Hi, tobben here, I'm taking some notes while reading the article at hardware-x.com

I'm loving the huge amounts of work put into this article and celebrate the idea of adding a recycling extruder. I'm trying to be a bit picky and help make the Hangprinter related paragraphs super accurate, since future builders might read the article by the letter.

Screenshot from 2023-03-02 23-08-07.png

Kinematic equations (1), (2), and (3) matches those that were used in Slideprinter (aka HP0 or 2D-Hangprinter). The equations used in ReprapFirmware today include z, spool buildup, and flex compensation. Slideprinter is useful for explaining the concept.

> Use of timing belts instead of gears reduces backlash in system.

Backlash was actually never a problem given that all axes are always in pre-tension. Belts are good because they move smoother & quieter than gears, and they introduce a weak link that breaks first to prevent any potential of a dangerous situation.

> • (1) Dorisea Spectra line 300 mm spool

This is not the standard HP line. I always get a bit nervous when the exact specified line from the HP BOM is not used since different spectra lines' flexibility vary so greatly, and this impacts performance a lot. Nowadays there's flex compensation in firmware, so one can specify the used line's springiness and get decent performance (except accelerations should be lowered if lines are more flexible than HP standard lines).

> Note: Depending on the nature of the assembly and angle of the lines coming from the ceiling to the floor anchors, a different kind of line roller may be needed (e.g., straight in place of tilted line roller anchors for the B and C anchors). Print more of these as needed.

Sry for this. We luckily have adjustable angle snail houses and adjustable tilted line deflectors now.

> The hangprinter chassis consists of simple wooden beams cut from MDF board

Wow, is MDF really stiff enough? Great if it is. Ah I see you made them thick and then put screws into the beams along the length direction, and got a nut in there to fine tune beam placement along length direction. That's nice. How much did the three MDF beams weigh?

> Furthermore, the end effector of this system is heavier than the filament extruder head used on the original hangprinter v4 design thus making Nema23 stepper motors a better choice than official v4′s BLDC motors for torque.

I don't think this is entirely true. I don't agree with the stepper/BLDC reasoning in the article, but I don't want people to buy ODrive v3's either. When running the 6HC version of Duet3, going for steppers is probably a sound choice, except when encoder data is unavailable you miss most post-2017 improvements of Hangprinter's software. Back when I started HP4 development I designed it with the (then unreleased) Duet 6XD and ODrive S1 boards in mind. We'll get there eventually.

I think there should be an earlier mention of the closed loop control, torque mode, and auto calibration features present in late HP3 (which used steppers) and HP4 (which uses BLDCs) but dropped in HP Hybrid.

Found mention of auto calibration in Future Work section, commenting on it here:

> Another improvement is to implement a computer vision based autocalibration algorithm that detects the various line length components of the A, B, C and D axes and sends those values to the firmware in the DUET3. Such a process has been implemented in hp-mark 106, that works with the latest Hangprinter v4. That approach only works with its closed loop BLDC controllers, the O-drives, and cannot be used with open loop stepper motors.

The auto calibration process is actually independent of hp-mark. The auto calibration lives in its own repo and can work with data points containing:

1. Motor positions (encoder readings, mandatory part of data point)
2. Effector positions (xyz-coords, hp-mark generated or hand measured, optional part of data point)
3. Line tensions (ODrive generated data has been tested but was too noisy to improve results, optional part of data point)

One could feed hp-mark generated data into the auto calibration algorithm, but I currently don't do that, since hp-mark isn't accurate enough. Instead I hand measure four effector positions on the print bed and leave the rest of my data points with only the motor positions part. (See eg this result posted on Twitter.)

The auto calibration process works with any Hangprinter, not just HP4. It was first introduced in 2017. It works with any closed loop motor, not just BLDCs. It's true that it doesn't work with open loop stepper motors.

Anyways, In Fig 42 the belt roller insert is flipped upside down compared to how it is in the original scad files. No worries if it works like that on the Hybrid. Would rub the belt on the original HP4. Just mentioning case it's unintentional.

Lots of things in the article makes me happy. Fig 51 definitely does.

> 2. The lines between the anchor and triangular chassis are perpendicular to the edge of the triangle.

That's not a requirement coming from Hangprinter geometry or the firmware. Have you introduced it in the article somewhere?


> A, B, C anchors are all placed equally the same distance from the triangle (+/-5mm).

This is quoted from "Determining line lengths, anchor placement, print area and heights" and in that context it's correct as an ad hoc requirement to make explanations easier. However, it's not a requirement or assumtion in the firmware or in the Hangprinter geometry. Many builders wrongly assume that ABC anchors must be placed along a circle around the origin so it's helpful to mention that no such requirement exists.

> Hangprinter parameters and calibration

This section explains very well how to manually calibrate a Hangprinter. I'm impressed but my heart also sank a little since it's been my goal for some years that no Hangprinter builders should have to go through this process. Seeing manual calibration re-introduced into HP Hybrid gives memories and mixed feelings. At least your understanding of anchor location parameters are stellar.

But as you say in your own Limitations section:

> The most notable issue from using the system for manufacturing is the need for manual calibration of anchor points and line lengths, which introduces errors in the print accuracy of Hangprinter effector due to errors in measurement.

Much of my work since 2017 has been about fixing this so I lost some opportunity for input on my more recent work when you went for open loop steppers on the Hybrid. With the Duet 6HC steppers superficially made sense though, sending many of us back to 2017. I'm blaming supply chain and board shortages, but both 6XD and S1 boards are now available, giving Hangprinter a lower price, better performance, and way easier setup, without sacrificing those post-2017 HP improvements.

> If movement is off by a lot or there is significant flex in the lines, re perform the calibration procedure with better measurements.

Better to say "slack", "under-tension" or "over-tension" here instead of "flex".

The Hangprinter FAQ mentions manual calibration. Any reason why it (or any of the links it refers to) made it into the list called Troubleshooting steps and helpful resources?

> The system also has no flex compensation in current firmware implementation so the end effector loses tension as it moves away from the print center, which reduces the accuracy at the edges of the print area as shown in the validation data.

This must have been written before flex compensation was announced.

Thanks for a good read. It's impressive how deep into the subject matter you have gotten.

Cookies help us deliver our services. By using our services, you agree to our use of cookies.