MDC / Manncorp 77xx(FV) expert questions

by vincentp on Sat Sep 01, 2012 5:56 am

hi !
I'm posting this topic for both solving a couple of placement issues I have and adding that knowledge in the handbook I'm writing for this machine.

For now, after running a calibration for the nozzle, flying vision and bottom vision, I have very consistent placement results and I'm very happy with the machine. Aside of a few feeders I miss for larger parts (I finally got them in reels) that I'll purchase later,
I have 2 parts I can't place automatically because I don't know where to start,
so I'd like to discuss this here

One part is a 10x10x4 mm square inductor
Image

it's provided in reels, I also made tray to simplify vision and placement testing. I couldn't find a way to test it correctly.

I tried defining it as a tantalum cap (TAN method, side vision), contrast and brightness would work however, the part is so large that the vision testing area is enlarged to almost all the camera FoV width (the vision "working area" is the part dimension multiplied by a coefficient) and I have a marker that is found in the right side of the camera image.

- is there a way to reduce that working area for that particular part (or for all parts, no suitable IMO but...)

- is there a way to test it with the bottom vision. Again, I tried.... but... the part is coming in the tape with the bottom pads view as up and down and for instance, if you want to check it like a QFN with the CSP methods you need pads on the left and right sides to have the symmetry working, so at the end, I would need the part to be presented in front of the camera with the pads L and R. I couldn't find a way to ask the vision to pre rotate the part of 90° before checking it, find the zero (which is actually 90°) and compute placement angle from there.
Worse, I have to confirm but I think that it might be other way around, that's because I have the part rotated 90° on the PCB that the part is pre rotated and can't be checked.
Basically I would like to define the presentation angle to the bottom cam, then have the placement angle applied.
possible ?

the second part I have is a micro SD card holder (*chime sound* "holy grail"). anybody was able to place them automatically on this machine ? if so, with which method ??

actually, for both the inductor and the SD card holder, I would "just" (ahah... it's always "just" needs which sound like "simple" needs I know) need a bottom cam vision technique that takes the bounding box of a part (I just need to setup the contrast and threshold to get it right). From there, estimate a center which can be a pondered center, I don't care, then match this with my placement location by trials and errors or simply by adding placement offsets in my part def. The inductor, in particular, has only 2 BIG pads and placement accuracy isn't important.

any idea how to do that ? the new software has a LOT more in the part definition menu but I haven't tried them all and I don't know what all of them do.

help ?

thank you !

[EDIT] ---------------------------------------------------------------------------------------------------------------------------
Adding a bug list to address to MDC to keep the software moving.

Here's my listing, I'll keep adding up thanks to other users ! I'm using v2.1.21.
BELOW IN RED WHAT HAS BEEN CORRECTED WITH v2.1.23

- Writing in the constants will generate an error after a few edited value, if you don't leave the constant screen and keep editing. Software crashes, you might need to edit the constants in several attempts - I haven't played much (again) with the constants but I'll keep trying with 2.1.23.

- cancel the tape index during manual assembly (F6). Trying placement (and vision) on a part sourced from a feeder, you might want to replace the part yourself instead of having the tape indexing. SW prompt for index but zero doesn't cancel it. FIXED

- Tray angle. I thought it was a bug but it's the MDC answer to my request of having the tray angle better handled. Now PCB angle is only the DESIRED angle / orientation of the chip and doesn't deal with how the chip is oriented in the tray (example in post #11 below). FIXED.

- 5 key should be able to return to camera mode after nozzle mode. It works when you do teach mode => nozzle mode (9). Press 5 you return to camera vision. This does NOT work after returning from a vision test (it would be useful so that you don't have to go with the double ESC ESC to discard feeder location change) - FIXED PARTIALLY. When you come back from a vision test 5 takes you back to camera mode but when you press ESC twice you'll jump again to NOZZLE MODE.

- When playing with part def during vision test. I leave my nozzle above the tray of the part I test and I test the part over and over. Calling F3 to check (and edit) part def works the first time. Return from vision check (eventually put the part down or keep it sucked), press F3 to tweak part def again, head returns to origin. NOW WORKS - JUST DO NOT SAVE (with F2), this seems to cause the head return to the origin next time you press F3 to get the part def, for some reason. Save data once part testing is finished and when you are back to the datasheet (PCB, feeder/tray def etc).

- Assembling a panel. Once the panel is finished, I should be able to assemble another one. If I press 0 to check the bias on the new PCB I install, I move to the origin and the bias is shifted. BEWARE, if you don't correct it, it will be kept until you terminate assembly. Even if you terminate and start over, bias is lost. I need to return to PCB list, edit mode, F8 on bias point (which stayed correct in the data list), head moves to the right point, I can assemble again. FIXED (it seems)

- in the DATA list submenu, the LowCam button (right next to the nozzle exchanger menu button) doesn't work. It's supposed to take you to the bottom camera for continuous part checking but it opens the nozzle exchanger data window instead. FIXED in 2.1.23 - an updated update of the 2.1.23 has been put online on the 23/0902012 - if you got the 2.1.23 earlier, please DL it again

- terminating the assembly in emergency (F10, STOP button) when validating "TERMINATE" will turn vacuum ON while it was OFF when you pressed STOP. As a result, you must manually turn the vacuum off (I go in a feeder list, or anywhere I can go in teaching mode, then press V). If vacuum was on when assembly was interrupted, it's turned off properly though. NOT FIXED YET

- The stroke (up down) axis of the machine is referred as Z pretty much everywhere in the manual including in the system constants. It's also name U (with the associated keyboard shortcut) in teaching mode. The problem is that in the teaching mode screen, the ANGLE is name Z which makes no sense. I asked MDC if they could replace the Angle name there by theta, A, angle, anything but not Z. U can stay, I got used to it as the stroke (possibly referenced as U due to the spelling of 'stroke' in Japanese ?). FIXED - thank you !

- in the tray programming (teaching mode), only the crosshair is displayed. The bounding box of the part should be displayed too, just like for feeder programming. FIXED - brilliant, thank you !

- Mechanical synchro problem. My last placement program line is a big inductor that requires a feeder indexing of 4 piston push. At the end of the assembly of panel, the machine tries to move the feeder indexing piston back to origin while it's still indexing the feeder (generally when the piston is out, pushing on the feeder lever). As a result, many steps are slipped / missed and I need to re-home otherwise the feeder system is totally lost. It's not good at all for the machine neither. FIXED in newest 2.1.23 (28/09/2012) see changelog to be sure you have the right version

- Terminating assembly while being in manual centering (Lv=25) : you can't quit the current centering operation and the mode and software is stuck (I need to force it to terminate) FIXED. If the manual ("semi auto") centering is canceled, the part is taken to the dump area and assembly stops.
Last edited by vincentp on Mon Oct 01, 2012 6:45 am, edited 13 times in total.
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by vincentp on Wed Sep 05, 2012 11:59 pm

Solved the problem for the inductor.
got it defined as CAP with bottom vision 26. Level was turned down to 0.4 as the part is highly reflective. Vision stayed to 26 (and not the darker 126) because the leads are dull (yeah, why should it be easy after all, I'm a nerd) so I had to keep a regular lighting.

CAP part def is "special", with the use of the Pt part as "scanning depth" for the edges. All detailed in the upcoming manual. Big thanks to MDC for teaching me all this.

For the SD, I yet have to try a part def for it. I think I can define it as a connector however, it doesn't have the leads on the edges but inside, so I'll see how it ends. I'm pretty sure it's possible to automate this. If it does work, I'll have 100% of parts done in full-auto.

More time to play pinball I guess :lol:
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by vincentp on Sat Sep 15, 2012 12:03 pm

I edited the first post. I thought that it would be useful to collect some of the bugs of the MDC software so that I can address a list. I've already started this and certain things are in the pipe, other things were corrected.

The version 2.1.21 is working nicely overall, aside of the tape index that is corrected on the next version.

If you experienced some repetitive bugs, please post with how to obtain the bug. Overall, I moved from 2.1.16 (which was on the PC I got) to 2.1.21 since late june. One of the bug I found was for instance the refresh of the bounding box when changing the angle in teaching mode (decrement was refreshing, not increment). MDC is very responsive to correct those things.
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 16, 2012 3:30 pm

Hi Vincent,
This is a great thread. Thank you. I'm a new owner of a 7712 and am slowly figuring this all out. None of the threads I have gone past only the most basic howto for a setup. The manual is missing lots of details where the reader has to assume or just know what to do. My first mistake was not realizing the parts type library is used by all projects/files. I deleted all the data in the one provided and regret it. Now I have to start over defining parts.
Now I have 10 of the following parts on my board and I'm not sure how to define them in the parts type library:
01. SOT23-5 low-dropout regulator 5v, Digikey part: 296-20717-1-ND
02. MSOP8 Low-Power 22-Bit ADC, Digikey part: MCP3550-60E/MS-ND
03. VSSOP8 Dual-supply bus transceiver, Digikey part: 296-17014-1-ND
04. QFN16 17v Synch Step-down converter, Digikey part: 296-19717-1-ND
05. SOT223 Regulator 5v-3.3v, Digikey part: LM3940IMP-3.3CT-ND
06. LGA14 ADXL345 3-axis accelerometer, DigiKey part: ADXL345BCCZ-RLCT-ND
07. QFN24 3-axis gyroscope Invensense ITG-3200, Invensense: ??
08. LCC16 3-axis digital compass, Digikey part: 342-1082-1-ND
09. SOT523 Small Signal Transistor, Digikey part: DTA143ZE-TPMSCT-ND
10. 3.2x2.5mm/4-pin Crystal 12.000MHZ 18pF, Digikey part: 887-1121-1-ND

I was wondering why we don't either create a shared on on this Wiki or share what we have? Since the library is shared for all projects if we had a massive contains everything library everyone could use it. I'm willing to share what I learn. Anyone willing to share theirs?

Any help with the above list of parts very much appreciated. I have all my parts for this board defined including the above but the above are defined right now as LV1, X&Y approx measurements and type chi. Which means the outline when I'm in the PCB parts menu is not really representative of the part.

I must say though, I'm really happy so far with the machine in that as I am learning these things I find the machine is fully capable of doing what I need. I'm just struggling to understand the manual and how to define parts.

Thank you Vincent for starting this thread. I will try and contribute to it as I learn. Thank you for sharing what you are learning about parts definitions. I'll be very interested in helping to write that new manual. Can we have a Wiki version of it so many people can edit and update it?

-BMW
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 16, 2012 4:00 pm

All,

I have a question about entering data into either the parts type library, the feeder or PCB lists.

I have Reels in feeder 1-10,
I have cut tape parts in blocks on the other side 28-51.

My questions:
1) The gaps in the feeder listing. 11-27 and those blocks not used. They are currently filled with values from a prior feeder entry because I don't really know the best way to add new lines.
2) How do you add new lines? If I go to 2,7 (edit, add) It asks me for an input value. What is this value? If I use 1 it seems to copy line 1 one time. The behavior seems common to many of the edit add menus but isn't explained at all anywhere.
3) I want to just add a new line after the last line. So say I'm in the PCB parts menu and I have part 1, feeder 1 as 100nF CAP that I place 16 times on a board. What is the best way to add those 16 lines? I input 16 to the Add option and it added 16 copies of an existing line. I then went in one by one modifying. Very time consuming.
4) For the cut tape block I see a pair of green paralel lines on the right side of the part. On the 8mm blocks they line up perfect with the holes for advancing the tape. However on the larger blocks they do not match at all. How do I edit this and train it?

Thank you again for helping with my questions. Let me know if theres a better way for me to format them. Should add screen shots or other details.
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 16, 2012 7:54 pm

Good news,
I used LV=1 and chi with the outer dimensions as X,Y. I was able to get all my parts input.
Using F6 to do test placements I go the rotations right. It appears from watching the placement that it picks the chip up, looks at it, then either rejects it or places it. It was rejecting one chip I setup as SOP but changed later to chi.
This is getting more fun.
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by vincentp on Mon Sep 17, 2012 1:38 am

bmw,

thank you for posting those. Most of your questions will be answered in the document I'm writing (now about 160 pages, dedicated to the 7722FV but most of it will be applicable to the 7712 I believe).

the part definition file is indeed shared, it's a common library, and it makes sense (all PnP software work with a library of some sort). I took a moment checking the file structure of the MDC software before doing anything, it ended up being useful. That very point is mentioned in the handbook (soon to be finished for its beta version at least).

part definition : adafruit has a tutorial for a LGA part which should match extactly your ADXL acell. I used the same definition (but with a different sizing) for the something similar to your gyros (another QFN, I use 16 and 32 pads packages).
It's all in the wiki here
http://www.ladyada.net/wiki/mdcpickandplace/qfn

A basic parts definition :
http://www.ladyada.net/wiki/mdcpickandplace/partdef

your SOT-23 : flying vision, defined as TR or tr depending on the orientation. You might have a rotation offset (and eventual X-Y offsets). WAIT until you read my document as I explain the calibration process.

Skipped parts : your tolerance could be too narrow (I use 70% in general). Your pixel to mm scale ratio for the side vision camera could be wrong too (calibration again), therefore part is rejected.

SSOP parts should be defined as SOP. I don't check them with flying vision but with bottom vision. However, flying vision works on them BUT you will have to add offsets. Once the offsets are defined in the part definition, placement is really consistent. MDC provided me examples of placement of a PLCC-32 with flying vision. Aside of an angle offset / correction, it works perfectly.

Feeders numbering : you don't need to add new lines. Your machine has 2 banks of possible feeders, leave the list as is, just tweaks the feeder data when needed (ie when you have a feeder installed). Unused feeder slots just remain here, they aren't called from the PCB data list, that's it.
Use the remark field of the feeder list to remember what is populated / used in your design, in case you have a lot of feeder swapping.

adding new lines to the program : I use add (or insert). The method you define is correct. I start with a 100nF, setup the stroke, place the part. Then add XX of them, they will be copied, then you place each one in teaching mode at the right place. That's not the method that is consuming, overall, programming a board is time consuming. Teaching mode is great but when it comes to large PCBs, it has a limit. It's not a software problem but a method problem. You can also use CAD data import to have the PCB program created automatically, if offers a good start base if you board has like 500 lines. Mine is like 30 parts (ish). Programming takes me like 1hr and that's it. Add testing, I would say that a good morning is required to program & test a PCB.

Cut tapes block : I don't know much about them. Could you post a picture of what you mean about the alignment with the green line and the dots ? is it a manual cut tape system or an auto cut tape ?

About how parts "look like" on the PCB (in teaching mode). It's the bounding rectangle most of the time, with an orientation mark. The orientation mark DOES NOT WORK for tray-sourced part, I've ask MDC if something could be done for that. Ideally, I'd like the red dot of a QFN or TQFP to reflect the actual expected orientation of the chip on the PCB. Tray angle should be taken in consideration for that. It is, but it doesn't shift the orientation in the right way it seems, or I got the tray angle thing wrong (possible).
Example : you want your part to have the QFN with the dot in the lower left corner (ie +90° rotation). Your tray is defined with an angle of -90° because the part is rotated 90° CW in it. Therefore the PCB data is be +180°. Placement works (proper orientation) BUT the red dot location (affected by the feeder angle) doesn't end at the right place. It's just "annoying" when programming, but once you know the PCB angle data is correct for the part, it's just working. This is less problematic with SOT or SOP which tend to have their orientation correct as there's only 2 possibilities. Issue is currently with square packages with leads / pins all around.

The manual will be really like a book even if I'm not a "publisher". I don't see a problem to provide the source DOC so that it's inserted in the wiki but I really see the document as something you print to have on hands all time when you use the machine. I do have the machine PC hooked up to internet to check the wiki BUT I really like printed material when I work as I find changing between windows very annoying.

I was thinking about a link to 2 versions of the PDF, one low res, the other one with printing quality. Note that some of the pictures I haven't captured again yet, so it's the stock MDC pictures, some of which are out dated. I'll keep fixing this in v2.0 as I'm running out of time for that job.
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Tue Sep 18, 2012 8:05 pm

Thank you Vincent,
Reading and digesting now.
Oh, before I forget. Another learning item. The casset feeders. I put all my 10 8mm reels side by side slot 1-10. Then found out that the tape was not being pulled back resulting in the head not finding the parts. What it appears is everyone uses scotch tape to hold the tape on the black spool. However that adds thickness and a surface to bind when the reels are loaded side by side tightly. So now I need to space them every other slot. Which means me going into the tape feeder screens and redoing the assignments. Not a real biggie but a hassle. How is this not mentioned anywhere? Am I the only one to have this happen?
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by vincentp on Wed Sep 19, 2012 12:42 am

Not sure what you mean about your 8 mm feeders (picture of the problem?)

I do load them all side by side (currently using slots 1-15 ) and I don't have such issue. To be accurate, I'm using the automatic feeders, not the tape / auto cut tape ones. I alternate up and low reel position with the back arm but that's it. I'm using tape as well (transparent 3M tape) to hold the clear cover tape on the black spool, things are working nicely. Bigger feeders (12,16,24 mm) off course need more spacing, sometime wasting a slot. By alternating up/low reel positions, I was however able to have only 1 slot between, like a 12mm feeder in slot 29, the next in slot 31

a picture would help to sort you out.

[EDIT] attached my "Jelly Beans" feeder group for reference
feederbank.png
feederbank.png (114.13 KiB) Viewed 3561 times
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by vincentp on Thu Sep 20, 2012 1:10 pm

I think someone requested some part definitions, here are mines, quite similar to adafruit ones. I didn't invent warm water here, stock part definition file is fairly correct. You often start tweaking it because of unwanted offsets that are not related to the part definition itself but to machine calibration.
The only problematic thing is the threshold thing (during bottom vision testing), if it's totally off and you don't know where to start, vision gives you an error (and the finger :lol: ) and you have to go with trial an error but most of the time, I use the +/- buttons during the vision to adjust things and I see where it starts to work, and it's a matter of minutes before you finalize the part.
My square inductor gave me a hard time but... it's working now.

my parts definition file (screen cap... I can post the file contents as it's just test AFAIK but I though the screen cap would be more user friend to visualize all the params).

you can see several fiducials. I now tend to abandon the shiny ones and replace them by holes (part #25)which I find to be more consistent and insensitive to the copper finish you get (and I'm not forced to buy gold or HAL finish).
Ignore the level in the definition of SOT parts, this parameter isn't used (it applies only to bottom vision).

parts.png
parts.png (27.22 KiB) Viewed 3546 times
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by vincentp on Sun Sep 23, 2012 7:27 am

MDC uploaded software version 2.1.23, many aspects of the bug list (post #1) have been corrected *thumbs up*.

If you plan to migrate from as low as v2.1.21 (or before), here's a warning. Tray angle is described in the manual as a way to indicate the main axis of the part (longest dimension) isn't on X. This has changed. Programming the tray angle indicated to the software how the part is oriented on the tray. Then comes the relationship between the tray angle and the part angle on the PCB.

EXAMPLES
For a TQFP, the zero (0°) orientation is top left corner. If your part is rotated CW (-90°) in the tray, so you put -90° in the tray angle.

PRIOR 2.1.22 : the part angle in the PCB data is the angle APPLIED to the part to get it oriented properly on the PCB. For instance, you want to get your part at 0°. As the part is -90° in the tray, you ask the placement program to "compensate" by +90° (90 CCW). As a result the part ends at 0°

This has the HUGE drawback that the angle programming relies on the way the part is in the tray. As a result (and to my demand I think), the PCB angle reflects the desired part placement orientation. Like parts sourced on feeders actually.

From 2.1.22 : define the part orientation on the PCB as it should be. If the part must be at 0°, put the angle at 0. Define then the orientation of the part in the feeder. The software will calculate the compensation angle.
It's then easier to change things as you don't need anymore to correct both PCB and tray angle data, but just the tray angle, if you get a tray in which parts are (for some reason) oriented differently.

Bottom line : if you were using tray angles in your designs, you'll need to correct those data to ensure your parts are placed properly oriented. If you never used tray angle (left it 0), your PCB data should remain correct as the angle you defined is already a compensation angle offset.

Overall, 2.1.23 is GREAT ! thank you MDC !
vincentp
 
Posts: 98
Joined: Tue May 15, 2012 2:29 am

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 23, 2012 4:14 pm

vincentp wrote:Not sure what you mean about your 8 mm feeders (picture of the problem?)

I do load them all side by side (currently using slots 1-15 ) and I don't have such issue. To be accurate, I'm using the automatic feeders, not the tape / auto cut tape ones. I alternate up and low reel position with the back arm but that's it. I'm using tape as well (transparent 3M tape) to hold the clear cover tape on the black spool, things are working nicely. Bigger feeders (12,16,24 mm) off course need more spacing, sometime wasting a slot. By alternating up/low reel positions, I was however able to have only 1 slot between, like a 12mm feeder in slot 29, the next in slot 31

a picture would help to sort you out.

[EDIT] attached my "Jelly Beans" feeder group for reference
feederbank.png


I Vincent,

Yes, exactly. That's how I had them (8mm reel feeders). I have since separated them by one space. I will see how it goes now.
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 23, 2012 10:22 pm

FWIW: I checked my software version: LUNA-Win version 2.1.19

I see 2.1.23 is available. How do I download, install it? Will it overwrite the old directory? Can I keep my part library?

I look on the QFN page here: http://www.ladyada.net/wiki/mdcpickandplace/qfn
The columns do not all match up with the ones I have. I will take a look and use the values you give Vincent. Thank you. Maybe it's older version vs newer version.

The irony. I got the machine to save me time but I have no time to learn to use it ..and once I do learn it I hope to have time again. :)
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 23, 2012 10:40 pm

Vincent, do you have a link to the download for the software? Do I write Mr. Isono directly? -D
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm

Re: MDC / Manncorp 77xx(FV) expert questions

by bmw330i on Sun Sep 23, 2012 11:32 pm

partLGA14s.jpg
Image to show the advancing unit guide is not aligned.
partLGA14s.jpg (462.96 KiB) Viewed 4423 times
Hello,
I'm not sure I've seen this yet here. Apologies if it is. There is a lot to read I may have missed it. This part is one of the ones not advancing. Basically any of the > 8mm cut tape do not advance properly. I'm pretty sure this image shows why. The pickup head/advancer is not aligned.

In this image there are two parallel green lines. I believe this represents the advancer guide. Notice is is far to the left of the holes it should be lined up with. How do I modify this?

-D
bmw330i
 
Posts: 51
Joined: Mon May 07, 2012 8:59 pm