.
No idea for the squaring station, but it is not important now, till I use 1206 SMDs.
I hope to find ( for free ) a refrigerator pump, just to see how much void it can create.
SMD pick and place....not so easy, but lot of fun !
Paolo
PS Is there a way to exchange pictures ?

Hi, Paolo

Using rack and pinion for the Z is not a good idea in my humble opinion. We do not need speed there, it will just move some milimeters to clear the right of the SMD components !! And we need precision there to handle different component heights.

If you couple the needle with a small piece of silicone tube you will probably have enough compensation for misslanignents and height differences..

The idea of making a hole on the stepper axis is a little too radical to me :-) It maybe a good idea but how to make a hole that small ? And what about the debris ? And the heat inside the motor ? I was thinking more in the idea of using something in front of the stepper that will hold the needle and have a hole in one side to get the vaccum tube in..

I think we can attach pictures to the messages or create a picture folder in the photo gallery.. I will try to take some pictures of my "bindomatic" junk this weekend..

Best regards,
Alexandre

Z axis: do you think to consider component high with software?
In any case I think to have some sort of " backlash" ( say 1 mm )with a small spring to counterbalance height.Can I have high precision with a piece of
silicone tube from motor to needle (how your experience ) ?
A small hole in the stepper axis is not a problem wit a lathe , and helps to solve the problem for +-180 degrees rotation.My problem is: is my stepper
used for rotation SMALL ( 39x39x32mm )enough ( weight) ? how small (mm) is your? Does it have bearings? Can you simulate time required for this part of the cycle( Z axis goes down, then sucks the SMD, then goes up again )?
What kind of movement are you using on Z axis ?
Paolo

Z axis again: what kind of adhesive do you think to use?
Has the machine to PUSH the SMD over the adhesive, or just to let it drop over? This will tell me if I ( we) need a small sping or not on the Z axis.
.....an adhesive dispenser on the same axis....no !!!!!!
This is really R & D.
I'm lucky to be the second builder, I can have a lot of infos from you !
Again
Paolo

Z axis again: what kind of adhesive do you think to use?
Has the machine to PUSH the SMD over the adhesive, or just to let it drop over? This will tell me if I ( we) need a small sping or not on the Z axis.
.....an adhesive dispenser on the same axis....no !!!!!!
This is really R & D.
I'm lucky to be the second builder, I can have a lot of infos from you !
Again
Paolo

Hi,

You just apply the solder paste before putting the parts on the board. The solder paste surface tension is enough to hold the components in place. No need for glue for small parts.

You can just cut the vaccum and the solder paste surface tension will hold the part in the right place.

My machine did not work.. I may be a bad example to follow :-) All the things I am telling you comes from experience of mounting thousands of SMD boards using simple hand tools...

Best regards,
Alexandre Guimaraes

You say: my machine did not work.
This is really NOT a problem. Your first prototype is a mine of ideas and
suggestions for a working machine: we will built it.
....and now some more stupid ideas of mine: why not put a pressure sensor on the vacuum tube? and a sort of potentiometer on the small compression spring on the " go down" Z axis? Just imagine:
PICK: Z axis goes down
Vacuum pump is ON
pressure in the tube is low ( but not VERY LOW)
when the needle reaches the SMD ( level does not matter , no level to program ) pressure goes VERY LOW, Z axis stops and goes
up
TRAVEL from PICK to PLACE position: if we loose SMD pressure goes LOW ( but no VERY LOW ), so we can have some alarm
PLACE :Z axis goes down
when SMD is on the PCB the small spring starts to compress, we know it from the potentiometer connected to the spring
motor stops, vacuum pump is OFF, motor goes up, no problems with level, no level to program.
Just an idea of mine, you know, I'm a little mad.....
........................................................................................................I have found the higher SMD component on the RS catalogue ( ...the Bible...)
it seems to be :aluminium low ESR series VFC 105° Panasonic capacitor:
the bigger ( for example 10V 470yF 0r 35V 47yF) is 10.2 mm high:
this means we need a max Z axis travel a little more than 10.2 X 2, I will
go for 25mm: what do you think ?
I think Z axis with all the accessories is 90% of the machine.
the problem is the rest is another 90%......
Regards
Paolo

Alexandre,

I know what you mean, hand-mounting SMT parts. I have to hand-mount 0805s, SOICs, SOT-23s, etc. on prototype boards. Once they work, we order-up a stencil, paste the panels and put them in the Zevatech FS-730s which places parts to the tune of 3 per second.

Until that happy point it's a pair of needle-sharp tweezers, a Nikon stereo zoom low-power microscope, a Hakko 851 hot-air rework station and a very, very steady hand. Thank God for solder surface tension; it corrects a lot of near-misses.

Mariss

Dear Mariss, can you make us a present? Some detailed pictures of the head
of your pick and place ? We will really apppreciate it. And any suggestion from a user is very important in this stage of developement:
We count on you, Mariss
Thanks
Paolo

...adding a turbo and some nitrous, maybe, we can.......

You say: my machine did not work.
This is really NOT a problem. Your first prototype is a mine of ideas and
suggestions for a working machine: we will built it.
....and now some more stupid ideas of mine: why not put a pressure sensor on the vacuum tube? and a sort of potentiometer on the small compression spring on the " go down" Z axis? Just imagine:
Paolo

Hi, Paolo

The pressure sensor is how it is done in comercial machines.. I think it is a great idea but.....

Nowadays I really became a believer in " Evolutive " development :)

Lets get the thing to work and then we can add many gadgets to make it better and faster. If we sophisticate too much at this stage we will not acomplish anything real because we can talk about it during months... And besides it gets to a point when the homebrew gets so close in price to the comercial machines that it is not worth it anymore. We have to keep focus on a machine that is a replacement for hand placement but not a replacent for the great pick and place machines in the market. If the volume goes up it is better to buy a ready machine.

I will grab a picture of my junk machine... And we can get ideas from my many mistakes...

Best regards,
Alexandre Guimaraes

Alexandre,

I know what you mean, hand-mounting SMT parts. I have to hand-mount 0805s, SOICs, SOT-23s, etc. on prototype boards. Once they work, we order-up a stencil, paste the panels and put them in the Zevatech FS-730s which places parts to the tune of 3 per second.

Until that happy point it's a pair of needle-sharp tweezers, a Nikon stereo zoom low-power microscope, a Hakko 851 hot-air rework station and a very, very steady hand. Thank God for solder surface tension; it corrects a lot of near-misses.

Mariss

Hi, Marris

Nice to see you in this thread.. Lots of real World experience with the pick and places machines combined with the best drivers for the price that I have ever heard off :cheers:

Placing by hand is a PITA most of the times and the mistakes are inevitable. You can imagine my situation. My small company produces about 10 to 15 boards everyday with hand placement. I do not have enough money available to buy a real machine... So the easiest path available is make a machine.. My idea is to make it really easy to program and use for small runs.. Maybe even for your prototypes :) What program do you use to design the boards ? Can you extract a small portion of the BOM with the placement information ? So I can check if it is similar to the one I have ?

Other function of the machine will necessarily place solder paste with a pneumatic dispenser. This way we can eliminate the need for the stencil for short runs. The pneumatic control is already working and I plan to integrate everuthing together once I have a machine that really works... And not the crap I made...

Best regards,
Alexandre Guimaraes

Uh, maybe that's not a good idea...

The machine is a 5' by 5' by 5' cube and weighs 3,000 lbs. It is a 3-head, gantry type machine. The heaviest part (X axis) probably weighs 100 - 150 lbs, moves at around 4 feet per second and accelerates at around 4 - 5G.

The machine is bolted to the floor. Were it not, it would chase me all over the shop like an insane, badly loaded washing machine. You don't know fear until you have had a 3,000 lb machine come looking for you. Ask me how I know.:-)

Mariss

Here is my material for Z axis test
We will try to build a longer axis for the " rotation " stepper, to put
needle on the lower side and an attachement for tube in the upper.
Paolo

Hi, is there someone on this web site?

Yes, why???

Yes, why???
Can someone help me with the following info?
squaring station : I need info and pictures
suction needle: internal diameter of the most used
suction needle: what is the value of vacuum ?
Thank you
Paolo

Can someone help me with the following info?
squaring station : I need info and pictures
suction needle: internal diameter of the most used
suction needle: what is the value of vacuum ?
Thank you
Paolo
Vacuum is generally measure in inches of mercury. I'm not sure about the size of the pickup needles most are very tiny...in the range of an inside diameter of 0.015 to 0.03, the ends have cups that are very pliable....I suspect they are a silicone.

I do not know if this helps or not, but I've worked with pick and place and chip shooters for over 7 years. One of the big things to consider is that on all the machines I've ever worked with the the machine will turn on the vaccum for picking up and carrying of the componet. But as the machine places the componet vaccume is reversed and the componet is helped off the nozzle with a gentle burst of air. Very important with smaller 0402's 0603, 0201's.

Nozzle size is dictated by what your going to try and pick up. The larger the part the larger the nozzle. Nozzles material varies much between differnt manufactures as does the pattern a nozzle can have. Panasonic for some of the smaller nozzles used a diamon epozy tip that supposedly lasted longer. The opening was an "X" shape. But that was for the smaller chips. After that they went over to straight tubes. For the really large parts you could get into custome nozzles down to robotic gripper like arms for some connectors.

Fuji, the nozzles were aluminum, but they would mushroom over time. So the switched over to stainless steel. I still have a windmill off a chip shooter. Missing a nozzle or two for it but I'll try and post a picture of it Friday. I have finals tommorrow and should be studying.

Their use to be a company called manncorp
Did a goggel for them. They are still in buisness. http://www.manncorp.com/

They use to put out a brochure that had alot of closeup of their equipment and you could see how it all worked. They had a very primative machine that used a tray at a angle to center the componets. It was to primative for the company I worked for then's needs but their constant bombardment of myself with sales liturature made me atleast remember their name.

Take a look at their site and fill out the web form so you can get the brochures for the machines. I don't think they offer the really primative SMT placer any mre since I didn not see it on their site, but a call to them might yield an old brochure.

Important even for 0805's and SOT-23's.

We were having placement problems with one of the three heads on one of our Zevatech FS-730s. Turns out the service technician found a problem with a solenoid valve that wasn't giving that little "puff" of positive pressure immediately after part placement.

Valve was replaced, problem went away.

Mariss

I've seen that and the wrong nozzle being used for smaller chips. Sucks them up into the head and on some older machines you had to take the head off to get the compont out.

are teh pictures of he CP6-4 Windmil. Think they were over $1k new.

You'll probably hate me for this but I have to say it.

Of all the CNC projects to assume, this is the least renumerative. Before I go on, let me say when I was younger, I also designed a pick and place machine in ACAD; I still have the .dwg files to remind me.

These are enourmously complex machines. CNC-wise they are very simple point-to-point X,Y positioners. That is where the easy stuff ends.

Modern machines 1, 2, 3 or more heads each use laser measuring heads to detect the Z and theta axis measurements of the part you are about to place. It uses these measurements to adjust the final X,Y trajectory of the part about to be placed as well as its thickness and rotation from the pick-up point.

Parts come on either 7" or 12" reels; lots of 7" or 12" reels. On my machine it's 40 locations in front, another 40 in the back. Each reel requires a mechanism (solenoid plunger) to index the reel 4 or 8 mm after a part pick-up. That's 80 solenoids right there.

About the parts that are on a reel. The pocket (stamped paper tape reel) is much bigger than the part (resistor, capacitor, SOT-23). The part in the pocket can have an angular misalignement of +/- 15 degrees and an X,y error of 0.03" in either axis. When placed, the part must align within a degree and can't be off by more than 0.002".

If you are building a machine, what are you going to do about the reel-feeders? Use Juki, Fuji, Seimens, Panasonic or your own design? Either way you will need a lot of them.

How about ICs? Tape and reel or linear feeders? I have linear feedes on my machine and each is a work of art CNC-wise. Several servo motors and a computer in just the feeder.

My first pick and place machine was a Manncorp EMC-97. It was a servo driven tooth-belt transmission piece of crap that gave me the grey hair I have now. It could chug along at at one totally misplaced (>0.025") part every second and a half. The machine was named Ursulla for some wicked queen by my daughter. I paid $35,000 for it to make my aquaintance.

Not a part she was supposed to place landed even remotely near the solder pads on the board. Service techincians came and went; they pronounced that was the best Ursulla could do.

It was my sanity or Ursulla. I chose sanity and got rid of her for $1,000 to a machinery reseller. She lasted for the longest 2 years of my life. God protect whomever she wound up with.

For the same money ($35K) I bought a near zero-time Zevatech FS-730. A dot-com failure company that put up a $250,000 machine for near 10 cents on the dollar. Earlier this year I bought another one, similar source but a little dearer ($50K).

Now I have two honest 15,000 parts per hour machines. When they are tooling, they put down 3 parts per second. That's for 0805s and SOT-23s. It's slower on SOICs.

My point is this. Don't even consider making a machine that can't place a part every second and a half. If it can, have the parts land someplace near the pads. If it's even close, solder surface tension will take it the rest of the way.

Second, don't bother unless you have thought the problem all the way through. That means laser parts measuring and how to deal with a lot of parts feeders. Without that it's a non-starter.

I told you you would hate me for this.

Mariss

Thanks to all, to CNCzone, to Vipertx, to Scotsss, to Alexg and Mariss for help and suggestions.
Mariss, I agree with you: if I had to build a lot of SMD populated PCBs, then I'd buy a GOOD pick n place, or I will have it done by a company: it is the same as for good PCBs.
But I have to build very small runs, or prototypes, so I can:
place by hand : very very inexpensive
buy a " hand placer" : 1500 to 5000 USD
buy a old, useless pick and place : 500 to 10,000 USD
buy a " useless" new machine ( say 20.000 USD + accessories)
buy a new or recent-used machine ( more than 40.000 USD )
PLUS feeders, nozzles, and a very expensive import, plus 20%VAT, not to say about service....
I'm a one-man-business, and speed is not a problem for me, consider time to talk with customer, time to design a working circuit, to design a PCB, to find components, to build the PCB, to solder, to seelk-screen the box, to drill it , etc.: so to place a SMD component every some seconds is really not a problem: no need for speed.
Of course I'm talking about prototype or very small runs, low-tech big-SMD circuits.
Feeders: no problem, as I have to buy in any case:new machine, used machine, home-built machine.
Precision: this is the problem! This is where I need a lot of help, suggestions, pictures, and the help of people like you ( using or trying to build) is invaluable !
Paolo

You'll probably hate me for this but I have to say it.

My point is this. Don't even consider making a machine that can't place a part every second and a half. If it can, have the parts land someplace near the pads. If it's even close, solder surface tension will take it the rest of the way.

Second, don't bother unless you have thought the problem all the way through. That means laser parts measuring and how to deal with a lot of parts feeders. Without that it's a non-starter.

I told you you would hate me for this.

Mariss

No hate... Nice advice is always good to hear...

I am making about 25 boards mostly every day with 2 people working on them using hand held vacuum pickup needles and with components in small trays already out of the tapes.. And it works !!

I guess that my needs for a pick and place machine and Ciccio needs are not for a fast machine or one that can handle big varieties of sizes and shapes but just something that can automate a tedious and erro prone process.

I am planning on no parts fedder at the "first phase" of the machine !! I will just cut the tapes to the right size and put them in a support over the table. I am planning on just using it for 0805 passives and SOT-23. All other sizes will be hand placed. 1 part every 3 seconds is enough for me to equal what I have in hand placement today.. And I guess that Ciccio requirements are similar to mine..

If we keep it simple I really believe it is quite doable. If we loose track and try to make a complex machine we will just not be able to accomplish it or will end up with a cost that is the same as a commercial machine.

In this way IMHO, we can keep the parts fedder out and just use the tape cut by hand and put inside a "container rail". This rail can be quite easily done with a mill or even with small pieces of plastic glued together.

The pick-up head is a simple mechanism with a hypodermic needle that can be rotated and makes contact with the parts. No need for pushing the components out at low speeds. The solder paste surface tension is enough to pull the component from the needle. When placing bu hand we do not even take the vaccum out !!

If kept simple we can get it to be a simple XY gantry with a little more complicated Z and rotation axis. It will work slowly and with low precision but it is still much better than doing it by hand.

Best regards,
Alexandre Guimaraes

Thanks to all, to CNCzone, to Vipertx, to Scotsss, to Alexg and Mariss for help and suggestions.
Mariss, I agree with you: if I had to build a lot of SMD populated PCBs, then I'd buy a GOOD pick n place, or I will have it done by a company: it is the same as for good PCBs.
But I have to build very small runs, or prototypes, so I can:
place by hand : very very inexpensive
buy a " hand placer" : 1500 to 5000 USD
buy a old, useless pick and place : 500 to 10,000 USD
buy a " useless" new machine ( say 20.000 USD + accessories)
buy a new or recent-used machine ( more than 40.000 USD )
PLUS feeders, nozzles, and a very expensive import, plus 20%VAT, not to say about service....
I'm a one-man-business, and speed is not a problem for me, consider time to talk with customer, time to design a working circuit, to design a PCB, to find components, to build the PCB, to solder, to seelk-screen the box, to drill it , etc.: so to place a SMD component every some seconds is really not a problem: no need for speed.
Of course I'm talking about prototype or very small runs, low-tech big-SMD circuits.
Feeders: no problem, as I have to buy in any case:new machine, used machine, home-built machine.
Precision: this is the problem! This is where I need a lot of help, suggestions, pictures, and the help of people like you ( using or trying to build) is invaluable !
Paolo

Hi, Paolo

Let's keep the feeders for a second phase... They are complicated....

Precision is not a big problem with 0805 components and SOT-23 ! They self align nicely of you put enough solder paste on the pads. No need for a squaring station. We are doing it by hand here and the girls do not even use the magnifying lenses that I have bought for it. They just pick and place the component.. The machine will do better then them... 0.2 or 0.3mm should be enough and visual inspection and hand retouching of some solder joints will be necessary anyway if you do not use a big reflow oven. If the machine misses one in 500 components it is still much better than hand placement.

We are on the same boat about speed. It is nice to have but not critical..

Best regards,
Alexandre Guimaraes

In the picture some material for my Z axis
Any good idea for the best way to build Z axis control ( lead screw ) ?
Paolo

..and my attachement ( picture ) is lost in cyberspace.......
Paolo

To all the DIY CNC machinists
Merry Christmas
and a very successfull
new year

by ciccio

Hi to all,
i have made the conversion of a lexmark 3200 printer to a small pick and place
system.
I was able to utilize the cleaning part inside the printer as Z-rail with a
travel of appozimate tree cm. This rail and the C axis is piloted using a common rc mini-servo.
External parts used and costs:
lexmark printer, 0 Euro, received from a friend
tree way solenoid: 30 Euro
two smd pick-up pen including vacuum pump: 70 Euro
small fan: 4 Euro
pcb bottle for concentrating the air flow produced from the fan: 0 Euro.
silicone tube for vacuum pen, 0.2 Euro
rc-servos, 2x 15 Euro = 30 Euro ( i know, this can be sourced for 5 Euro/pc)
electronic (pic,..., stepper motor controller used from the printer board) 10 Euro.
USB optical mouse, 12 Euro, used as high speed, low cost vision system.
linear guides as used for drawers, 12 Euro
two wood plates, some wood spacers and rubber foam, 5 Euro.
The hw is in place, for the electronic i need to make a pcb, currently it´s
wire wrap and on experimenter boards.

Hi to all,
i have made the conversion of a lexmark 3200 printer to a small pick and place
system.
I was able to utilize the cleaning part inside the printer as Z-rail with a
travel of appozimate tree cm. This rail and the C axis is piloted using a common rc mini-servo.
External parts used and costs:
lexmark printer, 0 Euro, received from a friend
tree way solenoid: 30 Euro
two smd pick-up pen including vacuum pump: 70 Euro
small fan: 4 Euro
pcb bottle for concentrating the air flow produced from the fan: 0 Euro.
silicone tube for vacuum pen, 0.2 Euro
rc-servos, 2x 15 Euro = 30 Euro ( i know, this can be sourced for 5 Euro/pc)
electronic (pic,..., stepper motor controller used from the printer board) 10 Euro.
USB optical mouse, 12 Euro, used as high speed, low cost vision system.
linear guides as used for drawers, 12 Euro
two wood plates, some wood spacers and rubber foam, 5 Euro.
The hw is in place, for the electronic i need to make a pcb, currently it´s
wire wrap and on experimenter boards.

Sound's way cool!!!!
Could you post some pictures?

I am interested in hearing more about the optical mouse as a vision system. Are you using the mouse to detect if a part is even on the nozzle? How much rotation you need on the nozzle to align the part correctly? or some other reason. Yes, please post pictures

Hi nisma:
congratulations ! You did it!
A lot of interesting solutions !
We need more infos and a lot of pictures.
...and that damned optical mouse :
that is really art .
Paolo

...sorry, nisma, I forgot :
what about result / precision ?
What about feeders ?
What about speed ( not so important in this beta stage )
very very interested !
PS are you italian ? I live near Cuneo.
Saluti
Paolo

The mouse sensor gives from 64x64 pixel output to 160x200 pixel for some
sensors at high speed, from 200HZ up to 2.5 KHZ.
Search for "hacking optical mouse". I use it for aligning the pcb and the
placement coordinates and do detect the components to pick including the
displacement.
For the feeders, i have a solution, even it´s needs four cm every reel, because i use RC servos.
I have some modifications in mind and xxx.
The precision is 4 mil without microstepping, and the speed is 1.5 second/component for the planned design modification.
The planned changes is to make two sliders, one for the pcb and annother for
the components, in order to optimize the pcb sliding and to improvere the
efficience. Additional this allows preparing the components (reels strips,...)
on a panel and loading in on the pcb tray. Then the pick and place machine
moves the component from the pcb tray to the components tray.
The overall speed including loading the components will be between 5 and 10
seconds. For real production systems, it´s supposable to use one printer
for loading the parts from reels/feeders and annother for pick and place parts
and interconnects this two printers using a small turn table, concretly a cd mounted on a rc servo.
Then one or more pcb´s can be assembled using the components stored on
the components rail. Additionally four reels feeders are planned for generally
used resistors, like 10k, and so on. It´s possible to modify one use and trash
35mm film camera to handle the reel transport system. Ask photo developing
shops to get some from it for free instead of trashing it after removing the
film. The drawback of this system is, that it needs
4 cm every feeder, so only four reel feeders are planned.
Some have a flash inside it. You can use it as error signals, like the industrial
control signals if something need user intervention.

Someone has one idea about needle tooling and changing ?

I can make photos only in then days, because i receive back my digital camera from my sister after she returns from asia trip.
Sorry for the longer delay.

Hi Nisma,
....so let me see:
4 mil WITHOUT microstepping
2.400 components/hour ( 1,5 sec/comp)
It seems to me there is a lot of work and developement in this machine.
Waiting for pictures !
About nozzles : have you seen:
http://cgi.ebay.com/Assembleon-FCM-nozzles-PA-2747-00-2747-20-2747-3-etc_W0QQitemZ120029343513QQihZ002QQcategoryZ58293QQrdZ1QQcmdZViewItem
I can send you a nozzle PA 2747/20 Philips-Assembleon ( 0603-1206-sot23), I think nothing is better than a professional nozzle, especially if I can buy it at an inexpensive price on ebay.
It seems to me you used "off the shelf" components: an old printer, rc servos, and so on: this means ,with a little help, now we can start thinking to build our pick and place .
Thanks, Nisma, and please do not forget the pictures; we are counting the days !
Paolo

I have the nozzles, but i need a system, that lets the machine change
the nozzles (tooling) without operator intervention.
The 2.4000 components it´s only true for placement from the reel,
otherwise 1.200 components/hour is realistic.

I think the form of a nozzle allows to put and take it from some sort of wide-toothed comb. Can someone help us with a picture from a P&P ?
I also think this needs some work for milling machines.
Some question again:
do you modify rc-servos?
For example on the theta (rotation) axis ? A servo only rotates about +- 130 degrees. It needs to rotate endless on the feeder.
Do you use some sort of "spring" on Z axis to compensate different high ( feeder-PCB)?
Please tell us more !
Paolo

I have seen some photos from a P&P machine, but i need a cheap idea,
how to hold and separate the needle adapter.
The nozzle proposed from ciccio don´t work, because the printer need
more height clearance and this nozzles are basically designed for
using it with solenoids, but this are too heavy as to be used on low cost printers without modifying the linear sliding rails on the printers.
If someone has a dot-matrix printer, then it will maybe work, but today,
this are either too costly or not easy to find for dismissed parts.
This type of nozzles can be used on a x/y/z table with several nozzles and
it will works really nice, no nozzle tooling is required, because several (different) nozzles can be used in parallel. Further this type of P&P eqipment is really fast, because the increased count of nozzles automatically multiply the placement capacity. It´s obvious, that this is a totally different design and costs at least 10 times more then utilizing a printer. My approach is to use low cost
cots components as mutch as possible and eventually using multiple machines
in order to increase the troughtput.
To respond to the questions,
for the Z axis, i have used the cleaning assemble from the lexmark printer.
This has a spring inside and i have replaced the two internal springs with
the springs from the paper feeder in order to eliminate the backlash when
moving the Z axis. The axis is moved with a rc-servo and the cleaning
assemble convert the rotational movement of the servo to a linear movement
of 3cm maximal travel. The height differences including the height clearance for the components on the pbc should be in the range of 3cm in order that
the pcb can be assembled with the machine.
It´s possible to extend the travel to 5cm, but this needs some work with a real cnc machine.
The C rotation (theta) currently uses unmodifyed servos.
For now, this is sufficient. Later, the servo can be replaced with a modificated one, when this is needed or desired.
For the rc-servos on the feeders, i have removed the electronic and power the servo with 2V (3V on the negative pole, 5V on the positive).
For now, the feeder motor is activated for a certain amount of time, and
then it stops. When the head picks up the component, the optical sensor
measures the position and rotation including the advancing error.
This works pretty well. It´s possible to add optical or mechanical sensors
for measuring the advancing of the reel (1-3$), at the moment, this part
is not fitted.
Basically this are the same as using one small geared motor, but servos are
cheap, readily available and with a nice servo holder, everyone can replace
the servo motor. Further the roll extracted from the printer can be
screwed directly on the servo.

Hi to all,
have someone a idea about automatic nozzle changing between these two
different nozzles ? I appredicate any help or hint.
By the way, one commercial and low cost manual smd reel feeder that
i have used as base for the motorized feeder using rc servos.
This images comes from the vendor site.
Regards,
nisma

very interesting thread :-)

I just had a thought: the feeders seem to be a costly and difficult part, but since we seem to be aiming at a speed that's near or just above hand placement, you could use manual feeding. if your x and y are sufficient, then you could manually pull through something like 10 or 20 components and the software could just keep track of where the next part should be, instead of advancing the tape one component after picking a component. You could order the placement so the software switches to the next component after placing up to 20 of the previous component, so it can place 20 of each component unattended, and after that you need to do manual feeding of the used reels. just a thought.

better yet: add a small pin to the placement head. after picking and placing one component, go back to the reel, insert the pin into the reel, and pull it through one component using the placement head.

jaap.

IMHO, RC Servos are not the best solution for reel feeding. True, normal RC servos only rotate ~180 degrees. They have a potentiometer attached to the shaft which provides feedback for the position. You can detach (remove)and "fool" the servo to continually rotate. You have a normal gear reduction motor at this point. However, you have no accurate control how much it rotates. It seems like stepper motor is a better solution

Or your could use a rc servo modificated for continuous rotation for advancing
the feeder. Go to the next office store and select one white taper canceller. There come with with of 8mm and 4.2 or 5mm.
This parts are perfect for being used as part of cheap feeders in addition to
modified rc servo.

rc servos are cheap and have already the necessary electronic to drive the
motor. Principally i have used it, because i had some spares laying around.
The position is checked with a fork type infrared sensor removed from
the printer, that checks the perforations, that are present every 4mm on
the reel.
I only advance in multiple of 4mm. The internal rc potentiometer is repleaced
with a fixed resistor, as necessary for the continous rotation.

That is good info. An IR sensor looks at the holes in the tape and shuts off the motor. The RC servo is also geared for slower speed and higher torgue. This seems like a less complicated way rather then using a stepper motor. I retract my previous comment.

I think we are talking about a small PnP machine used for DIY or R&D purpose. So it does not have to be fast but must be cheap and versatile.

Recently I am also planning such a machine for my own using. I was fed up with soldering hundreds of small capcitors manually for just one board. I have got a small X-Y CNC bed without any controller, to which I still need to add Z axis and and a R axis for rotating the head.

I'm going to add one or two webcams as a vision system. There will be a training zone on the bed, so that the aligment between the head and the camera will be done at the run-time.

There will be two types of heads: one for the paste dispensing another for PnP. I'll just discuss the PnP one here.

Basically that head is like a syringe and takes that type of needle for small devices. For a large package like QFP, that needle will pick up a bigger adaptor first. The adaptor is a plastic or rubber nozzle. The head can put the needle into it under the guide of the vision system, and the nozzle will stay with the needle for rest of the operation. To take off the nozzel, put the nozzel on the nozzel tray (a smooth even surface), turn the vaccum on, and the head puts up to get rid of the nozzle, then, turn the vaccum off.

For the feeding, there will be a component area with a variaty of slots and trays. The slots are for the SMD strips, e.g. up to 50 components. The strip is pre-cut and peeled mannually according to how many is needed for this board. For more than 50 piecs, just use a second slot. For a few less than 50 pieces they can be compined. However, there is no need to use strip and slot for just a few pieces, say < 10. There are also square trays, e.g. 20mm x 20mm for loose SMD. Again, the vision system is used to locate the components.

The head is rotatable. The head socket is like a cylinder with teeth on the outside, and a stepper will drive a timing belt to rotate the socket. The head socket may measure 20mm ~ 25mm in diameter and fixed to the Z platform via a ball bearing. The rotating motor is also no the Z platform.

I haven't decided how to put the head on the socket quick and easy, and also compatible with the paste dispenser head.

I know there are a lot for work for it. So it has to be well planned and ideally if people can cooperate on it, e.g. control software, vision software, system interface to a PC, making of the mechincal parts, etc.

ta
David

I think we are talking about a small PnP machine used for DIY or R&D purpose. So it does not have to be fast but must be cheap and versatile.

Recently I am also planning such a machine for my own using. I was fed up with soldering hundreds of small capcitors manually for just one board. I have got a small X-Y CNC bed without any controller, to which I still need to add Z axis and and a R axis for rotating the head.

I'm going to add one or two webcams as a vision system. There will be a training zone on the bed, so that the aligment between the head and the camera will be done at the run-time.

This will not work. It´s better to measure the offset and set it manually.
For the first time, you can use a paper that your have fixed with
tape to the base. Next put the needle on a stamp-pad and now, with
the camera and manual Z control, you can measure the offset. It remains
constant. With the camera, you can now measure the offset accuratly.
You need a led on the X/Y table for the purpose of reference point for
the camera. This allows you to automatically measure the backlash and
other type of meccanical errors, that can be compensate by SW.


There will be two types of heads: one for the paste dispensing another for PnP. I'll just discuss the PnP one here.

Basically that head is like a syringe and takes that type of needle for small devices. For a large package like QFP, that needle will pick up a bigger adaptor first. The adaptor is a plastic or rubber nozzle. The head can put the needle into it under the guide of the vision system, and the nozzle will stay with the needle for rest of the operation. To take off the nozzel, put the -nozzel on the nozzel tray (a smooth even surface), turn the vaccum on, and the head puts up to get rid of the nozzle, then, turn the vaccum off.


Generally, with two nozzles (of different diameter and with different rubber)
you can make all components excepts big electrolyt condensators.
Basically, you have two indipendent Z axis, one for the P&P needle and
the other for the dispenser. This can be one Z axis and two solenoids as
example. Otherwise you don´t be able to dispense and place at the same
time. Maybe you don´t have necessety to do so.
At the same time, you can think of using two needles on the same head,
maybe one fix and the other interchangable, either manually or automatically.
The C Axis can use the same timing belt and motor to drive both axis at
the same time.



For the feeding, there will be a component area with a variaty of slots and trays. The slots are for the SMD strips, e.g. up to 50 components. The strip is pre-cut and peeled mannually according to how many is needed for this board. For more than 50 piecs, just use a second slot. For a few less than 50 pieces they can be compined. However, there is no need to use strip and slot for just a few pieces, say < 10. There are also square trays, e.g. 20mm x 20mm for loose SMD. Again, the vision system is used to locate the components.

The head is rotatable. The head socket is like a cylinder with teeth on the outside, and a stepper will drive a timing belt to rotate the socket. The head socket may measure 20mm ~ 25mm in diameter and fixed to the Z platform via a ball bearing. The rotating motor is also no the Z platform.

I haven't decided how to put the head on the socket quick and easy, and also compatible with the paste dispenser head.

I know there are a lot for work for it. So it has to be well planned and ideally if people can cooperate on it, e.g. control software, vision software, system interface to a PC, making of the mechincal parts, etc.

ta
David

I have written some SW for the linux system in order to control the P&P
including vision recognizion. This implies, that the used webcam
is full supported on linux.

Ok... add me to the list of people building a pick'n'place :)

Been acquiring some gears/motors to start planning the system, and this thread has been very useful to generate some ideas.

For vacuum pickup, the aquarium pump is what I was planning to use. I have see these vacuum tools available...
http://cgi.ebay.com.au/Pick-and-Place-Vacuum-Pens-for-SMT-SMD_W0QQitemZ280039971649QQihZ018QQcategoryZ36327QQssPageNameZWD2VQQrdZ1QQcmdZViewItem (http://http://cgi.ebay.com.au/Pick-and-Place-Vacuum-Pens-for-SMT-SMD_W0QQitemZ280039971649QQihZ018QQcategoryZ36327QQssPageNameZWD2VQQrdZ1QQcmdZViewItem)
Which looks like a basic pump with pen.

I plan to use an aquarium pump and for the needles use solder paste syringe needles. They come in a good range of sizes, have flat bottom and you can get them in plastic too. Then use a hand pick'n'place tool to get some cheap silicon heads for the bigger parts.

I have considered a crane system, whereby the base would be at the centre rear of the table and rotate the crane across the board, and then have the head run along a track. Maths would be a complication. I'm keen to have a full XY with static PCB, as it is far more flexible for retrieving parts. In addition, I want to have one XY for the PnP and other for the solder paste dispenser.

I want to have approx 12 tapes at the front, feeding with servos via the tape holes. The idea with the tape plastic is to part it from the tape, rotate it 90 degrees to allow component pickup and then rejoin it with a common roller to the tape.

Using an optical mouse sensor, I'm thinking of using one to correct the datum for the pickup tool... and it's position is right in the path for component pickup. Hence once a component is collected, it passes over the sensor before placement. The head can therefore correct rotation of the part on the fly, or pause to correct it with a simplier software alogorithm determine the flat edge.

The vacuum pick'n'place tool would be operated by a servo for the Z plane, using a spring loaded catilever arrangement. Therefore the head will go down to the component and then the spring would allow the servo to rotate further without placing much more force on the component.

I'd like to consider in the future having a seperate basic XY table to hold the hot air station, or adaptor a heat gun element to have a system that reflows the board. So a pre-heater underneath, and perhaps a temperature sensor on heating element to track/control a soldering temperature profile. Rather a seperate table for this to avoid over complicating the pick'n'place, and its really an additional feature that can wait.

Currently I'm only looking to place the 0805, 1206, SOT23, SOT223 components. These take the longest to place and I find the easiest to hot air solder. I may include an area to hand place components for the pickup head to use. For example, spots for all the common footprints, plus some for SO8,14,16, TSSOPs, diodes, DPAK/D2PAK etc... and have an LCD for prompting the component required. Have a tiny light sensor under each component depot would make the system quite nice to use, particularly if the tape feeders are not up and running at that stage. Ideally I'd put the dremel in the XY head to route out these component depots hahaa..

Only just starting on this project and putting ideas together...

Ok... add me to the list of people building a pick'n'place :)

Been acquiring some gears/motors to start planning the system, and this thread has been very useful to generate some ideas.

For vacuum pickup, the aquarium pump is what I was planning to use. I have see these vacuum tools available...
http://cgi.ebay.com.au/Pick-and-Place-Vacuum-Pens-for-SMT-SMD_W0QQitemZ280039971649QQihZ018QQcategoryZ36327QQssPageNameZWD2VQQrdZ1QQcmdZViewItem (http://http://cgi.ebay.com.au/Pick-and-Place-Vacuum-Pens-for-SMT-SMD_W0QQitemZ280039971649QQihZ018QQcategoryZ36327QQssPageNameZWD2VQQrdZ1QQcmdZViewItem)
Which looks like a basic pump with pen.

I plan to use an aquarium pump and for the needles use solder paste syringe needles. They come in a good range of sizes, have flat bottom and you can get them in plastic too. Then use a hand pick'n'place tool to get some cheap silicon heads for the bigger parts.

The solder past needles don´t do the job. The manual smd pick and place
needle have a small silicon head. This rubber is really important for a
pick and place when only moderate vacuum are available, like a aquarium pump. Further this reduce vibration and allows for some errors on the
Z axis, because the rubber compensate it. You can buy this needle as replacement. Maddel as example sell it for 1.5$ if i remeber correctly.
With two needles of different size, you can make over 94% of all components.


I have considered a crane system, whereby the base would be at the centre rear of the table and rotate the crane across the board, and then have the head run along a track. Maths would be a complication. I'm keen to have a full XY with static PCB, as it is far more flexible for retrieving parts. In addition, I want to have one XY for the PnP and other for the solder paste dispenser.

Can you have enought resolution with a cran system without using
costly gears without backlash.


I want to have approx 12 tapes at the front, feeding with servos via the tape holes. The idea with the tape plastic is to part it from the tape, rotate it 90 degrees to allow component pickup and then rejoin it with a common roller to the tape.

Sorry, i don´t have understand it, maybe my bad english. Can you explain
it again eventually a bit more explitly.



Using an optical mouse sensor, I'm thinking of using one to correct the datum for the pickup tool... and it's position is right in the path for component pickup. Hence once a component is collected, it passes over the sensor before placement. The head can therefore correct rotation of the part on the fly, or pause to correct it with a simplier software alogorithm determine the flat edge.

The vacuum pick'n'place tool would be operated by a servo for the Z plane, using a spring loaded catilever arrangement. Therefore the head will go down to the component and then the spring would allow the servo to rotate further without placing much more force on the component.

I'd like to consider in the future having a seperate basic XY table to hold the hot air station, or adaptor a heat gun element to have a system that reflows the board. So a pre-heater underneath, and perhaps a temperature sensor on heating element to track/control a soldering temperature profile. Rather a seperate table for this to avoid over complicating the pick'n'place, and its really an additional feature that can wait.

Currently I'm only looking to place the 0805, 1206, SOT23, SOT223 components. These take the longest to place and I find the easiest to hot air solder. I may include an area to hand place components for the pickup head to use. For example, spots for all the common footprints, plus some for SO8,14,16, TSSOPs, diodes, DPAK/D2PAK etc... and have an LCD for prompting the component required. Have a tiny light sensor under each component depot would make the system quite nice to use, particularly if the tape feeders are not up and running at that stage. Ideally I'd put the dremel in the XY head to route out these component depots hahaa..

Only just starting on this project and putting ideas together...

For the C rotation, just today i have found a nice system.
Two 12V stepper motor, one for the Z axis and annother for the C axis,
both driven with one ULN2803. The stepper motors should be 12V and
have 1.8 degre step or 200 full steps / rotation. The motor is driven with half
steps.
One plastic bevel gear with a reduction of 2:1 is used to rotate the nozzle
and the component when the Z axis is lifted up completly. The C axis have
some type of friction (antivibration nut as example) that removes the possibility of rotation during the Z axis movements.

The solder past needles don´t do the job. The manual smd pick and place
needle have a small silicon head. This rubber is really important for a
pick and place when only moderate vacuum are available, like a aquarium pump. Further this reduce vibration and allows for some errors on the
Z axis, because the rubber compensate it. You can buy this needle as replacement. Maddel as example sell it for 1.5$ if i remeber correctly.
With two needles of different size, you can make over 94% of all components.

I'II have to look into that, thanks. I want to buy one of those hand vacuum pickup tool pen's, which come with 4 silicon nozzles. I don't know how they go with 0805 components... never used one. I did experiement with a syringe, just with the plastic tip, and I could reliably pickup a SO8. Certainly finding someone local for the actual pickup needles and silicon heads is the preferred method.

Only thing I'm not sure on yet is how to create a system for both the suction and then slight blow action to help the component detach. I'm thinking of making some sort of solenoid that will cut off the suction which also result in a pulse of positive pressure.


Can you have enought resolution with a cran system without using
costly gears without backlash.
That was my concern also, and with the concept was looking at motor control strategies that would approach the cooridinants with one direction and deadbeat to the other direction with slight residual pressure to obtain consistent operation. Certainly the mechanical side requires precision to get the needed placement accuracy. I plan to develop a few system 'parts' to test movement speeds, relative accuracies (step resolution) under loads and prove design ideas. Then I'm going to fork out some money and have a machine shop make some of the assembly components accurately.


Sorry, i don´t have understand it, maybe my bad english. Can you explain
it again eventually a bit more explitly.

Certainly... hard to explain without a picture, however...
The idea is to have the tapes on a common axial holder. The tapes are spooled via servos pulling the tape via the tape holes. The pickup head always picks up from the component tape in the same location. At this location, the clear plastic tape cover must not be covering the component, but I don't want to strip is off and have it collect seperately. Rather I want to have it re-joined to the tape. Therefore, the section where the pickup head takes the component the clear tape is peeled to the side (perpendicular to the actual tape) so the pickup tool has clear access. The clear tape is ~20mm further down, the tape rotates back onto the component tape. Both before and after the pickup location there will be a roller to hold the tape down, and also help with the rejoining of the clear tape to the component tape.

For the C rotation, just today i have found a nice system.
Two 12V stepper motor, one for the Z axis and annother for the C axis,
both driven with one ULN2803. The stepper motors should be 12V and
have 1.8 degre step or 200 full steps / rotation. The motor is driven with half
steps.
One plastic bevel gear with a reduction of 2:1 is used to rotate the nozzle
and the component when the Z axis is lifted up completly. The C axis have
some type of friction (antivibration nut as example) that removes the possibility of rotation during the Z axis movements.

The Z axis I'm looking at a servo with catilever operation and spring arrangement to allow excessive travel after the component has been touched. The C axis I have not given much though to yet, however was thinking of another servo. Might not give full rotation however, but also possible to pre-rotate the head prior to component pickup. I do like your approach with a stepper though.


I haven't decided on stepper or DC motors yet. I like the thought of DC motors with optical feedback. They can move faster and with more torque, which can therefore provide greater absolute positioning, faster. So I may use DC motors for the XY. Even if I used steppers, I may still use optical feedback... or at least allow for it in the design.

Thanks for the feedback... this is going to be a fun project :)

Only thing I'm not sure on yet is how to create a system for both the suction and then slight blow action to help the component detach. I'm thinking of making some sort of solenoid that will cut off the suction which also result in a pulse of positive pressure.

If you limit the components to 603, and slow down the speed,
waiting 500ms-700ms instead of 10ms for releasing the component, you simply can cut-off the suction. For this, you must have a small hole
in order that when the system is off the vacuum don´t remain.
The needle for a siringe (small diameter) is a good starting point.


That was my concern also, and with the concept was looking at motor control strategies that would approach the cooridinants with one direction and deadbeat to the other direction with slight residual pressure to obtain consistent operation. Certainly the mechanical side requires precision to get the needed placement accuracy. I plan to develop a few system 'parts' to test movement speeds, relative accuracies (step resolution) under loads and prove design ideas. Then I'm going to fork out some money and have a machine shop make some of the assembly components accurately.

I have found some cheap linear guide system from igus, dryline serie,
and as such i prefer a ordinary x/y table.


Certainly... hard to explain without a picture, however...
The idea is to have the tapes on a common axial holder. The tapes are spooled via servos pulling the tape via the tape holes. The pickup head always picks up from the component tape in the same location. At this location, the clear plastic tape cover must not be covering the component, but I don't want to strip is off and have it collect seperately. Rather I want to have it re-joined to the tape. Therefore, the section where the pickup head takes the component the clear tape is peeled to the side (perpendicular to the actual tape) so the pickup tool has clear access. The clear tape is ~20mm further down, the tape rotates back onto the component tape. Both before and after the pickup location there will be a roller to hold the tape down, and also help with the rejoining of the clear tape to the component tape.

If you pick off 20 components, this system will work ?


The Z axis I'm looking at a servo with catilever operation and spring arrangement to allow excessive travel after the component has been touched. The C axis I have not given much though to yet, however was thinking of another servo. Might not give full rotation however, but also possible to pre-rotate the head prior to component pickup. I do like your approach with a stepper though.

I have a arrangement with two servos, it works, but ...
When picking up components, you should check the height of the component.
When placing the component, you must release the component above the
solder past, and not overtravel and pressing down the solder past.


I haven't decided on stepper or DC motors yet. I like the thought of DC motors with optical feedback. They can move faster and with more torque, which can therefore provide greater absolute positioning, faster. So I may use DC motors for the XY. Even if I used steppers, I may still use optical feedback... or at least allow for it in the design.


The resolution is greater with stepper motors, and the speed is not a problem.
Generally, the motor must be slow down because the linear guide don´t support the speed.
For the normal gantry table, optical feedback is not needed for component
size 402 and above.

If you limit the components to 603, and slow down the speed,
waiting 500ms-700ms instead of 10ms for releasing the component, you simply can cut-off the suction. For this, you must have a small hole
in order that when the system is off the vacuum don´t remain.
The needle for a siringe (small diameter) is a good starting point.
Ahhh... this makes sense, and a wait period hear is quite exceptable to simplify the design. If fact, having the blow function is something that could be considered at a later stage, if I wanted to trim the time down. Thankyou for the tip! You saved me a good hour of procrastination. How small are you talking? 0.3 to 0.7mm? Confused between needle only pickup, or needle with silicon head.

I have found some cheap linear guide system from igus, dryline serie,
and as such i prefer a ordinary x/y table.
cut-paste... equipment to look for ;)
Edit: that stuff is quite expensive. Had a look, and it would offer the linear guide, but not the mechanism for controlled movement. I've been thinking of using a long large guage thread to support the gantry and move it.


If you pick off 20 components, this system will work ?
I'm hoping it will, but will only know by giving it a go. For the start of the reel, there is always plenty of leader tape to use for the slack. Otherwise, I'II just need to sit there and feed it through while running. My concern with this setup is with the reels not being in a cassette, but then again these reels will most likely have the common components anyway.


I have a arrangement with two servos, it works, but ...
When picking up components, you should check the height of the component.
When placing the component, you must release the component above the
solder past, and not overtravel and pressing down the solder past.
I had planned to stamp the component right down on the board. I can see why that would be counterproductive. I always lean towards feedback design, rather than program the height. What about having an infrared switch on the XY head, where the pickup tool lifts the component to standard clearance height which the infrared switch detects has the bottom of the component. Placement is then an offset to this detected component bottom? Or is this just taking it too far?


The resolution is greater with stepper motors, and the speed is not a problem.
Generally, the motor must be slow down because the linear guide don´t support the speed.
For the normal gantry table, optical feedback is not needed for component
size 402 and above.
I don't yet have a feel for the effective gear ratio vs movement for the mechanisms yet, so appreciate such comments. My previous experience with steppers has taught me that torque is lost with speed, hence my interest with DC motors with optical feedback.


I will have to start putting some of these ideas into CAD to get a better perspective on how it is going to come together.

cut-paste... equipment to look for ;)
Edit: that stuff is quite expensive. Had a look, and it would offer the linear guide, but not the mechanism for controlled movement. I've been thinking of using a long large guage thread to support the gantry and move it.


In the past weeks, i have searched for sourcing the parts.
Now, i´m near finished. For some items i still search some cheaper alternatives. The resulting price is 120-170 Euro including the motors and
electronic for the microstepper driver.
For a setup for placing BGA´s or uBGA, two usb cameras are additionally
needed, add maybe 30-40 Euro for it and 160 Euro for a linux controller
with ethernet, 2xusb host and lcd display for standalone operation.



I'm hoping it will, but will only know by giving it a go. For the start of the reel, there is always plenty of leader tape to use for the slack. Otherwise, I'II just need to sit there and feed it through while running. My concern with this setup is with the reels not being in a cassette, but then again these reels will most likely have the common components anyway.

For stripes of reels, i have used a small sugar panel, on that i have designed
some lines in order to align the stripes correctly.
I have fixed one side with a needle, peel off the protection, and fixed the
other end with a second needle. Further i have one mark for the first position,
on that i align the center of the first component. This works great for
small productions. The stripe don´t be necessary of 50 items, but can be
tree or eight components only.


I had planned to stamp the component right down on the board. I can see why that would be counterproductive. I always lean towards feedback design, rather than program the height. What about having an infrared switch on the XY head, where the pickup tool lifts the component to standard clearance height which the infrared switch detects has the bottom of the component. Placement is then an offset to this detected component bottom? Or is this just taking it too far?

This is the way that some machines uses, but it´s only used for verification
of the coded height, because the resolution is poor.
I have used a switch to detect the height of component from the pick-up,
with the servo doing aftertravel, but it´s not very affidable.
So i´m returned to codify the height, and using one switch on the ground plane to measure the height of the
component. For the switch, i have used a piezo operated switch made
from a piezo buzzer.

Only just starting to realise how little I know. I have a lot to learn about the mechanisms and mechanical requirements for a CNC before I get to the details for the pick'n'place requirements.

Only just starting to realise how little I know. I have a lot to learn about the mechanisms and mechanical requirements for a CNC before I get to the details for the pick'n'place requirements.

The meccanical requirements for a pick and place is quite different as for
a cnc.
The pick and place has to move a small weight with no lateral forces.
The cnc have to move a big weight and need to be really stable and have
big lateral forces.
Further the move rates for P&P is mt/s, on a cnc it´s mm/s.

Wit drawer linear guides it´s possible to lower the price to 1/2 for the P&P.
If you have two spare 1/8 microstep driver, then it´s possible to make it
under 50$.

Good point!

I've had a look at a lot of the designs, and a common theme is to have two smooth rods and a single thread drive down the middle. I had initially though of having two thread drives that were geared together, and therefore driven by a single motor. This is in an effort to reduce the mechanical requirements and also weight. I would still use a solid size for good measure, but I'm planning on a small setup around a foot long.

Given I'm interested in two gantries, this mechanism would most likely benefit from sharing the same rod sliders and seperate centre thread.

Looking at the igus products, I can now see how these guides could be used instead of rods for the y and z axis. There are only two dealer in Australia for igus, one of which is only 10mins away. So early next week I'II be taking a visit to get an idea of what is available and costs. I would really like to find these types of linear slides with a thread already built in. This would be absolutely idea, and would certainly be worth the extra cost!!!

I'm having trouble with the combination of Z and C. Ideally, a 1.8 degree stepper motor with a hollow shaft would be perfect. However I'm trying to use a smaller/lightweight 7.5 degree stepper, using a hollow tube mounted on the shaft. The tube is a T-piece, to allow the external vaccum/blow off connection. It does restrict the C to ~180 degrees, but 360 degrees with rotation before pickup.

Trouble is, I want to have the Z spring loaded. Such that I can touch a part and push slightly (to get a good vaccum pickup) and detect when to stop. This would also provide for a height calibration.

I had tried many variations in CAD of how to implement all this. Some using the igus N-rails, others with a fine pitch belt pulleys and more with fine pitch bolts for screw driven. Sometimes the problems result in resolution issues, either two highly gears (5 second traversal) to not enough precision (>0.2mm).

Has anyone got some ideas, suggest parts to try or webstore to look at, or seen a good design I could use as a reference?

I'm having trouble with the combination of Z and C. Ideally, a 1.8 degree stepper motor with a hollow shaft would be perfect. However I'm trying to use a smaller/lightweight 7.5 degree stepper, using a hollow tube mounted on the shaft. The tube is a T-piece, to allow the external vaccum/blow off connection. It does restrict the C to ~180 degrees, but 360 degrees with rotation before pickup.

Trouble is, I want to have the Z spring loaded. Such that I can touch a part and push slightly (to get a good vaccum pickup) and detect when to stop. This would also provide for a height calibration.

I had tried many variations in CAD of how to implement all this. Some using the igus N-rails, others with a fine pitch belt pulleys and more with fine pitch bolts for screw driven. Sometimes the problems result in resolution issues, either two highly gears (5 second traversal) to not enough precision (>0.2mm).

Has anyone got some ideas, suggest parts to try or webstore to look at, or seen a good design I could use as a reference?

What about this;

Your stepper and "pick-up: tube are mounte on a small single linear rail. That giges you your Z movement. then parallel to it is a stepper driven leadscrew. The nut on the leadscrew has a horizontal bar that "raises" the stepper and pick-up tube. The bar could lift by sitting unter the stepper motor mounting flange.

With this arrangement the Z axis is only raised. Gravity is used to lower it, as it rests on the bar.

That way you won't drive it through the PCB. You may need a spring to control the downward pressure.

Cheers,

Peter.

Thanks for the reply phomann. This is very close to a design I did try. Part of the problem I had was in resolving the nut fixture, while keeping it aligned. I did use a spring in this arrangement which gave it both the free play to not bind and also the desired cushion for contact with a part. It was certainly easier with a linear bearing like the igus N-27 series and I very much liked the solution. The head arrange was however large and I had trouble figuring out how to hold the top of the screw to minimise movement and pressure on the steppers axis. I tried the thread in between the carraige and the C-axis, however the length became a bit of a problem. Then I tried it above the carraige assembly, however this was getting quite high. I have been trying to create a compact head, in an effort to have three heads (2 PnP, 1 paste).


The variant to this design was to use the igus N-40 series and use a 2.5mm pitch belt running top to bottom to move the carriage. The pulley gears were embedded in the rail, which made for a neat axle holder. Size was still a problem, as was step resolution. I had to use larger pulleys to clear the rail, and then introduce a gear or worm drive. This was a nice solution, though difficult to make it compact.

I have been drawing everything in CAD with great detail to work out the particulars of the design, ensuring the necessary clearances and operating movements.

Reading back through the thread, it has re-kindled my interest in using mini servo motors. I've pulled one apart, and it's easy to remove the rotation restriction. However I will be looking at an alternate hardware driver and an alternative method for feedback.

The approach is to use a igus N-27 rail for the Z-axis, with one servo on top with a long arm, connected to a push rod, which is then connected to the carriage. Not a linear arrangement, but something easily corrected in software. It has the additional advantage of having higher speed mid-stroke, and slower at the extents.

On the carriage, I am going to try another mini-servo to directly drive a T-piece tube (where vacuum connects), then use a solder paste nozzle with silicon pad for pickup. So far, the design looks very compact. I do have to resolve how to get the accurate feedback I want, as well as a spring loaded arrangement for measuring/detecting object collision on the Z-axis.

Ok... I back on track. Using servos is a much much better solution to using steppers. They are quite quick too! Moving the distance I need in less than half a second. With some magnetic positional feedback, the solution is so much easier on the mechanics, and far more compact. They are a little noisy though <zheet...zheet........zheet...zheet>.

I am including a spring mechanism, which will allow relief for vertical downward pressure. Electronics will detect this, allowing for automatic height control calibration.

Now the Z-axis is so fast, I may have to re-consider the use of steppers for the X & Y. I have NMEA 1.8 degree bi-polar steppers. Not sure have fast I can rotate these with the necessary torque, however I don't think they will be as accurate and as fast as a good mid-sized DC motor with worm drive to belt arrangement. I don't think the backlash will be much of a problem given I'm looking at a 40 tooth gear (driven by worm gear). It is noisy though driving the worm drive at speed. The worm drive will also be used for an optical sensor, which will give some great position precision.

Maybe some good idea can be taken from new Madell
pick & place , see
http://cgi.ebay.com/New-Automatic-Table-Top-SMT-Pick-and-Place-Machine_W0QQitemZ3804674511QQihZ017QQcategoryZ58293QQcmdZViewItem#ebayphotohosting

Very nice... thanks for the post.

I'm still working on mine, though it is designed for a much smaller working area and without tape feeders (use tape strips instead). I'd really like to see a close up of their PnP head!

If you go here
http://espares.assembleon.com
you can find some interesting pictures
Paolo

Fantastic site! Some really good diagrams. Now that I've seen them, I realise they aren't that useful afterall. The placement heads on them are quite complex from a hobby perspective. Not something easily designed and manufacturered. I'II certainly be continuing with my current approach.

Good site though... appreciate the link!

I'm actually working on the reflow side of the equation. I've look at the toaster ovens, but just not happy with them. I'm going to try my own design using nichrome wire. Have a cheap multimeter with thermal probe and optical output being feed into a micro, which by next week should be controlling the temperature. I'II will end up with something that can reflow a board following a temperature profile.

I built a reflow oven based on a very low cost oven
It is in some way similar to Elektor project
It works , I can have good SMD boards
...now I need an inexpensive P&P...
I buyed on ebay
http://cgi.ebay.com/Pick-Place-Rotary-Vacuum-Tool-w-Encoder_W0QQitemZ300046358555QQihZ020QQcategoryZ58293QQrdZ1QQcmdZViewItem

How did you control the temperature (traic? relay on/off?).
Did you use temperature feedback?

I liked the toaster oven idea, cheap, but concerned the thermal response would be very slow or not hot enough.

inertia of the system makes it slow.
In my opinion there is no need for feedback.
It starts, goes to setted temp, then stops
and "beeps" to remember to open the door to cool .
Elektor oven has feedback and very fine software.
I started with triacs, but now I use relay.
It works, no problems, and it is good for me.
Why do'nt you try elektor oven?
paolo

I don't really like working with high voltage, but also concerned that these ovens would limit the ability to fine tune the process.

I really want to following a reflow soldering profile, and also ensure the board heats uniformly. Not sure if this is possible with a small toater oven.

If you need an oven for professional use: buy it
If you need an oven fore some SMD board: elektor oven ( or similar)
If you want to test, try a 20 Euro oven , you can have good result even without electronic ( just to see what happens)
There are good inexpensive toaster oven with a fan
You can buy TWO low cost ovens, take heaters from one oven and double heathers on the other.
If you do not like mains voltage, please ask for help from an experienced technician.
Paolo

I would really like to follow a reflow curve, and know that it is in fact performing accurately. I can't risk having even prototype boards "well done" and have potential problems. I have done trials with a heat gun and multimeter thermocouple, and was quite successful with reflowing double sided boards. However I did find it difficult to control the temperature manually. The first attempt I made I accidentally went well over 300'C. Hence, prefer to have it automated.

The small/cheap toaster oven I looked at had a top and bottom element, with a metal slotted shield around it. Simple mechanical timer, with option to have top, bottom or both element on. I'm not sure if it contained a fan for convection.

What wattage oven have you used?

I would really like to follow a reflow curve, and know that it is in fact performing accurately. I can't risk having even prototype boards "well done" and have potential problems. I have done trials with a heat gun and multimeter thermocouple, and was quite successful with reflowing double sided boards. However I did find it difficult to control the temperature manually. The first attempt I made I accidentally went well over 300'C. Hence, prefer to have it automated.

The small/cheap toaster oven I looked at had a top and bottom element, with a metal slotted shield around it. Simple mechanical timer, with option to have top, bottom or both element on. I'm not sure if it contained a fan for convection.

What wattage oven have you used?

here is my SMD oven
I think it was +- 1200W

Very nice! :)

I was reading up on this thread...
http://www.elektor-electronics.co.uk/default.aspx?tabid=29&forumid=16&tpage=2&view=Topic&postid=697
Lots of people have mentioned trouble getting peak temperatures.
But I might buy the Jan06 article for some background material before making a decision about using an oven vs trying nichrome wire.

You can buy two low-cost toaster and add heathers from
one oven inside the other you convert to SMD oven.
This is difficult for home-made SMD ovens here in Italy, because we have only 2.700W ! So 1500W for oven, 500W
for PC, two lamps, solder iron....
I can have enough temperature without problems, time seems OK. Not tested with RHOS compliant solder paste.
But I can buy another oven (20 euro) and build my 40 euro -2400W smd oven. Where can I send you a file?

Here are temp:
1' 00 100 C°
1' 30 135 C°
1' 45 155 C°
2' 00 170 C°
2' 15 185 C°
2' 30 197 C°
3' 00 222 C°
3' 15 230 C°
3' 30 245 C°
3' 45 252 C°

I would like to acheive a system that can peak to 270'C for 10 second, though from practical experience I've found that 250'C to be sufficient.

The profile I would like to be able to acheive is...
0 25
60 150
120 200
150 260
160 260
220 50


I plotted this against your results, and believe it to be favourable.
Are your results from maximum heat, or have you controlled the rate of heat rise?

.

The profile I would like to be able to acheive is...
..........
150 260



.
Are your results from maximum heat, or have you controlled the rate of heat rise?

Yes, max power, relay always ON
I think you need a lot of power, at least a better toaster than mine!
please see post on your address
Paolo

I should add, I haven't had convincing results using the nichrome wire yet. I have been able to reflow solder, however using a 15V power supply at 3A is barely enough to power for a 20x20mm area at 5mm of height. I really need to setup a pre-heater under the board to assist.

You can buy two low-cost toaster and add heathers from
one oven inside the other you convert to SMD oven.
This is difficult for home-made SMD ovens here in Italy, because we have only 2.700W ! So 1500W for oven, 500W
for PC, two lamps, solder iron....
I can have enough temperature without problems, time seems OK. Not tested with RHOS compliant solder paste.
But I can buy another oven (20 euro) and build my 40 euro -2400W smd oven. Where can I send you a file?

Below is a link to an artile on the SMT oven I built.


http://http://www.homanndesigns.com/SMDToasterOvenProject.html

IT may give you a few tips.

Cheers,

Peter.

The link re-directed to something a little more x-rated than expected haha... but found it via the main page ;-)

http://www.homanndesigns.com/SMDToasterOvenProject.html

Had a good look through the report, very interesting thankyou!

I'm going to continue with the nichrome wire for now. With further tests, I believe a proper reflective metal enclosure and a decent transformer will provide the power needed for small boards. Will need a slight airflow and also a resistor bank to assist and help with pre-heat, but the temperature control has a good response.

Hi guys,
Any progress with your P&P machines?
I'm also thinking of building one later this year, must get the mill modified and running in cnc 'mode' first.

About the P&P...I can't really decide how to drive the XY axes. I'd prefere to have the servos drive the axes by timingbelts, 'rack&pinion' style.
For better accuracy and repeatability the servo's encoder could be replaced with a linear scale placed along the axis...expencive, messy, and potential problems with servo stability/hunting

Yet another other option is to use ballscrews, either direct drive, or geared down with timing belts. Screws would have to be fairly high lead, probably 10mm or so, and this will obviously require preloaded ballnuts. All in all an expencive solution.


Anyone got far enough with your machines to advice for or against any of the above solutions ?

Well... my small reflow oven works FANTASTICALLY!!!!

Forget the toaster ovens. Too big, too hard to control. Small reflow box running nichrome wire from a 20A 13.8V power supply and some CPU fan/heatsinks. Works really well. Be using it for production work. In a basic comparision, a board that took 29mins with the soldering stick took 18mins with the paste and reflow oven. But when I build multiple boards at once it feels more like double. The first "build for hours" session I built twice the number of board I ever have!

I did have to implement a PID control loop to get really good temperature profile tracking. The nichrome wire needed to be of a specific length to match the power source and the windings need to be varied to even the heat (otherwise centre gets too hot). In addition, and very importantly, you mush have airflow to get consistent reflow. Otherwise, stratisifcation becomes evident.

For the pick'n'place... now that I have the reflow working, the pick'n'place with solder paste dispenser becomes very realistic!

I'm using DC motors geared from worm, to pinion to belt pulley. I tried steppers, but it was constantly a problem to get the resolution, torque and speed. The trade off is alway causing a problem. The DC motor however doesn't have that problem. Will have to implement a PID control loop to get fast performance, as well as an optical encoder coming from the worm drive for precision.

For the PnP head... I'm actually using remote car servos! Mechanism is far more precise, and I can easily use a servo saver to prevent the needle from pressing too hard. They push/pull on linear bearings. I have a dual head arrangement, one for components, one for solder paste dispenser.

I'm up to design revision 16 on the CAD model. I have tried LOTS of different configuration, linear sliders, bearings, screws etc etc... But I'm very happy with revision 16, it's looking a realistic solution!

Now I just need to get my time machine to work so I can make it!

Well... my small reflow oven works FANTASTICALLY!!!!

For the PnP head... I'm actually using remote car servos! Mechanism is far more precise, and I can easily use a servo saver to prevent the needle from pressing too hard. They push/pull on linear bearings. I have a dual head arrangement, one for components, one for solder paste dispenser.

I'm up to design revision 16 on the CAD model. I have tried LOTS of different configuration, linear sliders, bearings, screws etc etc... But I'm very happy with revision 16, it's looking a realistic solution!

Now I just need to get my time machine to work so I can make it!

Hi,

I've been really keen to build a P&P machine that in in end I purchased one. It's a 17 year old Yamaha 84. (Philips 84CSM). Works well, but is very expensive for what it is.

I just couldn't afford the time to design and build a beast. I am still interested in building one, as they still apear overpriced for what they are.

Cheers,

Peter.

Hi,

I've been really keen to build a P&P machine that in in end I purchased one. It's a 17 year old Yamaha 84. (Philips 84CSM). Works well, but is very expensive for what it is.

I just couldn't afford the time to design and build a beast. I am still interested in building one, as they still apear overpriced for what they are.

Cheers,

Peter.

OMG!!! Homann Designs.... (looking at website) in Highett. That's less than 10 minutes away from me :eek: I'm in Parkdale!

I am slowing optimising the design so that it uses 3mm sheet aluminium and L shaped extrusions. This is in an effort to obtain the best assembly accuracy. It does take a lot of work to figure out exactly how everything assembles, including provision for the bolts. Dimensionally, the model is accuately very accurate. I hope when the design is finished, I can have the necessary components CNC'd.

Before anyone starts blabbing at me for certain blacklash error, servo inaccuracy and bearing slop, please don't. Everything is going to work with feedback control! Besides... if you look closely at the design you should be able to pick out certain concepts I have applied to obtain accuracy. Such as the precision of the servo is greatest at it's extent of travel (when closer to the PCB). The servo actuates in 0.2 sec for full swing, but its velocity is slower as it approaches the board.

I have gone through a lot of design approaches. This is revision 16... aka, the 16th design approach I have taken. It will be quite fast, but made for the hobby level assembly.

I tried concepts for tape feeders, but just too complex and expensive. So I will be placing a tape strip manually, and the system will pickup components in sequence until the tape is empty. Realistically, this is the most flexible and fastest way to getting this operational.

The system is so light weight, the best design kept coming back to using a big solid single X axis. One with a carriage that has adjustability for tracking. Dual rails was always a nightmare! The X-axis is long, to accommodate a decent length of component pickup area. The Y-axis is kept as short as possible, to minimise error. The rail used is quite solid though.

I've used a worm drive from DC motor, as I believe it has the lowest backlash error and provides a high rotation area to provide the optical sensor pickup. I can also change the counter gear to suit the tradeoff between speed and precision. This setup by far has been much easier to deal with than 1.8 degree steppers driving the pully direct. The torque with micro stepping just doesn't allow the speed, and I'm far more happy with optical positional feedback.

Only trouble I have at present is getting the vacuum line to the needle head, while providing the flexibility for rotation.

http://www.eclipze.com.au/files/picknplace/PnP_Revision_16.jpg

OMG!!! Homann Designs.... (looking at website) in Highett. That's less than 10 minutes away from me :eek: I'm in Parkdale!



If you give me a bit of warning, I'll set up a viewing of the machine operation. It will give you a lot of ideas. I'm just off Nepean Hwy.

A couple of points though.

1. Run the Y axis out the other side so you can have feeders on both sides. You WILL need them.

2. Consider having 2 X rails, one on either side of the PCB. If you don't want this, then at least have a dolly wheel to support the cantilevered weight.

3. Not sure why you want to make the machine light. The machine I have is 600+ kgs. The weight helps with the accuracy of the machine.

4. Condider making the Z axis a Pneumatic slide. Or at the very least, have the servo only lift the axis to stop it dropping down. Allow it to go down via gravity or a weak spring. THat way you won't RAM it into the PCB by mistake. Another thing is that my maching only has about 8mm of travel in the y-axis.

5. The head only needs to rotate +/- 180 degs, so you can use a small universal pneumatic coupler for the air to the read.

6. Not sure how you are planning to do the feeder advanchment. For what it is worth, my machine has a plunger mouned to the side of the head, so that when it lowers the head to pick up a part, the plunger pushes down on the feeder advance mechanism. As the read raises up of the feeder, the feeder advanches to get the next part ready.

Just a bit of brain food. Keep up the good work.

Cheers,

Peter.

Peter,
please help us with detailed pictures!
You know what we need.
Detailed pictures, dimensions, ideas for Z axis from
your P&P: everything can be useful!
Thank you
ciccio

I’d be keen to have a look at that machine phomann :)
Early April is best, as I won’t get a chance to work on this until then.

I don’t wish to run the Y-Axis to both sides, as I don’t want to allow components to fall on the rail. It also complicates the design quite a lot. I only want to have a small placement area. This is a baby PnP ;-)

The weight is not an issue. The rail with such a short length is just not going to flex with such lightweight component, and such a short travel. A second rail becomes an alignment nightmare. I’ve tried this with several variations, mounted bottom, top, and even on their side. I dolly also provides drag and variation in the design. I’m placing my bets that this will be the optimal chose for both accuracy and simplicity, given its small scale and adjustable X-axis bearing.

The servo has a servo saver on it. If the servo pushes too hard, the mechanism will just spring back. I can control the torque by controlling the DC magnitude. I have tried spring arrangements too. Pneumatics I’m not familiar with, and prefer to keep the mechanism controlled by motor. 8mm travel (assume you mean the vertical Z-axis). Much shorter than I’d planned. I’m accommodating 20mm.

Very interested in seeing out this pneumatic coupler works. 180 degrees is certainly fine!

I’m not doing feeders. Too complicated and after a lot of though, not that practical for this small design. I want the machine to PnP only a few boards at a time. A strip of tape manually feed into place gives plenty to work with. For 0805’s, it will give around 80 components. Larger components have multiple tapes. The reels will be on a common rod, and manually feed through to expose the next strip of components. The machine will let me know when it has run out of a component and requires further feeding. Perhaps in future versions a feeder as you suggest would be worth looking into. I’m trying to narrow the scope at present so I can get something up and running first.

Cheers,
Tony

I’d be keen to have a look at that machine phomann :)
Early April is best, as I won’t get a chance to work on this until then.


Hi Tony,

Early April should be OK. Just let me know when you are ready.

As you your comments. The way you are proceedijng seems fine. At least you are actually building something. I never got that far.

Cheers,

Peter.

I'm not sure whether you have seem this video:
http://www.youtube.com/watch?v=bbBr9cbMqq4

It is a CNC mill doing SMT pick'n'place. At least there is an interesting way of component alignment :)

That's an awesome effort!!!

Quite interesting that it is operating without a rubber pickup. Without any spring back with the head, you'd want to be confident of the pickup hieight. I would have though using just the pick pin head and an alignment strategy as used would have resulted in a lot of dropped components. Obviously not the case!


Good video!

That's an awesome effort!!!

Quite interesting that it is operating without a rubber pickup. Without any spring back with the head, you'd want to be confident of the pickup hieight. I would have though using just the pick pin head and an alignment strategy as used would have resulted in a lot of dropped components. Obviously not the case!


Good video!

There is no need for rubber pickups. My Yamaha 84 has 3 heads, all without ru