Five bar Parallel SCARA print area problem

JoergS5

@rutku said in Five bar Parallel SCARA print area problem:

G91 G1 H2 X90

ok, first, I made a fault for testing: G91 means relative, but I wanted the arm to rotate to the 90 degree position, so this must be G90 G1 H2 X90.

The relative rotation means 53 degree plus 90, which is your 143 degree (the Count value is about 132T). But the arm seems to rotate into the wrong direction, so 53-90 = -37 degree, which looks correct of the arm position in the image*). Please change the S0 / S1 value for the stepper definition and retry.

If it's
M569 P0 S1 ; Drive 0 (X) goes forwards
then please change to
M569 P0 S0
or other round.

*) on the image the arm rotated to position 141, but it should be 143, so you can make the M92 value a bit higher. But I don't know the reason, can be some error in arm length or hinge play.

rutku

@joergs5 Hi, yes the problem is solved. Completed the movement as you say. But when I want to draw a square, the machine works on the back section.

@joergs5 said in Five bar Parallel SCARA print area problem:

M569 P0 S1 ; Drive 0 (X) goes forwards

sqaure.gcode

G90
G1 X0 Y90
G1 X0 Y140
G1 X50 Y140
G1 X50 Y90
G1 X0 Y90

SQUARE VIDEO

@joergs5 said in Five bar Parallel SCARA print area problem:

on the image the arm rotated to position 141, but it should be 143, so you can make the M92 value a bit higher. But I don't know the reason, can be some error in arm length or hinge play.

There is some curvature in angle gauge. showing two degrees missing.

JoergS5

This post is deleted!

JoergS5

This post is deleted!

JoergS5

This post is deleted!

JoergS5

@rutku

The following tests are to verify that the arms are rotating correctly and an idea about the reason for the curved lines. Arm 1 is the upper arm in the images which is connected to the wheel, arm 2 the lower arm, so only watching the proximal arms. With hotend I mean your pencil position in your case.

You're still using 128 microsteps, so please add F200 to all the following G1 commands, so the processor has more time to calculate values and to generate the steps. Hiccups (= lost steps) will probably not being reported by your hardware.

Step1: (check direction and M92)

run homing, arm 1 shall be 53 degree, arm 2 90 degree
run G90 G1 H2 X90 F200 and verify that arm 1 is 90 degree now and rotated counterclockwise
run G90 G1 H2 Y70 F200 and verify that arm 2 is 70 degree now and rotated clockwise

If one or both arms rotated wrong, change S0 to S1 or S1 to S0 for the corresponding stepper (P0, P1) and redo the test until the arms rotate as expected. If the resulting degree is wrong, search the reason. One possible reason is if the belts have too low tension. Another one is too much play in the hinges. A third possibility is that the arm approaches the endstop too fast, so the angle is not set correctly. Then set the F parameter lower in the line which is commented "approach endstop slowly". The endstops must be approached from the same side always (always from left or always from right), because they trigger at different positions. You may want to change G1 H2 X-2 Y-2 F900 to a higher value (like X-10 Y-10) also, because you may have moved too far before.

Step 2: (check X0 symmetry)

homing
run G90 G1 H2 X90 F200 and G90 G1 H2 Y90 F200 and verify that both arms are 90 degree and the hotend is on the X0 line
run M114 and note X and Y positions. X should be near 0, Y somewhere at 140
run G91 G1 H2 X20 F200 and G91 G1 H2 Y-20 F200 and verify that arm 1 rotated counterclockwise, arm 2 rotated clockwise and the hotend is still on the X0 line.
run M114 and note X and Y positions. X should be still near 0, Y somewhere between 50 and 140
check visually that the hotend is on the X0 line and the reported Y value is correct (mm distance from the X axis)

Step 3 (check for curved line)

homing
run G90 G1 H2 X90 F200 and G90 G1 H2 Y90 F200 and verify that both arms are 90 degree and the hotend is on the X0 line
run G90 G1 X0 Y100 F200 and verify that arm 1 rotated counterclockwise by about 10 degree and arm 2 rotated clockwise by about 10 degree
run M114 and note X and Y positions. X should be 0, Y at 100. The hotend should be on the X0 line (middle between wheels) and 100 mm away from the x axis in Y direction.

Step 4: (check whether low segmentation is the reason)

homing
run G90 G1 H2 X90 F200 and G90 G1 H2 Y90 F200 and verify that both arms are 90 degree and the hotend is on the X0 line
run G90 G1 X0 Y50 F200 while drawing a line. If it's a curved line, then do this:
take a new sheet of paper or a pencil with different color and reset position to home and G90 G1 H2 X90 F200 and G90 G1 H2 Y90 F200, then run: G90 G1 X0 Y100 F200, then G90 G1 X0 Y90 F200, then G90 G1 X0 Y80 F200, then G90 G1 X0 Y70 F200, G90 G1 X0 Y60 F200, G90 G1 X0 Y50 F200. In other words, instead of one movement, use mutliple small movements. Compare the lines. Is the line of small movements more straight than the one-movement-line?

Please tell me the results of steps 1 to 4, before we continue.

rutku

@joergs5 said in Five bar Parallel SCARA print area problem:

You're still using 128 microsteps, so please add F200 to all the following G1 commands, so the processor has more time to calculate values and to generate the steps. Hiccups (= lost steps) will probably not being reported by your hardware.

Step1: (check direction and M92)

run homing, arm 1 shall be 53 degree, arm 2 90 degree arm 1 = new 51 degree

run G90 G1 H2 X90 F200 and verify that arm 1 is 90 degree now and rotated counterclockwise ✓

run G90 G1 H2 Y70 F200 and verify that arm 2 is 70 degree now and rotated clockwise ✓

If one or both arms rotated wrong, change S0 to S1 or S1 to S0 for the corresponding stepper (P0, P1) and redo the test until the arms rotate as expected. If the resulting degree is wrong, search the reason. One possible reason is if the belts have too low tension. Another one is too much play in the hinges.

Step 2: (check X0 symmetry)

homing ✓

run G90 G1 H2 X90 F200 and G90 G1 H2 Y90 F200 and verify that both arms are 90 degree and the hotend is on the X0 line ✓

run M114 and note X and Y positions. X should be near 0, Y somewhere at 140

m114

M114 
X:9999.900 Y:9999.900 Z:0.000 E:0.000 E0:0.0 Count 83200 83200 0 Machine 9999.900 9999.900 0.000 Bed comp 0.000
ok

The M114 command does not give X and Y positions for some reason. Your help is thanks.

JoergS5

@rutku according to RepRap.cpp source, 9999.9 means: Ideally we would report "unknown" or similar for axis positions that are not known because we haven't homed them, but that requires changes to both DWC and PanelDue.
// So we report 9999.9 instead.

[ Maybe I made an error in programming somewhere, that after a G1 H2 move the M114 doesn't know the position. Please go on without the M114 and look whether the hotend position is correct my measuring the distance from the X axis. ] => the C parameter reason is more probable

I see another problem: your C parameter is C-90:90:-90:90
which means the arms 1 and 2 cannot have an angle greater than 90 degree. The Step 3 and 4 tests will not work, because they need bigger angles. Please set the C parameter to C-90:180:-90:180. Even your Step 2 test is at the boundary of the 90 degree limit, maybe this is the reason for the M114 9999.9 values.

I plan to rework the firmware for a detailed object model (asking for the paramenters in json format), it would be useful imho to report the reasons for angle limit errors also. It's easy to fall into the angle limit traps.

rutku

@joergs5 said in Five bar Parallel SCARA print area problem:

which means the arms 1 and 2 cannot have an angle greater than 90 degree. The Step 3 and 4 tests will not work, because they need bigger angles. Please set the C parameter to C-90:180:-90:180. Even your Step 2 test is at the boundary of the 90 degree limit, maybe this is the reason for the M114 9999.9 values.

I did what you said. but gave the same mistake. I wrote 270 degrees, gave the same mistake.

config.g

M669 K9 L1 X-95:93 Y0:0 P126:133 D210:213:0:0 B51:90 A15:165:0:360:0:360 C-90:270:-90:270 
M208 X-1000:1000 Y-1000:1000 ; set axis minima and maxima

m114

M114
X:9999.900 Y:9999.900 Z:0.000 E:0.000 E0:0.0 Count 101689 64711 0 Machine 9999.900 9999.900 0.000 Bed comp 0.000
ok

I'm going on by skipping the M114 command.

JoergS5

@rutku yes, please skip. I am curious about the results of your tests!

rutku

@joergs5 said in Five bar Parallel SCARA print area problem:

Step 3 (check for curved line)

homing✓
run G90 G1 H2 X90 F200 and G90 G1 H2 Y90 F200 and verify that both arms are 90 degree and the hotend is on the X0 line✓
run G90 G1 X0 Y100 F200 and verify that arm 1 rotated counterclockwise by about 10 degree and arm 2 rotated clockwise by about 10 degree

This command is very strange sounds. I repeat the same action four times. The same sound continued to go out.

Strange Sounds

debug

G90 G1 X0 Y100 F200

constraintsOK => cachedInvalid:0 cachedX0:-29.16 cachedY0:307.49
getInverse => not cantilevered proximalL:126.00 distalL:210.00 xOrigL:-95.00 yOrigL:0.00 x_0:0.00 y_0:100.00
getIntersec firstRadius: 126.00 secondRadius: 210.00 firstX: -95.00 firstY: 0.00 secondX: 0.00 secondY: 100.00
getTurn => turn:-16759.64 = (x2:-206.06 - x1:-95.00) * (y3:100.00 - y1:0.00) - (y2:59.51 - y1:0.00) * (x3:0.00 - x1:-95.00)
getTurn => turn:16759.64 = (x2:-29.87 - x1:-95.00) * (y3:100.00 - y1:0.00) - (y2:-107.86 - y1:0.00) * (x3:0.00 - x1:-95.00)
getTheta => workmode: 1 thetaA:151.82 thetaB:301.12 proxturnA:-16759.64 proxturnB:-16759.64
getTheta => use: 2 x2: -29.87 y2: -107.86 thetaB: 301.12 x1: -206.06 y1: 59.51 thetaA: 151.82
getInverse => not cantilevered proximalR:133.00 distalR:213.00 xOrigR:93.00 yOrigR:0.00 x_0:0.00 y_0:100.00
getIntersec firstRadius: 133.00 secondRadius: 213.00 firstX: 93.00 firstY: 0.00 secondX: 0.00 secondY: 100.00
getTurn => turn:-17592.40 = (x2:21.18 - x1:93.00) * (y3:100.00 - y1:0.00) - (y2:-111.94 - y1:0.00) * (x3:0.00 - x1:93.00)
getTurn => turn:17592.40 = (x2:209.85 - x1:93.00) * (y3:100.00 - y1:0.00) - (y2:63.52 - y1:0.00) * (x3:0.00 - x1:93.00)
getTheta => workmode: 1 thetaA:237.32 thetaB:28.53 proxturnA:-17592.40 proxturnB:-17592.40
getTheta => use: 2 x2: 209.85 y2: 63.52 thetaB: 28.53 x1: 21.18 y1: -111.94 thetaA: 237.32
getInverse => cachedInvalid 0 cachedX0:0.00 cachedY0:100.00 cachedX1:0.00 cachedY1:100.00
getInverse => cachedXL:-29.87 cachedYL:-107.86 cachedXR:209.85 cachedYR:63.52 cachedThetaR:28.53 cachedXR:209.85
getInverse => cachedYR:63.52 cachedThetaL:301.12 cachedXL:-29.87 cachedYL:-107.86
getInverse => cachedInvalid:0 x_0:0.00 y_0:100.00 xL:-29.87 yL:-107.86 thetaL:301.12 xR:209.85 yR:63.52 thetaR:28.53 x1:0.00 y1:100.00
constraintsOK => cachedInvalid:0 actuatorAngleLMin:-90.00<0 && thetaL:301.12>actuatorAngleLMax:270.00
constraintsOK => thetaL(-58.88) -=360
constraintsOK => cachedInvalid:0 thetaL:-58.88 < actuatorAngleLMin:-90.00 || thetaL:-58.88 > actuatorAngleLMax:270.00 return false
constraintsOK => cachedInvalid:0 actuatorAngleRMin:-90.00<0 && thetaR:28.53>actuatorAngleRMax:270.00
constraintsOK => cachedInvalid:0 thetaR:28.53 < actuatorAngleRMin:-90.00 || thetaR:28.53 > actuatorAngleRMax:270.00 return false
getAngle => x1:-29.87 y1:-107.86 x2:0.00 y2:100.00 x3:209.85 y3:63.52
getAngle => angle:88.32 = angle2:350.14 - angle1:261.82
constraintsOK => achedInvalid:0 headAngle:88.32 < headAngleMin:15.00 || headAngle:88.32 > headAngleMax:165.00 isnan(headAngle):0 return false
getAngle => x1:-95.00 y1:0.00 x2:-29.87 y2:-107.86 x3:0.00 y3:100.00
getAngle => angle:320.70 = 360 + angle2:81.82 - angle1:121.12
constraintsOK => cachedInvalid:0 angleProxDistL:320.70 < proxDistLAngleMin:0.00 || angleProxDistL:320.70 > proxDistLAngleMax:360.00 isnan(angleProxDistL):0 return false
getAngle => x1:93.00 y1:0.00 x2:209.85 y2:63.52 x3:0.00 y3:100.00
getAngle => angle:321.61 = 360 + angle2:170.14 - angle1:208.53
constraintsOK => cachedInvalid:0 angleProxDistR:321.61 < proxDistRAngleMin:0.00 || angleProxDistR:321.61 > proxDistRAngleMax:360.00 isnan(angleProxDistR):0 return false
constraintsOK => cachedInvalid:0 cachedX0:0.00 cachedY0:100.00
CartesianToMotorSteps => motorPosX = cachedThetaL:301.12 * stespPermm[X_AXIS]924.44
 motorPosY = cachedThetaR:28.53 * stepsPermm[Y_AXIS]924.44
CartesianToMotorSteps => machinePos[0]:0.00 machinePos[1]:100.00 motorPosX:0.00 motorPosY:924.44
CartesianToMotorSteps => XYZ_AXES:3 numVisibleAxes:3ok

JoergS5

@rutku said in Five bar Parallel SCARA print area problem:

88.32 = angle2:350.14

you can try the same test with L2 mode, but I fear, there is an error in the firmware and I have to setup your configuration and test it myself. The debug code looks like the calculated angle is strange and incorrect. Step 3 is exactly where workmode 1 changes to workmode 2, this may confuse the firmware. It should behave logic however even then, leaving the arms at the top. But your debug seems trying to rotate the arms to the bottom.

Please give me your config.g content and the current homing file. Please tell me which firmware version you use. I'll setup a printer like yours. Give me a few days please until thursday.

If you want to further test your printer, use L1 mode and stay in the positive X region with Y between 50 and140. This should work ok. You can make Step 4 test to test whether segmentation is the reason for your curved lines. If you want, you can experiment with the M669 S and T parameter, please see https://duet3d.dozuki.com/Guide/Five+Bar+Parallel+SCARA/24#s103

When I know the reason, I'll tell you which part of the source to change.

rutku

@joergs5 said in Five bar Parallel SCARA print area problem:

Please give me your config.g content and the current homing file. Please tell me which firmware version you use. I'll setup a printer like yours. Give me a few days please until thursday.

Ok I'm waiting for you. Thank you so much.

m115
FIRMWARE_NAME: RepRapFirmware for STM32F4 based Boards FIRMWARE_VERSION: 3.2.2_2 ELECTRONICS: STM32F4 FIRMWARE_DATE: 2021-05-02
ok

config.g

; General preferences
G90 ; Send absolute coordinates...
M83 ; ...but relative extruder moves
M552 S1 ; Turn network on

M586 P0 S1 ; Enable HTTP
M586 P1 S0 ; Enable FTP
M586 P2 S0 ; Enable Telnet

; Drives
M569 P0 S0 ; Drive 0 (X) goes forwards
M569 P1 S0 ; Drive 1 (Y) goes forwards
;M569 P2 S1 ; Drive 2 (Z) goes forwards
;M569 P3 S1 ; Drive 3 (E0) goes forwards
M584 X0 Y1 ; set drive mapping

M350 X128 Y128 I1 ;S3 ; Configure microstepping with interpolation
M669 K9 L1 X-95:93 Y0:0 P126:133 D210:213:0:0 B51:90 A15:165:0:360:0:360 C-90:270:-90:270 
M92 X924.444444 Y924.444444 ; Set steps per mm

M203 X10000 Y10000 ; maximum speeds mm/minute

M566 X15 Y15 ; Set maximum instantaneous speed changes (mm/min)
M201 X40 Y40 ; Set accelerations (mm/s^2)
M906 X1200 Y1200 ; Set motor currents (mA) and motor idle factor in per cent

M84 S0 ; Set idle timeout

; Axis Limits
 M208 X-1000:1000 Y-1000:1000 ; set axis minima and maxima

; Endstops
M574 X1 S1 P"xstop"   ; X min active high endstop switch
M574 Y2 S1 P"ystop"   ; Y min active high endstop switch
;M574 Z0 S1 P"zstop"   ; Z min active high endstop switch

home5barscara.g

G91              ; relative positioning
 
G1 H1 X100 Y300  F400 ; move quickly to endstop and stop there (first pass)
G1 H1 X300 F400
G1 H1 Y300 F400
G1 H2 X-2 Y-2 F400      ; go back a few degrees
G1 H1 X100 F90  ; move slowly to endstop once more (second pass)
G1 H1 Y100 F90  ; move slowly to endstop once more (second pass)
;G1 H2 X-5 Y-5 F400      ; go back a few degrees
 
G92 Z0
G90              ; absolute positioning
;G0 X0 Y90 ; move to a reasonable position

JoergS5

@rutku thank you.

My idea is to introduce a new mode L5, who joins working modes 1, 3 and 4 to one print area. This would allow you to use most of your area of positive Y.

rutku

@joergs5 An excellent idea. I was thinking about this idea. I didn't mean to do that without running the system. Also are not required offset for X and Y axes?

JoergS5

@rutku the offsets for x and y are required, because the firmware has to know where the actuators are placed.

I will make a new thread in the firmware development space, you'll get a first version for testing when I implemented L5. The thread is here: https://forum.duet3d.com/topic/23483/parallel-scara-printer-improvements
When you and I tested L5 successfully, I'll make a pull request to the official firmware.

rutku

@joergs5 I'm following the topic. I'll be happy to test new updates. Thanks, I'll continue with the other thread.

JoergS5

@rutku we should discuss your specific config here, and the firmware development there. I install a printer with the same properties like yours, so we'll find a working solution for your printer.

JoergS5

@rutku I thought about merging work modes. This is difficult, because while changing the mode, the position of the hotend changes and during the change and rotations, time is lost. So it must be while not printing and should probably be placed in the slicer, not the firmware. If the firmware decides, it will break the print process at places one doesn't wish.

I concentrate on code verification, i.e. whether the code is correct for work mode 1 for your test scenario (later for the other modes too), and improve the existing code like I describe in the other thread.

rutku

@joergs5 said in Five bar Parallel SCARA print area problem:

@rutku I thought about merging work modes. This is difficult, because while changing the mode, the position of the hotend changes and during the change and rotations, time is lost. So it must be while not printing and should probably be placed in the slicer, not the firmware. If the firmware decides, it will break the print process at places one doesn't wish.

It doesn't matter. I will think of something in the future. My priority is the work of the 3d printer. After working, I will understand better the codes you wrote.

@joergs5 said in Five bar Parallel SCARA print area problem:

I concentrate on code verification, i.e. whether the code is correct for work mode 1 for your test scenario (later for the other modes too), and improve the existing code like I describe in the other thread.

Yes, I'm looking forward to the result. My 11 month old baby started breaking my 3d printer. I can only debug on my small laptop. My eyes are dying. Tomorrow I will take my 3d printer to my office. I'm going to review your actions on the big monitor in my office. My eyes will be relieved. Have a nice day...