Mini Kossel - WORKING! (changed from Kill Me Now)

Eddiie

A couple of suggestions:

1. Have you checked that for each pair of rods, the spacing between the bearings on the carriage is the same at the the spacing between the bearings on the effector?

2. Does the tilt still happen if you remove the Bowden tubes?

–-------------------------

1. No, I have not checked. And this might be related. So, the rods should not only be the same length (they are), they should be straight as possible between the carriage and the effector?

2. Bowden tubes are not yet connected.

Eddiie

@DjDemonD:

Hi Eddie, hang in there, I have a mini kossel and whilst these machines can cause immense frustration you've seen that when they're working they are wonderful.

Try your single nozzle effector and test again, if its much more consistent then its a problem with the other effector. As David points out usually spacing between the rods (add/remove washers).

Have you considered a z-probe? Whilst they have their complications and limitations the time spent setting one up and ensuring its reading correctly is less than the time spent doing manual calibrations over and over, which you have to do if you are changing and developing your machine.

I've been inclined to go away from z probes for machines which are basically stable and just churning out parts, but with machines in active development/upgrading anything that can cut down on the time taken to do the necessary tasks is well worth considering.

–--------------------------------

Thank you for the encouragement. Really need it.
I am going to investigate the spacing.

I do have a IR z-probe but the tilt of the effector makes it calibrate incorrectly.

Cheers, the spacing gives me some hope.

DjDemonD

My tip with the IR probe, is try to mount it as close to the nozzle as possible. I will add that to do this you need a well insulated heater block (silicone covers now available from e3d). Its also worth insulating the IR sensor with kapton tape. This reduces the magnitude of effector tilt measured by the IR probe, making it more useful.

Of course sorting out the effector tilt is also important, but presuming you've done everything to prevent this (search Nebbian's posts on reprap forum, hes obsessive about it), if it is still too great for the IR probe then calibrating manually is one option, trying some sort of nozzle based probe is another (FSRs/Piezo/nozzle to metal bed electrical signalling).

dc42

If you do resort to manual calibration, you can use the calculator at http://www.escher3d.com/pages/wizards/wizarddelta.php which implements the same algorithm as the firmware.

Eddiie

Update…

I've found a new diamond hotend for mini kossel effetor and have been working with the person on tweaks. He has been very responsive and is a great help. Here is the link to the thing: http://www.thingiverse.com/thing:1844678/#comments

The rods are now parallel to the hotend and carriages, things work much better now, but there still seems to be a tilt somewhere.

Still working on the IR probe mounts, I made a working effector and probe that seems to work but there seems to be deviations on the measurements it is getting. For example the sensor/probe is mounted on the y/Z plane the close the sensor gets to the X plane, the pressure on the print bed is reduced. Like there is a 2mm lift of some sort. How do I fix this?

Also, how do I account for the space/measurement of the effector? (the distance between hotend and rods)

Same when doing a manual calibration.

Eddiie

Also, does G28 (home) wipe out auto calibration data?

dc42

The distance between the hot end and the rod bearings doesn't affect auto calibration. But a large effector reduces print area, by preventing the nozzle from getting close to the towers.

G28 does not wipe out auto calibration data. However, if the homing switches do not trigger at exactly the same position each time, then re-calibrating after G28 to get the new endstop corrections may be worthwhile.

Eddiie

Update:

I find that my endstops were not screwed down from when I was reprinting frame pieces. hehe
Fastening them to the chassis so they do not move has greatly improved calibration. Go figure.

Besides that, I found a crack in one of the vertical carriages. This led to printing 3 new one's and has improved the stability of the hotend.

I am working with the creator of the shroud to reduce the amount of surface area on the rod bearing mounts (for lack of a better term from my vocabulary). Less surface area will give a greater degree of movement freedom for the rods to move, thus increasing the print radius and strain on the rods - the strain causes the hotend to twist a bit as it reaches the vertical towers. It also put strain on the carriages, causing it to crack one of the old one's.

Tonight I hope to actually print!!

DC42, I'd like to include a picture of the shroud with the IR sensor to get your feedback on the height.
Diamondhead.

Eddiie

On to next topic, extruder calibration…

I read somewhere that if you extrude 10mm of filament, you should get 10mm of filament. So I have made the necessary tweaks to make sure xxmm of filament is xxmm of filament that is extruded from the bowden tube. The article was not clear if they were referring to xxmm of filament out the other side of the extruder (bowden tube) or out of the hotend. They are 2 completely different things.

10mm of filament @ 40mm is too fast for the hotend to extrude and the filament jams / or stops moving.

The default settings did not seem to extrude enough filament in testing. The new settings to extrude 10mm of filament out the bowden tube seem very large.

Here they are.

Default

M92 X100 Y100 Z100				; Set axis steps/mm
M92 E98:98:98:98:98:98 	           ; Set extruder steps per mm explicitly for all the extruders - 6 for Duet085+Dt4 test (663)

Modified

M92 X100 Y100 Z100 E150:149:151:200:200:200:200

I have 3 extruders all are of the same type.

Manufacturer's settings taken from Marlin for the extruder are:
configuration.h

DEFAULT_AXIS_STEPS_PER_UNIT = 150 
DEFAULT_MAX_FEEDRATE = 200
DEFAULT_MAX_ACCELERATION = 9000
DEFAULT_RETRACT_ACCELERATION = 3000

configuration_adv.h

MANUAL_FEEDRATE = 60

T3P3Tony

HI Eddie

You are right, when calibrating your extruder steps you want to get to the point that commanding 10mm causes the extruder to push 10mm of filament through. You are measuring filament in not out!

Generally I start with 20 or 30mm to get in the right ball park and the fine tune at 50 and then 100mm. that way if you are 0.5mm out at the end, you are 0.5% in error which is small enough in my experience.

Depending on the type of extruder, espcially the gearing and hobbing diameter you can get widely different steps/mm for extruders. ~650 is what we use on the RepRapPro style geared extruders, ~420 on a Titan…. it varies.

DjDemonD

Whilst this is absolutely the way it should happen, its worth being aware that a fair number of us especially those with titan extruders have found, and the reason is still unknown, that we need to use far fewer steps/mm than we ought to. What I'm basically saying is that just because you push in 10mm of filament when 10mm are requested doesn't mean the objects you print will have optimal extrusion. Maybe its discrepancy in nozzle size or something like that.

The 10mm in for 10mm commanded is a good starting point. What I'd do after that is print an object which can show you both under extrusion and over extrusion. I designed such an object http://www.thingiverse.com/thing:1619976 the idea is you print two cubes, one with infill 15% the other 95%. As you print it you watch to see if the 15% one has stringy infill, or the 95% one has no tiny gaps between the infill lines. Use the flow rate modifier (or filament modifier) on paneldue to adjust up or down (or you can use M92 Exxx) to find the optimum between these two points. Don't start looking until layer 6 or so, as up until that point its heavily dependant on whether your first layer is squashed down or not. This fine tunes your extrusion.

In reality with time, you can tell just by looking at an object printing whether it needs +1% or -1% etc… but it takes a little while to tune into that. I sometimes go -1% flow during the first few layers and then back up to normal after that to compensate for first layer squash.

T3P3Tony

DjD, having followed Ian's long thread on this I don't disagree that extrusion modifications is needed in some cases. I think that this modification should be done once the steps/mm is set right, by setting the flow rate modifier rather than messing with the steps/mm however both have the same ultimate effect

deckingman

I'm with Tony on this one. Once set to extrude the right amount of filament, the steps/mm should be constant and therefore treated as one. There are just too many other variables to throw another one into the pot. One thing I managed to glean from all my testing was that where an object being printed appears to exhibit signs of over extrusion, it isn't because the steps per mm are incorrect but for some other reason. So, IMO the best way to compensate for that is through the use of extrusion multiplier as this can easily be changed from print to print or even "on the fly" whereas changing steps per mm is more of a global setting and less easy to do. Ultimately of course, both will achieve the same result.

T3P3Tony

P.S. Eddie I am looking forward to the day you say we can change the name of this thread to something more positive!

Eddiie

Thank you all for the support.

Last night I did a test print of the popular 3dbenchy boat. At least it printed without the head crashing into the bed for the first time. The new problem is definitely filament extrusion.

The print job also stopped in the middle, like it was finished, head homed and all, but it was only about 50% complete.

MY Slic3r is setup to use the colormixing from the Repetier firmware, it has some M codes that are in the startup of every file to set the color. These are reported as errors in Reprap firmware.

The codes are -
; Cyan Extruder0
M163 S0 P1
M163 S1 P0
M163 S2 P0
M164 S0

How do I change extruders (colors)?
Sending a print job starts immediately, it does not wait for the hotend to reach proper temperature.
I think it only waits for the bed to reach temp.

deckingman

@ Eddie. In conjunction with Tony, I've been working on a document for eventual inclusion in the Wiki and/or a blog. It's a bit long and detailed and still only a draft but it'll explain what you need to do with regard to settings for the Diamond hot end. If you PM me with an email address, I'll send you a reply with the document attached.

Ian

Eddiie

Here are some pictures -

Benchy Test Print -

My Kossel -

Eddiie

DC42, here are some pictures of the IR probe, it seems to me there might be a better place/angle to mount it? Yes, I know you urge to mount with the electronics facing away from the hotend but I think in this case it is OK due to the angle of the head. I believe the tilt is caused by the rods colliding with the mounts on the shroud. I have a new one printed to test tonight that has less surface area for the rods to hit.

Eddiie

for some reason the pictures are being rotated..

dc42

Hi Eddie,

1. Your print may have stopped in the middle because you uploaded it to the Duet before slic3r had finished writing it, so you uploaded an incomplete file, Check whether the file size on SD card is the same as on your PC.

2. To set the colour mixing, see M567 and M568 in the gcode wiki at http://reprap.org/wiki/G-code.

3. If it starts printing before the hot end has finished heating up, that usually means that you didn't have a tool selected before you started heating.

HTH David