Fiducials

We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are:

Does anybody else have this problem? Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)?

Thanks for all inputs.

Bob

reply »

The circle is not a "good" fiducial. The gradually change in contrast of a circle versus a cross makes it difficult to find the center of the fiducial. By pre-tinning the fiducials there will be a convex mirror surface distorting the edges even more. The black circle you see on the screen is the result of a very high contrast, but the edges are fuzzy. Copper fiducials have less contrast to begin with and are therefor less sensitive against contamination. If the fiducials are taught with a medium resolution there should be no rejects.

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? | Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob |

reply »

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface?

| Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob | Bob,

I have went through this same problem. First off I would not suggest copper fiducials unless they are treated with an OSP (Organic Solder Preservative). If you use bare copper fidicials they will oxidize. Bare copper fiducials will also cost you more from your PCB vendor. Fiducials will need to be masked before HASL. When I started with the company I am presently working for they were using bare copper fiducials. Operators would pretreat every fiducial with a tarnish remover before pick and place operation.

Making the fiducials bigger will allow for a thinner deposit of solder on the fiducial. You should determine what type of Hot Air Leveler your PCB vendor is using. If your ckt brd is double sided and you have a fiducial recogition problem on one side only, then your PCB vendor is using a Horizontal Hot Air Leveler. The air knives, which blow the solder off the PCB, thinning the solder on the PCB, are above and below the PCB. The bottom of the PCB, or the side of the PCB closest to the bottom lower air knife on the hot air leveler, will always have more solder than the top of the PCB. This is due to gravity. Never the less, your PCB vendor should be able to control the solder volume to eliminate your vision problem.

I assume you know the good old erasure trick. This method has been used for years as a temporary fix to infrequent fiducial vision problems. Just rub a pencil erasure on the problem fiducial. This will take the shine off fiducial.

Thanks,

Chris

reply »

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? | Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob | We have also had trouble with fiducials, sounds very much like your problem. We used .050 circlular HASL fids. We have since changed to .060 fids and it has all but disappeared. What type of equipment are you using? Using an eraser is not a good idea as the eraser shavings can plug apertures and also get into solder joints. Short term - use something that will not leave any particals, plugs for the end of tube components do not leave much and it does not take much to rough up the surface. Only "stroke" it 2-3 times, that should do it. Fiducials should also be placed where there is other copper plating, if they are isolated the solder will attract more to them during the HASL process. I think circular is very popular in the industry but if you wanted to try something different I would recommend diamond shape. The diamond shape will force the solder off. It is to much hassle to mess with OSP or copper fids when the solution is simple.

reply »

If the fiducials are rejected from the vision system, I would not relay on the accuracy of the position. Again, I would like to point out that the circle shape has a somewhat sinoid contrast in X- and Y- direction ( same as a diamond shape, but rectangular is better ). The shape of this sinoid may be so distorted that the vision system will not find the center. Low cost boards may not have fine pitch components on it and it is worth to spend the extra cost for different fiducial shapes. With 15 mil components a circle as a local fiducial may not be trusted and you only waste time looking for the fiducial. In order to reduce some of the reflections from the shiny fiducials I once covered the light source of the indirect light from the fiducial camera, only leaving the direct light source with good results.

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? | Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob |

reply »

Hi Bob,

I know what you're talking about, it seems that most machines that I've worked with, have at one time or another some problem seeing a fiducial. Some machines are worse than others, and it really didn't make a difference what kind of vision system it was, although I do think that binary fiducial vision systems (like Fuji's) are a bit more picky than gray scale systems.

Most boards I see usually have circles on them, they're the easiest and simplest fiducials to use...and most of the time they work good. But I've found that the donut or ring shape works better. You don't have the near reflection problems that you do from the dome effect on the donuts like you do with plain circles. Even if the HASL is on rather thick and shiny, the camera can still see the ring well enough to pass it. The problem with the circles is that the reflection most of the time will cause the circle diameter to appear to be smaller than what you've told the machine it's supposed to be. Also sometimes, if you put a polarizing filter over the lense of the camera, that'll help....there's a lot of machines that have that standard, but there's some that don't...

-Steve Gregory-

reply »

Hi Bob,

I know what you're talking about, it seems that most machines that I've worked with, have at one time or another some problem seeing a fiducial. Some machines are worse than others, and it really didn't make a difference what kind of vision system it was, although I do think that binary fiducial vision systems (like Fuji's) are a bit more picky than gray scale systems.

Most boards I see usually have circles on them, they're the easiest and simplest fiducials to use...and most of the time they work good. But I've found that the donut or ring shape works better. You don't have the near reflection problems that you do from the dome effect on the donuts like you do with plain circles. Even if the HASL is on rather thick and shiny, the camera can still see the ring well enough to pass it. The problem with the circles is that the reflection most of the time will cause the circle diameter to appear to be smaller than what you've told the machine it's supposed to be. Also sometimes, if you put a polarizing filter over the lense of the camera, that'll help....there's a lot of machines that have that standard, but there's some that don't...

-Steve Gregory-

reply »

You don't say what kind of machine you are using but several just use the edge of fiducials to determine area (and centers). If you are teaching in binary and have the ability to select a decision level (threshold), use a level half way between loss of edge at the high and low level. Don't worry about what the center of the fiducial looks like. Recognition rates should go up pretty well. Try to avoid any rework because of time/process. Especially try to avoid erasers because of contamination....flux pens usually work ok if you really have to use something.

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? | Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob |

reply »

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? | Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob | Bob ,

We have recently went through the same problem and ended up curing this by changing the diffuser on the p&p vision light source . The previous model gave more of a "straight" light that shone down on the board and gave lots of glare . The new diffusers are cut in such a way that the light is now coming more from 45 degrees onto the board - with much reduced glare . These worked immediately on fitting them ... ask your p&p manufacturer if they have such an option (we're using Panaosnic) . You could try the DIY approach and fit a section of white perspex under the diffuser first - this will give you an idea if the above is going to cure your problem ( or a polarising filter as suggested earlier ) .

reply »

We are currently using circle, cross and double-cross(#) fiducial patterns with SiPlace equipment. Most of the problems seem to occur when, in the case of a HASL fiducial mark, a PCB is topside reflowed, goes away for thru-hole and comes back for bottomside glue and placement. The previous reflow action has caused solder on the fiducials to reflow and solidify with an uneven surface, most noticable in the case of cross or double cross patterns, where the solder seems to congregate at the crossing points. This, of course, plays havoc with the vision recognition as light gets reflected all over the place.

At present, we are experimenting with leaving the fiducials unplated and covering with solder resist. Results so far look good, not just on SiPlace but also on Dek and Camalot vision systems.

It's quite simple to try this out before getting your PCB supplier to change the fiducials; just select a unique track feature on the PCB, teach that as a fiducial (obviously you will need to measure it accurately), and see what happens!

Best of luck.

reply »

We have resorted to adding a ring (or donut) around our 0.040" round fiducials. The donut acts as a "plating thief" to minimize the thickness of the copper plating (in the case of HASL finish) and of the solder plating (in the case of solder-plated finish). The donut is especially helpful when the fiducial is isolated from the rest of the metal features, since the isolated features are the ones that get plated the thickest.

The donut should also provide some amount of scuff protection, which will help if your boards ever get stacked or mishandled.

| We are using .040" diameter fiducials (global and local) on our boards. They are solder covered and HASL. I get batches of boards where the p&p vision system has trouble reading the fiducial. The fiducial looks dark on the screen. The board vendor says it is the uneven solder plating. My questions are: | | Does anybody else have this problem? | Does anybody use bare copper fiducials? If so, how has that worked out as far as contrast against the laminate and potential oxidation of the copper surface? | Would using a larger diameter circle ensure flatter plating (although I think this is going to be directly related to how well the fabricator controls his HASL process)? | | Thanks for all inputs. | | Bob |

reply »