[flippersplace]

I have a working DIY pick place machine

State as at July 2011:

[ame=http://www.youtube.com/watch?v=4DFtt7RTR2k]DIY Homemade PCB SMT Pick Place machine 2011 July - YouTube[/ame]

I'm fairly active with incremental improvements

I keep track of the details of the machine are on my own webpages at Home made pick place SMT machine

This thread is for any discussion re this machine

[flippersplace]

This change is the first attempt at converting from recycled DVD drive heads stepper mechanisms to pneumatic actuation

Goal is to reduce complexity, and vibration (DVD steppers are half stepped and are a bit 'clicky' contributing to parts moving around during up / down strokes of pickup heads )

Also, I now have cool as 'pssst' sounds from pneumatic solenoid exhausts

Details here Changes 20110829

Before & after pics below, as well as rear of machine after a bit of rearranging to make room for the pneumatic valves etc

Video to come soon

[James Newton]

wow... Questions that pop to mind: What sort of steppers and drivers are moving that? You program pic's and live in NZ? I know several people who fit that description, do I know you? How about 4 alignment boxes so the parts on all 4 heads can be aligned at once (assuming they are the same size parts)? Other random queries will certainly follow...

[RomanLini]

Good to see your build thread up! It didn't take you long to get those air pistons in there.

How expensive does it work out for the pistons and also air solenoids to control them? I was lucky already having a nice fast 12v air solenoid in the junk box, and I've been dreading buying more solenoids although a multi-pickup tool like yours looks real tempting!

[flippersplace]

@James
No we do not know each other

The steppers are generic Chinese Nema 23 76mm driven through a generic Chinese TB6560-3V2 via parallel port

Screws originally were 1605 but due to warp the bottom one has been replaced with a 1/2" 10tpi thread as that's what I had spare

Re 4 bump pits, I do have 4 for each head but they are not quite aligned to the space between the heads. Well, actually, I have 2 per head, one large & one small.

I wasn't too worried about getting them precisely spaced to match the heads as even my heads are not precisely identical, (I can't fabricate things ie brackets etc to that level of precision)


@Roman, I bought 20 sets of more or less identical pneumatic assemblys which gave me the premanufactured pneumatic actuator including linear slide, festo cylinder, 4 way valves, SMC speed controllers etc etc. When I saw them initially I could see how easily I could use them for this purpose.

A lot of parts for NZ$65 each, shame I had to buy so many though. I figure I'll sell off some of the spare bits as I now have 20 x 4 way Festo 24V valve assemblies, 80 SMC AS1000 speed controls, 20 SMC rodless cylinders etc etc etc

I was amazed with the precision of the micro linear slide shown

Photos are:
- Pneumatic assembly as purchased
- Donor actuator from above
- Dissembled parts I reused
- A new head ready to bolt in

[flippersplace]

Today. second DVD head in 2 weeks that didn't retract

So I just bolted in & calibrated one of the heads removed a week a go from right side bank in the interim (took literally 5 mins as they are all interchangeable)

I had been considering upping DVD drives from 5V to 6-7.5V to try avoid the above. Had played a bit with that in the past and they get a bit warm at anything over 5V however my PIC firmware powers down a head pretty quickly after a move (in anticipation of the above voltage increase) so it was still an option

Instead I'm just gonna move ahead and convert the second bank of heads to pneumatic

I'd already made another 4 brackets 2 days ago and had almost finished modifying the linear slide brackets etc per the other set of 4 heads yesterday anyway, just a small bit of filing the plastic left

However, I am short a 4x Mosfet driver PCB for the second bank of 4 pneumatics valves. I have a new 4 x mosfet PCB (all SMD except jacks in the post at the moment from iteadstudio.com (US$9.90 + post for 10 50mm x 50mm PCB's, just like seeedstudio, probably the same company for all I know) and the relevant SMD optoisolator & Mosfets shipping from Mouser as well, but they might be days away for all I know

So reluctantly got out the laser printer / hot iron / toner transfer and made another of the above this morning old school like and will have all pneumatic heads Monday.. I did take the opportunity to use some bright component colours however (its good to find additional creative outlets)

Still tweaking supply pressure vs additional springs (is a one way festo cylinder) as occasionally, one of the new pneumatic heads will bounce adjacent parts out of the exposed tape.. but not too often. Its really important to pad a peeled tape underneath so there is no give to minimise this

Basically trying to balance down stroke speed/impact/downward pressure vs upward start abruptness/ speed / stop abruptness

Machine was placing the 0.5mm pitch SSOP IC's for the current PCB but not always accurately. Seems some issues to resolve with the alignment of each PCB origin.

When I first ran this machine I used the downward looking camera to align to the edge of the PCB at the PCB origin. However I noticed that the guillotining is really variable in the same batch (I am assuming each PCB is at least square...) so I changed to use the downward camera to visually align to the centre of a mounting hole (all my PCB's have mounting holes 4mm in from each corner) but either the SSOP footprint is not centered or the PCB layers are not aligned accurately enough. I'll investigate this over the next day or so as well

Regardless, I think I'll end up nominating a pad from each PCB and use that instead

I do have to align for each rotation at the moment but it's very quick, ie Click VB dialog move button, jog a bit, click set VB dialog button

I did buy some smaller steppers with the idea of bolting a new pneumatic head assembly to one each of these but between the trade off of additional mass on the gantry, extra control lines per head etc etc I still think any effort I put into that area will be rotating the PCB instead. I have a small order of various T5 belt and pulleys coming from the UK so I can perhaps experiment with options. Additionally my Excel to GCode generation is sequenced by rotation so it should fit nicely.

Long term if I one day get to a vision system for centering parts, whether I rotate the head or the PCB its pretty much the same maths anyway

Roman, in your own thread you mentioned about dropping the part into a shaped pit for centering, iI had tried that but I can't make the pits accurate enough. However at the time I thought this was an ideal time to rotate the part as well. Not gonna happen for me but maybe for someone else.

I did download a trial of RoboRealm last night but it looks too complex to spend time on right now

[RomanLini]

Roman, in your own thread you mentioned about dropping the part into a shaped pit for centering, iI had tried that but I can't make the pits accurate enough. However at the time I thought this was an ideal time to rotate the part as well. Not gonna happen for me but maybe for someone else.
...

I think I will still give that a go, for common sizes like 0805 and 1206 anyway. My CNC is pretty good for fine precision work although I need a 1mm diameter ball end mill, currently I only have 1mm dia straight end mills. If the smallest part of the pit is say 10% larger than the 0805 it should locate the part to within +/-5% which should be plenty accurate enough for 0805.

Thanks for sharing your work with the pneumatic heads, it is interesting hear about tuning the pressures and the parts tape bounce caused by overspeed etc. I'm still sticking with RC servos for head lowering because I have the servos and because I have some experience tuning them for speed and position in small robotics applications.

Regarding your PCB alignment, you really need to work with aligning to the top copper as all other layers (holes, silkscreen etc) may be out of place in the PCB manufacture. You might be able to get away with holes for alignment, although I have received PCBs back sometimes where the vias are drilled noticeably off centre (so all holes were) and that might be enough to upset a TSSOP alignment.

Can you give more info on your tape holders? Are they machined to hold the tapes snug to allow accurate blind pickup, as that is the system i was hoping to use. I generally use lots of 0805 and 1206 and hoped for good enough blind pickup and placement without really needing to bump align. That's the theory anyway!

[stewi]

Air activated component feeders used to be trouble.
Back in the time, we had successful assembly houses in US, I designed and patented a feeder concept. Apparatus for feeding electronic ... - Google Patents
At a low speed you are going with a single nozzle head, you would not require the fast acting air piston motion for the feeder. If you are worried that the components flip out of the tape pocket during advancement, you would need a component cover, which pulls back before component pick up.
However, this is an interesting thread. I started working on 4,000 component an hour placement machines and they do easily 40,000 now.

[RomanLini]

Thanks for the link to the patent Stewi. It doesn't look that much different from the feeders that were in commercial use since the 80's? What were the feature(s) original to yours that had never been done before?

[stewi]

There are a few things unique to other feeders.
The feeder has a base with one advance motor and several solenoids in order to engage only the required feeder. The feeder itself has only a few inexpensive plastic components and is meant to remain on a half empty tape, for minimal change over time. The cover tape is peeled off and than guided towards the empty tape. Empty tape is cut off by most of the placement machines and chip shooters.
The feeder base was meant to adapt to different machine vendors, such as Siemens, Universal and others.

[flippersplace]

Can you give more info on your tape holders? Are they machined to hold the tapes snug to allow accurate blind pickup, as that is the system i was hoping to use. I generally use lots of 0805 and 1206 and hoped for good enough blind pickup and placement without really needing to bump align. That's the theory anyway!

Hi

My holders are the generic chinese made things from Ebay like this one

1 set 5 way SMT SMD Feeder for DIY Prototype Pick Place | eBay

However, regardless,yes the tapes are held firmly, and if not I insert pins to ensure they are. I can calibrate each feeder using the downwards USB camera very quickly as I have a custom VB dialog just for that purpose

However the primary issue is that the fit of the parts in the tapes is such that the parts can move around. So the accuracy of your place is determined by this slop.

Picture is 0805R

Do a few R's or LED's in a row and it soon looks pretty untidy & unprofessional

[stewi]

If you look at the spec than tape width can deviate by +/- 0.1 mm. All automatic feeders advance the tape with pins in the perforation holes, because holes and component pockets are punched out together more accurately. Tape pockets are over sized to accommodate component size variations.
Together with the fact that the solder pads are larger and the re-flow process does some pulling as well, an automatic pick head with +/- 0.1 mm accuracy should be doing blind pick and place with 1206 and 0805's without optical or mechanical component alignment.

[RomanLini]

Thanks Stewi for the info on the tape feeders you designed.

Flippersplace, thanks too for the info on those feeders. I didn't know you could buy tape guides pre made, I was just going to machine slots in some plastic to make my own tape guides.

Thanks too for the photo. The 0805 tapes I have here don't seem to be quite that "loose" yours looks like almost 20% clearance. Mine seemed to be 15% at most and I was going on that assumption...

Basically the idea was to use the solder pad reflow centering (as Stewi said) and pick and place blind, ideally there should be less than +/-10% position error on a 0805 and my machine should position to better than 0.05mm.

I lathed up my tape feed pin yesterday, I made it 1.54mm which is a "perfect" fit in the tapes with no slop I can notice. That should be able to position the tapes very well (hopefully). It's easy to be confident of the machine before it's tested haha.

Time will tell I guess, but I was really hoping to be able to place 0805 and sot23 etc blind and quickly. If it's going to need a camera and/or repositioning that might change some design decisions...

[stewi]

Assuming you are doing this for low volume application, you could be using strips of tape and not the entire 7" tape reel. In this case you could position all required components in one carrier plate and place this plate on alignment pins closed to the board you want to populate. Remove all cover tapes.and program your machine that it picks up the components from one start position and the next +/- 4mm off the first position (I guess this is what you had in mind anyway).
If you want to do blind pick than plastic tape has some advantage versus paper tape. The component pockets from plastic tape have a draft angle and the component is centered.
If you do low volume, you can likely not specify paper or plastic tape. If you are using strips of tape and the cover tape is peeled off you could slightly tilt your carrier plate to move all component into one corner of the tape pockets, before you insert the carrier plate into the machine.
Keep in mind, if you make your own tape carrier plate to use alignment pins for the tape's perforation holes. The outer dimensions of 8 mm tape are too inaccurate.

[stewi]

I lathed up my tape feed pin yesterday, I made it 1.54mm which is a "perfect" fit in the tapes with no slop I can notice. (

I only find the spec for larger tape, but the holes are the same 1.5 +0.1 -0.0
http://www.ec-central.org/PDF/Engine...ia-481-2-A.pdf

[stewi]

If I can’t hold you up designing your own automatic tape feeder here are some of the challenges:
The permitted deviations on tape width, tape pocket sizes and perforation holes are not precise enough for reliable pick up of components smaller than 0603.
8 mm feeders have to accommodate paper and plastic tape with very different physical characteristic in fast feeding motions.
Pull forces on the cover tape deviate too much. In order to peel the cover tape reliably, more force is required, which will move the tape itself forward against the advancing pin.
With the use of a sprocket wheel in order to advance the tape, the tape has the tendency to wrap around the wheel and will require a peeler to release the tape from the wheel.
The sprocket teeth exiting the tape create a vibration to the tape, which can eject the components out of the exposed pocket.

For all the above known issues machine designers came up with solutions:
Each component is inspected by a camera vision system. X, Y, Z, deviations from the pick up nozzle are then corrected upon next pick up.
The sprocket wheel and the cover tape are driven by two separate motors. The sprocket wheel advances the tape forward, than turns slightly backwards in order to engage the advancing pins with the rear edge of the perforation hole, this way the peel off force of the cover tape are not moving the tape out of position.
An additional solenoid activated slider covers the component to prevent rejection out of the pockets during fast motion.

[stewi]

Adding some pictures to the feeder design, I made.
As mentioned the feeder contains only some mechanical parts. Motors, solenoids and control logic are located in a base.
The component is covered during tape advancement. The empty tape is disposed together with the peeled cover tape. Most of the modern chip shooters are cutting the empty tape, with a fast acting knife.

I spent quite some money on patent and development. It didn't pay off.

[RomanLini]

Wow Stewi thanks a lot for the detailed info on the tapes and tape feeding. I'm a bit worried that we are hijacking Flippersplace's build thread so maybe he can comment on that? We can always discuss this in another thread?

My machine (you can see half finished here; http://www.cnczone.com/forums/rc_rob...ight_pnp.html?) will be using a "feed pin" for tape feeding.

This is mounted on the XY head and the pin will auto-insert in a tape hole and feed the tape (X or Y) forward any required distance. For simplicity the tape cover would be peeled back manually, or maybe with a small weight and aligator clip as successfully used by others.

I was going to machine the tape guides to be a little snug on the tapes so they stay in position after being moved by the feed pin. Then parts can be picked up from about a 80mm exposed length, so MANY parts can be picked, then a single (up to) 80mm feed movement performed.

Possibly the feed will be done after the PCB is finished, so feed and cover tapes can be done and inspected by me before the next PCB starts.

The design of my machine was to do blind work; lots of SOIC, 0805, SOT23 etc at reasonable speed. I was hoping to avoid cameras and alignment procedures for most components. I appreciate your voice of experience on the tapes and feeding issues!

Sorry to hear about the patent not working out too. And Flippersplace, please speak up if you want this moved to another thread!

[stewi]

We could be moving, opening another thread or request the administrator to get moved. On the other hand, this thread doesn't have much attention anyway as CNC folks are likely not interested in surface mount technology.
However, some quick notes on your concept.
First, you are not mentioning any theta axis. You'll have to layout resistors in 0 and 90 degree and SO IC's likely in all 4 different orientations.
For blind pick and place this is actually a better way to go since the rotation of a component will double the offset error.
Second, when you advance the tape with a pin, I would recommend you go in opposite feeding direction for 0.1mm or so (and make the pin dia only 1.3 or 1.4 mm). Else, you pull the pin, while it is reaming against a perforation hole and all components will go flying.

[stewi]

I'm reading your posts with interest, although I feel going back in time 20 years.
I know quite a bit about these machines [ame=http://www.youtube.com/watch?v=ur0njbsHLi0&feature=related]Siplace Multistar - YouTube[/ame]
With fast movements like this one, vacuum shut off and parts blow off from the nozzle is quite interesting. Air moves unfortunately only at sound speed.
Let me know, how I can contribute to this thread.