Controller Board for DIY Pick place Project

[Alphatronique]

Hi

still progress firmware was now 99.9% debugged
but like any project when you think it done you just find unexpected stuff

placement speed have been increased
and homing function was now much better and reliable

when i will got next pc software i will able to make new video that assemble a real board

Best regard

[Alphatronique]

Hi all ...

ok i still increase speed seem that 400mm / second was max that machine may take
actually whit acceleration/deceleration curve it 333mm/second (scope say 1,5sec for moving 500mm)

from motor stop to take or place part it ~100ms (75 whit no rotation ,90 whit rotation ,95 whit mechanical centering)

so i got close to original machine spec

basic firmware work now only remain 1 small bog for be able to place a complete board
bog was that a each part placed head lost 0.2mm , on X,Y so after 5 part part trouble begin whit feeder
i knot that it not mechanical missing step since it happen to any speed
so i think that is during the direction change that i lost a step , will hook logic analyzer on it for cont pulse send to stepper

i really happy whit that project noting beat a machine that was build right for you
and was able to make change you like when you like and i was still well under new machine price

Best regard

[Alphatronique]

Hi

finally found the missing step issue (a stupid programing error).
so now head position no more drift

so i will try to build a pcb today ...

Best regard

[Alphatronique]

Hi

assembly of 48 closely spaced resistor whit view of GUI and close-up on head / pcb ..

http://www.youtube.com/watch?v=MOtb3EI0DEg

Best regard

[tastewar]

Very cool!!

[Alphatronique]

Detail of GUI operation

from cad system to production in <3 minute
whit PCB panelization in machine software whit PCB preview :cheers:

.ini file hold feeder pick point and last feeder setting on machine

http://www.youtube.com/watch?v=4VNK7XBh_ls

Best regard

[blogger]

Dear Alphatronique

Would you like to write proper usable software for the MDC pick and place?

[Alphatronique]

Hi

i knot some one that have make it ,but it also redo most of mechanical
and it on older machine

problem whit new machine is that it much more complex
and reply on vision for "fix" mechanical defect
my old zevatech was build like a thank since i born to be blind machine
so i very repeatable but rebuilt it today will really expensive

and i not knot how MDC layer will react to a hack of it current machine ?

but on the good news side i really think to make a kind of universal
modular controller that may fit many machine
but this will later once i will finish to integrated vision to my software
and fully defogged it , is sill have some small issue to fix before make it "commercial" grade

p.s. last night i have increase speed by 30% so 1300 part /hour whit feeder at 12" from placement zone

Best regard

[Alphatronique]

HI

during long week-end i have fix\add

1-) move from old pm360 machine to a newer 460 machine whit ATC ,add the cognex camera to machines
and build 5 controller card so now all my machine have the new controller (ya i buy 2 more machine few week back)

2-) ADD auto tool changer support ,for now we will only handle first 5 bit ,bit #6 pose problem whit negative coordinate

3-) bootloader in firm and software

4-) fix cut-tape feeder operation on software

5-) handle part rotation in 1.7 deg increment , this still need mechanical calibration

6-) begin to work whit machinist for make my own nozzle ,CAD draying was done waiting for first prototype

7-) cognex vision system integration in software
ok for vision software was ready but still have lot of work to put on optic side
i have problem whit lighting ,mirror reflection and gosh image ,uniform focus ,etc etc
will need front surface mirror and box for control illumination and
probably help for some one that better that me in optic ,so if you may help me not be shy

on image above RED circle was the machine bit position ,green cross was calculate part center
and text on top was data send trough rs232 to pc software

in this case "X:300.666 Y:410.057 T:-3.830 S:44.391"

on PC side it do ...

X = ((X - 240) / 21) and Y = ((Y - 320) / 21) "21" was scale value for pixel to mm
T = rotation in degree
S = score < 40 vision result was considered as void

this will convert in +/- value based on center position

in this example part was X=300.6 - 240 = 60.6 \\ Y= 410.0 - 320 = 90.0 pixel off center
whit -3.8 degre rotation scale value will be ~21
x=60.6 / 21 = 2.88mm
y=90.0 / 20 = 4.28mm

problem was that this was a still image take whit perfect illumination and focus on stopped machine
but on a working machine it hard to make it reliable and handle variable lighting condition


Best regard

[Alphatronique]

Hi

this was ghost effect caused by reflection on mirror
so i just order 38mm front surface mirror from Edmund Optics

think i finally found right camera distance and lens set that work well
it have to got resolution for fit a 100 pin TQFP and a soic-8 whit same optic ..
that explain wly expensive machine use 2 camera whit 2 set of lens...

Best regard

[Alphatronique]

Hi

look that IC whit marking on bottom was not may confuse system and see it as a pin ...
so seem that i will need to work again on the vision system programing
red circle = machine nozzle
big green cross = calculated part center
small green cross = IC pin found in image doing mean of all pin coordinate and rotation = part center
for now i not find better way to have one programing do all

a easy way may teach every single part but i what to keep it simple and one thing do all
or at least as more as possible

the good thing is image resolution was fine whit this setting so it may have hope

[Alphatronique]

HI


this is how now look the .ini file ,so here all major machine config was stocked
i build a box for camera so i will able to test it back vision part of the software


[Settings]
CoordUnit=128 ;for old pro format
SpeedSlow=1000 ;speed value for slow
SpeedMed=4000 ;speed value for medium
SpeedFast=8000 ;speed value for fast
LaneMaxX=361.00 ;in mm
LaneMaxY=510.00 ;in mm
TrickX=309.00 ;in mm ATC Tes PAD
TrickY=127.00 ;in mm ATC Tes PAD
CameraX=319.00 ;in mm
CameraY=171.00 ;in mm
CameraPpmmX=21.00 ;in ppmm (pixels per mm)
CameraPpmmY=21.00 ;in ppmm (pixels per mm)
JunkX=0.00 ;in mm Junk bin for reject Vision part
JunkY=0.00 ;in mm Junk bin for reject Vision part
PCBZoneX=54.80 ;in mm PCB 0.0 cordinate
PCBZoneY=145.80 ;in mm
PCBZoneWidth=300.00 ;in mm
PCBZoneHeight=300.00 ;in mm
CameraTimeout=200 ;in ms
TakePlaceTimeout=2000 ;in ms
MotorCheckInterval=250 ;in ms Motor position pooling delay
TakePlaceCheckInterval=100 ;in ms
CurrentBit=-2 ;-2=no info, -1=unknown bit, 0=no bit, 1-5=bit number
COMPort=1 ; Modbus
CameraPort=2 ; Cognex camera port
EnableVision=1 ;set to non-zero to enable
DebugMode=1 ;set to non-zero to enable

[Default Part Components] ;lesser number has higher priority
1=AD8132ARZ,SOP8P ;Assing AD8132ARZ to SOP8P pakage
2=R*A,0805 ;Assing any Resistor finish by "A" to 0805 pakage
3=R*B,0603
4=C*A,0805 ;Assing any CAP finish by "A" to 0805 pakage
5=C*B,0603

[Default F1 Lane Coordinates] ;all in mm (Default pick point for feeder)
X1=15.70
Y1=34.90
X2=37.20
Y2=34.90
X3=58.70
Y3=34.90
X4=80.20
Y4=34.90
X5=101.70
Y5=34.90
X6=123.20
Y6=34.90

[Default F2 Lane Coordinates] ;all in mm
X1=170.70
Y1=34.90
X2=192.20
Y2=34.90
X3=231.70
Y3=34.90
X4=235.20
Y4=34.90
X5=256.20
Y5=34.90
X6=278.20
Y6=34.90

[Default L1 Lane Coordinates] ;all in mm
X1=1.40
Y1=151.30
X2=1.40
Y2=172.80
X3=1.40
Y3=194.30
X4=1.40
Y4=215.80
X5=1.40
Y5=237.30
X6=1.40
Y6=258.80

[Default L2 Lane Coordinates] ;all in mm
X1=1.40
Y1=306.30
X2=1.40
Y2=327.80
X3=1.40
Y3=349.30
X4=1.40
Y4=370.80
X5=1.40
Y5=392.30
X6=1.40
Y6=413.80

[Default B1 Lane Coordinates] ;all in mm
X1=12.80
Y1=505.10
X2=34.30
Y2=505.10
X3=55.80
Y3=505.10
X4=77.30
Y4=505.10
X5=98.80
Y5=505.10
X6=120.30
Y6=505.10

[Default B2 Lane Coordinates] ;all in mm
X1=167.80
Y1=505.10
X2=189.30
Y2=505.10
X3=210.80
Y3=505.10
X4=232.30
Y4=505.10
X5=253.80
Y5=505.10
X6=275.30
Y6=505.10

[Default R1 Lane Coordinates] ;all in mm
X1=356.85
Y1=145.50
X2=356.85
Y2=159.50
X3=356.85
Y3=173.50
X4=356.85
Y4=187.50
X5=356.85
Y5=201.50
X6=356.85
Y6=215.50
X7=356.85
Y7=229.50
X8=356.85
Y8=243.50
X9=356.85
Y9=257.50
X10=356.85
Y10=271.50

[Default R2 Lane Coordinates] ;all in mm
X1=356.85
Y1=300.50
X2=356.85
Y2=314.50
X3=356.85
Y3=328.50
X4=356.85
Y4=342.50
X5=356.85
Y5=356.50
X6=356.85
Y6=370.50
X7=356.85
Y7=384.50
X8=356.85
Y8=398.50
X9=356.85
Y9=412.50
X10=356.85
Y10=426.50

[ATC] ;all in mm ATC cordinate
PADX=307.00
PADY=127.00
AX=327.15
AY=116.35
BX=327.00
BY=91.50
CX=327.00
CY=66.45
DX=326.90
DY=41.50
EX=326.90
EY=16.45

[Components] ;format = {Name}={Bit},{Speed},{WidthMm},{HeightMm}
0603=2,Slow,16,8
0805=1,Fast,20,11
1608=1,Fast,16,8
2125=1,Medium,21,25
3216=1,Medium,32,16
MLF16W=1,Medium,16,8
MLF4W=2,Medium,16,8
MLF8W=1,Medium,16,8
QFN=3,Slow,80,200
SOP16P=3,Medium,120,60
SOP8P=3,Medium,80,60
SOT23=5,Medium,20,16

[Part Types] ;format = {Name}={NeedVision}
3000=1 ; 1= vison enable on that part
744066100=0
AD8132ARZ=0
AD8137YR=0
ADC122S655=0
AT45DB161D=0
ATMEGA128=0
ATMEGA644-20AU=0
C100GA=0
C103GA=0
C104GA=0
R102A=0
R102B=0
R103A=0
R104A=0
R104B=0
R120A=0
V9MLA0805=0
X10.00D=0
X12.00D=0
X14.745D=0
X25.00D=0
X8.00D=0

[Alphatronique]

Hi

wow new camera bod work really nice

i use now optic grade front surface mirror ,ok it bit expensive but no more reflection
and new lighting also help allot , it test also a new algo for TQFP i now try to find i corner
then extract center based on corner position and rotation

this solve i part my lens problem since camera only need to see part corner not the whole part
so whit same lens i make TQFP-100 to 0402 resistor ,i setup new box\lens for "see" ~25mm x ~30mm

in this image calculated center was on a pin so it not to bad for a first try

[Alphatronique]

Hi

whit few mod i have something working now

centering was right on bouldzaill

need to implement that change on pc side software

[Alphatronique]

Hi

base machine function was now really close to be ready for production (less vision)
so i will release file format and modbus implementation

for now seem that future of the project will be commercial way ,i put lot of money on this hardware
so hardware / software will be not released to source level
but instead the hardware will sold to fair cost and software supplied for free
since modbus protocol and file format was public any one may make it own hardware specific to it machine

first batch on prototype was a viable for beta tester on ebay now .. 260794241847

so in my future post here i will focus only on vision system , i will as i do in the past put public the .job file
so any one will be take advantage of my work on this field ,so vision part will remain open source

Best regard