[Eclipze]

Yes, indexing something like 5 at a time would certainly help speed things up, though would increase the complexity of the software quite a bit. Would also be problematic with initial condition expectations to know where the machine is up at power on. Certainly an efficiency gain could be had.

I looked at the igus N series linear rails, which are definitely the cheapest. Unfortunately, there is just too much play in them. Accuracy would be sacrificed in a big way. The T series however is much much better, and the ones I have acquired for the job.

The X and Y are pretty easy, but creating the head with the Z and C is difficult. I've gone through 18 designs in CAD. I looked an belt driving system and a range of linear steppers and methods of implementing it. So far, I'm using igus T rails for the Z and servo motors to move it, as it has presented the most practical solution. The C is done with a tiny 1.8 degree bipolar stepper. The indexing by a separate servo motor actuating a pin, nice and fast. I'm only targeting one pickup head, indexing pin and a paste dispenser. For larger components, I can change the head, but will probably just have the machine paste the footprint and place it myself by hand... unless there are a lot of heavier components where changing the pickup nozzle is worth the trouble.

I kept getting stumped by being able to actually make it, so I diverted to putting together a small CNC first which will be used to machine the parts for the pick and place.

[aaron800]

I'm curious what free (open-source would be best) software people use to do pick and place? Anything special?

[atlab]

Hi, guys.
Is there any progress?

[Eclipze]

Believe it or not, but the project for me is still going. Gone through so many revisions, it just not funny anymore. All in CAD. But I've finally settled on a design for the vacuum nozzle and solder paste head. Trying to add a bracket for two webcams that is also adjustable.

Also been working on the driver electronics design, and are playing with the Trinamic drivers. Had one moving a stepper, and so far so good to continue with them.


I'm only spending a couple of hours a week on it, so progress is slow.

[nisma]

Nothing special, just a usual modified cnc softwar including optical pattern matching.

[atlab]

Hi, Eclipze.
You are the only one who keep on this project.
I thought of small PnP machine for my home projects, but I have begun with the software and electronics.
I made some steps:
1. I have thought up and have tested the data exchange between the computer and the controling electronics (1 channel of 5). I have used RS232.
2. I wrote and have tested the program for microcontroller to control axis. There will be 4 microcontrollers to control axises and 1 to control the valves.
3. I wrote and have tested the software to correct the position (angle) errors. I have used the USB webcam 640*480 pics.

I develop the scheme and PCB for the control electronics now.
The project moves very slowly but it is very interesting.

[Eclipze]

Sounds fantastic!

My background is definitely electronics and embedded software, but it has been the area I've spent the least amount of time so far. I have found the mechanical considerations to be quite challenging. I have very much tried to limit the weight and target a reasonable speed for all axis. I'm not looking for anything really fast, it's more of a machine for short production runs than a manufacturing solution. Anything over a couple of hundred boards I get loaded externally anyway.

I'm now on the 20th revision of the CAD design, and believe I finally have the base solution that I can manufacture. Having a CNC machine has certainly changed the approach from the earlier models, so now a lot of it is about cleaning up the and modifying the brackets to be easily made.

[nisma]

For the mechanical design, don´t forget the pcb loading/unloading station, this will
simplify your life costing very little, It´s composed of 6 motors, two chains, some
spockets and four lead screws.

[atlab]

... Trying to add a bracket for two webcams...

Do you have an experience of using of two webcams in one program simultaneously?
VFW (video for windows library) does not support such possibility. So, i am planning to place webcam on base only.

P.S. I am listening Stewart Dudley now: "Memories Of Australia". Great music!

[Eclipze]

nisma... mine is a small desk setup, designed for prototyping or small production runs. A board loader would be way overboard for what I'm doing. As long as I can load a 1/4 panel and remove without any fuss, it will suit my requirements.

atlab... I hadn't realised this limitation existed. My intention was to have a webcam for placement and another for solder paste dispension. It is really more so for convenience and initial setup etc... and helpful for inspection.

I could easily remove the one for the solder paste to avoid complicating the software. It's not a high priority to have this, and I can always have an inspection mode (paste, then move to inspect). It will save weight on the head too

[phomann]

nisma... mine is a small desk setup, designed for prototyping or small production runs. A board loader would be way overboard for what I'm doing. As long as I can load a 1/4 panel and remove without any fuss, it will suit my requirements.

atlab... I hadn't realised this limitation existed. My intention was to have a webcam for placement and another for solder paste dispension. It is really more so for convenience and initial setup etc... and helpful for inspection.

I could easily remove the one for the solder paste to avoid complicating the software. It's not a high priority to have this, and I can always have an inspection mode (paste, then move to inspect). It will save weight on the head too

My Yamaha pick and place has a board feeder. It is basically a conveyer belt that consists of 2 10mm wide flat belts that run under either side of the board. A pneumatic stop pops up to stop the board, then 4 pneumatic pins locate into holes and lift the board up pushing it against overhanging tabs. Its quite simple.

Cheers,


Peter.

[nisma]

Atlab, i use linux SW that don´t have this limitations. Because the usb lib is
portet to windows, it´s possible to drive one or more cameras with it without using
the WFW interface, if the camera is supported under linux. Annother approach is
to have separate programms, that manage separate cameras and using pipes, shared memory, shared dll, whatever communicate with annother programm.
If you need help, ask offline.

[atlab]

nisma, thanks for your reply!
What is SW? Is this library for Linux only?
I don't want to write a program for Linux because this mean to start from zero.
Yes, I have thought about multiprogram data exchange, but for the present it is not a very important problem, i use 1 webcam to correct chip position and angle.

Another approach is to use DirectShow and filters, but i haven't any experience with it.

[nisma]

Hello Atlab, i suggest to discuss it offline, either icq 179855421 or with private messages.

[cnczoner]

Hi all,

I came across this thread when searching for Pick & Place machines, and I'm really looking to purchase one, but I find this thread interesting. And perhaps I can convince myself to build one instead.

Question -- it seems like moving the head is the easiest part, and the suction/pickup, nozzle-change assembly, and feed mechanisms are the hardest parts. Why not get a manual pick-and-place machine and add the X-Y mechanism to it, with appropriate software?

Also, here is another company I've not seen mentioned yet, who sells low-end PnP machines -- http://www.ntscope.com/m2-2.html . I've read some reviews that people have problems with these machines, but perhaps it can give you some ideas for your DIY units.

Cheers,
-Neil.

[nisma]

Hi all,

I came across this thread when searching for Pick & Place machines, and I'm really looking to purchase one, but I find this thread interesting. And perhaps I can convince myself to build one instead.

Question -- it seems like moving the head is the easiest part, and the suction/pickup, nozzle-change assembly, and feed mechanisms are the hardest parts. Why not get a manual pick-and-place machine and add the X-Y mechanism to it, with appropriate software?

You could get a suitable 12x18" xyz table for 900$ or less.
The same table allow you to make the feeder adapter plates, the nozzle holder plate, the manual feeder stripes, and other minor part.
Total cost, 1000$ including pneumatic parts.
Annother 500$ goes away for feeder 5-8, either new (clones) or
used, 250$ for the rest, including a used dispencer.
I have build Feeders banks, 130mm wide, component cost is a bit over 100$,
but i don´t want missing the single feeders. Hovever, it´s possible to start
only with the feeder bank and then buying single feeder as needed, or of
course, with the manual feeder only. This explained setup is for a small
p&P having a typical maximal setup of 30x 8mm lanes, one 130mm
feeder bank for populating single feeders, one jedec tray exchanger,
and one automatic batch loader for pcb´s in the size from 5x5cm up to
20x30cm. Leaving the batch loader gives you additional 15 lanes.
The P&P head has two nozzles, one of fixed size and the other can be
automatical changed. The real speed is between 2000 and 3000
Placements/hour and 9000 dots/hour and the batch system needs 12 seconds for unloading an loading a new panel/pcb.
It´s really hard making such a device without a cnc that allow you to make
the required parts, if not impossible, and for the single 8 and 12mm feeders,
it don´t really make sense building it.
The question for meccanical p&p, unfortunatly this systems don´t have the
required precision. Before the ROHS, most meccanical relay on the self aligning feature of the solder mask and if you read the tecnical docs,
this type of device can place components down to a placements accuracy of +- 10-16mil with a repetibility of generally 25mil, far outside the need for a
vision system. Presumly, it can be optimized to about half the error, but this
give you only the possibility to place 0.5mm pin spacing instead of 0.8mm pin
spacing, not a big advance. Further, it can need a long time to get out the
way to drive the board at low level without there software and how to
correct the hw alignment errors in sw.

Also, here is another company I've not seen mentioned yet, who sells low-end PnP machines -- http://www.ntscope.com/m2-2.html . I've read some reviews that people have problems with these machines, but perhaps it can give you some ideas for your DIY units.

This device are not so cheap if you consider the complete pricing including the feeder. Furhter there sw is not so smart and don´t allow optional assembling as example.

[nisma]

A question to all active participant. Should i post the files for building such a device,
mostly dxf files, then eagle files and firmware for mips architecture.

[klmjr]

YES!

[nisma]

If there is only small interest, i can supply the alu parts, otherwise, because the concept is based on a wood router, max advance on alu is 0.15mm/min.
This requires a lot of time to make the parts, days instead of hours.
I have a small proxxon mf70 converted to cnc for this purpose.

[parceval]

Hello nisma,
i am still intrested in building a p&p machine.
Can you upload some pictures of your machine and your designs?

Greetings,
Karsten