> Hi, Roman
>
>
> > Hi everyone, has anyone built a home made
> > CNC pick and place machine to do SMD work?
>
>
>     I have the first one built from printer parts and it works but I am
> going for the second one with real guides. The linear guides are cheap
> enough and will give better results. The one I have built is too big, too
> ugly and too slow :-(
>
>     Take a look at http://www.igus.com and you will find nice linear guides.
> Subscribing to the http://groups.yahoo.com/group/CAD_CAM_EDM_DRO  will

> printer. If you decide to go to belts instead of screws for driving the
> table take a look at http://www.smallparts.com . They have nice parts.
>
> > I have an idea for a new product that will
> > require lots of 0805 SMD parts and I like the
> > thought of using a machine to do glue spots,
> > then pick and place, then once parts are glued
> > I can just hold them into a solder cauldron
> > (with bamboo tongs!) to solder them in one go.
>
>     I prefer using the machine to put solder paste and then the components
> and use the toaster oven method. You do not need any glue !! Even if you

> if you have a better idea on this areas.
>
>     I have been looking at this problem for a while and if you need more
> information I might have it...
>
> Best regards,
> Alexandre Guimaraes
>
> --
> http://www.piclist.com hint: The list server can filter out subtopics
> (like ads or off topics) for you. See http://www.piclist.com/#topics
>
>

> put on the components pads. The paste holds the components in place. A
> vacuum twezzer will also be very helpfull. With the right temperature
> profile and good board layout you can get results that are so nice that no
> one can tell the boards were not soldered in a big convection oven :-)
>
> Best regards,
> Alexandre Guimaraes
>
> --
> http://www.piclist.com hint: The list server can filter out subtopics
> (like ads or off topics) for you. See http://www.piclist.com/#topics
>
>
>

> >toaster oven? solder PASTE? wha?!?!?
> >woefully unaware,
>
>         Look for the page of Kenneth Maxon and you will understand everything ;o)
>
> ---8<---Corte aqui---8<----
>
> Alexandre Souza
> TakeThisOuTtaitoEraseMEspam_OUTterra.com.br
> http://planeta.terra.com.br/lazer/pinball/
>
> ---8<---Corte aqui---8<----
>
> --
> http://www.piclist.com hint: The PICList is archived three different
> ways.  See http://www.piclist.com/#archives for details.

> Yep, oven was my first idea. I'm currently doing
> solder paste by hand, and toaster oven. But, after
> some tests it is a LOT slower than glue and dip
> technique. Modern VCRs are made this way, they use
> a heatproof red glue spot, pick and place SMD, then
> wave or dip solder the board. Soldering takes about
> 4 seconds, even with my crude test attempts. That's
> much faster and easier than in a toaster oven.
> You don't need paste, which is expensive and goes
> off, and it's easier to build a glue applicator
> than a paste one as glue flows better.

> > I have a suggestion here. Get the parts loose,
> > not on rolls. Sprinkle them on a white tray,
> > then use a cheap video camera into the PC and
> > just get the picker to identify the part
> > orientation and pick it up. You only need a video
> > still capture board,
>
>     Hey, that is a really cool idea ! The main problem would be to correlate
> the real position of the components to the distorted camera view and to
> solve the problem of refilling the trays and making sure that you do not
> have one component over the other or on the back side when using SOT23
> transistors ! It is pretty hard to get the components out of the rolls and
> not flip half of them on their back side. I am really not sure what would be
> easier, a roll feeding system or the camera.

>
> THANKS Alexandre for the excellent info you
> provided,
> and sorry I didn't reply sooner but I have been busy
> in the workshop. :o)
>
> > > it has good power, high speed and 1/28th mm
> > > resolution is fine for 0805 SMD stuff.
> >
> >     How about the repeatability and backslash ?
> Those are the problems I
> > have seem on the printer mechanisms I used. The
> printing head support was
> > too loose without the belts being too tight.
>
>
> It depends on the quality of the printer mechanism.
> I have some 15" NEC Pinwriters from mid 80's that
> are VERY rugged, full steel chassis and 2 large
> steel
> rails. Some of the smaller printers are not good at
> all. One of my 15" printer chassis cost me $5 and
> has a 25mm hardened rail with PB bearings! Don't
> ignore good old printers.
>
> I just chose a chassis to make a CNC PAP (PNP?)
> machine,
> it's a 15" unit with steel cable drive from motor
> to carriage, even less backlash than a belt. Even my
> belt chassis get a lot less than 0.1mm backlash, the
> cable driven unit will be good for 0.05mm
> repeatability.
> How accurate do you need for 0805 placement? I'm
> sure
> even 0.1mm will be ok.
>
> >     Do you need just one machine ? I plan to have
> a solution that can scale
> > up with my production. That is another reason to
> go for comercial new parts.
>
> Sure, but you are obviously doing the volumes to
> need
> good machinery. I'm still sticking parts on boards
> with tweezers and a magnifying glass! The thing i'm
> making will do 300mm x 120mm, perfect for pick and
> place on small boards.
>
> >     Loctite has some nice solutions in that area
> and you can even use 454
> > gel cyanoacrilate !! It works great !! The only
> problem is that you need a
> > good dispenser for it. Dispensing paste is not
> that critical, specially if
> > you use water at the end of the system. Water does
> not compress so it is
> > easier to control the time and pressure as the
> syringe goes down. You just
> > have to control the time and pressure and it is
> quite repeatable.
>
> OK, you've convinced me to drop the glue idea and
> just dispense solder paste direct to the board. :o)
> How do you mix the water with the paste, what sort
> of applicator nozzle, pump etc are you using?? I
> thought
> of standard size rubber hose in a simple peristaltic
> pump turned by a stepper motor. That should dispense
> dots and also lines of paste quite well.
>
> >     Wave solder is ok if you glue and have a very
> controlled fluxing and
> > cleaning of the board but I would never trust
> dipping a critical board ! You
> > will stress every component over their specs !
>
> Not totally true, but i'm going to trust you and
> stick to the paste and oven method for now as you
> suggested. You helped me so much before with my
> manual toaster oven system. :o)
>
>
> >     If the production is too big I would go for
> the comercial machinery or
> > let a factory do it for you. It may not be worth
> the risc. My production
> > levels do not allow sending them out and I cannot
> handle it without some
> > automation, that is the reason I went to the house
> made solution. It is
> > dependable and repeatable but not too fast or
> energy efficient.
>
> My oven system is working well, thanks in part
> to your temperature charts etc. I'm approaching
> that point where it's a pain to tweezer tiny parts
> by hand, and a machine that does the SMD placement
> will be a big help. But I don't need high speeds
> or quantities.
>
>
> > > I have a suggestion here. Get the parts loose,
> > > not on rolls. Sprinkle them on a white tray,
> > > then use a cheap video camera into the PC and
> > > just get the picker to identify the part
> > > orientation and pick it up. You only need a
> video
> > > still capture board,
> >
> >     Hey, that is a really cool idea ! The main
> problem would be to correlate
> > the real position of the components to the
> distorted camera view and to
> > solve the problem of refilling the trays and
> making sure that you do not
> > have one component over the other or on the back
> side when using SOT23
> > transistors ! It is pretty hard to get the
> components out of the rolls and
> > not flip half of them on their back side. I am
> really not sure what would be
> > easier, a roll feeding system or the camera.
>
> You don't have to buy on the roll. In 1k
> quantities you can often choose loose or rolled.
> I see the only real work is pouring the parts
> on a tray, making sure they are right side up.
> Parallex error shouldn't be too much problem if
> the camera is 40cm or so from the tray. I can live
> with some manual parts fiddling, anything is much
> better than placing with hand held tweezers. :o)
>
>
> >     Believe me on this one :-) It is easier with
> the vaccum tweezer. You
> > just turn the vaccum on and them rotate the
> component to whatever position
> > you need. With the mechanical tweezer you have to
> get the right force and
> > will have to rotate the head 2 times to get the
> job done. The first time to
> > grab the component and the second to put it at the
> right orientation at the
> > board.
>
> Ok, you've sold me on the vacuum tweezer. Are tips
> commonly available? What type of vacuum pump are
> you using? I have a vac solenoid from a bike fuel
> injection system, a high speed one. Any suggestions
> on how to set it up? Does it ever drop parts?
>
> > > Gluing is also safer for
> > > double sided boards??
> >
> >     I also thought that way and got glue from
> loctite for the components on
> > the bottom of the board. I still have the glue on
> the fridge and never used
> > it after the first day ! The only components that
> insist on falling once in
> > a while are the AVX surge supressors. They are
> pretty dense and fall once in
> > a while. If you get the right solder paste the
> surface tension is more than
> > enough to hold the components on the bottom side.
>
> Good information, thank you again.
> Any help you can provide on the CNC PNP stuff
> will be very much appreciated. :o)
> -Roman
>
> --
> http://www.piclist.com hint: The PICList is archived
> three different
> ways.  See http://www.piclist.com/#archives for
> details.
>
>
>
>

> Don't forget that printers are pretty hot on relative accuracy, but in terms
> of absolute positioning of the image on the paper tolerances of >1mm are
> considered the norm.
>
> You can see my 1st attempt at
> http://www.mailmaster.freeserve.co.uk/ncdrill/drill.htm Boy it's hard work
> without machine tools! The main problems I had were belt backlash and the
> 12mm rods flexing. Actually, because prototype PCBs are so cheap now I just
> use it for routing cut-outs in front panels.

> > I thought
> > of standard size rubber hose in a simple peristaltic
> > pump turned by a stepper motor. That should dispense
> > dots and also lines of paste quite well.
>
>     I use a paiting air compressor. They are cheap and work very well. Just
> have a big air cylinder and you can turn the compressor on just once or
> twice a day. I live the air line with just 40 PSI on it so I have very low
> risk. I tried many small pumps and they all make too much noise and end up
> costing more than a good membrane pump. Just go to the hardware store. For
> the cylinder to hold air I use disposable freon cylinders. Very cheap ( free
> most of the time) and reliable enough for 40 PSI ! The solenoid air valves
> should not be hard to find also.

>
> Hi,
>
> > Do you have a picture you could share of this smd needle pickup doohicky?
> > It doesn't have a vacuum break valve on it?

>  I think I will even try ascii art for the first time in my life :-) It
> is as simple as it can be. Just put a neddle that you have cut and filed
> flat inside a silicon tube of appropiatte diameter and it is done. It is
> even quite comfortable ! You do not need the break valve if the vacuum is
> just strong enough to just barely hold the parts. They will fall off with
> the surface tension of the solder paste. The parts are much lighter than we
> can imagine. The only problem is with round parts, that we usually grab with
> mechanical tweezers or use stronger vacuum and a break valve. This decision
> I live for the operator, they usually use the mechanical tweezer for round
> parts.
>
>     ----------------
>                 ----
>                     \-------
>                     /-------
>                 ----
>     ----------------
>
>     If the vacuum is too strong and the parts are not "gluing" to the solder
> paste just make a small hole on the side of the needle and you should be
> fine.
>
>     By the way... I forgot to mention that the needle eventually "glogs"
> with the solder paste and you need either to change it or "deglog" it with a
> small "accupunture" needle ! These needles are quite strong and
> exceptionally thin..

> >
> > Hello,
> >
> > I have a source for membrame pumps.
> > These are used in professional printing press /
> > publisher.
> >
> > These are very suitable for providing air to
> .......
> > (your paste dispencer ??)
> > The problem would be how to turn the airflow off
> !?
> > (A valve ???)
> > And they work on 220 VAC.
>
> Thanks Simon. :o)
> Whats the source for the pumps?
> I already have a vacuum solenoid (12v)
> from a Suzuki motorcycle fuel injection
> system, it is a SPDT airflow if that makes
> sense.
> -Roman
>
> --
> http://www.piclist.com#nomail Going offline? Don't
> AutoReply us!
> email TakeThisOuTlistserv.....TakeThisOuTmitvma.mit.edu with SET PICList
> DIGEST in the body
>
>
>

>
> > You are a legend Alexandre. :o) I have an aquarium
> > air pump and some needles I bought to use with
> > the solder paste syringe. Are these the right
> > type? So you just use the metal tube needle,
> > filed flat on its end, as the bit that touches
> > the SMD? I thought it had to be rubber or
> something
> > but your way sounds very easy to make. The
> aquarium
> > pump hose will match up to my vacuum solenoid too,
> > same size rubber hose.
>
>     Hey, you make me blush :-) These crazy ideas is
> what I always liked the
> most about the piclist.. We are not afraid of doing
> it in some different
> way.
>
>     The parts are so small and the vacuum so little
> that you will never hurt
> the parts not using rubber. Your description is
> perfect. Just file flat the
> needle. If you make the vacuum small enough the
> parts will fall off the
> needle as soon as they touch the solder paste !!!
> Works so good that you
> might want to forget about the automatic pick and
> place alltogether :-(
> Please don't.......
>
> > Ok, I understand now. With regulated pressure and
> > controlled time you can get fairly exact quantity
> > of paste dispensed. But is this suitable for the
> > repeated tiny paste spots needed to use 0805 SMD
> > parts?? And does the consistency of the solder
> > paste change over time and cause problems with
> > dispensing an accurate spot each time?
>
>     You can have more than 30% errors without any
> trouble at all. Just make
> the pads for the components bigger than the ones
> your software tells you.
> The consistency do change with temperature, pressure
> and number of
> repetitions but not enough to cause problems.
>
> > Will do. :o) But i'm hoping that people aren't
> > objecting too much to this SMD CNC topic? I would
> > think there must be a lot of us doing small
> > quantity manufacturing that would like a cheap
> > SMD pick and place machine.
>
>     I am curious of how many people actually have
> this kind of problem... I
> always fell that we are a group that basically has
> the hobbyist, the
> independent consultors and the guys with the big
> companies that can spend
> lots of money on equipament. Maybe I am wrong. If no
> one objects I guess if
> on topic enough for the [EE:] tag. At least we do
> not have any cows involved
> yet :-)
>
> Best regards,
> Alexandre Guimaraes
>
> --
> http://www.piclist.com#nomail Going offline? Don't
> AutoReply us!
> email RemoveMElistservspamBeGonemitvma.mit.edu with SET PICList
> DIGEST in the body
>
>
>

> > Hi Alexandre, this "glogging" might be a problem

>     The problem would not happen at all with a CNC setup. It is caused by
> the operator misplacing the needle when putting the component on the right
> place. It is also very dependent on the operator, some have to clean the
> needle many times a day and others once a week !
>
> > Would the addition of a vacuum solenoid, as I was
> > planning, to turn the vacuum off before the tube
> > is removed from the top surface of the part, fix
> > this problem?

>     On the CNC setup you have to have the valve and it would indeed keep the
> paste away from the needle. The component can be move a little further away
> from the paste and dropped over it. When you do it by hand you actually
> press the component over the solder paste.
>
> > Since my vacuum solenoid is a SPDT type I can
> > feed one tube with vacuum and the other with a
> > small pressure

>     I do not think you would need the blow. It should fall down by it's own
> weight. If you use the blow it needs to be really slight or you will blow
> the paste and the component away.
>
> > PS. And I would want the machine to be able to
> > pick and place round parts like SMD diodes too,

>     No need to change needles. Just use a stronger vacuum and the solenoid
> valve. That is how some of the operators do with the manual setup. Most
> prefer to pick the round components with a mechanical tweezer and some use a
> stronger vacuum and the solenoid valve.
>
>     The manual and automatic setup are quite different beasts. In the manual
> setup using a stronger vacuum is not the best way because you need to use
> the valve and use both hands or a foot switch to drop the components over
> the solder paste. If you use a very light vacuum the solder paste surface
> tension will "suck" the component from the needle. You can get the job done
> with just one hand. On the automatic setup the vacuum can be much stronger
> and round components are no problem at all.

>
> Thanks again Alexandre. :o)
> I've been checking some stuff and I'm getting pretty
> keen on this machine. The problem of getting high
> speeds and good accuracy I think is sorted by using the
> nice 15" printer chassis with the cable drive, and
> 8th step or 16th step microstep drivers. I already
> have 8th step drivers (1600 steps/rev) but want to get
> some boards made because i'm sick of veroboarding
> them.
>
> Looks like i've decided to go the opposite way to your
> machine (in the picture) and keep the travel distances
> short and build a small, very light machine to try to
> keep speeds high, instead of a big sturdy mill.
>
> Imagine something NOT like a gantry mill, but more
> like a box on a rail. 300mm travel. The box has a short,
> light unsupported arm out one side, with a head travel
> of about 140mm. The arm is light and the head is lighter.
> I've been toying with the idea of using long bent tubes,
> one for the paste applicator tip, and a rotateable one
> for the vacuum tweezer. These have small lightweight
> stepper motors, (from disk drives maybe?), to lower and
> lift the tubes. Have you ever seen inside a pen/chart
> plotter machine? The long thin pen arms can move very
> fast, and no real strength is needed for PNP, only speed
> and accuracy. :o)
>
> So I get only 300 x 140mm work area, maybe less, but
> the work area is on the corner and can fit large boards
> if they are partially hanging out. The advantage is
> a fast action, i'm hoping for 1 to 2 PNP's per second,
> from a machine basically built from junk but with
> good microstep motor drivers. I don't see the small
> work area as much of a sacrifice to get a fast and
> portable sized machine. Something you could put on
> the back seat of a car and take to a friends place,
> plug it into their PC and it's running.
> -Roman

> So I get only 300 x 140mm work area, maybe less, but
> the work area is on the corner and can fit large boards
> if they are partially hanging out. The advantage is
> a fast action, i'm hoping for 1 to 2 PNP's per second,
> from a machine basically built from junk but with
> good microstep motor drivers. I don't see the small
> work area as much of a sacrifice to get a fast and
> portable sized machine. Something you could put on
> the back seat of a car and take to a friends place,
> plug it into their PC and it's running.

>Alexandre Guimarces wrote:
>
>>Hi,
>>
>>>I've seen the discussions about moving arms / printer
>>>mechanisms for the CNC.
>>>But is it not easier to move your PCB on the X and Y
>>>axis and let the pick and place part just move X and Z
>>>axis to get parts ????
>>>
>>    I am unable to make a mental picture of that !!! Could you sketch it for
>>me ? The board moving X and Y is ok but where do you get your parts from ?
>>Would not you end up with 5 axis to control ?
>>
>
>I was getting ready to suggest the same thing. My lab has a CNC mill
>that basically does this. The head only moves in the Z (up,down)
>direction while the table, which is long and narrow, moves in the X and
>Y (horizontal) directions. The tool caddy is at one end of the workspace
>and can move independantly of the table. To change out a tool the
>part-in-progress is moved out of the way while the tool caddy moves into
>place to do the change out. You probably don't need the complexity of a
>free moving part caddy, but you could easily make the work space large
>enough to hold both the board and the parts pick-up area.
>
>I'm not sure if this method is really applicable to this case though.
>AFAIK mills do this because they need everything to be rock solid stable
>to get high enough tolerances. The tradeoff is that the moving
>components are much more massive (which is why they're so stable under
>milling loads) and therefore slower. For a mill, accuracy is more
>important than speed. A PnP machine doesnt need anywhere near the same
>level of accuracy but does need speed. I this case I think it really
>depends on what's less-massive, the workspace or the head.
>
>--
>http://www.piclist.com hint: PICList Posts must start with ONE topic:
>[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
>
>
>

> I quoted some strong driving system to use just 2
> arms, one mounted over the other and the pickup head on the end of the top
> arm and the mechanical system would cost over us$ 3000 !!! The idea is great
> but maybe harder to implement then it seems. I decided to go with the gantry
> style because the parts are cheap and readily available and everyone knows
> how to workout the design. I will spend about us$ 300 to us$ 400 in parts
> that are ready available, except for the motors that I already have scrapped
> from old printers. I will send you in a private msg some pictures of a very
> interesting design with just 2 arms. If anyone else want's it drop me a note
> and I will send the pictures. They are small and should be about 100k.

> > So I get only 300 x 140mm work area, maybe less, but
> > the work area is on the corner and can fit large boards
> > if they are partially hanging out.

>     The 300 x 140 mm work area should be enough for most applications. You
> may not need all the speed you are running for. It is nice to have a fast
> machine but remember that if the machine is cheap enough you can just make
> one more when you need more production. I prefer slow, cheap and reliable.
> Working with the mechanics people for a while I lost some of my urge for
> doing the best all the time, sometimes the best is the worst enemy of the
> good. In the electronics we can usually make it cheaper, faster and more
> reliable. On the mechanics we have to settle for one of the goals and try to
> work the others out the best we can.