Wire up the retraction cable management system

Trikarus uses a very thick main cable with a lot more weight. To compensate the weight effect we wire up a self retracting dyneema cable system (see V2 Silent VR Cables by KIWI).

1. Wire up the six retraction cable units
   1. Put the first unit next to the B anchor where the main cable leaves the ceiling plate and attach it to the main cable. Keep around 50 cm distance between the mounting point and the ceiling plate
   2. Attach all other units to the main cable and leave 50 cm space between each mounting point. Follow the winding direction of the main cable
   3. The last unit will be attached at the mounting point of the really first unit because one retraction cable unit can only handle a maximum of 180 cm elongation
2. zero out the small silent stoppers - we do not need them
3. rotate the rubber fixations at the main cable until the face inward. This reduces possible effect of collisions with the movement system

De-spool unrequired D line amounts

The effector is mounted by three lines, which are all on the same spool (D spool). Due to different winding, the height or the tilt angle (or the horizontality) of the effector can always change somewhat. It's a good idea to have some more line on the spools to have some backup but too much line is bad for the movement system. For this reason, the followup calibration should be done with as maximum as possible despooled lines. To have maximum quality of the behaviour of spooling up /down lines while printing, its good to leave only the amount of line on spool wich is required to keep Hangprinter fully operational in it's build volume.

Note: our system has got 28 windings for each coil on sools A, B and C. That's more than enough for our dimensions.

1. Move effector down (Z direction)

   By sending an according move command the D-axis is going to lower the effector towards the floor. Move it until it hovers a few centimeters above the floor. Note that the ABC lines are dropping down to the bottom because they are not handled automatically by the drives. We do not perform classical XYZ movements in this step. We only move the D drive down. The ABC lines will loose tension while down movement until they are nearly flat on the ground. Please have close look that the lines do not jam with other elements like maintenance tools. In regular printing mode that might also be the print platform which could lead to line jammings. If the effector is moved in the wrong direction, the configuration should be checked. The direction of rotation of the motors can be changed in one of the following three ways: config.g in RepRap firmware, physical wiring of the phases or configuration in the Smart Stepper firmware. You can also pull the effector manually without any GCode movement. Then you will need to power down the machine or power off the D drive Smart Stepper or set the D drive Smart Stepper into "torque 1" mode.

   G0 H2 D-250 F500 ; repeat until the effector is at the desired height - 250 mm each. It does not yet matter if the moved height is not exactly what you entered as long as your effector moves down to the floor until it hovers some centimeters above

   

2. Pull all the line trough them while spooling down D drive. The effector should be kept at the bottom of on of the frame corners all the time. So you simulate the maximum possible required D length in normal printing operation circumstances. Leave at least two full windings on the D spool to have some good grip (less force on the blank knots).
   
3. When done re-tie the guitar tuner trimloks for knotting. Put the unrequired segment of line into the little line reservoirs (magnetic winders).
   

Equalizing the D line lengths (fine tuning)

Doing this step the effector is not level yet but you configure one of the hard requirements: lines D1 = D2 = D3. You will reach the condition that ceiling plate and effector are parallel to each other. But because the effector tilts around itself around Z axis we have make them parallel without dangling line and without any horizontal shifting first. So we need to wind up the effector back to top again until it sits in the three brackets (transportation arm fixations).

The effector needs to sit in the arms really tight. We use clockwise turning at the guitar tuners to tighten and counter-clockwise turning to losen the lines. Clockwise turning means when viewing at the guitar tuner's tuning wing points to your face (same rule is used for bottom anchors!). We are now going to "tune" all lines that they have same tension. We can check this by "playing" them like guitar strings. If they all sound equal then the tension is the same. Please not that the limit switch might create an extra offset between effector and fixation arm which should be nulled before doing this step.

Validate distance between nozzle tip and ceiling plate

While the effector is in top position it's good idea to validate some measurements. Move the effector to maximum Z position and take a ruler to measure the distance from nozzle tip to ceiling plate. This should give nearly or exactly the same values as the pre-set constants from theory:

We measure this again by hand because it's useful and because it can change if you do something with the nozzle (changing to another for example).

Remove effector tilt around Z axis and finalize frame levelness

The final horizontal alignment takes place using the three bubble levels installed on the effector, the three laser dots from the ceiling module and the machine frame.

There is a poisonous tilt around the Z axis. Omit self-twising misalignments which may come frome a effector which is idling perfectly horizontally but twists (around Z axis) by unequally created forces from effector's center of mass or unequal line lengths at the bottom anchors (if installed already). Due to the doubled lines this problem is reduced but still possible. Trikarus effector twists a lot because the extruder unit is heavy. The nozzle is in the center of the effector but the mass is not. You can see this in the photos. However, the self-rotation of the effector has bad influence on the line equality for the anchor line pairs ABC and the self-tilt also makes it impossible to get a horizontal level effector and therefore no level ceiling plate. So this step is really import. In short: the three D lines must be totally vertical.

1. move effector to the ground
   
2. put some foam pieces below the effector corners so it cannot hover freely. We want to adjust it's rotation while using the laser dots. In case the lasers were correctly preconfigured and the laser screws on the effector have correct distance, the laser light will hit the surface of the screw heads. You need to screw out the effector corner screws until they match with the laser spots. The positive effect using the laser pointers is that they project the light also directly below the screws so they appear on the ground or print platform. We can always use the laser lights to put back the effector to X0 Y0 - even at higher Z positions. This can be really helpful for resurrecting failed prints.
   
3. adjust the ceiling plate guitar tuners again until the effector spirit levels are fine
   

Adjust  to align effector axis and nozzle axis (zeroing out offsets)

The center of the effector is not the same as the center axis of the nozzle. There might be an XY offset between those two vertical axes or you can eliminate that offset by adjusting your hotend installation on effector. If your effector has two or more tools (extruders or whatever) use the effector axis instead of nozzle axis to do later alignments. For calibration of the machine it's essential to do the best on eradicating misalignments or redundant / unrequired measurements. To eliminate possible issues it's best practice to match nozzle axis with effector axis (Trikarus is a single tool machine). The Aero Extruder mounting assembly is designed to be reconfigured. If this was not already done while assembling the effector, you should it do while setting up the printer environment. The following drawings show the orientation of the effector in the coordinate system, the A1/A2, B1/B2, C1/C2 pivot points and the offset of the nozzle. A well placed nozzle can reduce errors which can appear on slack lines while printing (self-rotation around Z axis). Note that this step is not required for finding the origin on the print platform. To place the effector into origin we use the laser pointer spots instead only.

Get the maximum movable Z height

For creating a good slicer profile it's recommended to read the maximum movable Z height for the effector. This depends on the frame or room where the Hangprinter is mounted, and it depends on the configured height of the print platform (honoring platform additions like tapes or sprays). Please have a look at the drawing in general overview.

1. To find out the maximum Z height just lower the effector until the nozzle tip touches the print bed in origo (if that's not already the case).

2. Null the Z value

   G92 X0 Y0 Z0

   
   

3. Raise the effector to the ceiling. Moving into maximum will raise the encoder error because tension will build up if the effector stucks. That's exactly what we want in this case.

   G0 H2 D250 ;repeat until effector stops moving and Smart Steppers react on torque overload.

   
   

4. Interprete the values correctly:

   1. read the encoder values of the D axis Smart Stepper
      1. geterror()
      2. readpos() - this needs to be recalculated by line buildup compensation and spool radius

   2. read the recent Z value from current position

      M114 ;read current position

      
      

   3. compare with M114 value and readpos() value. You may also use a laser distance meter to check rough value.

Fine adjustments of ABC lines orientation and de-spooling of unrequired line segments

To have maximum quality of spooling up lines while printing, its good to have only the amount of line on spool wich is required to keep Hangprinter fully operational in it's build volume. To do this release the guitar tuners and pull all the line trough it while spooling down and while effector is at the bottom (print platform). When done re-tie the guitar tuner trimloks. This step is the same like D lines which already happened while building up. Put the unrequired piece of line into the little line winders (reservoirs). Do this step while the print head (nozzle) is touching the origin X0 Y0 Z0

• the position of the six ABC anchors must be adjusted so that the following requirements are met
  • the anchor pairs A1 + A2, B1 + B2 and C1 + C2 are each parallel to the effector
  • line pairs have same length (A1=A2, B1=B2, C1=C2). The line lengths per anchor pair have to match exactly as much as possible. The anchors can be rotated and moved losening or tightening the screws at the frame. We can use the laser pointers illuminating the horizontal effector screws to validate that the effector is still in un-tilted position while we adjust the line lengths.
    
  • each anchor pivot point pair has to have the same Z coordinate. For example if A1 Z coordinate is lower or higher than A2 Z coordinate this will mean that the line length of A1 ≠ A2. If the floor and/or frame parts are bumpy/uneven you will have to ensure "clean" Z height per anchor the mechanical way first. See also Building basics, checks, maintenance
    
• you can use tape and helper lines to perform measurements and temporary markings

This step is required to get matching line lengths by knotting the lines in the guitar tuners and adjusting the position of the anchors. Using the guitar tuner's tune function does not help to apply proper tension for each drive because every time the drive loses power it will lose it's torque. We use the tuners basically just for quick knotting of the lines. For having more secure knots we can apply some windings on the shafts of the tuners. We use clockwise turning to tighten and counter-clockwise turning to losen the lines - same way like already done at the ceiling module. Put remaining line on the magnetic line winders.

The lines cannot tear simply by tensioning them (49.9 kg per rope pulling force necessary). Rather, the stepper motors lose steps beforehand or the closed loop stepper motor drivers give way (maximum holding force exceeded).  The maximum holding force for the used Building basics, checks, maintenance are 8.2 kg.
This step needs to be done when Smart Steppers are enabled/active with simple PID mode. Do not put them in torque mode for this step. Otherwise it may happen that you will need to catch moving lines or to fix messed up lines from de-spooled drives because torque mode without knotted lines results in endless turning drives.

The guitar tuner acts like a knot between the line pairs. If the lines de-spool unequally it will mean uneven forces. While printing you might figure out that there are lines which have completely no tension.

•               floss each line pair A1+A2, B1+B2 and C1+C2 in the related guitar tensioning mechanism (use the small trim lock scew to fix them - marked with red circle in the following photo)
  
                  
                  
  
                
              

• "play" the two lines <A or B or C>1 and <A or B or C> 2 like guitar strings and listen to them. Or try to use a guitar tuner. The frequency or "the sound" of both strings should be the same. Then they have the same length!
  
• Put remaining line on the magnetic line winders
  

Notes on the knot:

Ideally for ABC drives you would have one long line each which is fed through the motion system parts. But most Hangprinters including Trikarus include multiple coils. So for example there is not one A line but an A1 and an A2 line. While the printer moves it always will happen that one of the line's tension drops to zero. This could be reduced if the knot between A1 and A2 could float (moving freely) because the tension would better normalize between then. The tension per spool changes due to unequal line buildup even if A1 and A2 are on the same spool with the same motor. It would not help to knot A1 and A2 separately. It would just give some better handling for (un)mounting the lines.

Clean up the spool winding quality (fixing line salad at the drives / optimize line buildup)

Before we continue with final measurements we re-adjust the lines on spools. Previously they were put on the spools by hand. Now we can use our electronics to make the best winding we can get. Additionally it may happen (often) that you lose lines because lines get slack while printing and the winding fails. It often causes manual troubleshooting at the drives. Then you need to climp up to the ceiling and fix it.

To adjust or fix problems easily

1. smoothen the failed drives using /opt/sms_modes/sms_smooth_*.sh scripts
2. unwind the spool completely by hand. If you have luck you must not climb
3. let the spool wind up again using /opt/sms_modes/sms_torque_*.sh while keeping the lines in your hand. This way the drive will perfectly spool up (important to have best line buildup)

<iframe width="560" height="315" sandbox="allow-same-origin allow-scripts allow-popups" src="https://videos.stadtfabrikanten.org/videos/embed/0c1f9c50-0234-4d88-952c-dbb81225e545" frameborder="0" allowfullscreen></iframe>

Calibrate the IMU (MPU 9250) at the effector

Once installed and leveled physically the IMU should be nulled to have some clean reference output for correct monitoring and calibrating tasks.

TODO

Move the effector back to the top

Now the effector should be moved up to the maximum height so that there is enough space to set up the printing platform. This can be done by

• GCode

  G0 H2 D250 F500 ; repeat until the effector is at the desired height - but at least about 2 meters above the floor.

  
  

• or with torque zup mode (/opt/sms_modes/sms_zup.sh).

Preparing the print platform (apply adhesion promoter)

Trikarus uses an IR probe for Z level sensing. Usually they have high precision when used with opaque bed surfaces, medium or low with transparent surfaces. Transparent surfaces (glass, PEI etc.) must be backed with a matt black surface and must not be coated with glue, hairspray etc. The platform is opaque and can be masked with blue tape for example. You may also use other trivial "standard" products like classic hair spray or glue stick, or more specialized stuff like 3DLac or Magigoo. There are a lot of different materials and adhesion types possible. In our case we use 19 mm thick chipboard with blue painter's tape on it (to give a cooler look, to better peel off the printed parts and to protect the bare surface). You need about 0.31 m² of tape per segment and per side to wrap it. So in total you have 6 * 0.31 m² = 1.86 m² (upside and downside) surfaces to cover. On a regular tape roll there are about 2.4 m² so you will need more than the half of a tape rolls. Use a roller or squeegee to apply the tape.

Print platform position and orientation - centering print bed axis (if there is one) and nozzle axis (alignment)

If your print bed is some random piece of platform (for an example of a random platform like Torbjørn does look the following photo) with no regular or symmetric shape you cannot really align some axis. But you can mark the point where nozzle tip on print platform is in X0, Y0, Z0. Some of the following steps cannot be applied to a such a platform.

Connect platform parts

Similar to the base frame, the printing platform can be connected with screws. It consists of individual parts so that it is easier to transport by car and by human. The platform frame is built directly into the printer frame already set up.

1. Clamp straight profiles into the corners and screw together

2. Place platform feet underneath

3. Insert short straight profiles on the side lugs of the long straight profiles and insert them in the center cross

Align the print platform and leveling

The platform should be located in the middle. We can make use of the three lasers which point on the surface of the middle beams. We also also make use of the premarked center point and put the center feet of the print platform on it. Adjust all four screws until the platform is level over complete beam surface structure.

Place the 5-point polygon platform segments on the frame and check levelness again

• Place 3x platform elements. Everything should attach flatly and nothing should wobble

• If individual segments wobble or have an uneven surface tilt, small pieces of tape can be put at the bottom site to compensate for this

• configure the horizontal six platform adjusters to fixate the platform from all sides (prevent lateral displacement)

• re-adjust the platform orientation according to the laser spots and the circular template sheet. The middle is formed by the three inner bed segment edges
• If the platform was levelled correctly and if the ceiling module is really horizontally the laser dots should perfectly match the markings on the PETG template.

Mark the platform position

Dragging around the effector and measurements of ABCD lines (line lengths and anchor coordinates)

Getting measurements of the anchors feels a little bit strange because you do not measure absolute XYZ coordinates like in a regular coordinate system. It's more like a "fused system". You measure the values of a XYZ vectors for each anchor (they are called "anchor points").  This step is highly important. Using a laser distance meter we can measure the line lengths between the effector and the anchor points ABCD. You can only get useful measurements with tightened lines. Please do this step in regular torque mode (/opt/sms_modes/sms_torque.sh). This step also can help to re-adjust anchors in case the print result should fail later on (due to happened changes in position of anchors or the leveling of the effector or levelness of ceiling plate). In this step we measure anchor coordinates XYZ and line lengths to validate those anchor coordinates. At least measurements for three of the six ABC lines are required. But it's better to take all six values to check against for each line pair. This will help you to eliminate other existing adjusting errors. By measuring all six bottom anchors we can see the rotation error around Z axis.

It's enough to measure with zero digits after comma because we will never get it more precise. Pre-defined anchor coordinate values and line lengths from theory do not help in practical use. We use the real situation and are going to measure the printer how it was built and configured physically. Note that there are small deviations for all anchors because i used ceramic inlets with inner diameter is 3 mm. Remember that lines are only 0.5 mm thick. So there is 2.5 mm of possible tolerance per inlet. Try to perform measurements using the center of the inlets. The following drawings show what we need to measure.

Measuring anchor coordinates

There are two differet ways on how to measure ABC XYZ coordinates. The preferred way is to measure everything relative to the print platform (which was made horizontal before). The second best way is to measure everything from ground (which may not be level). Using both methods we can determine the error of the ground. For this step we need tape measurement, caliper, laser distance meter, some temporary lines and tape and a lot of time and calm hands.

Measuring line lengths

In march 2022 we simplified the measuring routine a lot because we use nearly exact/symmetric frame. It allows us to calculate with default angle values 30/60/90 deg. Helpful tool: http://cossincalc.com/#angle_a=&side_a=1010&angle_b=90&side_b=&angle_c=&side_c=585&angle_unit=degree

 Table sheet for downloading: abc_measurements_2022.03.04.xlsx

( is not a fixed value because it depends on the print platform adjustment. The better the ceiling plate and bed were configured to be level, the more homogenious the measurements D1, D2 and D3 are.)

If they match well we are done and finally we can configure the K parameter of M666 which defines the coordinates for firmware. After entering them in config.g we need to reboot the controller. If not we need to remeasure. We get the required M669 command from Excel sheet.

M669 A0.00:-1166.00:-30.85 B1020.18:589.00:-28.90 C-1010.65:583.50:-33.90 D2617.00

Testing to reach the entered ABC anchor coordinates

For Trikarus it's no problem to move to outer coordinates where A,B and C anchors are defined (XYZ). We might get large errors if the buildup compensation is not done well (slack lines) - i measured about 26 mm at A anchor before i calibrated the printer well - but we can just move the effector to the outermost fork anchors like shown in the photo:

We can make use of this extremity moves because if the effector is placed there we can have another visual check for line lengths (A1 = A2, B1 = B2, C1 = C2). If we find any error we can re-adjust the guitar tuner trimloks this way.

Validation of line buildup compensation (M666 Q and R) for ABCD

The line buildup compensation algorithm is really important for exact movements and to prevent slack or overtightened lines of ABCD drives. It is not that important for small build volumes but if you want to print high, using your great build volume advantage, you need it. Because without compensation the error created by spooling up and down lines will be too large (see Building basics, checks, maintenance). This makes it impossible to get good print results and it will waste your expensive filament. We need to ensure to have well tensioned lines all over the build volume. This steps requires to have properly configured movement system (e.g. corrent micro stepping, transmission ratios between pulleys and gears, etc.). The buildup compensation factor is highly sensible and is the same for all four drives. While sourcing the parts of your Hangprinter you measured / modeled the spool radius from each drive (or at least from one drive), the coil width and line diameter and you precalculated a line buildup compensation factor. If you put your effector in origin and there is a lot of idling line on spools then you have to adjust the spool radii with the extra diameter which comes from the extra line. We need to perform this validation step after leveling the ceiling module / effector / frame and after measuring and entering the anchor point locations into firmware.

See Building basics, checks, maintenance for the calucation of the factors. There exist theoretical values but due to mechanical tolerances in line diameter, coil width and so on we get some corridor of possible numbers. Usually we will find more correct factors by manual experiments. 

Checking the factors:

1. Enter the calculated factor Q and the theoretical radii (R) values in firmware config.g
2. reboot the controller (each time you change M666 settings you need to do this)
3. Place printer in origin and apply sPID mode. We use /opt/sms_modes/sms_prepare.sh which does a lot of steps like sPID mode, setting home position and resetting the error values of the Smart Steppers
   
4. Make ABC smooth using
   1. /opt/sms_modes/sms_smooth_a.sh
   2. /opt/sms_modes/sms_smooth_b.sh
   3. /opt/sms_modes/sms_smooth_c.sh

5. Perform Z movements. Move slowly using F-Parameter in G0/G1 if you lose steps (check the error value on Smart Steppers). Maybe we need to adjust the motor currents before. See Building basics, checks, maintenance

   G1 Z<VALUE>; slowly raise up D axis to +Z coordinate

   
   

6. Measure the distance between the nozzle tip and print platform center using a laser distance meter. If the buildup was calculated and entered correctly (Q and R values of M666 command) there is +/- 0 mm deviation between targeted height and measured height. Note that we cannot use the position information of Smart Stepper because they don't know anything about buildup compensation values. They only know about sent amount of step commands. The same is for error (we can only compare targeted steps with reached steps). You can even reach highly accurate XYZ coordinates at height of 2200 mm and higher. This was succesfully validated with Trikarus. Having deviations we need to adjust Q and R values and to remeasure Z until everything fits and the lines are tensioned all over their way - regardless of the Z coordinate. To validate if you found the correct line buildup value you will have a set of constraints:

   1. targeted height = measured height. There should not exist any deviations Z coordinates at the ground level ( ~ 500 mm or less) and ceiling level (~ 2000 mm and higher)
   2. no or less wobble: the ABC lines are not slack at higher Z coordinates (> 1000 mm). The effector still feels stiff when touching because the lines are just tight like yourself
   3. the Smart Stepper error limit is not reached (trouble by force overloads)
      1. note that it might be neccesary to recalibrate Smart Steppers if the default error is too high
      2. effector will not "dance" if error value is below ~ 0.90 (tested)
7. Now place the effector to origin again and set all drives to sPID mode. Make some regular movements. Check if ABCD work well. If ABC drives create large forces at higher Z or the lines are just wobbling we need to adjust the radii.

If general machine calibration is bad, Smart Steppers have wrong calibration or tension gets too high due to misconfigured line buildup, the following video shows what can happen. The drives may get load and make some kind of cracking-noise sound and the effector starts to "dance" (it can happen while printing too the same way. It will make holes in your print surface or kick the print off the platform). The system can get overtightened and the servo motors try to fix the error by pulling back the lines.

<iframe width="560" height="315" sandbox="allow-same-origin allow-scripts allow-popups" src="https://videos.stadtfabrikanten.org/videos/embed/b5047066-d5a5-4c0c-b233-116aa743b872" frameborder="0" allowfullscreen></iframe>

Place the feeder

Check if all electronic parts act as exspected

• ceiling module
  • UPS
  • Power Supply
  • Raspberry Pi
  • Duet 2
  • Relay LED spots and stripes
  • Relay Power Supply
  • Webcam
  • Smart Steppers
  • Gyro sensor
  • LED stripe and LED spot
  • Z limit switch
  • Laser pointers
• effector
  • IMU
  • heater cartridge
  • encoder
  • extruder fan
  • filament fan
  • LED stripe
  • z probe

Uninstall the three quick releases

Otherwise the frame will block and cannot be pulled down to maintenance working height. If you forget those you will possibly destroy the threaded rods of the quick releases (tested painfully). So please do this first!

The remaining things

The disassembly works the inverted order it was buildup except that you can skip adjusting steps