Bottom camera offset

[kevin]

Hi everybody

In my case, My Bottom camera (Looking up camera) or Head is not vertical cause accuracy problem.

After Bottom camera align and placing , i move the top camera to that component on the PCBA, it is not center of top camera.

I can see the 'Transformation offset" feature and fill in the parameter, then my nozzle runs out the center of Bottom camera.

Can somebody tell me:

That is the current way we are using offset?

If i am correct, i think that is too difficult to calibrate the nozzle in center of Bottom camera daily.

Why need to run out the nozzle offset instead of using offset coordinate.

my mean that why we don't calculate offset parameter after alignment as example below.

Component location coodinate

R1 X10 Y10 Z-30 C0

After offset X+0.1, Y+0.1

R1 X10.1 Y 10.1 Z-30 C0

Thank you.

[SMdude]

Hi Kevin,

There could be several things going on here.

First being your z slide may not be 100% square to the bed, and the camera not 100% parallel to Z.

To overcome this error, I lower parts to the placement height for bottom camera imaging.

The other thing that might be causing you issues are your endstops.

Depending on temperature your endstops will be at different positions.

To overcome this, I have put a homing fiducial at my homing position and after homing, the machine homes to the fiducial. This is pretty good, but still requires checking and tweaking before a job.

The next step would be to put the homing fiducial near the bottom camera, then if there is any expansion/contraction the offset caused will be minimal.

Mick

[Marek T.]

Top camera must be ideally synchronised with nozzles only, to nozzle could pick up parts ideally centerly as you applied center of the fed part to the system. It's only to pick up the parts without problems.
Even if part is not picked up centerly it's aligned over bottom. To be placed ideally on pcb you must use properly offsets of bot-cam. Sure as Mick said homing must be ever made properly it's obvious.
When you later go with top-camera to the placed part it doesn't have to be alligned top-cam with part. Part must be aligned with pads not with top camera:-).
Then if you need see placed part centerly with top-cam - add offsets for top-cam. Or play with mechanical offsets correction (cameras positions and angles) what is almost not possible to do. But as lower software offset applied as better view effect on the screen (specially in case of large soft rotate offsets addition).

[kevin]

SMdude

I am using opto sensor PM L24 (sunx sensor), i also checked the accuracy after homing, it is no problem.

I understand you mean that the focus of nozzle same high with pcb is ideal, but it take more time for pick and place.

Hi Marek

Top camera must be ideally synchronised with nozzles only, to nozzle could pick up parts ideally centerly as you applied center of the fed part to the system. It's only to pick up the parts without problems.

Off course it is no problem.

Even if part is not picked up centerly it's aligned over bottom. To be placed ideally on pcb you must use properly offsets of bot-cam. Sure as Mick said homing must be ever made properly it's obvious.

But i am not clear about why need to offset like run out nozzle, we can't keep the nozzle in the center of bot camera, then very difficult to calibrate, for me every 1 hour, i check the center of nozzle 1 time.

When you later go with top-camera to the placed part it doesn't have to be alligned top-cam with part. Part must be aligned with pads not with top camera:-).
Then if you need see placed part centerly with top-cam - add offsets for top-cam. Or play with mechanical offsets correction (cameras positions and angles) what is almost not possible to do. But as lower software offset applied as better view effect on the screen (specially in case of large soft rotate offsets addition).

So, my solution that will mark in the Topcam screen, where to show the center of component on the PCB.

because, when I start with new program, it is too difficult to identify the center of component pad.

Thanks

Kevin

[Marek T.]

I'm afraid that don't understand description of the problem :(.

--------------------------

[SMdude]

For a start, do you have good alignment from your nozzles to top camera? Set it using some bluetack at your placement height and use the move to buttons on the right hand side of the jog controls.

Until nozzle calibration is a working thing, you have 2 options.

1 you need perfect nozzle holders and nozzles>>> Now for reality

2 you need to use prerotate and have the nozzles align with the bottom cam with the nozzle at 0 and park your nozzles at the end of each job before shutting off your machine as this will park them at 0 again.

3 Make nozzle calibration work.

With prerotate, after every placement the nozzle rotates back to 0, so when the part is picked it is picked up in the center, providing your nozzle to top cam offsets are correct.

[Marek T.]

C-axis runout elimination is the biggest challenge, specially when using genius Chinese-Juki nozzles as we do.
So I think that machines like that should have c endstops the same like xy to do their homing at start, and to put nozzles always the same direction oriented. I have this but have only ones c motor for 3 nozzles so the pain is smaller with this, however no problem to compile Smoothie as 6 homed axis.
Plus mentioned pre-rotation to this.

Or nothing above but nozzles runout auto-calibration function that stacked...

[kevin]

Hi Marek

In my case, before running machine, i perform to calibrate the center of nozzle with Bot camera.

One thing i am not clear why when i fill in the parameter of bottom camera offset, my nozzle move out bottom camera center.

in this case, if the nozzle pick up center of component, that is ok. How about if the nozzle pick up not center of component, then how the nozzle offset correct.

Because the nozzle move out the center of bot camera.

The nozzle can't offset correctly, is this correct?

Did you see it on your machine as picture bellow:

[Marek T.]

Hmmm, I have Y offset -55 but don't remeber already why did it so.....

Maybe I have told you wrong. If you have situation that part is alligned properly (in pipeline) and always placed moved with the same position error on pcb, maybe try make correction not of bot-offset but bot-location.

It is not important if part is picked centerly by nozzle or not, parts are different staying in tape or trays, parts are never picked up absolutely centerly. Bot-cam is to find it and add proper correction.

I don't do calibrations of the nozzle at all (did it 3 months ago at some new setup). Only always pul nozzle the same oriented into the holdera and do C-homing before the job (together with xy I mean).

[SMdude]

You guys use the bottom cam offset?

I use the rotation adjustment there, but that is it and it is automatically populated by the rotation angle set in camera location..(I'd better take a look at the manual and see what that offset is for)

I just use the "move nozzle to" button, then jog until it is centered and hit the "set to nozzle location" button.

When using prerotate, all parts are picked up at "0"(if the nozzle has been set to be centered over the bottom camera and your top camera to nozzle location agrees)

If the top cam/nozzle offset is correct, your parts should be picked up in the center.

Then, the machine rotates the part to the calculated placement rotation, bottom cam checks for rotation and rotates to correct rotation error.

Then when rotation is correct, it will apply the part xy offsets to the placement(you need to have your pixles/mm set right) and place it in the correct location on the board(provided the board has been set up properly).

When not using prerotate(just normal bottom vision), you need to center the nozzle axis over the bottom camera and the top cam/nozzle offsets need to be set to the center of rotation.

Then when picking parts if your runout is only very small and parts big, you won't see many problems, however, if attempting to pick small parts, the nozzle might not even touch the part due to the runout.

Nozzle runout calibration would fix this as the machine can then always apply the correct xy offsets to the nozzle to put the nozzle in the center of the part at picking and that part in the center of its placement location. For small passives then you can get away with just placing them direct without bottom vision.

Mick

[Jason von Nieda]

The offset in Kevin's screenshots move the image with respect to it's center. It is there to adjust the image if the lens is not mounted in the center of the image, or if the image sensor is not mounted in the optical center. It's not really something most people should ever have to mess with.

In general, you should only need to set the bottom camera location - not the offset.

Kevin, I would suggest you remove the offset, then post screenshots showing the problem you are experiencing. I don't quite understand what the issue is.

Jason

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/91d231cd-ad98-4ff9-801b-dd8f2f747134%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Marek T.]

Jason, what in following case? Top camera positioned over bottom. On bottom view the camera see center of the top lens perfectly. But on the top view the camera is not seeing center of the bottom camera same perfectly but little moved. Or some sensor is not centered with it's lens or some camera is not mounted perfectly vertical. Then we use mentioned offset to align it, correct?

[Jason von Nieda]

Yep, that's the condition where you'd use that offset. All the offset field does is translate the image by X,Y, so what OpenPnP sees as the "center" of the image is not really the center of the image.

For what it's worth, I learned a neat trick from the CHMT36VA software. To align the top and bottom cameras they include a PCB with a target at the same position on on both sides. You place the PCB on the up camera and position it so the target is in the center of the image. Then you align the down camera with the target on the top of the board. Then you know both cameras are perfectly aligned to each other.

Jason

On Wed, Aug 15, 2018 at 5:35 PM Marek T. <marek.tw...@gmail.com> wrote:

Jason, what in following case? Top camera positioned over bottom. On bottom view the camera see center of the top lens perfectly. But on the top view the camera is not seeing center of the bottom camera same perfectly but little moved. Or some sensor is not centered with it's lens or some camera is not mounted perfectly vertical. Then we use mentioned offset to align it, correct?

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/e568ee7b-193b-43f8-99d1-42d74109421b%40googlegroups.com.

[Marek T.]

So now reminded why used an offset in my setup :-).
Nice trick with pcb, simple but need really well done pcb with top and bottom not moved any cent.
I made a small glass 10x10mm and cuted with laser a crosshair on it. When picked by nozzle and sent to bottom also shows centering perfectly :-).

[Jason von Nieda]

Laser etched glass sounds like a good idea. I will try it!

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/2b77de74-b845-47f8-9272-4b2d49822c58%40googlegroups.com.

[kevin]

Thank you Jason/ Mark

I remove the bot offset all to 0, and calibrate the nozzle in center of Bot camera.

I tried to place 4 components with difference angle 0, 90, 180,-90,

After that, i move the top camera to see the center of component on PCB,

I can see all of 4 components on the right of top camera center about X+0.2mm

In this case, we only need to offset +0.2mm, is this correct?

You als can see my video.

https://www.youtube.com/watch?v=eRHo-VSOCmc

[kevin]

Sorry, off set x-0.2mm, not +0.2mm

Vào 12:51:27 UTC+7 Thứ Năm, ngày 16 tháng 8 năm 2018, kevin đã viết:

[Marek T.]

But the parts are placed centerly on pcb?

[kevin]

You mean the parts are placed center of pad. Yes, if we change x co-odinate of board by -0.2mm.

But when we do not use the bottom vision for another parts, it will be over 0.2mm again.

Due to the bottom camera is not align with top camera and nozzle.

if we have the 'really bottom offset''  every thing will be aligned very easy.

Thanks 

Kevin

[Marek T.]

Kevin forget top camera. Use it only to detect fiducials, temporary.
Focus on that to bottom camera align part and system placed it on pads. If placement is not accurate and at each try the same moved - change LOCATION xy of bottom camera, these 0.2mm. It must effect proper placement.

When you have it done, tuned, you will worry about offset of the top camera to it look centerly onto the parts.

[Trampas Stern]

You could just drill a hole or series of holes in a PCB to make cross hairs.  

[Marek T.]

Yes it's same good idea like pcb with double-sided crosshairs. In fact the only holes are enough to see and set required ofset.

Also good point is to observe on bottom camera an inside of the nozzle. If the cam have the center of optic moved aside to sensor or is not vertically mounted - it's clearly visible then.

[Mike Menci]

I use cheep plastic foil (apx 0.2mm thick) calibration ruler from microscope camera 

Mike

[Mike Menci]

Here some Vision shots - up camera and down camera to the same spot on plastic ruler: