Mhackney's Rostock Max

[mhackney]

for rev 1 it is 8 for rev 1.1 like mine, it is 16 micro steps as I showed in the previous post.

[mhackney]

These delta printers are a LOT easier to calibrate in Z and level than a cartesian printer.

First I set the Z height at 0,0,0. I moved the head using G0 X0 Y0 Z10 to make sure I didn't crash into the glass. Then with a sheet of paper (I use cigarette rolling paper - an old machinist trick, it is about .001" thick) under the nozzle I click down 1 mm at a time. In my case, I hit Z=0 before the nozzle touch the paper. So, I increased my X, Y and Z max lengths to 370 mm (a guess but I wanted them to be longer than the actual tower travel) in the Repetier firmware's configuration.h

#define X_MAX_LENGTH 370.0
#define Y_MAX_LENGTH 370.0
#define Z_MAX_LENGTH 370.0

Then I ran down to z=10mm again and clicked down 1 mm at a time until I was within a mm and switched to .1mm clicks. As I moved the paper back and forth under the nozzle, I clicked the .1mm button until the nozzle just touched the paper - you can feel it snag slightly. I looked at the Z display and it read 5.4mm. This means that the printer thinks the nozzle is 5.4 mm above the table glass but it is actually at 0. So, I simply subtract 5.4 mm from 370 in the config file:

#define X_MAX_LENGTH 370.0 - 5.4
#define Y_MAX_LENGTH 370.0 - 5.4
#define Z_MAX_LENGTH 370.0 - 5.4

This gives me Z=0 at the center of the table. I homed and went to Z=0 (G0 Z0) several times and checked with the paper to make sure it was correct and repeatable.

Next, I used the technique described at http://minow.blogspot.com.au to set the Z=0 at the X, Y and Z tower locations. I used the coordinates/tower points Gene lists in the manual.

Z Tower – X0 Y90
Y Tower – X63 Y-30
X Tower – X-78 Y-48

Starting with the Z tower, I issued the gcode: G0 X0 Y90 Z10 to make sure I stopped above the glass. Then I used the 1mm and .1mm buttons to lower the head until I reached 0 or touched the paper. You set Z=0 at the tower positions using the stop screw at the home switch. If the nozzle is above the glass when the firmware thinks it should be Z=0, then you need to screw the stop screw IN (clockwise from the top, the same motion as tightening a screw). If it reaches Z=0 but the firmware thinks it as above the table, you screw the stop screw OUT. This takes a little bit of practice. Start with 1 full turn and see what happens. After each stop screw adjustment, you Home and then issue the gcode to go to the tower location. Test and repeat the adjustment until the paper just snags the nozzle.

Do this with the Y and Z towers too. Once complete, test going to X=Y=Z=0 to make sure you are still at zero there and also retest the three tower location "0s". Mine was right on after one adjustment cycle.

[mhackney]

The Repetier firmware has an auto tuning command for determining the values (Kp, Ki and Kd) for the PID algorithm. The PID algorithm (proportional integral derivative) is a control mechanism with a feedback look that can manage the heating elements very effectively.

In Repetier, the M303 command starts the auto tune. Here is the command syntax:

- M303 P<extruder/bed> S<drucktermeratur> Autodetect pid values. Use P<NUM_EXTRUDER> for heated bed.

If P is left off, P=0 is assumed - which is normally the hot end. S is the target temperature and should be the temp you would normally want to maintain. So for ABS, 200°C on the hot end is a good target. For the heated bed, 100°C is good. Make sure the device is at room temperature before running the auto tuner.

I ran: M303 S200

After running the auto tune on my hot end, the last set of K values were used to replace the defaults in configuration.h:

/** P-gain. Overridden if EEPROM activated. */
//#define EXT0_PID_P 24
#define EXT0_PID_P 14.91
/** I-gain. Overridden if EEPROM activated.*/
//#define EXT0_PID_I 0.88
#define EXT0_PID_I 0.59
/** Dgain. Overridden if EEPROM activated.*/
//#define EXT0_PID_D 80
#define EXT0_PID_D 93.85

(when I change values in configuration.h I like to comment out the original default and create a new line for my value. That way I can always go back to a known state).

To calibrate the PID for the heated bed, I ran: M303 P1 S100

I also realized that my extruder steps per mm were off by a factor of about 2 (againd due to 16 microsteps for the version 1.1 RAMBo.) So I changed it in the configuration.h:

//#define EXT0_STEPS_PER_MM 292
#define EXT0_STEPS_PER_MM 584

[Polygonhell]

Next, I used the technique described at http://minow.blogspot.com.au to set the Z=0 at the X, Y and Z tower locations. I used the coordinates/tower points Gene lists in the manual.

Z Tower – X0 Y90
Y Tower – X63 Y-30
X Tower – X-78 Y-48

Nit picking here, but ideally you want these 3 points to be in a circle around the center of the platform and these 3 are not.
The reason is because if they are in a circle, it removes any contribution from convex or concave movement that you fix later with the changes to DELTA_RADIUS.
In practice the error is small, but since not introducing it is as simple as picking 3 points in a circle, why introduce it, the points suggested in the link you attached are in a circle, and any linear scaling of them would also be.

[mhackney]

Hey Polygonhell, You are correct. I realized that when I did the math after I posted. They are in the ballpark but not exact. The points published on the minow site I referenced are indeed 120° apart at the same radius (90mm) centered on the center of the bed.

G0 X-77.94 Y-45
G0 X77.94 Y-45
G0 X0 Y90

I rechecked Z=0 at these positions and I was still spot on. Not surprising since the original check was pretty close.

Thanks for pointing that out, I wasn't planning on posting this update as the recheck came out the same.

cheers,
Michael

[mhackney]

Sometime today: http://www.repetier.com/download/

Now supports cylindrical build volumes! Plus bug fixes.

[geneb]

The points that I chose were used because I've got a square bed.

g.

[mhackney]

Thanks Gene. As I pointed out, I did not see any difference in calibration when I used the "on the circle" points. In theory, using a points on the circle at the base of the towers gives the best results, in practice, somewhere in the vicinity is probably good enough!

Cheers,
Michael

[mhackney]

I've posted in a few other areas about my extrusion problems but I want to capture it here in my thread.

I am printing with ABS. When I attempt to print, I am getting very little filament extruding. I've gone through the usual debugging process (I have 2 other printers with Steve's extruder and earlier hot ends) and nothing is clicking yet. Basically, it feels as though the filament just gets jammed in the hot end and won't push out. I have checked my temps with a thermocouple. Here is the first weirdness:

I have my hot end temp set to 235°C. When I measure the temp at one of the little openings over one of the resistors it is dead on at 235°C. However, when I measure at the nozzle-hot end junction where you normally read the temp, it is 198°C. I've checked this multiple times and it is consistent. On my H-1.1 with the earlier nozzle design, the temp measured at this point is very close to the set temp in the host.

If I disconnect the Bowden tube from the extruder and try to manually push the filament I feel very little resistance until I reach the nozzle (approximately). At this point it requires a lot of force to get an extrusion. I have checked the teflon tube in the hotend and there is no binding in it cold.

I can adjust the extruder rollers to get good feed and can not stop the filament from moving by grabbing it and pulling. But, when I try to actually extrude, the filament will move a bit, some will extrude and then stall. The rollers are moving but the filament is slipping. Tightening up the adjustment screws does not improve traction but it does cause the filament to distort/squish into an oval.

I just noticed this morning that when I start to extrude 10mm at 60mm/min it takes - consistently - 2 to 4 seconds before any filament starts extruding from the nozzle and then after the extruder stepper stops, it continues to extrude for 4 seconds or so. One interesting observation - if I extrude a filament into the air (so it dangles straight down), wait the 4 seconds for it to stop growing, let it cool and then measure it, I can calculate the volume of plastic. I also know the diameter and extruded length of the source filament. The volume in is very close to the volume out, so I know things are reasonably calibrated. Example:

Source Filament: 10mm of 1.70mm filament: volume = 22.70 cc

Extruded Filament: 85mm of .56mm diameter: volume = 20.94 cc

I have a .50mm nozzle so this seems to make sense.

I have tried increasing the temperature to 245°C but didn't notice any significant difference. I am concerned about increasing too high and causing issues with the teflon and/or ABS.

At this point, my hypothesis is that "something" is restricting the flow of either molten or solid filament in the lower part of the hot end. The nozzle orifice seems to be clear and smooth. However, it is much longer than the earlier nozzles.

[Highcooley]

I can only comment on the delay of 2-4 seconds between extruding and filament actually flowing out of the nozzle. This is due to the pressure which has to build up through the entire bowden tube. Therefore, this is normal behavior. Usually, you can see the flow speed build up until constant flow during maybe about 1-1.5 seconds. When the extruder stops, the flow speed decreases again for about the same duration. I usually do tests @ 100 mm/min, so I don't know, if it also takes 1-1.5 seconds @ 60 mm/min or if the duration is a bit longer.

If you hit extrude just in the moment, when the pressure starts decreasing, it should build up instantly.
However, if you press extrude right after the flow stopped, it should definitely take far less than 2-4 seconds to restart flowing.
Usually, I experience 2-4 seconds, when I start printing after I let the hot nozzle ooze without extruding for longer than about 20 seconds.

All in all, the whole bowden setup is the reason, why I need 7mm retraction to eliminate ooze. The Repetier firmware also provides an advanced function to manage filament pressure during changes in print speed. It is quite complex to understand and get right, so I haven't touched this feature yet. However, this could be very interesting for bowden setups.

Your measurements of temperature are interesting. As I mentioned in another post, I have the feeling that the new nozzle design needs higher temperatures. I cannot explain to myself, why this is the case, as most part of the nozzle is fully integrated into the heating block, unlike the old nozzle.

[barnett]

I wonder if the nozzle is compatible with the other nozzles you might have on your other printers? If so, you might try one? Not sure if the teflon tube is transferable.

Is yours the no22le that they are sending as a replacement for people with the older one?

FWIW, I can report a similar 4 sec drip time after extruding into air. I was thinking about adding some gcode to the end of each job to retract about 25mm but I wasn't sure exactly what to type.

[mhackney]

I do have the newest nozzle design. I am not sure what it would take to use an older nozzle, I think just the inner and outer teflon tubes. I have plenty of that so it might be worth a try.

[mhackney]

Well, I am 1/2 way through printing a calibration cube. Things look reasonable so far for a first print. Turns out the issue was loose pulleys on the X, Y & Z stepper motors - believe it or not. Here's how.

Firstly, I replaced the screws in the kit with set screws. Turns out these were not effective and don't do a good job of really securing the pulley. When I calibrated Z=0 and the X, Y, Z towers, my axis movement was pretty slow. Things seemed fine and repeatable. However, when going to print it seemed that I was not getting extrusion flow. I pointed out a day or 2 ago that when extruding into air things seemed fine but when printing it wasn't. That should have been the necessary clue! Turns out when printing and the printer moves down to Z=0 for the first layer it was consistently going too far and really pushing the nozzle against the glass, blocking the outflow of plastic. I noticed this several times and attempted to compensate with the X Y and Z MAX_LENGTHs. Apparently that was not having any effect and I was not actually measuring the new Z height to validate the change. Apparently, there was just enough "give" in the pulleys that when all 3 axis move/accelerate quickly they would slip a little.

I've replaced the set screws with the original screws and tightened everything. I then did remeasure X=0 and crudely adjusted with the MAX_LENGTHS and verified that I had a little space. I also rechecked my first layer height in slicer - for some reason it was set at .2mm and I have a .5mm nozzle (I think I mixed configurations from my other printer that has a small nozzle opening). I reset that and re-sliced and now I was actually able to put down a first layer and continue printing.

The cube has finished the bridging of the openings and should complete in a few minutes. It ain't pretty but that's typical for the first print. Now begins the fine tuning process! It does look like I have to adjust carriage backlash and recalibrate Z and the towers, the cube looks a little ratty. Filament flow is not perfect and I noticed the "morse code" effect on the top layers. I also noticed that the filament was very tight coming off the spool so I unspooled some by hand and it fixed this. I think the spool does not spin easily on the laser cut "axel" so I'll make a nice insert for the spool and a round axel to replace. i'm going to wait a minute before posting so I can include a photo here...

and here it is:

[img]http://mhackney.zenfolio.com/img/s8/v77 ... 1814-4.jpg[/img]

The red mark shows me how the part was aligned on the table, that's the corner near the X tower.

You can see that the extrusion gets progressively worse and the top layers are pretty bad but now I have something to work with and start tweaking!

[mhackney]

I was doing a 2nd calibration cube test print with my heated bed at 100°C. Just as it heated up to 100, I heard a very load SNAP and when I looked over saw that my new glass bed snapped. The fracture is very symmetrical:

[img]http://mhackney.zenfolio.com/img/s2/v72 ... 5760-4.jpg[/img]

It looks like it originated here:

[img]http://mhackney.zenfolio.com/img/s8/v83 ... 5640-3.jpg[/img]

There was no evidence of a defect or scratch at that point on the glass. This plate was 3/16" thick. I now think that is too thick and the glass can't relieve thermal stress. Most folks seem to be using 3mm and .125" window glass. I also had not added an aluminum layer to spread the heat more evenly yet so there is a hot spot in the middle and colder around the edges.

A bit of a bummer and a bit of excitement!

[Polygonhell]

Sometimes glass just does that, the glass on my fireplace front just shattered one day when cooling down, no cracks or scratches, just bang.
I have some 1/4 inch glass that I've used without issue, though I usually use thinner glass because it's easier to get to temperature.

[mhackney]

Ok, I am now starting to make some good progress on improving print quality AND not wasting a lot of time or plastic.

First, some observations:

• The filament spool holder on the Rostock is a nice feature but I find that there is too much friction with the stock Spool Axel. I realized that some of me "filament starving" issues were due to the filament not coming off the spool easily. I've been keeping an eye on it and helping it manually for now. I think an axel with bearings will be much better and a good first "real print".
• I've been wasting too much time with the calibration cube and not really getting good feedback. I just downloaded The Essential Calibration Set from thingiverse and using the .5mm thin wall shell. It prints fast and provides great feedback. Once I am through with it, the other calibration objects also look interesting.
• Calibrating Z=0 and the tower lengths is essential to do hot! This made a big difference in my first layer thickness and, therefore, quality. Now I am putting down a first layer at the expected thickness.

Starting with the .5mm thin wall calibration object I started a print with these basic parameters:]

10mm/s for all speeds except first layer speed at 30%
200mm travel
.36mm layer and 1st layer height
3mm retraction
210°C nozzle 100°C bed
random starting points
2 skirt loops

I immediately discovered that the nozzle drags across the previous layer and causes problems (delamination for instance). Since a delta printer can move in any cartesian direction pretty quickly, I simply added a 3mm Lift Z (on the Printer Settings tab in slicer). This made a huge difference. I completed my first ever single wall print on the Rostock by adding the Lift Z. Here is a photo of that print (purposely unflattering - see below):

[img]http://mhackney.zenfolio.com/img/s3/v44 ... 1948-3.jpg[/img]

Note that the skirt loops stuck nicely and are even thickness and right on the money at .036mm! Only part of the first loop printed as the extruder was getting prepped. Also note that the single wall is nicely layered (although maybe a bit wavy from mechanical backlash) and complete all the way around. The blobby back wall is where the layer ends/starts were randomized. I'll need to fine tune the Retraction parameters on the Printer Settings tab.

I'll do a little more fine tuning on this part before trying one of the others. But, I am now starting to get somewhere!

cheers,
Michael

[mhackney]

I am doing a set of very controlled experiments and keeping notes and photos. I'm labeling each print at the same location so I'll know its orientation. I've started with the .5mm shell object and have it basically nailed. Some observations:

I started with retract and Z Lift. I found that the nozzle design is a lot different than the earlier pointier design that it drags and sticks to the filament a bit. With a delta design, doing a quick Z lift at the end does the trick. I also saw little blobs at the start of a layer so I experimented with retract. Here is the sequence of tests:

10mm/s speeds, 220°C nozzle:
0mm retract, 3mm Z lift - blobs (see photo in previous post)

10mm retract, 3mm Z lift - 2mm gaps in the layers
[img]http://mhackney.zenfolio.com/img/s2/v72 ... 3956-4.jpg[/img]

5mm retract, 3mm Z lift - smaller blobs than no retract
[img]http://mhackney.zenfolio.com/img/s4/v62 ... 4088-4.jpg[/img]

I also test 3mm retract (blobs) and 4mm (very slight blobs) so 5mm retract was optimal for this filament and other parameters.

I then started lowering the nozzle temp by 5°C. 210°C looked ever so slightly better (less stringing) but 205° was very bad - it wouldn't stick to the heated bed or previous layers.

I did a quick retest of retract with a 210°C nozzle. 5mm remained optimum.

I'll next lower Z lift, this is really just an optimization for speed. Then I'll start ramping up the print speed and fine tune as I go.

So far, looking pretty good. One thing I've noticed is the string after the print is done. It is formed as the nozzle goes home and is the full length! It doesn't break, just forms a tiny thread.

cheers,
Michael

[Polygonhell]

When you're seeing gaps like that with a lot of retract, I would look VERY carefully to make sure the extruder isn't stalling, or the filament slipping on the prime part of the cycle.

[mhackney]

Polygonhell, I'm curious about why? I see the same phenomenon on my H-1 with high retracts too. I actually tried 10mm, 8mm and 7mm retracts and it was very consistent narrowing of the gaps.

[Polygonhell]

Polygonhell, I'm curious about why? I see the same phenomenon on my H-1 with high retracts too. I actually tried 10mm, 8mm and 7mm retracts and it was very consistent narrowing of the gaps.

FWIW I don't like the Steve's extruder design, it's massively over designed with way too high a gear ratio for 1.75mm filament and suffers from stalling (IMO because of the high extruder RPM) with large, fast retracts, and the pinch roller design tends to grip filament well to a point, but slip sometimes intermittently.
Here's why you commonly see stalls on the prime portion of long retract/prime cycles.

Retract has minimal resistance, with PLA in particular what tends to happen in the prime is that the soft PLA you just pulled up the teflon tube starts to cool, reaching the rubbery state, and starts to mushroom (like a PLA jam) greatly increasing the force required as it's forced back into position. The longer the retract the worse this is.

Now you would think that the high gear ratio of the Steve's extruder would help with that increasing the available torque, but stepper motor torque drops off with RPM and at some point you end up with less overall torque, I believe that because IMO we underdrive these motors with only 12V, the Steve's extruder is past the sweet spot when trying to retract/prime at 50mm/s.

There is a secondary issue where if the pinch roller isn't adjusted perfectly the filament can also partially slip, it's the only extruder I have ever seen that does this, the common designs usually either stall or just eat filament, but the Steve's appears to somehow still push filament but at a reduced rate.

These are just my observations, you can get good prints out of them, but it's not my favorite extruder by a long way.
I intend to print out a direct drive extruder this weekend to play with, I also have a couple of Wades extruders here I might try if I get to it.

[mhackney]

Thanks for the clarification Polygonhell. I only have experience with Steve's extruder but I do know what you mean about slipping - especially on fast retracts. I've gotten into the habit of putting a black mark (Sharpie) on one of the extruder gears and a mark on the filament when I'm doing this kind of testing so I can make sure nothing is slipping.

Do you have the new nozzle? It really does seem to snag and drag more than the more "teet like" (sorry for the scientific term) earlier nozzle. It seems to string more too - although that could be a Bowden phenomenon also - this is my first printer with that feature.

I did find it satisfying though that changing the retract amount over a range of increments had a predictable effect on the size of the gap.

regards,
Michael

[mhackney]

I got the LCD Controller with my Rostock and RAMBo. I'm using the instructions and information that John posted: http://www.reprap.org/wiki/RamboLCD and the threads here (which I'll reference when I get to the configuration). The first thing I needed was a 2x4 and 2x10 header to solder the cut cables to and plug into the RAMBo. Rather than order just these, I decided to build them from 1x4 and 1x2 headers - I have LOTS of those! Here's the layout:

[img]http://mhackney.zenfolio.com/img/s8/v83 ... 6868-4.jpg[/img]

I have 2 pins of overlap on the 2x10 header. I used a medium CA glue - it's easier to work with than the normal thin superglue. Here they are glued up and ready to solder:

[img]http://mhackney.zenfolio.com/img/s8/v83 ... 7184-4.jpg[/img]

This is my project tonight.

[Polygonhell]

Do you have the new nozzle? It really does seem to snag and drag more than the more "teet like" (sorry for the scientific term) earlier nozzle. It seems to string more too - although that could be a Bowden phenomenon also - this is my first printer with that feature.

I did find it satisfying though that changing the retract amount over a range of increments had a predictable effect on the size of the gap.

regards,
Michael

I added myself to the list to get one sent out (it may be in my mailbox for all I know), but I'm currently using the original nozzle, it too had a lot of resistance to extrude FWIW, I run abut 0.3mm of Zlift@200mm/s, but for a single walled test print I'd disable that and probably retract as well. I find any delay at either end of a move increases the tendency for blobs/strings. Kisslicers wipe setting is really good for this, and it's smarter in general on when retract occurs (it wouldn't occur at all in that single wall test), but Slic3r has nothing equivalent.

FWIW I think you are doing the right thing, change one thing at a time, print calibration pieces and document.
I don't see the gap introduced by retract, but I haven't printed PLA in a while, I do have a couple of issues where if I have large retract the filament buckles at the exit from the rollers (solved with a printed piece) and if I print something with a LOT of retracts (10's of times per layer) I do see filament starvation, and in some cases extruder jams. The buckling issue doesn't happen with PLA, only softer plastics like ABS and worse still Nylon.

[mhackney]

I completed hooking up the LCD controller. One comment - the wiring connection diagram in the PDF file is based on looking down on the connector from the backside where you will solder the wires.

I used both of the ribbon cables and measured the distance from the RAMBo - through the center of the base - to the LCD Controller mount area between the Y and Z towers - as 24". So I cut the connectors off one end of each cable to leave 24". Then I split the individual wires back about 3".

[img]http://mhackney.zenfolio.com/img/s8/v79 ... 3424-3.jpg[/img]
(the look like hydra!)

I tinned each of the leads on the connectors and the wire ends. I also used a short (3/8") section of shrink tubing to insulate the connection after it was soldered. Pay careful attention to the wire chart. I soldered all of the leads on one ribbon in order first and then the other.

[img]http://mhackney.zenfolio.com/img/s8/v85 ... 3316-3.jpg[/img]


I tidied the connectors up with a couple of zip ties.

[img]http://mhackney.zenfolio.com/img/s8/v76 ... 3380-3.jpg[/img]

Then I turned off and unplugged my RAMBo and installed the cables. Again, pay attention to the photos. Note: the photos in the PDF are upside down! Make sure you get this right.

Once I had the connectors hooked up I plugged in the USB to the RAMBo and held my breath. I was greeted by this:

[img]http://mhackney.zenfolio.com/img/s8/v76 ... 3240-3.jpg[/img]

I did not do anything special to the Repetier firmware from Halopend that I am using. I understand that it may not have the memory card reader enabled but I'll get there eventually.

All in all, the installation took about 30 minutes. I played around, homed, etc and everything seems to work fine.

I can imagine that lots of folks will customize the display and controls to do all sorts of cool things!

cheers,
Michael

[geneb]

I'll agree that the Steve's Extruder is an animal. About 3 hours into a PLA test print, the filament jammed (I suspect the PEEK hit the sweet spot temp for PLA and it was all over but the crying at that point) and because I was stupidly fiddling with the extruder tension and had it too high, the drive literally destroyed the metal teeth inside the push-fit connector. Tiny metal bits, etc.

Yet another lesson learned.
g.