Tips for accurate placement?

[alanambrose]

Hi,

Does anyone have any tips for accurate placement? Anything to watch out for? I'm talking in this case about component-to-component mis-alignments rather than the overall calibration onto the PCB. (That might be a separate useful thread  ) I usually end up doing a fair amount of nudging of components and that's OK with a small number of chips ... But I have an IR LED board coming up with ~500 very expensive LEDs on it that I would like to be accurate without me nudging each damn one.

Mike, I remember, for instance some images of an 0402 pattern that looked nicely regular. How did you do that?

What are the key factors - vacuum, nozzles, RVPlace settings, power?

Alan

[trev]

CDF file needs to be good. I found that if it did not recognise the part well then it would drift off its position for picking up the part and also placement. But when the part is properly recognised the machine would actually tune itself in to perfect placement.

The nozzle pickup position needs to be calibrated properly. That would knock off the position a bit and mess up placement.

Nozzle also needs to be clean and okay. An old nozzle can cause the component to rotate between pickup and placement.

I used to nudge quite a bit but once the cdf was sorted and nozzle pickup position calibrated the machine really delivered well.

As a side, for best speed having the most common parts in a feeder close to the camera speeds things up. Likewise having the board orientation such that the most parts are closest to the camera. If you have a lot to do this time saving starts to mount up.

Trev

[alanambrose]

Thanks Trev,

I've taken to watching the values that are calculated for body size by the camera and entering those in the cdf. One thing I don't understand is whether the X/Y position for pickup is critical or whether it is adjusted successfully after the 1st pick or two. I set the pick height by moving the nozzle down until the component moves the tape down and then backing off one 10 mil notch. Does that sound right? That's an interesting thought re nozzle wear - do you have some way of refurbing old ones?

Alan

[Mike]

Centring of the feeder pick location isn't a huge deal as unless feeders are locked in the CDF, it will tweak the pick position over multiple picks until it's centred.  This can be an issue if the vision isn't set up optimally, such that the centre position isn't very consistent. Inductors with solder on the bottom are a regular problem.
Locking feeders can be helpful in some cases, but the first thing to do is optimise the vision thresholds and blob size.   

[trev]

The camera centering was the other thing I meant to say. I used to have to lock all the feeder positions because it would drift off. I think what was happening was the camera was not centered and when picking a tool up, if that position was not good it could knock the centering out as well. So randomly I would experience loss of feeder position. A friend looked at the machine and almost instantly said the tool position wasn't great. He used to run several of these machines and knows them well. We re-positioned the tool pickup location and re-centered the camera and it has been great since.

He said they used to re-center every morning and after lunch just to make sure all was good.

I do not have any ideas for working old nozzles. Sorting out stock of new nozzles is on my list.

He also disabled merge moves as he believed this created loads of problems. I actually had it on a large capacitor and could see it almost sliding into position as it would move across and down at the same time. I think it was just going too quick but turning merge moves off was advised so I stuck with that for all parts.

Trev

[alanambrose]

Ah thanks for the info. I just did the tool position, using the patented ball bearing method , and the camera cal. Everything was surprisingly accurate, and I last did those a couple of years ago. Once thing I didn't quite get - in 'set ratio' it says 'if the readings are 468-478 or a greater distance return to 'set focus'. Does that sound like a grammar mistake? Mine were 452 for x and 453 for y.

Alan

[trev]

I wondered if they meant greater distance apart? So within 10 counts of each other?

[alanambrose]

You may well be right. For anyone looking at this in the future, here were the changes (attached). Pink highlight is old version, green is after tool position & camera cal. This is on an RV1S. The tools will move a bit in the holders anyway, so the tool positions can't be fantastically exact. The camera centre display looks a bit odd to me - it reports only tiny shifts of the centre but the graphic on the screen seems to show much bigger movements.