Assembly line update

[jmelson]

OK, reading more, it sounds like what you have is alternate PANELS are getting mis-registered.  What I'd do is make a one-component test program.  Put a piece of double-stick tape on the board, and run it, and check alignment.  keep working with this until the alternate misregistration issue is solved.  I don't know how much debug ability you have in the vision system for identifying how the fiducials are picked up, but clearly something is VERY wrong.

Oh, you say you DON'T use a panel fiducial.  Maybe this screws up the software, and it requires you to do a panel fiducial first?  I could easily believe this could happen, and such a case was NOT properly TESTED by Yamaha.  They tested it with just ONE panel with no panel fiducials, it worked fine, and then never ran a second panel.
Totally believable if that was a non-standard way of running, that they never though to test it that way.  Certainly worth a shot.  Hope you have SOMETHING on the panel you can use as a fiducial.

Jon

[mrpackethead]

on pretty much everything i do, i have panel fids, and pcb fids.    It is possible to run with just pcb fids, or in fact no fids at all.

The fact that is a 5mm offset every 2nd time is suggesting to me that something in the conveyer system is not being reset when the 2nd board comes through.  A program restart will bring everthing back to zero.

I dont' rely on the board clamp, and always use the location pins. 

Have you checked that the machine actually thinks the location pin is at 0,0?

[kylehunter]

OK, sorry for the confusion! Yes it's really hard to explain this..

So, the panels that I am running consist of 9 individual PCBs. There are fids on the panels, and on the PCBs. From CAD, I exported the centroid data on the entire panel itself. So in essence, the pick and place treats the ENTIRE panel as a single board. I did this because it is so easy to export out the data on one file, and I don't have overly fine pitch components, so I don't need the PCB fids. But if I ever need, I can use the PCB fids and just set them as a local fiducial.

Every ODD PANEL gets missplaced (0, 1, 3), while every even panel is fine. I'm sorry, but I don't see how it is possible that the conveyor or anything is the problem. The machine finds the fiducials every time, so even if the panel wasn't on the rails properly, the fiducials would either error, or they would still be found, and it would correct for it. Also, it wouldn't be every odd board, it would be random.

[mrpackethead]

So, the panels that I am running consist of 9 individual PCBs. There are fids on the panels, and on the PCBs. From CAD, I exported the centroid data on the entire panel itself.

This is not a good idea, for a whole bunch of reasons.  The first being when you get X-outs ( failed pcbs in a panel ), you'll have no way of dealing to that, except for not using that entire panel.     You should be just doing one pcb, then setting up the panelisation data ( offsets, rotations etc ).     If you dont' accept X-outs from your pcb-fab you'll be payhing a premium you dont need.  You can even set up ignore markers on the pcb, so the machine will check to see which pcbs on the panel to miss.

The machine will first look for panel fids. ( once ), and then will look for each set of pcb fids.

Every ODD PANEL gets missplaced (0, 1, 3), while every even panel is fine. I'm sorry, but I don't see how it is possible that the conveyor or anything is the problem. The machine finds the fiducials every time, so even if the panel wasn't on the rails properly, the fiducials would either error, or they would still be found, and it would correct for it. Also, it wouldn't be every odd board, it would be random.

Becuase your machine is not cleanign up the conveyor after it transports the first board.  Resetting it, forces it to. 

Check. (a) is your location pin at 0,0    ( as the machine see's it ).  (b) are you using the push in at the end?

I would export my cad data as a single pcb, and set up  panel config.. use both panle and pcb fids.   

[jmelson]

OK, sorry for the confusion! Yes it's really hard to explain this..

So, the panels that I am running consist of 9 individual PCBs. There are fids on the panels, and on the PCBs. From CAD, I exported the centroid data on the entire panel itself. So in essence, the pick and place treats the ENTIRE panel as a single board. I did this because it is so easy to export out the data on one file, and I don't have overly fine pitch components, so I don't need the PCB fids. But if I ever need, I can use the PCB fids and just set them as a local fiducial.

OH, OK, only NOW do I understand what you are doing.
So, it only picks up 2 (or 3) fiducials for the whole panel, then.  This is how I do my boards, so far.  So, no board data, only panel fiducial data.

Every ODD PANEL gets missplaced (0, 1, 3), while every even panel is fine. I'm sorry, but I don't see how it is possible that the conveyor or anything is the problem. The machine finds the fiducials every time, so even if the panel wasn't on the rails properly, the fiducials would either error, or they would still be found, and it would correct for it. Also, it wouldn't be every odd board, it would be random.

Well, check the field of view of the camera, and the settings for fiducial recognition, to see if there is any possibility of it picking up something else in the camera image and misidentifying the centroid of the fiducial.  Are you absolutely sure it is EXACTLY every other panel?  That's why I suggested a one-component test build with double-sided tape.  If it truly is, EXACTLY every other board, then it pretty much has to be some kind of software bug.  Do you have any sample programs from the machine's previous owner?  Maybe there's something in your placement file that is triggering this foulup.

Jon

[mrpackethead]

Well, check the field of view of the camera, and the settings for fiducial recognition, to see if there is any possibility of it picking up something else in the camera image and misidentifying the centroid of the fiducial.  Are you absolutely sure it is EXACTLY every other panel?  That's why I suggested a one-component test build with double-sided tape.  If it truly is, EXACTLY every other board, then it pretty much has to be some kind of software bug.  Do you have any sample programs from the machine's previous owner?  Maybe there's something in your placement file that is triggering this foulup.

My machines will potnetially find a fididual within +/- 10mm or so.   I have it set so it fails, if the fid is not within 1.0mm of where i expect it.  if its 10mm out of whack, something is very wrong.   i would imagine that the fids would always be within .2mm of where i expect them, if not closer.  Some of the variance can be how accurately the boards where routed.. that can be .1 - .2 out.

[mrpackethead]

Well, check the field of view of the camera, and the settings for fiducial recognition, to see if there is any possibility of it picking up something else in the camera image and misidentifying the centroid of the fiducial.  Are you absolutely sure it is EXACTLY every other panel?  That's why I suggested a one-component test build with double-sided tape.  If it truly is, EXACTLY every other board, then it pretty much has to be some kind of software bug.  Do you have any sample programs from the machine's previous owner?  Maybe there's something in your placement file that is triggering this foulup.

My machines will potnetially find a fididual within +/- 10mm or so.   I have it set so it fails, if the fid is not within 1.0mm of where i expect it.  if its 10mm out of whack, something is very wrong.   i would imagine that the fids would always be within .2mm of where i expect them, if not closer.  Some of the variance can be how accurately the boards where routed.. that can be .1 - .2 out.

picking up the wrong fids would only be a thing if (a) there was something nearby that was being detected (wrongly) as a fid.  and the board was somehow getting in the wrong place.   I wished i was there, i bet i could diganose this in minutes.    Its so difficult by formum!

[kylehunter]

...

...

OK!

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

What I'm still confused about, is even though the board wasn't at the proper end stop location, the camera still found the fids. Shouldn't that have have been able to fix the location issue? I mean, it obviously wasn't, but why not? That's how fiducials help to align the board in general no?

Another question, does anyone have any info/manuals/documents for how to service/maintain/calibrate CL feeders? I don't have any knowledge of how this should be done.

Thanks!

[mrpackethead]

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss-aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

Well, glad you've found that issue, but it still sounds like your fids are either not being properly located or ignored.    I'd be guessing that your placement is reliant on the board's actual geometry..   ROuting is not 100% accurate and has some variation so you might get a panel that is not 100% one day, and get a small misplacement.   If you're board doesn't have tiny parts this might not be an issue. It is amazing how 'badly' parts can get placed and still reflow their way into place..  ( not something you should rely on! ). I'm constantly amazed at how what looks like a slightly misaligned part will find its way into place though.

To test if its really looking for fids,   modify your setup so all the parts apart from  1  jellybean part is removed from placement.   Stick a little bit of double sided tape on where they should be placed...    Get a black vivid marker pen and draw completely over the fid's..     THen run it though...

If it is really looking for the fids, it should stop when it cant' locate them.  ( my machine, squeels annoyingly, and the red tower light comes on ).

If it just goes and places the part, then we know we have a config problem.     If  it does fail on the fids, then we have to have a look at something else.
 







If you wanted to send me a vios file, i could have a look at it, and see if i can see an issue with the way you've configured the board set

[kylehunter]

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss-aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

Well, glad you've found that issue, but it still sounds like your fids are either not being properly located or ignored.    I'd be guessing that your placement is reliant on the board's actual geometry..   ROuting is not 100% accurate and has some variation so you might get a panel that is not 100% one day, and get a small misplacement.   If you're board doesn't have tiny parts this might not be an issue. It is amazing how 'badly' parts can get placed and still reflow their way into place..  ( not something you should rely on! ). I'm constantly amazed at how what looks like a slightly misaligned part will find its way into place though.

To test if its really looking for fids,   modify your setup so all the parts apart from  1  jellybean part is removed from placement.   Stick a little bit of double sided tape on where they should be placed...    Get a black vivid marker pen and draw completely over the fid's..     THen run it though...

If it is really looking for the fids, it should stop when it cant' locate them.  ( my machine, squeels annoyingly, and the red tower light comes on ).

If it just goes and places the part, then we know we have a config problem.     If  it does fail on the fids, then we have to have a look at something else.
 







If you wanted to send me a vios file, i could have a look at it, and see if i can see an issue with the way you've configured the board set

It definitely looks for them each time before it places the parts. But I'll test covering them tomorrow and see what happens.

[mrpackethead]

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

Whats the input air pressure on your machine and what is the regulator on your machine saying..   Its almost suggesitng that the accuators are not running fast enough.   

[kylehunter]

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss-aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

Well, glad you've found that issue, but it still sounds like your fids are either not being properly located or ignored.    I'd be guessing that your placement is reliant on the board's actual geometry..   ROuting is not 100% accurate and has some variation so you might get a panel that is not 100% one day, and get a small misplacement.   If you're board doesn't have tiny parts this might not be an issue. It is amazing how 'badly' parts can get placed and still reflow their way into place..  ( not something you should rely on! ). I'm constantly amazed at how what looks like a slightly misaligned part will find its way into place though.

To test if its really looking for fids,   modify your setup so all the parts apart from  1  jellybean part is removed from placement.   Stick a little bit of double sided tape on where they should be placed...    Get a black vivid marker pen and draw completely over the fid's..     THen run it though...

If it is really looking for the fids, it should stop when it cant' locate them.  ( my machine, squeels annoyingly, and the red tower light comes on ).

If it just goes and places the part, then we know we have a config problem.     If  it does fail on the fids, then we have to have a look at something else.
 







If you wanted to send me a vios file, i could have a look at it, and see if i can see an issue with the way you've configured the board set

Welp, you were right again..

I had the panel fiducials set to block fiducials and not PCB fiducials, and I wasn't using any block offsets. Surprisingly it still looks for the fiducials even though they are set to block, and just won't use them if it can't find them..

So now I'd say mystery is solved. The reason for the delay being required anyway, is that the conveyor speed is set pretty slow, so it needs to wait like an extra 0.2 seconds..

Any thoughts on the feeder maintenance side? I'm getting a bit of a stockpile of feeders that need work.

[mrpackethead]

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss-aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

Well, glad you've found that issue, but it still sounds like your fids are either not being properly located or ignored.    I'd be guessing that your placement is reliant on the board's actual geometry..   ROuting is not 100% accurate and has some variation so you might get a panel that is not 100% one day, and get a small misplacement.   If you're board doesn't have tiny parts this might not be an issue. It is amazing how 'badly' parts can get placed and still reflow their way into place..  ( not something you should rely on! ). I'm constantly amazed at how what looks like a slightly misaligned part will find its way into place though.

To test if its really looking for fids,   modify your setup so all the parts apart from  1  jellybean part is removed from placement.   Stick a little bit of double sided tape on where they should be placed...    Get a black vivid marker pen and draw completely over the fid's..     THen run it though...

If it is really looking for the fids, it should stop when it cant' locate them.  ( my machine, squeels annoyingly, and the red tower light comes on ).

If it just goes and places the part, then we know we have a config problem.     If  it does fail on the fids, then we have to have a look at something else.
 







If you wanted to send me a vios file, i could have a look at it, and see if i can see an issue with the way you've configured the board set

Welp, you were right again..

I had the panel fiducials set to block fiducials and not PCB fiducials, and I wasn't using any block offsets. Surprisingly it still looks for the fiducials even though they are set to block, and just won't use them if it can't find them..

So now I'd say mystery is solved. The reason for the delay being required anyway, is that the conveyor speed is set pretty slow, so it needs to wait like an extra 0.2 seconds..

Any thoughts on the feeder maintenance side? I'm getting a bit of a stockpile of feeders that need work.

I'm glad that its resolved..   Its amazing how some small gotchas can cause you so much greif.  But once you get past your machines ( if you keep them maintained ) will just keep on keeping on. 

For the low cost 8mm feeders,  if they get seriously out of whack. ( bad accutator ) i just biff them now, a new feeder is US$55, its not worth the time and hassle.         Theres minor adjustments you can make to them.   For more expensive feeders it def is worth repairing them.    Let me see if i can find the book which has the details in it.

[kylehunter]

Sorry about the delay, got swamped with design work, but was assembling boards the last couple of days. It looks like you guys were right, it was an issue with the panels getting to the proper location on the conveyor. The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss-aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

Well, glad you've found that issue, but it still sounds like your fids are either not being properly located or ignored.    I'd be guessing that your placement is reliant on the board's actual geometry..   ROuting is not 100% accurate and has some variation so you might get a panel that is not 100% one day, and get a small misplacement.   If you're board doesn't have tiny parts this might not be an issue. It is amazing how 'badly' parts can get placed and still reflow their way into place..  ( not something you should rely on! ). I'm constantly amazed at how what looks like a slightly misaligned part will find its way into place though.

To test if its really looking for fids,   modify your setup so all the parts apart from  1  jellybean part is removed from placement.   Stick a little bit of double sided tape on where they should be placed...    Get a black vivid marker pen and draw completely over the fid's..     THen run it though...

If it is really looking for the fids, it should stop when it cant' locate them.  ( my machine, squeels annoyingly, and the red tower light comes on ).

If it just goes and places the part, then we know we have a config problem.     If  it does fail on the fids, then we have to have a look at something else.
 







If you wanted to send me a vios file, i could have a look at it, and see if i can see an issue with the way you've configured the board set

Welp, you were right again..

I had the panel fiducials set to block fiducials and not PCB fiducials, and I wasn't using any block offsets. Surprisingly it still looks for the fiducials even though they are set to block, and just won't use them if it can't find them..

So now I'd say mystery is solved. The reason for the delay being required anyway, is that the conveyor speed is set pretty slow, so it needs to wait like an extra 0.2 seconds..

Any thoughts on the feeder maintenance side? I'm getting a bit of a stockpile of feeders that need work.

I'm glad that its resolved..   Its amazing how some small gotchas can cause you so much greif.  But once you get past your machines ( if you keep them maintained ) will just keep on keeping on. 

For the low cost 8mm feeders,  if they get seriously out of whack. ( bad accutator ) i just biff them now, a new feeder is US$55, its not worth the time and hassle.         Theres minor adjustments you can make to them.   For more expensive feeders it def is worth repairing them.    Let me see if i can find the book which has the details in it.

Thanks! That makes sense, I was thinking the same. What are your thoughts of used genuine Yamaha vs new China clones? I've heard the China clones have issues with the slider that prevents the parts from jumping out, and we do a lot of LED boards.

[jmelson]

The "Pre Fix Timer" was set to 0.1 seconds, which was too little time for the board to fully get to the end stop, causing the board to be miss aligned by a few mm. I made the timer 1 second, and ran 60 panels without a single issue! Hooray!!

What I'm still confused about, is even though the board wasn't at the proper end stop location, the camera still found the fids. Shouldn't that have have been able to fix the location issue? I mean, it obviously wasn't, but why not? That's how fiducials help to align the board in general no?

Yes, if the board was just a little out of position while the fiducials are located, and then MOVES when the first part is placed, you will certainly have a bad assembly. 

The setup on my CSM84 has a main stop, and then a push-in plunger presses the board against the main stop.  Then it does a "dance" tapping the push-in plunger and a pair of plungers opposite the fixed rail, it reliably fixtures the board on all 4 sides.

Another question, does anyone have any info/manuals/documents for how to service/maintain/calibrate CL feeders? I don't have any knowledge of how this should be done.

The guy who had my machine before me made a single holder/base for a feeder.  I added a bridge over the top of this with a hole, and a crosshair over the pick-up location of the feeder.  When I started using 0603 parts, I found a number of feeders were not reliable (0805 were just fine).  After using the bridge and crosshair, I could see the tapes were coming to different positions.  There is an eccentric on the advance pawl that sets the index position of the sprocket, and I adjusted this to get the pockets all to line up under the crosshair.  Then, if the pocket is left or right of the desired position, I check the bottom of the feeder for dings or debris.  After using this, all the feeders soom to work fine on 0603 parts.  The professional alignment fixture has a microscope there, but I found getting a crosshair just above the feeder was good enough.

Jon

[mrpackethead]

Thanks! That makes sense, I was thinking the same. What are your thoughts of used genuine Yamaha vs new China clones? I've heard the China clones have issues with the slider that prevents the parts from jumping out, and we do a lot of LED boards.

The feeders i've been buying from Ksun seem to be really reliable, i've had no real issues with them..    I've not keep track of how many cycles they have done, but some of them would be 100's of reels.

[kylehunter]

Another question for my Opal. Is it possible to have one part set to several different feeder locations? Like for the LED board, we're placing like 600 LEDs each panel. If we could have the machine use like 4 feeders for the LEDs and simultaneously pick the parts, I assume that would cut the time down by a lot. Also would make us need to replace parts much less often.

[mrpackethead]

Another question for my Opal. Is it possible to have one part set to several different feeder locations? Like for the LED board, we're placing like 600 LEDs each panel. If we could have the machine use like 4 feeders for the LEDs and simultaneously pick the parts, I assume that would cut the time down by a lot. Also would make us need to replace parts much less often.

Yes, you certainly can. I have done this exact job, in the past, and had six 8mm feeders side by side with the same part on it.  It means the machine can pick up six at the same time, and it really does speed things up..  Of course. you end up replacing all six feeders potentially at the same time!

[kylehunter]

Another question for my Opal. Is it possible to have one part set to several different feeder locations? Like for the LED board, we're placing like 600 LEDs each panel. If we could have the machine use like 4 feeders for the LEDs and simultaneously pick the parts, I assume that would cut the time down by a lot. Also would make us need to replace parts much less often.

Yes, you certainly can. I have done this exact job, in the past, and had six 8mm feeders side by side with the same part on it.  It means the machine can pick up six at the same time, and it really does speed things up..  Of course. you end up replacing all six feeders potentially at the same time!

Seriously?! How do you set that up in software?

[Mangozac]

Hi Kyle,

I find your heat and cooling issues to be quite interesting. Sorry if I missed it but what kind of volume of boards are you doing? If it's not a high volume would you be better off operating the line in a semiautomatic manner, only firing up the oven periodically and put them all through at the same time?

When I visited our CM in Shenzhen they showed me that they only operate the oven once a day - and they were even doing a high volume with a PNP doing 60k CPH. They would stencil and place all of the boards and at the output of the placer they would feed back into magazines. Once they had a sufficient volume ready the conveyor reflow oven would be switched on and everything run through. It saves excess heating of the factory and I suspect uses less power. That said, it may only be a worthwhile saving because of the lower labour cost...

[kayvee]

Another question for my Opal. Is it possible to have one part set to several different feeder locations? Like for the LED board, we're placing like 600 LEDs each panel. If we could have the machine use like 4 feeders for the LEDs and simultaneously pick the parts, I assume that would cut the time down by a lot. Also would make us need to replace parts much less often.

Yes, you certainly can. I have done this exact job, in the past, and had six 8mm feeders side by side with the same part on it.  It means the machine can pick up six at the same time, and it really does speed things up..  Of course. you end up replacing all six feeders potentially at the same time!

Seriously?! How do you set that up in software?

Not sure for the Opal but in mine you assign 6 feeder lanes to the same part, before running the optomiser, which then assigns them to the preferred feeder locations.  Load all 6 feeders with that same part, fit the feeders to the optomiser assigned positions, and off you go.

[mrpackethead]

Another question for my Opal. Is it possible to have one part set to several different feeder locations? Like for the LED board, we're placing like 600 LEDs each panel. If we could have the machine use like 4 feeders for the LEDs and simultaneously pick the parts, I assume that would cut the time down by a lot. Also would make us need to replace parts much less often.

Yes, you certainly can. I have done this exact job, in the past, and had six 8mm feeders side by side with the same part on it.  It means the machine can pick up six at the same time, and it really does speed things up..  Of course. you end up replacing all six feeders potentially at the same time!

Seriously?! How do you set that up in software?

Not sure for the Opal but in mine you assign 6 feeder lanes to the same part, before running the optomiser, which then assigns them to the preferred feeder locations.  Load all 6 feeders with that same part, fit the feeders to the optomiser assigned positions, and off you go.

This, is exactly what you need to do.   Let the machine do the optomization for you.

[kylehunter]

Another question for my Opal. Is it possible to have one part set to several different feeder locations? Like for the LED board, we're placing like 600 LEDs each panel. If we could have the machine use like 4 feeders for the LEDs and simultaneously pick the parts, I assume that would cut the time down by a lot. Also would make us need to replace parts much less often.

Yes, you certainly can. I have done this exact job, in the past, and had six 8mm feeders side by side with the same part on it.  It means the machine can pick up six at the same time, and it really does speed things up..  Of course. you end up replacing all six feeders potentially at the same time!

Seriously?! How do you set that up in software?

Not sure for the Opal but in mine you assign 6 feeder lanes to the same part, before running the optomiser, which then assigns them to the preferred feeder locations.  Load all 6 feeders with that same part, fit the feeders to the optomiser assigned positions, and off you go.

Hmm. Just tried it and the optimizer just ignores the additional duplicate parts. Even if I set the duplicate parts to different feeder numbers, it will set different parts to the same feeder location, and just pretend the additional ones don't exist.

Is this because the vios file references the part number (1-xx) rather than part names?

[jmelson]

Thanks! That makes sense, I was thinking the same. What are your thoughts of used genuine Yamaha vs new China clones? I've heard the China clones have issues with the slider that prevents the parts from jumping out, and we do a lot of LED boards.

I only have Yamaha/Philips feeders.  When I started with LEDs, I found that my nozzles did not handle domed LEDs well at all.  So, I got flat-top LEDs, and those worked fine.
I can't do 0603 LEDs, as the flat top is just too small, much smaller than the flat top of an 0603 passive component.  But, 0805 LEDs work fine.

Jon

[kylehunter]

Thanks! That makes sense, I was thinking the same. What are your thoughts of used genuine Yamaha vs new China clones? I've heard the China clones have issues with the slider that prevents the parts from jumping out, and we do a lot of LED boards.

I only have Yamaha/Philips feeders.  When I started with LEDs, I found that my nozzles did not handle domed LEDs well at all.  So, I got flat-top LEDs, and those worked fine.
I can't do 0603 LEDs, as the flat top is just too small, much smaller than the flat top of an 0603 passive component.  But, 0805 LEDs work fine.

Jon

Ahh, yeah. My LEDs are all flat and seem to place fine.

Another option for the multi pick of the LEDs is for me to make the four different parts, name them LED-1, LED-2, etc. Them make 25% of the placed LEDs use the LED-1, 25% use LED-2, etc.

Then let the optimizer happen with those. It'll work, but obviously since I'm choosing which 25% are grouped together, it won't be as fast as if the machine had full flexibility over all LEDs, but I assume it'll still be way faster