Fiducial Camera Offset and rotation

[spiyda]

I'm making progress on getting this RV4s back into placing chips...
but I have a problem with the fiducial camera

Its been several years since any replies on this..
Can anyone give me the information needed to set it up?

I have looked at the previous posts on this and they seem to be conflicting.
This may be because the OEM procedures differ to what can actually be done with string and guessology.

I have a good calibration file but the camera has been knocked out of its bracket in transit.

My intuition tells me that it should be possible to set up the camera independently of the main arm calibration procedure.
and that there will be a certain location on the base at which setting the offsets should be carried out.
and it seems logical to assume that the rotation is set at the same place.

I could really do with some help on this.. I think it is the last issue I need to address.

Chris



A

[SteveW]

I can't help directly - but bear in mind you can get by without it. If you've got a job to do, just go with manual fiducials and fix it later?
(Mine is broken, I don't miss it enough to chase very hard. Maybe one day...)

[Mike]

I think the big complication is that the position and rotation of the camera relative to the nozzle varies depending on the arm position. For manual fid'ing, the rotation is probably not a huge deal, but the position is, as the offset between camera and nozzle will have different X/Y values depending on the relative angle.

You should be able to get an initial position by using the setup PCB function, lowering the nozzle into some blu-tak then switching to camera mode and aligning on the impression.
You may need to repeat this for a range of positions and adjust to the closest avarage.
I generally find I need to add manual offsets, typically 5-25mil, which vary between baord designs due to different fid locations. Presumably this is due to changes from my original cal file.

As regards rotation, although this would at first thought appear straightforward, it may well be the case that the centre the image as seen by the vision system might not be precisely aligned with the mechanical centre of rotation of the camera body, and as the rotation of the camera varies with arm position, this would cause varying offsets. I suspect this may be the reason they use a calibration grid. I don;t recall if there is an adjustment in the cal screen for camera centering, but if so, then you'd want to set this first, to get the centre of the image at the centre of mechanical rotation in the camera mount. 

One thought I had to potentially get around all of this would be to make a little camera that fits onto the head in place of the nozzle. Or possibly a laser or LED that shines a tiny spot. That way you  avoid all the offset issues.

[spiyda]

One thought I had to potentially get around all of this would be to make a little camera that fits onto the head in place of the nozzle. Or possibly a laser or LED that shines a tiny spot. That way you  avoid all the offset issues.

I don't yet understand the system well enough to work out how I can use the fiducial camera without it being properly set up..
but perhaps knowing what the offsets are at different places on the machine, I can compensate with a second mark on the pcb
so I haven't given up with the existing camera,

From my experience, laser dots are not easy to use or set up..
a pair of fine laser lines from two sources would work much better,
they can be aligned to always have the crossing point in the right place.
still not as easy to see as a properly set up camera...

but you might be onto a winner there Mike..
Cameras are so much smaller these days and a tiny composite video cam would be small enought to fit in the end of the arm..
In fact replacing the existing camera would be an overall mass loss.


How about a small front silvered mirror on the back of the moving mirror assembly...
then the camera on a mount at 90 degrees to the nozzle (exactly opposite the existing arm cam)
It would need some means of putting the mirror in place to view the fiducial
I wish I had access to the software source, all sorts of things might be possible !

(edit)
you could actually use the existing camera, just remount it and add a second fixed mirror,
I checked and there is more than enough room inside the sheet metal "bump stop" that wraps around the head

[Mike]

I don't yet understand the system well enough to work out how I can use the fiducial camera without it being properly set up..

What I do is a dummy placement (on sticky tape) of a part nearest each fid.
Start with fids central in the cam, place, judge the error of each corner by eye and apply these offsets from the nominal centred position, repeat if necessary - rarely takes more than 2 passes.
I then have offsets for that pair of fid locations, which I apply to each board in the batch. I think there is a way to enter these so it does it automatically but I just use the up/down/left/right jog when fid'ing each board in the job.
Obviously this only works with manual fid mode, but IME auto fid mode is fiddly anyway due to lighting etc., and manual doesn't take much longer

[spiyda]

What I do is a dummy placement (on sticky tape) of a part nearest each fid.....

Thanks Mike, that makes some sense..

regarding adding a new relocated fid camera,
I measured up and it looks like a 1/4" Cmos board camera with a 25mm lens should do the trick.
It will be a little larger field of view, but a longer lens gets harder to source.
front silvered mirrors 1.3mm thick are also easily available..
the existing cameras are powered via the signal cable, so that needs a little thought
                     (edit - actually not the case, they are powered seperately from a 9VDC connection)
I will have a play in the next week or two and report back

If it works it will make aligning to fiducials easy peasy...

[SteveW]

and apply these offsets from the nominal centred position,

I'm being dim - how / where do you do this?
(All this talk of Fid cameras made me mend my broken one (snapped power wire in the crimp coming off the arm board) - so, now I've fixed it, there's about a 5-10 thou difference between manually guiding the nozzle onto the fid (and squinting with a torch from various angles - I'd be quite happy to not need to do that any more!) and using the camera. It's not _far_ off, but if I can apply an offset on a per-PCB (design, not stuffing instance), I'd be delighted.

[Gopher]

It's been a while now but there is a window that allows to you enter x y & z offsets, I think it might be in the same place as where you teach fiducial locations.....
The suggested official routine was always to place a few parts at the PCB extremes to get some idea as to the offsets that may be required and then whack them in there.

[SteveW]

Cheers, will have another look tomorrow. (I don't teach fiducial locations - they come in as components, so this is probably a window I've never tried)
(Speaking of Z offsets - I just found out why my big (well, 0805 and above) parts have been annoyingly refusing to pick up - there's a good mm difference in height between tool 1 and the others. No amount of careful setting up of pick heights was doing any good! Looks like I either need to shim or adjust the tools (maybe just the brass ring?) Sigh...)

[Gopher]

When I had nozzles made, they misread the measurements and left almost no plastic between the top and the bottom, this made the tool about 1mm too short, my solution was to whack an M6? washer underneath the brass shim which made the tool pretty much exactly the right height, same might work for you....

[Mike]

and apply these offsets from the nominal centred position,

I'm being dim - how / where do you do this?

I have a vague feeling there is somewhere you can enter offsets, but I just write them on a Post-It, then for each board in the batch, I start with the cross dead-on the fid, then step it away the required amount in each direction for each one.

[Mike]

(Speaking of Z offsets - I just found out why my big (well, 0805 and above) parts have been annoyingly refusing to pick up - there's a good mm difference in height between tool 1 and the others. No amount of careful setting up of pick heights was doing any good! Looks like I either need to shim or adjust the tools (maybe just the brass ring?) Sigh...)

You could set a height offset for the appropriate feeder lanes (setup-> feeders->offsets) - just set those lanes up using the actual tool for that part instead of tool 1