Neoden 4 pick and place

[Spikee]

Machine is not assembled square maybe ?
Or they use the wrong pulse per step settings / slack in belt / camera is not mounted / calibrated correctly.

[noras]

We use the third generation, how much?

[rom67]

Eclipze, thank you very much!
Yes, I did it. But the misalign still is appear.

When I try to calibrate the smallest nozzle (without rotation) the hole has a big displacement from the axis of rotation.
If I aligned after the rotation, the smallest nozzle doesn't able to take the SMD 0402 (2 times missis from 3).
If I aligned the nozzle without the rotation (just at the center of the hole at the current position) then I can take 9 SMD's 0402 from 10. But at the end I get a constant offset.

[rom67]

Could anyone tell what the last stable (available) firmware?

[Eclipze]

The lookup camera is central to the four nozzles, so it will not have a perpendicular view at the nozzles. 
I don't know the nozzle size I have, but if I know the next largest won't work for 0402.

1. Check the nozzle is clear for good suction.  There was one time I had an 0402 stuffed length ways in a nozzle obstructing the air pressure.   
2. Check the nozzle end is clean.  I clean the tips before every build.  Mostly in case any paste/flux residue causes stickyness. 
3. Check the accuracy of the tape feeding.  If not enough, the tape indexing might have a little bit of variation. I often use a feed strength of 40, as when too high, it causes the parts to bounce out of the pockets.
4. May sure to index twice, then set the pickup alignment for that feeder.  Very important... set the alignment target for the pocket square, not the component.  Index and double check it's lining up.

[TheSteve]

I don't use the size correct option.  It seemed related to vision checking, but I found it just dumps components unnecessarily.

I also do not use this option because it is not described anywhere.

I've place a lot of 0402.  You really need to set the pickup height with care, otherwise you'll just bounce them around.  I use different height offsets, as the pickups aren't precise enough across the entire machine.

The pickup heigt has nothing to do with the xy coordinate System. I have placed graph paper on the workbench and did some measurements with the camera. In the X axis I have an accumulating error of >= 0.2mm per 100mm. In the Y axis round about 0.15mm/100mm.

I think all NeoDen 4's have travel errors - the only way to help limit it is to use good fiducials on your boards. We did have one nozzle which had to be replaced, during the rotation calibration there was no center hole visible. We don't use anything smaller then 0603 and use the camera for all parts. Keeping the nozzles the clean is important. We have also had parts slide on the nozzles so we slowed things down. Lastly we have had errors in one axis that we can't explain, I think there is a bug in the math calculations somewhere, or rounding error etc, the error typically gets larger the bigger the board is.

[LitlePCB]

Lastly we have had errors in one axis that we can't explain, I think there is a bug in the math calculations somewhere, or rounding error etc, the error typically gets larger the bigger the board is.

see attached pics - this is the Y axis while 300mm traveling.

[LitlePCB]

I am currently writing my own software to create the file that the NeoDen4 pick & place machine Needs. Every time I go a step further, I encounter a new Problem 

Now I have successfully compensated the positioning accuracy. At the "Vision Align" (after automaticaly get the fiducial positions) the crosshair of each component is exactly at the middle. But if I start the Job, some components and by the way always the same on all panels, are placed a bit too far to the left. 

[Smallsmt]

Your problem looks like a Z height variation.
Maybe bend pcb or nozzle height uneven?

[LitlePCB]

Your problem looks like a Z height variation.
Maybe bend pcb or nozzle height uneven?

Always only with the one and always the same component? I do not believe that. The components are all from the same feeder and place height was testet visually.
Placementspeed was reduced to 10 (percent? Apples? Eggs? Milliseconds?)

@neoden

Why do not You update the coordinate list in this form after detecting the fiducials? This would greatly help to determine the error.

[Eclipze]

10% speed, really?  I only slow down for parts that have a bit of mass, so they don't slide on the nozzle from rapid movements.  Full speed, but use vision to correct for everything.

[LitlePCB]

10% speed, really?  I only slow down for parts that have a bit of mass, so they don't slide on the nozzle from rapid movements.  Full speed, but use vision to correct for everything.

This is for testing only! If I am able to run a full job, I will increase the speed of course

[Eclipze]

Try full speed... if no difference, then speed isn't your problem and you're going slow for no reason.  If there is a difference, then speed is a problem, but your running so slow that the error isn't as noticeable.

[Smallsmt]

Sometimes 0.1mm Z height change makes the difference!

The problem depends maybe on PCB or PCB holder so this area is a little bit lower than left side.
If you use same head same feeder and same nozzle this will be the most recent point.

Try to reduce the component height setting in steps of 0.05mm and check.

[LitlePCB]

Try full speed... if no difference, then speed isn't your problem and you're going slow for no reason.  If there is a difference, then speed is a problem, but your running so slow that the error isn't as noticeable.

I have already increased the speed with the slider above the Buttons. This has the consequence that the travel speed has 100% but only the picking and placing process remains at 10%
As I wrote - the reduced speed is only for my tests.

Edit: I changed the Rotation of the misplaced component from -90 to 90. The hell with it, now it fits!!!!

[Suntee Sun]

Dear Sir,

This is Suntee from NeoDen Tech, thanks so much for your attention and support to us.

Our newest version of NeoDen4 is v4.1.2.3, please kindly email to support@neodentech.com, we will send our upgrade file to solve your problems asap.

Yours sincerely!
Suntee

[LitlePCB]

Our newest version of NeoDen4 is v4.1.2.3, please kindly email to support@neodentech.com, we will send our upgrade file to solve your problems asap.

Dear Suntee,

please have a look at the attached Picture 

[LitlePCB]

@neoden I have a tip for you here.

For example: You are using opencv 2.4.9 in your Software. Why do not use the histogram equalization function at the image that you can see at the section "System setup"/"System configuration"/"Nozzle X individually alignment"?

See attached pictures before and after for reference. This would make it easier to position the nozzle visually. If you would still allow positioning via the keboard while the image is visible, that would be a great relief.

Further suggestions will follow.

[jr408]

Would you guys be interested in a dedicated forum/wiki on this machine? I've been thinking to write up some notes on getting started with the machine and it might be a useful to form a dedicated community.

[rom67]

1. Check the nozzle is clear for good suction.  There was one time I had an 0402 stuffed length ways in a nozzle obstructing the air pressure.   
2. Check the nozzle end is clean.  I clean the tips before every build.  Mostly in case any paste/flux residue causes stickyness. 
3. Check the accuracy of the tape feeding.  If not enough, the tape indexing might have a little bit of variation. I often use a feed strength of 40, as when too high, it causes the parts to bounce out of the pockets.
4. May sure to index twice, then set the pickup alignment for that feeder.  Very important... set the alignment target for the pocket square, not the component.  Index and double check it's lining up.

Thank you very much for advices.
I found the Samsung nozzle CN040 which has some bigger central hole than the NeoDen's nozzle.
Unfortunatly for 20 mounted SMD 0402 I losted 7 components (and it's the best result).
Usially I'm losting from 10 to 20 components for 20 mounted with NeoDen's nozzle and from 7 to 15 with the Samsung's. It's very regretfully.

[Eclipze]

That is quite unlike what I've experienced with 0402 placement.  I tend to only get missed parts if the nozzle is dirty (component sticking) or the pick height is too low and tape bounce dislodges the part from the tape pocket.  I use up facing vision for all parts to get good production placement.  Angles in pick and place data area always positive.  Avoid negative angles.  Avoid negative coordinates too (that's a nightmare).

[LitlePCB]

Does anyone know anything about the sequence of the fiducial recognition?

I know so far that they use the first chip position as reference Position to locate the fiducials because when this is changed approximately +/-5mm the machine is not able to find the fiducials.

After the first fiducial has been located, where is the origin of the z-axis? First Chip? First located fiducial?
And when the second fiducial has been located, will the z-axis origin moved to another position or it remains on the last Location?

[LitlePCB]

After approximately 20h operating time the first three LEDs of the upward camera are defective 

     We should never have bought this machine!     

[Spikee]

Have you bought this machine trough the local distributor ?
If so they should properly support you. If you bought it direct and really hate it you could ask them if you can return it.
You would have to pay the return shipping but at least you get something back ...

[Kjelt]

After approximately 20h operating time the first three LEDs of the upward camera are defective

Not good but also no reason to panic, at least that is something easy to fix and to improve upon.
If the camera or motors start failing that is the time to panic.