Neoden 4 pick and place

[thommo]

Ichan - from what I understand 'rail alignment' can have no effect on your outcome.

It doesn't matter if they are parallel to the travel axis or not. The N4 will determine the position and angle in the software.

You only need to be able to identify the 'mark' or fiducial accurately, and then input the number of rows and columns in the panel, and then the 2 offsets which represent the 'step and repeat' dimensions of your boards in the panel, relative to the 'first component'.

eg - if your boards are 50 x 100, and you have no panel borders in between each board [they butt up to each other on both the x & y axis, then the 'step and repeat' offsets are 50mm and 100mm respectively.

I do really understand how a standard fiducial will be much more preferable, but the fact is this board has none of them - i have to find the solution for this board, even the machine now already fully loaded with all the components for this board.

Has no time with the machine today, plan for tomorrow:

- check and if required realign the rails
- verify the board size on the machine coordinate
- try single board
- try TankSpark recommendation
- slow down the speed
- try manual mark point recognition

The last one is Neoden recommendation i got by email.

-ichan

[elmood]

ichan: The edge of board detection shouldn't matter too much. It's really mostly used to know roughly where to find the mark points.

As for alignment problems, I fought with this for ages before finally trying some more controlled tests to see what was going on. I was convinced that there was a bug or some problem with my data.

It turns out that using only 2 mark points is not enough. I moved to 4 spaced evenly as far to the corners of the board as possible and this made it work better. But two of my points were mounting holes of about 3mm diameter. Although they LOOK centred in the image during detection the machine doesn't actually show where it's zeroing in on. It turned out that probably lighting, etc. was making these points not really reliable. I'm now using small vias and this works MUCH better. But be careful to choose holes that don't have other ones very near them... one of our holes often detects the adjacent one and this causes all kinds of problems. (like putting parts on the rails)

Neoden said basically not to have any other holes within 3x the diameter of the mark point. I'm not convinced that this is correct and think it might need to be even further. But if you use the standard 1mm pad in a 3mm square cutout from your solder mask with some space around it I'm hoping that this will be acceptable most of the time. We found that by temporarily covering the adjacent hole with something helped detection... we just removed it before running the actual job.

[mrpackethead]

This is a fiducial layout i use on the rails of any panels i make.  ( See attached )

It is a 1mm dot of copper, and then a 6mm diameter copper track around it.   THe purple area is solder mask, ( inverted ), so there is never any solder mask anywhere near it.
This works very well across a range of machines.

[rwb]

@Elmwood So how is the machine running for you now that you have figured out how to use it? How many boards have you made with it so far?

And I will share this for the guys interested in manual stencil paste machines: https://www.tindie.com/products/CYBRES/cybres-sp2421-stencil-printer/

[Ichan]

I think i found where is the problem .

First as planned i check the alignment of the fixed rail, X changes along 336mm Y travel is only 0.07mm - that is almost perfectly aligned!

Then I verify the board coordinates on the machine, below is the points measured.



And here is the result.



See the difference in red, it is about a confirmation of the worst shifted amount of the right bottom board like on the close up photo before.

While doing this test i found that the automatic fiducial (mark point) recognition on the 1 mm drilled pad is almost exactly the same with the best my eyes can do.

Next i will try to re-align all the components manually (yes for all 88 parts), and try to use panelized mark points.

-ichan

[Ichan]

Just remember that on my LS60V the fiducial defined for each board (= panelized).

-ichan

[metalphreak]

Did you create the panel yourself or was it done by the board fab house? If they created the panel, they might've added a small gap between each copy of the board, to allow for the V-score groove?

[Kjelt]

Just a p&p noob question, if the fiducial is multi axis symetrical like the two rings example then the machine can not determine the board orientation (say 10x5cm pcb is placed as a 5x10cm pcb)
With an asymetric fiducial the machine can automatically correct the orientation or is this when multiple fiducials are used to make this decision, or what is the logic behind this?

[ar__systems]

Just a p&p noob question, if the fiducial is multi axis symetrical like the two rings example then the machine can not determine the board orientation (say 10x5cm pcb is placed as a 5x10cm pcb)
With an asymetric fiducial the machine can automatically correct the orientation or is this when multiple fiducials are used to make this decision, or what is the logic behind this?

Machine can't determine orientation of the board by two fiducials. It is told a general area to look for those and only use them for precise positioning. It does not scan entire area looking for them.

Doing automatic board orientation also serves no purpose, imho.

[mikeselectricstuff]

One reason for using 3 fids is so it can detect a wrongly oriented board

[Jefferson]

Pcb orientation isn't important. Two fiducials enough to ensure high accuracy. It's provided to Samsung, DEK, Topaz, ... and other industrial professionals placer without problems. Provide enough precision to supply board assembly line and the movement of the nozzle unit.

[mikeselectricstuff]

Pcb orientation isn't important.

Tell that to anyone who has accidentally loadad a PCB upside down.
The panel spec from every subcontractor I've dealt with shows 3 fids, and explicit instructions to NOT put a fid in the 4th location, as an orientation check.

[Jefferson]

Violation enough. Few observes IPC standards among fans. If you are familiar with the IPC standards, 2 pcs fed asymmetrically placed necessarily,  the feed and the head are oriented correctly.

[TankSparks]

@Ichan   

did you try my method?
I looked at your file,  its not going to work
Look at the sample file from neoden attached 
Notice where I circled.... how 272.0 =  272.0  if these values are not equal it will not work.
 use align on bottom left only
then enter the values manually for x,y where I crossed off the align button by adding board height and board width

Your board does not have to be perfectly aligned. 

[mrpackethead]

Pcb orientation isn't important.

Tell that to anyone who has accidentally loadad a PCB upside down.
The panel spec from every subcontractor I've dealt with shows 3 fids, and explicit instructions to NOT put a fid in the 4th location, as an orientation check.

This is wide spread industry practice to do this..

[mrpackethead]

@Elmwood So how is the machine running for you now that you have figured out how to use it? How many boards have you made with it so far?

And I will share this for the guys interested in manual stencil paste machines: https://www.tindie.com/products/CYBRES/cybres-sp2421-stencil-printer/

Thats a fun kind of way to make a stencil printer, with easily avaialble bits..  Its nearly awesome, but there are a few little issues;

- it might be tricky to do double sided boards, as the pcb is sitting on a flat platfrom. its needs some standoffs
- The way of adjusting the pcb with those "paddles" while possible would be unbeilably fustrating and slow.. Particually when you are getting to fine pitch...   A little bit more work required.

[mikeselectricstuff]

@Elmwood So how is the machine running for you now that you have figured out how to use it? How many boards have you made with it so far?

And I will share this for the guys interested in manual stencil paste machines: https://www.tindie.com/products/CYBRES/cybres-sp2421-stencil-printer/

Thats a fun kind of way to make a stencil printer, with easily avaialble bits..  Its nearly awesome, but there are a few little issues;

- it might be tricky to do double sided boards, as the pcb is sitting on a flat platfrom. its needs some standoffs
- The way of adjusting the pcb with those "paddles" while possible would be unbeilably fustrating and slow.. Particually when you are getting to fine pitch...   A little bit more work required.

The pin-locating system used on Eurocircuits' stencil printer  avoids the need for any adjustments, though you  really need more of an up/down motion to engage than a swing-up mechanism. A simple lift-off  frame with locating pins may be all that's needed

[Ichan]

did you try my method?

Yes, but the problem is still - verified by vision align.

I just tried single panel, the result is pretty good, not perfect yet but enough - will show the video and photo later.

I tried the panelized mode of mark points, but do not work. Anyone tried this?

From coordinate verification i found that may board units of length is about 0.002mm/mm larger than the machine, it is 0.2mm/10 cm - not good.

Panelized mark points will keep this error per-board not accumulated across the panel, eg the placement error for each board will be about the same as single board.

-ichan

[thommo]

Hi Ichan,
I guess you must be getting fairly frustrated by now.

I think you're trying to do something that can't be done - at least not as it's currently set up.

My suggestion is that you get from NeoDen the same copy of the software application that TankSparks is using. Once you've tried it with the same software as TankSparks, if you get a great success you'll know it was in your particular Software version. If you get the same unsatisfactory result with the TankSparks software version, using the same method, you've been using to date, you'll know to look at the Hardware, or it's in you usage of the software with that PCB / Panel.

If you check TheSteve's post 'reply # 601  6th March' you'll see the N4 is definitely capable of the results you're seeking, when the right combination is used.

BTW - how many of those blank PCBs do you have on hand? The sooner you can get your hands on some boards with Fids the better I think.

Peter

did you try my method?

Yes, but the problem is still - verified by vision align.

[Ichan]

I guess you must be getting fairly frustrated by now.

Not really, quite motivated today .

Below is the video and photo of single board test today.





My problem is more on the fact that 1mm on my pcb is not 1mm on the machine, the pcb is about 0.2mm/10cm larger than the machine, which one is true i do not know yet.

BTW - how many of those blank PCBs do you have on hand? The sooner you can get your hands on some boards with Fids the better I think.

I still have several hundreds of them. On the video above 1mm drilled round pad is reliable enough to be used as fiducial, even on dirty pads caused by the use of transparent adhesive.

I also tried the panelized board with mark points in panelized mode today, will post it soon.

-ichan

[rwb]

@ichan  It's nice to see some proper placement from your machine now 

It's also good to see it properly placing the  VFQFPN 16L chip accurately even though you do not have the machine place file perfectly setup yet.

I'm starting to lean towards buying the Neoden 4 now. Just need to see how all your boards end up going.

Hoping you get this all worked out soon. 

[rwb]

Also since were talking about 2 vs 3 Fudicals on PCB boards I figured I would share these pictures of a Evaluation board that I have here from TI.

TI has a company called Krypton do all their Evaluation board designs and manufacturing. Krypton assembles tons of boards for TI and use state of the art machines to do it.

If you look at all 6 of the EVM boards I have from TI you will see they use 3 Fudical marks on both the front and back of the boards.

[ServoKit]

@rwb: you will have to make up your mind some day and take the plunge ;-) - Good to see that ichan is making progress. I think it has become clear that one cannot expect plug-and-play or turn key with that kind of equipment. You must be willing to experiment and learn as you go.

Regards, Axel

[ar__systems]

My problem is more on the fact that 1mm on my pcb is not 1mm on the machine, the pcb is about 0.2mm/10cm larger than the machine, which one is true i do not know yet.

Sounds easy to fix, just make the correction in the export script when you extract the data from the CAD.

[Jefferson]

Also since were talking about 2 vs 3 Fudicals on PCB boards I figured I would share these pictures of a Evaluation board that I have here from TI.

TI has a company called Krypton do all their Evaluation board designs and manufacturing. Krypton assembles tons of boards for TI and use state of the art machines to do it.

If you look at all 6 of the EVM boards I have from TI you will see they use 3 Fudical marks on both the front and back of the boards.

Hi rwb! Old IPC-7351 described three points on two straight lines. And these lines are perpendicular. This is on TI pcb. However, other standard allows for diaganal two points, but with a shift or be asymmetrical PCB midline.

These pcb of not the best example, as must also ring as in the att. photo.