MDC Luna 7722FV Bottom Vision Problem

Chat about pick and place machines, reflow ovens, assembly techniques and other SMT tips & trix

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
gsattler
 
Posts: 48
Joined: Tue Nov 24, 2009 3:44 pm

MDC Luna 7722FV Bottom Vision Problem

Post by gsattler »

Hey all,

We have a 7722FV and have been using it to place primarily 0805 and 0402 components since we bought the machine about a year ago due to lack of feeders. Initially we continued to mount all of our leaded chip parts by hand, but would now like to do those with the pick & place as well. We recently upgraded from the MDC Luna NX 2.0.3 software to 2.1.9 software and we are trying to get the bottom vision to work more reliably for placing leaded chip parts. We have tried LV 126, LV 26, and LV 226 for the smaller chips as well as LV 127, LV 27 and LV 227 on the larger chips. If we are able to pick up the part in exactly the middle, then the part will place well, but due to the variances in the feeder advance mechanism or tolerances of the tray or tube, we are not always able to pick up the part in the exact center of the part. It is interesting that when we changed from the 2.0.3 software to the 2.1.9 software that the machine began to run more slowly. We have made these changes either to the machine or to the FconstXX.dat file:

1) We replaced the power supply with a MeanWell RS-100-24 and we no longer have any power supply problems, even when running the X and Y axes at full speed.

2) We slowed down the machine as mentioned in in the Adafruit SMT/MDC troubleshooting wiki in the hopes that placement accuracy would be improved. This didn't seems to affect the positioning speed of the machine.
http://www.ladyada.net/wiki/mdcpickandp ... oubleshoot

3) We also implemented the Closed Loop Vision as mentioned in the Adafruit SMT/MDC wiki
http://www.ladyada.net/wiki/mdcpickandp ... loopvision

When we update the FconstXX.dat file with those changes, either together or individually, we do get an improvement in the angular placement of the chips, but the X-Y accuracy of the chips is still lacking. The only other change we had to make to the constant data to enable 2.1.9 to work was changing K37 #3 from 2 to 0 to indicate the correct starting position of the cut tape feeder. The 0805 and 0402 components are still being placed very well, and we are using the largest nozzle allowable for each chip. We have tried to go through and understand the settings in the FconstXX.dat file with some limited success, but would love to compare FconstXX.dat files with other MDC Luna NX 7722FV users. I have attached our Fconst7l.dat as well as a matrix that I made to more easily understand the FconstXX.dat file. I hope that one or both of them are at least helpful to the community at large. I don't think that it would be a good idea for you to run your machine with this FconstXX.dat file since some of the file may be machine specific, I really don't know. If we are able to fix this problem we will be immensely happy.
Attachments
MDC Luna NX 7722FV Fconst & chart.zip
MDC Luna NX 7722FV Fconst.dat & Explanation chart
(59.78 KiB) Downloaded 242 times

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by adafruit »

can you post some screenshots of what goes wrong? we've been placing TSSOP FT232RL's and QFP atmega32u4's without difficulty (SOIC's are a breeze) and also bigger pitch leadless parts such as mma7455 altho with leadless there's always a wierd offset we account for in the pcb layout file
since we went closed loop life has been much better

gsattler
 
Posts: 48
Joined: Tue Nov 24, 2009 3:44 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by gsattler »

MSOP Bottom Vision Try #1
https://picasaweb.google.com/lh/photo/P ... directlink

MSOP Placed Top Vision Try #1
https://picasaweb.google.com/lh/photo/s ... directlink

MSOP Bottom Vision Try #2
https://picasaweb.google.com/lh/photo/J ... directlink

MSOP Placed Top Vision Try #2
https://picasaweb.google.com/lh/photo/- ... directlink

QFP Bottom Vision Try #1
https://picasaweb.google.com/lh/photo/k ... directlink

QFP Placed Top Vision Try #1 (used handheld camera to get whole chip clearly)
https://picasaweb.google.com/lh/photo/U ... directlink

QFP Bottom Vision Try #2
https://picasaweb.google.com/lh/photo/i ... directlink

QFP Placed Top Vision Try #2 (used handheld camera to get whole chip clearly)
https://picasaweb.google.com/lh/photo/m ... directlink

User avatar
ktownsend
 
Posts: 1447
Joined: Thu Nov 05, 2009 2:18 am

Re: MDC Luna 7722FV Bottom Vision Problem

Post by ktownsend »

How are you handling the fiducials, out of curiousity? I sometimes end up putting the fiducials in manual mode on certain boards to make sure they are perfectly centered ... I've never seen a QFP off be one pin entirely in both X and Y.

blogger
 
Posts: 43
Joined: Tue Nov 24, 2009 5:59 am

Re: MDC Luna 7722FV Bottom Vision Problem

Post by blogger »

Personally I've never seen a QFP properly placed by this machine, period.
No matter how accurate the position was defined or how many fiducials or whatever.
The fact that the machine in assembly mode needs TWO variables (set offset... WTF FOR??) *and* fiducials will mean there's never going to be proper alignment.

What I'm seeing in gsattler's pics for the thing just randomly throwing a QFP down on the board is exactly the kind of stuff I get with my machine.

Also, having to re-position tape/strip feeders on each machine power-up is getting boring but something I just have to deal with.

User avatar
ktownsend
 
Posts: 1447
Joined: Thu Nov 05, 2009 2:18 am

Re: MDC Luna 7722FV Bottom Vision Problem

Post by ktownsend »

I just built a bunch of boards with QFP48 today and had no problems (and I always inspect them under a microscope before putting them in the oven since I still add a couple oddball parts by hand)??? Closed-loop vision made a huge difference for me with QFP and QFN since now it makes 2-3 passes with the bottom vision if necessary before getting it right. Sometimes it might be off by 1 or 2 mil, but nothing that wouldn't be fixed in reflow.

User avatar
ktownsend
 
Posts: 1447
Joined: Thu Nov 05, 2009 2:18 am

Re: MDC Luna 7722FV Bottom Vision Problem

Post by ktownsend »

Just did a quick test placing a QFP48 on a test board using partial assembly (F6 in the PCB part screen, after setting the fiducial alignment first). It's off by maybe 2 mil on one axis but that would snap into place during reflow. With closed-loop vision this is pretty much the results I'm getting every time. Sorry if the photo is BANNED ... I overexposed it two stops but I'm too lazy to put the camera back on the microscope.
Attachments
QFP48Test.jpg
QFP48Test.jpg (59.66 KiB) Viewed 5916 times

isozee
 
Posts: 15
Joined: Sat Mar 27, 2010 11:29 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by isozee »

sattler wrote
When we update the FconstXX.dat file with those changes, either together or individually, we do get an improvement in the angular placement of the chips, but the X-Y accuracy of the chips is still lacking.
From your photos attached, vision system looks work fine so only concern is the component is placed on PCB with correct pressure and no jumping up after placement.
You can watch it carefully but you may use step motion by pressing function key F4 during assembly then head stops after every motor motion and pressing Enter key it advances (Q to quit). This way you can see the placement action more easily.
If component tends to jump up then check nozzle surface (any paste etc.) or increase puff time or increase Z stroke.
Re motor slow in V2.1 09 from V2.0.3,
When motor speed builds up usually motor speed increased by 4 increments but now it is 16 increments as default to improve missing pulse problem of stepping motor.
To use original motor setting then change system constant k20 #5 value to 4.

gsattler
 
Posts: 48
Joined: Tue Nov 24, 2009 3:44 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by gsattler »

Can the 7722FV do X-Y correction with bottom vision? We are getting chips placed with correct angular orientation but the X-Y placement is still off. If we pick up the chip in the middle, then the chip places accurately, if we pick it up off-center, then the chip will place off by about that much. Ktownsend, are you reel feeding any small pitch parts? Do you tilt your chip tray to put all chips biased to the same corner before calibration? How do you deal with the variability of chip placement in the holder?

User avatar
ktownsend
 
Posts: 1447
Joined: Thu Nov 05, 2009 2:18 am

Re: MDC Luna 7722FV Bottom Vision Problem

Post by ktownsend »

I normally use trays for QFP and just rotate the parts before placement (90, 180 or 270 degrees). I'm not intentionally rotating them for placement. I haven't really had any big problems with QFP placement myself, though, at least not more than maybe 2-3 mil max and they usually come out fine during reflow. Can you confirm that you're using closed loop bottom vision, though? When you watch the QFP part on the bottom vision camera, does it make more than one pass examining the corners? If it makes 2 or 3 passes than closed loop is enabled and if the first match isn't great it tries again (I think up to 5 or 6 times but this is a constant somewhere and I forget the default). If it only makes one pass, that's likely your problem there. Honestly, I think closed loop vision should just be enabled by default. The extra 250ms it takes is trivial relative to a misplaced part.

The only largish parts I'm feeding from tape af 32-pin QFPs on 12mm tape (using a cut tape feeder). I haven't had any real problems with them, though there isn't much wiggle room on that reel either.

I hadn't though of air pressure, but what Isono san definately makes sense. When I had the air pressure too high the parts were spit out a bit to violently and it caused some offset since they bounced back. It takes a bit of trial and error to get the air pressure just right ... enough to advance the tape and pick up heavy parts, but not enough that your slamming parts onto the board with the final puff to eject them.

The one part the is always hopelessly off for me is some 3216 tantalums, but I think it's the package since the leads are a bit off. I can probably fix it if I mess with the definition a bit.

You didn't mention how you were handling fiducials either. Try using manual fiducials just as a test to see if that's part of the problem, manually centering them on startup.

It doesn't sound to me like you have closed loop vision enabled, though, and it's only making one pass with the bottom vision and going with those results. How large is the QFP as well?

gsattler
 
Posts: 48
Joined: Tue Nov 24, 2009 3:44 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by gsattler »

Hey all, we are using manual fiducials on an older board design and were and are using closed loop vision and also were not having problems with component jumping after placement. The air pressure is set to get consistent pickup and not much more. We seem to have finally prevailed in the battle to fix the bottom vision! In the end we messed with the settings in the FconstXX.dat file in these locations:
K31 #3, #4 - X & Y offsets for 90° with Bottom Camera
K32 #3, #4 - X & Y offsets for 180° with Bottom Camera
K33 #3, #4 - X & Y offsets for 270° with Bottom Camera
K35 #3, #4 - X & Y offsets for 0° with Bottom Camera
K46 #3, #4 - X & Y offsets for 45° with Bottom Camera
K47 #3, #4 - X & Y offsets for 135° with Bottom Camera
K48 #3, #4 - X & Y offsets for 225° with Bottom Camera
K49 #3, #4 - X & Y offsets for 315° with Bottom Camera

The offset for all X's was the same, and the offset for all Y's was the same. Blogger, you might try a few test placements with some double sided tape and see if this fixes the problems you were having with your closed loop bottom vision setup. Before changing all the offsets we were getting really bad placement offsets and couldn't make heads or tails of how to fix it. Now we get consistent placement of LQFP-100 parts. Our other small parts like some MSOP-8 regulators, a TSOT23-8 and various SOT23-5 parts are also consistent and work well.

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by adafruit »

you know what is wierd. we have pretty much no problems with leaded parts but for all our leadless they are recognized perfectly but placed offset, with a really predictable offset (we hand correct) but offset nonetheless. its a bit of a mystery we'd like to fix one day :D

gsattler
 
Posts: 48
Joined: Tue Nov 24, 2009 3:44 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by gsattler »

Adafruit, we had a similar offset problem until we found the LV27 bottom vision camera crosshair calibration menu. Now all is good.

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by adafruit »

hmm! that sounds interesting!

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm

Re: MDC Luna 7722FV Bottom Vision Problem

Post by adafruit »

We updated our tutorial for troubleshooting to add how to realign the bottom camera

http://wiki.ladyada.net/mdcpickandplace ... _placement

Locked
Please be positive and constructive with your questions and comments.

Return to “SMT (Surface Mount Tech)”