LitePlacer.com forums

Hey.

I have tried to get liteplacer work but no results so far.
Machine accuracy is like one millimeter... when camera corrections is used.
Steps I have done:

* measured steps per mm very acurately using dial gauge with 0.01mm scale
* measured pixels per mm
* measured and corrected offset between downlooking camera and nozzle
* found suitable speed and acceleration. Max travel speed was 25 meters/s but i use 8 meters/s; acceleration is 2000 xjm for X and 1700 for Y
* found suitable currents for motors. Steppers are warm, not hot
* re-checked these values multiple times

System sees optical zero point very well. It detects fiducials and corrects differences to...incorrect.
When I position pcb zero perfectly to pcb copper and switching off all corrections, machine can go to center of any component without any problem.
Waching a pcb component with camera and then picking it up works well, placing it back works too.
Now trying to place a component. system detects tape hole (white paper tape), and takes component with offset. And places it with offset, running multiple times component placment error increases.
It always places component a slight Y+ offset, and X- offset.
Using full visual aid, fiducial calibration, nozzle calibration, offset is even worse - SO8 is totally misaligned
My nozzle adapteris bit off center, like 0.1mm but using nozzle calibration, error increases, like uplooking camera is wrong direction.

And sometimes when going thru stored ponts, machine lost zero point about 0.5mm
Im using software version: 07.03.2020

Im completely lost here, don't know what to do.
I have lot of experience with cncs, have built some cnc mills, with mach3, some linuxcnc machines, few plasma cncs and so on but here im stuck, looks like video processing algorithms
does not cooperate with me.

Lets try to pinpoint the issue. I assume you have marked your A axis and check, that it is in zero position on startup. Also, you have tuned the cameras with the mirror so that they look straight down. Please check your assembly, that your pickup tube is not excessively off from vertical. Also, let's use one nozzle only at first to make things simple.

"Probe (n.c.)" tries to place the nozzle at the location shown on camera, without using nozzle wobble correction. Use A axis in zero position. If the tip doesn't land on the target, your nozzle nominal offset is wrong. Rerun the camera locations routine (https://www.liteplacer.com/setting-camera-locations/) or correct it manually.
When this is correct, go to up camera location and klick "Nozzle Down". If the nozzle is not in the center of the up camera image, jog it there and click "Set" on up camera location.

At this point, you might want to recalibrate this nozzle. On nozzle page, dump calibration data and see, how much offset the system thinks there is. Does this correspond to real world?

You can disable the nozzle correction on the run job page. Wen you do this, your system should now work with that nozzle, and have whatever error there might be from nozzle runout.

Enabling nozzle correction should improve things. "Probe down" button does the same that "Probe (n.c.)", but does use the calibration data and should work at all rotation settings.

Please try these and report, what you experience and where errors appear. We'll continue from there.

> ...like uplooking camera is wrong direction.

It should be mounted with the arm towards the front of the machine.

> And sometimes when going thru stored ponts, machine lost zero point about 0.5mm

This is likely a separate issue. Loose set screw somewhere, perhaps? Belt end point not sitting snugly?

Hey,

Mechanically everyting is fine. After tensioning all belts to correct levels, distance per revolution is exactly 40mm.
Now cameras and optical corrections.

Cameras are well calibrated, pixels per mm are correct.
Camera to nozzle offset is correct. I used plasticine to measure offset precisly - first I made a small mark on plasticine then moved camera to same place.

When using "Probe n.c" nozzle goes exactly center of camera image, so that is correct.

Problems starts with fiducials and placement.

When I set pcb zero to left bottom corner of copper, component nominal locations are correct.
But when I measur efiducial locations strange things will happen
My board has 4 fiducials randomly placed. Corrections will correct some locations, but some ohter will go worse. Example is C11.
Correction corrects it to left (X-)
I have feeling that every component on the right of rightmost fiducials will have same amount of negative correction eg if correction is x+0.5mm and Y+0.3mm then C11 will be placed position X-0.5mm; Y -0.3mm
here is C11 nominal location and this is C11 measured location And last one is C11 nominal placed. You see a huge misplacement.
Using measured location it will be even bigger offset to component center.
Using Nozzle calibration data is turned off.

What will make this error - it will happen component pickut phase. Tape hole will be measured, then system will calculate component location and picks a component. And there is second very big misaligment
Component will be picked not center but offset: X-0.15mm Y+0.25mm and now when systm needs to rotate component it goes to very wrong place. It happen with C11: placement offset + component pickup offset, component will be rotated 270 degrees. Tape aligment is X+

Last image is C6 placed and SO8

Try with some paste or double sided tape or something similarly tacky...there could be some movement after placing as the component/vacuum is released.

Without bottom camera adjustment, the movement inside the tape pocket will be a factor. (using slightly larger footprints may make up for that offset)

It seems that C11 was placed more or less on the measured position. The measured positions should be accurate. But you have other issues also, so I suspect that is just one symptom, not the real problem. Still, it is worth checking that the measurement targets real fiducials, and not a nearby via or pad.

Please check the camera cable. The cable needs to be fixed so, that when the machine moves and the cable is flexed, the force to the camera mount doesn’t change and bend the camera out of alignment.

Maybe you are experiencing the same problem that i had. https://liteplacer.com/phpBB/viewtopic. ... 6131#p9304