Fuji parts out Signal for poka-yoke component replenishment

Hi,

I think the only solution would be to contact Fuji and ask if there is an interface to the machine to write software for it. I read somewhere that FujiTrax allows a IT system type connection which software could be written for.

Another solution I would use is to write some software that knows what parts are used per product, and then with a sensor on the in feed, you could estimate how long for parts to run low. You would know low, but not really out, as you might get some parts reject that throw your estimation out.

Then as you load reels, the operator clears the reel qty on a computer as they load the reel on the feeder. Then you could work it out.

Is this the kind of thing your after?

Regards,

Grant

reply »

Totally agree on what you said, but there's some other things involved:

There's not enough training, here in Juarez, Mexico, the level of pesonal turnover is very high so when you train on certain procedure, they leave for some reason and always is you have new people training.

here, production departament say, claim, exige for something engineered "poka-yoked" so the personal dependence is at the lowest level; and the worst thing, they have the management support (but does not want to spend money, of course) I know that following the actual procedure, the incidence of wrong component will be low. A general overview of procedure:

1.- take out the old reel and scan in a own program where you specify product, device and scan part number. program have component device locations for every product. the same program gives you the kanban where component is located.

2.- Take the same reel and give to inspector from the end of the line so she/he can verify that is the same reel (to avoid scan the same reel twice)

3.- Back operator go to kanban location and pickup the reel, then go to computer an scan the new reel, if for some reason he pick another reel, computer will tell him that reel is wrong, if it's ok then he load it into machine and mark the board.

4.- Once the marked board leaves the oven, inspector had to have component identified on board so he can compare it to first article (in case of wrong labeled component), so she can release transaction on the program.

Is not a bad system, BUT you can bypass all the procedure and just load the new reel and let the machine continue, allowing the possibility of a wrong component. I Know that personel ownership, training and discipline is the best way to solve this, but from a egoist engineering standpoint is hard to fight against all the situations involved and I prefer to look for alternatives.

Also, if somehow I can add to the actual software that we have the feature of "force" to the operator to follow the procedure in order to allow machine continue running, i will sell the application to other companies with the same problem (LOL), nah I'm just joking.

We have a product with 489 placements, 106 reels, 1 tray component (QFP of 256 leads, 0.5 mm of pitch) some QFN, a bunch of 0402, appart of a bunch of other "normal" components, we have scheduled about of 125% of the machine capacity, and anyway have to ship on time. so you can imagine the sweetish of the matter.

reply »

Seems to be easy and effective, I will ask for further information with them...

Regards

reply »