Delta Printer bad prints

Hey!

I have build a delta printer (picture) but have quite some issues with the bowden extruder. It includes a z probe to do auto calibration. That seems to work.

I have some experience with a normal "cartesian" printer, but I do not have experience with the issues I am now having with printing with my delta printer.

When I print a cylinder, my first layer looks like this:



Obviously this is not what I want :-) What I find odd is that

• The lines are not touching eachother (evidently)
• The lines seem not to be parallel
• The lines do not touch the outside circle...

Also, I am having issues with the PLA not sticking to the bed. However, as it's not a heated bed (yet), I can understand that.

Lastly, when I move to the next level, it seems that the extruder hits the extruded filament. Should I increase the layer height in the slicer software, or is this working on the symptoms rather than working on the actual cause of the problems?

Thanks a lot for all your insights!
Steph

It sounds to me that when the firmware thinks the printer is at Z=0, the nozzle is actually a little way off the bed.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hey dc42!

thanks for your advice!! I did another calibration which took the better part of this evening :-) and now it seems to already be much improved, although still not where I want to be.



It is really interesting to see the difference in the calibration! These are the parameters of the first calibration:

21:52:52.973 : echo: Endstop adjustment (mm):
21:52:52.974 : echo:  M666 X-3.94 Y-2.24 Z-4.65
21:52:52.976 : echo: Delta Geometry adjustment:
21:52:52.980 : echo:  M666 A-1.77 B-1.17 C2.95 E-0.50 F0.00 G0.00 R89.25 D202.50 H286.82 P0.00
21:52:52.982 : echo: PID settings:
21:52:52.983 : echo:   M301 P22.20 I1.08 D114.00

and these are the ones of the second calibration:

23:18:44.914 : echo: Endstop adjustment (mm):
23:18:44.915 : echo:  M666 X-4.88 Y-2.92 Z-4.09
23:18:44.917 : echo: Delta Geometry adjustment:
23:18:44.921 : echo:  M666 A-1.60 B-0.11 C1.71 E-0.50 F-0.50 G0.00 R90.37 D207.50 H285.50 P0.00
23:18:44.922 : echo: PID settings:
23:18:44.925 : echo:   M301 P22.20 I1.08 D114.00

I must say that I tweaked the firmware to only use 5 points to probe, as it would otherwise take waaaaaaay too long :-) I will however try when I have a bit longer the more precise method to see if that helps...

However, looking at the above parameters, and given that I did not change anything real to the machine, I wonder if this calibration is not a bit too "at random". It seems like it will change the D and R parameters just indefinitely until it happens to find some match with the other parameters...

Also, it still seems that the nozzle hits the layer below, even though I have increased the layer height in Slicer to 0.3 instead of 0.25. Could this also have something to do with the above settings, or is this normal behaviour?

Thanks!

BTW: the H parameter I manually set before to 285.5. I had now acurately measured the position of the probe, so this parameter now seems to be set correctly...

I think you may have the print speed too high and/or the extrusion width set incorrectly. What size is the nozzle on your machine? How did you generate the calibration piece, and what slicing parameters did you use?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hey!

I think the nozzle is 0.4mm (I forgot to specify it when I bought it, and the guy sent me a random one)... The extruded thread is ~0.7mm thick (diameter).

The speed is not too high (I think) for the first layer, but indeed may be a bit too high for the next layers. I can however try to print at a lower speed for the first layer to see what happens.

I kept the default Slicer parameters for the extrusion width:

Default width : 0 (auto)
First layer : 200%

Other slicing parameters that I can see in the Slicer software are:

layer height : 0.3mm
layer 1 height : 0.35mm
nozzle : 0.35mm

What do you mean with calibration piece? I used the guide at this location to calibrate manually. However, now I use the automatic calibration included in the Marlin firmware. As a first trial, I tried to print a cylinder of 50 diameter and 30 mm height.

Steph

What do you have the filament diameter set for in Slic3r? Does having Slic3r believing it is using 3.00mm filament when it is really printing 1.75mm filament cause this type of under-extrusion problem? Did you calibrate the steps/mm on your extruder by measuring how far the filament moves?????

Hi!

The setting in Slic3r is set to 1.75mm, the filament is also 1.75mm. I will try to set it to 3mm to see the difference this evening after work :-)

The calibration of the motors (using the Z axes movement to set the steps) and the Bowden extruder (measuring the filament when transporting it) has been done. Just to be sure, I will do it again this evening as well.

Hi!

I think I found one reason why things go strange over here... Thanks for vreihen, I double checked the extrusion speed of the Bowden Extruder. It appears that if I extrude 100mm filament, I actually move 118mm.

So I thought simply to change the parameter:

#define DEFAULT_AXIS_STEPS_PER_UNIT   {0.500445674*160, 0.500445674*160, 0.500445674*160, 0.251256281*439.5}

Now, for some reason, whatever I input in this parameter, it does not change the extruded length.

I thought this was simply a matter of changing this parameter, and uploading the firmware, but obviously it isn't? Anyone an idea what I am doing wrong?

Thanks!

You are changing the E parameter (0.251256281*439.5) correct?

I am not familiar with the need for the multiplier in the code. I understand multiplying the value by 60 to go from mm/s to mm/min, but not the fractions being used in your sample code.

It is possible that even after changing the firmware, you are pulling values from the EEPROM. Double check your values, reload the firmware, and then try an M502 to reload the firmware settings and then M500 to save them to the EEPROM.

Hey Kurzaa!

Indeed! Thanks! In my queeste to fix my delta printer, I indeed stored the parameters for the Delta Geometry in EEPROM. I didn't notice that it would also load the E parameter...

The correct parameter would now be:

E = 93.58231821

That is based on measurements of the filament.

Now up to testing!

Thanks again!

Finally :-)

It seems like we're coming somewhere!

Now, I have been forced to change the E parameter again (M92). For some reason, while the parameter I set before resulted in the correct movement of the filament, the E parameter had to be increased with about 1/3 to give an acceptable result in printing.

Now I have set it as such, I get the following results:


This already looks acceptable :-)

I still have the problem though that once the first layer is finished, the hot end travels fast over this layer, and hits it quite violantly (the whole effector vibrates). I tried setting the layer height to 0.5 even in Slic3r, but that didn't change a thing. (the layers weren't even higher it seemed)... For the other layers, the problems seems not to exist...

Well, I will continue my tests :-)

Is the hotend hitting ridges left in the first layer of the filament or just kind of dragging along the top of the extruded filament?

I had the ridge issue myself before I gave up and took a break. I think it is caused by the nozzle being to close to the bed, so the filament overflows and forms a valley whose edges are higher than the layer height. Occasionally, if I could just manage to print the second layer, the issue resolved itself and finished fine.

Now that I am working on my printer again however, I have had the nozzle dragging along the top of the extruded filament though on one or two tests. In that case, I believe the issue was caused by the nozzle not lifting during retracts. As there is no dedicated motor for the Z height like there is on cartisian printers, I believe it is always beneficial to enable Z lift in your slicing software for delta style printers. Hopefully someone else can confirm that however.

Can it be that there is some kind of buffer issue if you print too fast? It seems that my printer can follow quite well, but still it blocks every once in a while (= a few times per layer) for about 100-400msec or so. It seems to wait for a commands from the PC.

Maybe something to do with the baudrate (set to 250000)?

It's not really a problem, but now it seems like it's not really printing smoothly...

Edited 1 time(s). Last edit at 02/17/2015 03:02PM by Stephanel.

Quote
Kurzaa
Is the hotend hitting ridges left in the first layer of the filament or just kind of dragging along the top of the extruded filament?

I had the ridge issue myself before I gave up and took a break. I think it is caused by the nozzle being to close to the bed, so the filament overflows and forms a valley whose edges are higher than the layer height. Occasionally, if I could just manage to print the second layer, the issue resolved itself and finished fine.

Now that I am working on my printer again however, I have had the nozzle dragging along the top of the extruded filament though on one or two tests. In that case, I believe the issue was caused by the nozzle not lifting during retracts. As there is no dedicated motor for the Z height like there is on cartisian printers, I believe it is always beneficial to enable Z lift in your slicing software for delta style printers. Hopefully someone else can confirm that however.

Hey Kurzaa,

I also didn't have this issue with my previous printer. Either the printer was much much slower than this delta printer (and if the nozzle would hit the plastic, it would simply melt it and move on), or the nozzle indeed retracted a bit (or both). Not sure though how to setup the Slic3r to ensure that it retracts the nozzle if it moves across the product. As you say, it might not be practical, given that there's no dedicated Z axis motor...

I must say, the Delta Printer design is a challenge. Less so mechanically than software (calibration) :-) Glad that it now starts to work, now printing my first reach test object!

Quote
Stephanel
Can it be that there is some kind of buffer issue if you print too fast? It seems that my printer can follow quite well, but still it blocks every once in a while (= a few times per layer) for about 100-400msec or so. It seems to wait for a commands from the PC.

Maybe something to do with the baudrate (set to 250000)?

Do you have the option to print from an SD card? Do you have a GLCD connected? Unless you can see the USB buffering stats, I'm inclined to think that you are maxing out an 8-bit controller with delta calculations. There's a reason why delta printer owners are leading the charge to 32-bit electronics, like Smoothie, Duet, RADDS, etc.....

Aah :-) Now I start to understand the difference between the RAMPS and Smoothie, Duet etc :-)

Do you have any preference as to which to try out next? I have only knowledge of the RAMPS board, but didn't try out the others yet. They also seem to be significantly more expensive than the RAMPS board...

I do have an option to add an SD card. I didn't yet try to do it this way, as I first focused on getting the printer to work :-) Will try that this evening! If this increases the max speed, I will make it that all G code is uploaded to the SD card first and that then the print starts. This way, I hope to (partly) circumvent the limitations of the RAMPS/Arduino board :-)

Quote
Stephanel
Aah :-) Now I start to understand the difference between the RAMPS and Smoothie, Duet etc :-)

Do you have any preference as to which to try out next? I have only knowledge of the RAMPS board, but didn't try out the others yet. They also seem to be significantly more expensive than the RAMPS board...

I do have an option to add an SD card. I didn't yet try to do it this way, as I first focused on getting the printer to work :-) Will try that this evening! If this increases the max speed, I will make it that all G code is uploaded to the SD card first and that then the print starts. This way, I hope to (partly) circumvent the limitations of the RAMPS/Arduino board :-)

Stephanel

I have both boards and to be honest they are much of a muchness (OK Firmware is different but) the smoothie is easier to update the firmware (But the Duet isn't hard in fact it is easier than a Ramps set up) The duet has a simpler Config File but both are easy.
Both have digital pots for current setting both have a web interface though at this moment the DUET one is far superior The Duet also has FTP Access to the SD CARD (Which is where the configs are held for both DUET and Smoothie for that matter).

The duet has 4 drivers on the main board with an extension board available for an additional 4 if required the Smoothie is up to 5 maximum.

The Smoothie does however have breakout ports for its 5 drivers to allow for external drivers if you need high current ones or in the unlikely event that you blow one of the onboard ones.

The smoothie I believe will take a std type display as the ramps will The Duet does have the capability of running the PanelDue by DC42 which is extremely nice but not ultra cheap (This has its own 32 bit controller and is not just a semi dumb display)

as for costs weill I don't think that either of them are significantly more costly than a good Quality Arduino / Ramps Board / 5 Pololu driver modules (I am not talking about Cheap Chinese copies here).

HTH just a little bit (or confuse the issue lol)

by the way where are you located?

Quote
dougal1957
The Duet also has FTP Access to the SD CARD (Which is where the configs are held for both DUET and Smoothie for that matter).

FTP?!? How 1980's!!! One of Smoothieware's great features is that it presents the on-board SD card as a removable drive over the printer's USB cable. No ethernet connection required.

Regarding the differences, Smoothie's delta support has been well tested, and used for over a year by many of the people over in the Google Deltabot forum. Delta support for the DUET board is less than a month old, and was developed by forum user dc42. (Not a criticism of his work which I applaud, just a caution that it has not been tested as well as Smoothie at this point in time.) The DUET is a little bit cheaper at the moment, and supposedly has a much cleaner web interface than Smoothie does if you hook it up to ethernet. Either board is a good choice, and will most certainly perform better than an 8-bit RAMPS.

My personal choice is an Azteeg X5 Mini ($109) running Smoothieware, although I do have a commitment to buy a genuine Smoothieboard on the Deltabot forum's group buy.....

The Duet delta support may be relatively recent, but unlike Smoothie and Marlin, it does not generate approximate delta motion by segmentation. Instead it generates precise delta motion (to within a few microseconds accuracy). File uploading is normally done via the web interface, but FTP is also supported so as to make complete control of the SD card file system available via Ethernet.

The Azteeg board costs less than the Duet and the Smoothieboard, but lacks the Ethernet port. Personally, I wouldn't want to be without the web interface, and I use the USB port only for uploading new firmware and debugging. But if you are looking for a budget 32-bit board, there are few others available at present.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hey guys!

Thanks for all the response! It seems like these boards are indeed quite good, and I already noticed the excellent LCD screen of dc42. Sadly, living in Belgium, it is hard to come by 3D printer stuff, and as such I can only buy this online in the US or other countries. I have had some bad experiences with customs who put a hell of a lot of import taxes on products I bought online, so I will have to look into that :-)

I will this evening try out the SD card and see if that already helps a bit. If it does, I will try to upload the G code wirelessly, so I can at least continue until I find a suitable replacement for my board.

Quote
Stephanel
Hey guys!

Thanks for all the response! It seems like these boards are indeed quite good, and I already noticed the excellent LCD screen of dc42. Sadly, living in Belgium, it is hard to come by 3D printer stuff, and as such I can only buy this online in the US or other countries. I have had some bad experiences with customs who put a hell of a lot of import taxes on products I bought online, so I will have to look into that :-)

I will this evening try out the SD card and see if that already helps a bit. If it does, I will try to upload the G code wirelessly, so I can at least continue until I find a suitable replacement for my board.

Well you are in luck the smoothie board is sold from france By RobotSeed and the Duet in the UK Duet boards and the PanelDue also by Think3dPrint3D so you will have no customs etc to pay at all. just the carriage.

Doug

Ah I thought import taxes were always if it came abroad :-) Must then only be from outside the EU...

Quote
Stephanel
Ah I thought import taxes were always if it came abroad :-) Must then only be from outside the EU...

Yes as part of the EU it is free trade between member states so no import duties between the member countries of the EU I think a single extruder Smoothie is about Eur 110 and the Duet would be about Eur100 the Smoothie is supplied with a SD Card and all the needed connectors (but not the onboard 5V regulator which is an option at around Eur 4 IIRC well worth having) the Duet depending where it comes from will have Screw terminal's (Or if you can get an older one Headers) the header version would need to come from RepRapPro but I see they are out of stock so the screw terminal one is probably the only option now you also need to provide your own SD Card (Well I did don't know if tony (T3P3) includes one) but they are not a lot of money anyway. They both use Micro SD Cards.

Doug

Quote
Stephanel
Ah I thought import taxes were always if it came abroad :-) Must then only be from outside the EU...

In the EU, the import rules are as follows:

1. As a consumer, you can buy physical goods from any other EU member state at the VAT-inclusive price of that state, and then pay no import duty or import VAT. So if you buy a Duet from the UK, you will either pay the T3P3 price of GBP78 including VAT, or the RepRapPro "sale" price of GBP69 + 20% UK VAT. Likewise if you buy a Smoothieboard from an EU supplier, you will pay VAT at the rate in the supplier's country.

2. A VAT-registered (and VIES-registered in the case of Italy) EU business can purchase goods VAT-free from a supplier in another EU member state, but must account for input VAT at his own state's rate.

3. If you import goods from outside the EU, then normally you will have to pay import VAT at your own country's VAT rate. Your country's postal service may also charge you a fee for collecting the VAT. In some countries, import VAT is waived for low-value (up to GBP15 in the UK) goods imported for personal use.

If you do decide on a Duet, you may find my blog entry at [miscsolutions.wordpress.com] and the following 3 entries useful. They cover converting a delta from Arduino/RAMPS to Duet electronics.

Edited 4 time(s). Last edit at 02/18/2015 12:53PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hey!

Thanks for that explanation! My bad experiences indeed were more with products coming from the US or China than with EU. Seems like the Smoothieboard is out of stock at Robotseed. The Duet board seems good as well, and indeed, it does have the option for the LCD screen which makes it interesting :-)

It appears I have lost my SD card module, so testing that will not be possible today :-( Anyways, I can continue with optimizing the printer and learning the particularities of Delta printers :-)

I so have ordered the Duet Printer Board, but sadly, I will only be able to work on it in 2 weeks (will be delivered at parents home)... Hope it won't require a huge learning curve again as it did with the calibration of the RAMPS :-)

Quote
Stephanel
I so have ordered the Duet Printer Board, but sadly, I will only be able to work on it in 2 weeks (will be delivered at parents home)... Hope it won't require a huge learning curve again as it did with the calibration of the RAMPS :-)

Does your printer have a Z probe, and if so, what sort? If it does, and the type of probe is already supported by the firmware or I can easily add it for you, then you need only get the endstop switches approximately even and set the homed height slightly higher than actual. The auto-calibration facility in the firmware will then sort out the endstop corrections, delta radius, and homed height.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I continued my tests and am back to an old issue: when the hot end moves across the print, it appears to slightly hit it, damaging the print, and sometimes even cause that much vibration that the next prints are off by a few millimeter.

I did not have this issue with a cartesian printer, although it did hit the printed product, but it simply melted its way through is. As this Delta printer is significantly faster, I think the hot end simply does not have time to melt through the material.

I assume this is not a new issue. Is there a solution to this? Eg, forcing the print head to move up by ~0.5mm or so when it moves to a different location?

I tried decreasing the M666 Hxxx value by a few tens of millimeter, but this significantly worsened the print. Decreasing the print speed is an option, but why then move to a delta design in the first place?...

As you can see in the picture below, the print head has hit a few of the printed lines and simply tries to rip them apart, risking loosening the print and/or causing vibrations in the effector.



As a side mark, you can see that the quality of the print is still not fantastic. Especially to the edges, the inner fill does not always properly touch the circle. I think this still has to do with the imperfect calibration of the delta printer...

Edited 1 time(s). Last edit at 02/18/2015 03:19PM by Stephanel.

Quote
dc42
Does your printer have a Z probe, and if so, what sort? If it does, and the type of probe is already supported by the firmware or I can easily add it for you, then you need only get the endstop switches approximately even and set the homed height slightly higher than actual. The auto-calibration facility in the firmware will then sort out the endstop corrections, delta radius, and homed height.

Thanks! :-) I use a very simple Z probe Just a switch mounted to my effector.



It seems to work quite well, although I will most likely move to another solution which is less bulky. I intend on using a heated bed though, so apparently that limits the choices of Z probes... Will have to look into that soon...

In the end, I used the Z probe to do a first calibration, and finalized this manually using the paper trick.

I don't recognize that Z probe. Does it work anywhere on the bed? How do you deploy it?

The only limitation I am aware of with heated beds is that force sensitive resistors mounted under the bed supports need to be shielded from the heat, and may be less sensitive due to the extra weight of the heated bed.

Regarding the print hitting the bed, are you sure you have the delta radius calibrated correctly, so that the printing plane is flat? If it is slightly concave then that may contribute to the problem.

I was considering adding software auto-lift during non-printing moves to the Duet firmware, but someone pointed out that some slicing programs can already do this.

Edited 1 time(s). Last edit at 02/18/2015 04:01PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].