Eclipze's SMD Pick'n'Place Build....

**Eclipze** · 01-20-2010, 05:18 PM

I wanted to make sure I had a step resolution close to 0.01mm, but the actual repeatable accuracy is certainly more difficult to put a number on. I've be very happy with 0.1mm. If it was 0.5mm, I think I'd be pushing it to place 0805 reliably. TSSOP packages with 0.5mm pitch would need operator assisted adjustment/release... come to think of it, that would be needed for SOIC's too. I'II be tweaking/tightening the carriages up before I accepted 0.5mm hahaa...

The rails/carriages are overkill with respect to the load they carry. Keeping the weight of the head assembly low will hopefully contribute to the overall accuracy and speed. The Y-Axis rail is quite stiff with respect to the load, and with the weight of the head assembly being low I'm not concerned about it bobbing around under acceleration forces. If there is bobbing, I would be more suspicious about the flex in the baseboard first. However the design is for prototyping and pre-production, where speed isn't as important. I'd be more than happy to get around 500..800cph. But have no expectation of hitting 4000-5000cph compared to a production loader. Servos are expensive, and for such a small design these steppers (particularly that massive NEMA23 on the X-axis now!) is more than ample to move the load quickly.

The X-Axis belt plates are done! The belt is going to be cut to length and both end secured at the carriage. The belt ends will loop up through the hole (as shown in post #7) and retained by the mating track in the plate on the top. So they cannot slip out with use.

This top plate with the fine pitch tracks was a pre-test to make sure it was a good fit before trying to the same with the Y-Axis carriage. It's a good fit too, very happy. The depth of that track still leaves the belt slightly higher, so I have some compression when it's bolted in place. I also made the track wider than needed, so I had some sideways adjustment and also because the square edges of the tracks won't butt up nicely to the pockets machined with a round end mill tool.

The plate that sits underneath has two purposes. The first is to hold the belt nice and straight, pointing the belt at the pulleys. The second is the vertical part is going to be used for the opto-switches that will be mounted on the baseboard. This plate will pass through those at the limit of the travel (limit switches).

Now I know the track profile is good... next is the Y-Axis platform bracket.

**scsi** · 01-21-2010, 02:59 AM

I think we are misusing the term accuracy here. This particular machine does not have to be accurate or precise (within reason) but repeatability is paramount for the basic PnP functions. The thing will be traveling from point A to point B and it does not really matter how straight and true that motion is. What is important is that the business end hits the destination coordinates every time within 0.1mm. Does not have to be linear motion either - it's just cutting air so to speak.

I'm saying all this assuming that the machine that Eclipze is building will have some teaching function when you can define those points A and B manually for every part. Typically they use down looking camera with crosshair.

On the other hand, the industrial machines must be more or less accurate as they rely on motion linearity a lot. While many machines have manual teaching, the destination point of every placement motion is getting calculated by the computer based on the fiducial correction and part position correction factors.

I don't think Eclipze is going after fiducials with this machine or going to calculate the coordinates on the fly for every part. With this in mind, I'd suggest you target repeatability and not precision of the steppers. In other words, there is no reason to use 10x microstepping trying to get 0.01mm positioning precision if you cannot repeatably position the machine due to sloppy bearings, belts or weather fluctuations. Keep in mind that with high microstepping rates your torque reduces dramatically and the step loss probability increases. Try it with 1x or 2x stepping and see if repeatability is there. If it's not, solve that problem first and then go after higher precision.

For reference, on my machine I was playing with stepping parameters and ended up using 0.02mm/step and doing it throug "macrostepping" if there is such a term. Essentially, the servo drive has software gearing function when it can subdivide incoming step/dir pulses (microstepping) or it can multiply them. In my case I'm using 200x *multiplication* because the native travel distance of the linear servos is 100nm/step and that's a little bit too precise. That's nanometers...

-scsi

**atlab** · 01-21-2010, 07:18 AM

Originally Posted by Eclipze

...
...The X and Y are using NEMA 17 motors... All are bipolar 1.8 degree step.
...2.5mm pitch, 6mm wide belts... 10-tooth pulley...
X & Y step resolution is 0.0125mm

You will receive repeatable resolution worse than 2.5*10/400=0.0625 mm, not 0.0125mm.
But, IMO, this is suitable for 0805.

I've tested automatic fiducial recognition with webcam and calculation the coordinates on the fly.
Webcam was mounted on the old pen plotter. Movement resolution was about 0,1 mm (half step).
SMD footprints was 1206.
Webcam movements imitate:
- movement to nozzle changer
- movement to feeder
- movement to place point

"http://www.youtube.com/watch?v=JIJinWAhytc"]YouTube- auto2.avi

**scsi** · 01-21-2010, 11:59 AM

Atlab - that fiducial hunting video is awesome! Nice work! Can you operate two webcams in Windows?

privet,
scsi

**Xerxes** · 01-21-2010, 03:52 PM

Just for reference: some Yamaha high end ball rail + servo linear axis specify repeatability of +/-0.04 mm. And that is only repeatability, the accuracy is different topic and its added to repeatability error. So the total end point position will probably have larger error than +/-0.04 mm, maybe twice of that. From that I think its difficult to get even +/- 0.1 mm with drylin + stepper.

Very impressive video atlab! Did you use some library for feature recognition?

BTW, where do you get cost efficient cameras with suitable optics? What are the camers in Eclipze's and atlab's systems?

**Eclipze** · 01-21-2010, 05:51 PM

I'm not that keen to have a teaching mode. I'd rather write something to translate the pick'n'place generate generated by Protel to suitable GCODE. I'm going look at Mach3 for the actual PC software, as it seems quite a flexible tool and a combination of macros and the generated GCODE may provide a usable solution.

The stepper drivers are the Gecko G540. These are spectacular, and ramp from 10-bit stepping through to full step at speed, very smoothly.

atlab, I'm unsure as to your calculation of "2.5*10/400=0.0625 mm". The steppers are 1.8 degree, or 200 steps per revolution and the stepper driver is 10-bit. So I understand it to be 2.5*10/(200*10) = 0.0125mm. I will be able to test this when I have the Y-axis running.

For machine vision, it's really not something I've put much thought into yet. Always considered it more of an upgrade. It's also likely to be expensive... so I'II wait until the machine actually works before spending too much

I have a webcam in the design to watch the point of placement at an angle, although I could mount this camera perpendicular to face the PCB. This would probably be more beneficial for assessing the performance. Either way... vision correction is definitely of interest!!!

Machined the Y-Axis Platform

**Eclipze** · 01-21-2010, 11:43 PM

Added index pin sleeve and mounted the platform... I'm going to change this to a plastic sleeve, but need to shave a fraction off the holes before I can fit it.

**atlab** · 01-22-2010, 10:14 AM

Originally Posted by scsi

Atlab - Can you operate two webcams in Windows?

Hello (privet) scsi!

Yes i can with MS DirectShow, but i had some troubles with webcam selection, picture size and image buffer.
USB passband is touble too.
So, in fiducial recognition i used more simple MS "avicap32.dll" and WinAPI, but avicap32.dll can operate with one webcam only.

You can experiment too: try my little prog (DirectShow) in attacment. Click by left mouse button on required small webcam picture.

Originally Posted by Xerxes

atlab, did you use some library for feature recognition?

No. I spent much time solving tasks of chip angle/position correction and fiducial recognition.

BTW, where do you get cost efficient cameras with suitable optics?

I used inexpensive (US$ 20-25) webcam "Chicony 5136" (Panda 8D) 640*480@30fps.
Optics quality isn't very crucial issue (IMO).

**atlab** · 01-22-2010, 10:18 AM

Originally Posted by Eclipze

atlab, I'm unsure as to your calculation of "2.5*10/400=0.0625 mm". The steppers are 1.8 degree, or 200 steps per revolution and the stepper driver is 10-bit.

Stepper have rotor's stable points at step and halfstep positions, so 360/1.8/2=400.
Microstep is used only to make movemet smoother, not for positioning.
My driver have 1/4, 1/8, 1/10, 1/16, 1/20 microstep modes, but you can see in video that i can't positioning precisely (i used 1/16 mode).

Read this about microstep:
http://www.romanblack.com/stepper.htm

**Eclipze** · 01-22-2010, 10:48 PM

Originally Posted by atlab

Stepper have rotor's stable points at step and halfstep positions, so 360/1.8/2=400.
Microstep is used only to make movemet smoother, not for positioning.
My driver have 1/4, 1/8, 1/10, 1/16, 1/20 microstep modes, but you can see in video that i can't positioning precisely (i used 1/16 mode).

Read this about microstep:
http://www.romanblack.com/stepper.htm

This description of the stable points for half and full step are with respect the holding torque and applied load. This holding torque is significantly lower when microstepping, however you can still move the stepper to a microstepped position. Microstepping is not just for smooth movement, but you do have to have sufficient torque for the load. If you do have insufficient torque to hold a microstepped position, then the stepper will slip to a half or full step position where higher holding torque can maintain the position. This means you need a stepper with higher torque for that application.

The only time I've read material on microstepping error is with respect to the actual held angle. The higher the microstep fraction, the higher the error (steps are not equidistant).

**Eclipze** · 01-22-2010, 10:57 PM

Left Z-Axis carriage plates. Bearing is a nice press fit. I had to machine down the height of the pulley - something that wasn't originally going to be a problem, but the datasheet on the stepper has a much longer shaft in the drawings than than the actual part that arrived.