Neoden 4 pick and place

[rwb]

Ichan, so your planning on making 2000 of these boards on the Neoden 4 right?

[Ichan]

Ichan, so your planning on making 2000 of these boards on the Neoden 4 right?

Hopefully more,  the job is 6000 board currently worked by my spoiled LSV60, attached is the photo of the panelized board - populated and unpopulated.

The panel size is 204x144 mm, i am a bit worry because the N4 (with rail) spec for max board width is 140mm if waffle tray used.

This board has no fiducial so i will try to use drilled holes, also i want to try the bad board detection on  the coming machine.

-ichan

[rwb]

Well I would say your about to put the machine to the test 

I'm excited to see how this turns out for you as I'm sure everybody else is also.

I would say most guys are leaning towards the machine not being able to get the job done but only time will tell.

I think you can make a custom waffle tray to increase your working area.

[elmood]

Hey folks,

I like the idea of collecting together our observations. I've put the stuff I already sent to Neoden and some other notes and ideas on a collaborative text editor here:

http://collabedit.com/hfe7r

It's not pretty but you should be able to edit without signing up for anything. Please feel free to edit / add / rearrange things. It would be great to see everyone's experience.


elmood

[elmood]

Oh, by the way... we made a few successful boards, but the crappy oven wrecked them so we haven't run anymore jobs... we found a small Essemtec oven on ebay which we are hoping to have up and running this week. (need to put in a proper exhaust for it)

Here's a video of our first mostly successful test so far:

https://youtu.be/Wyn8Cg2i6Wo

[rwb]

Thanks so much for posting that video Elmwood.

It looks like it placed the IC parts pretty well.

Have you ran it any quicker with the same results?

[rwb]

Here is a video of a guy going over all the PNP machines he came across in the under 10k range after which he built his own:

https://youtu.be/dvwZeb2uwkg

[mikeselectricstuff]

Thanks so much for posting that video Elmwood.

It looks like it placed the IC parts pretty well.

Have you ran it any quicker with the same results?

The camera dwell time looks like it could lose a few hundred mS....

[Martin Hodge]

Do you really need to use vision on every part? My TM220 reliably places 0604 parts at maximum speed without vision.

[rwb]

You can turn the vision On or OFF on the Neoden 4.

I would rather have vision ON than OFF just to be safe unless I was under a time crunch where I needed maximum speed.

So how many boards have you successfully ran using the TM220 machine? What are its limits as far as fine pitch parts?

[elmood]

rwb: By no means was my video showing an attempt at the highest speed. I was just trying to get a successful placement of a board. I used vision on all parts and ran the entire job at 50% speed. Still seems fairly okay to me for a board with only 63 parts.

[rwb]

Elmwood - Yea I had nothing bad to say about the speed. That speed would work fine for me as long as it placed all my fine pitched parts accurately.

I just wonder how well it holds its precision at the higher speeds.

How long did it take to to get to that point of placing everything correctly?

[mikeselectricstuff]

You can turn the vision On or OFF on the Neoden 4.

presumably on a per-part basis, not globally?

[rwb]

If I remember correctly yes on a per part basis.

A Neoden 4 owner will have to confirm.

[elmood]

Yes you can adjust the vision per feeder. Same with speed of placement and a few other things.

It took longer than I thought to get going mostly because of external things regarding my oven... I wasn't going to place a full board with no way to keep it.

Also this old board has no fiducials so it doesn't have reliable marks to use for alignment. I spent a long time trying to work out why some parts would place well and others not so well. It turns out that using larger features is not a good idea. The machine doesn't show the centre point it thinks it got when detecting the marks... It looks perfect on the camera crosshairs but there was obviously some error.

I found that at least 4 points were required in all corners of the board for perfect placement. Keep in mind that I have a lot of other work going on so it took a Saturday with a colleague to verify all my coordinates and file managing code was okay when exporting the placement job file. (No rounding errors, etc.) Once we confirmed that we started experimenting with different mark points and found that it made a huge difference! It seemed reliable after that.

I'm pretty sure that these kinds of tweaks would be needed on any machine, but the bad UI and lack of how-to docs makes it extra difficult.

[Martin Hodge]

So how many boards have you successfully ran using the TM220 machine? What are its limits as far as fine pitch parts?

0604 is the smallest I've tried. Occasionally it will place one of these 0604 resistors on it's side. That is a result of the tape advance being too fast and shaking the part a little so that it's picked up that way. The software really needs to be able to allow adjustment of the vacuum sense and tape-advance speed for these situations. The finest pitch part is an FT231X which it places within tolerance every time. That did surprise me. I've run ~65 boards so far. It has been a tremendous benefit to me even with it only placing 80-90 percent of the parts.

[rwb]

Martin - Yes the Charmhigh version with vision looked to be a killer little machine. Same thing yours does except it has vision for part correction.

[thommo]

Ichan,
There is another post in this forum where an owner was frustrated with issues related to alignment drift and accuracy.
This maching is designed [as are all PnP machines] to access a clean reference marker on a board, or panel border - fiducial.

There are two problems with using a hole [although it is also round] :
1.
It is 3 dimensional, which means the light source will likely have a substantial effect on the 'shape' it sees once the gamma is set high - eg no grey tone, just black and white.

At best you will end up with an ellipse - certainly not a circle. What is 'seen; through the hole will likely also have some effect on the outcome. Would highly recommend that you place fids on all boards, and also on panel boarders, especially if there is no space available on the PCB.

I believe once that issue was recognised, the placement drift and accuracy issue disappeared.

2.
Although the through holes are usually very reliably placed, each PCB house works to different tolerances.
This means that the actual through hole 'could' be offset from the footprints by the amount in the 'tolerance'.
What this means for component placement is that you are now placing in reference to a location that is 'not known' relative to your footprints.

I recommend you place fiducials on all future runs of your boards if you are going to use a PnP machine.

Your contract manufacturer most certainly would have done so in the past, if only on the panel borders which may have been removed prior to delivery.

This board has no fiducial so i will try to use drilled holes, also i want to try the bad board detection on  the coming machine.

[thommo]

Well done guys - despite the steep learning curve, I reckon if you were to run that board for some hours, you would be a lot more relaxed with it, than without it.

Keep the info and pic/vids coming please - it's a major confidence booster for all of us considering desktop PnP.

Oh, by the way... we made a few successful boards, but the crappy oven wrecked them so we haven't run anymore jobs... we found a small Essemtec oven on ebay which we are hoping to have up and running this week. (need to put in a proper exhaust for it)

[rwb]

Another short video of a Neoden 4 running slowly.

https://youtu.be/nIFlSOgJ3n0

I asked this guy to come on over here to discuss 

[rwb]

Have any of you guys calculated out how much the NeoDen 4 can save you over sending the boards out to be populated & reflowed? That is something I have heard little about but am interested in hearing about.

[mikeselectricstuff]

Another short video of a Neoden 4 running slowly.

https://youtu.be/nIFlSOgJ3n0

I asked this guy to come on over here to discuss 

I just spotted something really dumb that it's doing - it's obviously been told to use a lower speed on the ICs in the trays, yet it's also using the slow speed when moving to the feeder with nothing on the nozzle 

[TheSteve]

It also desperately needs an independent speed control for the Z axis.

[mikeselectricstuff]

Have any of you guys calculated out how much the NeoDen 4 can save you over sending the boards out to be populated & reflowed? That is something I have heard little about but am interested in hearing about.

This would be different for everyone, and would vary by job type and costs of potential subcontractors.
And it's not always about cost - can be convenience, speed etc.

[mikeselectricstuff]

It also desperately needs an independent speed control for the Z axis.

Seriously? you can't set this independently...?