Seems like every time I get my hopes up that I may have a fix, something breaks. However I am getting closer.  I swapped the V6 clone hot end with genuine E3D parts (except for the heatsink which is retained from the clone).  The new Genuine E3D heatbreak looks better but not perfect, though I don't expect it to have any issues feeding.  I bought the E3D V6 heat block simply because it included the cartridge type thermistor which I expect will be hassle free and it also has some neat silicone socks included.  There are some additional marlin configs needed but they are well documented here. 

I also added a 3D Touch probe, because when I am just trying to get the damn thing working, why not double down on the possible points of failure (but I had the printer and wiring loom already in a minor state of disassembly). So whats the problem now?  I lost the damn nozzle (doh).  So now I'm waiting for a pack of those, but otherwise the printer is back together and ready to hopefully rock and roll.

To date this is where it stands currently.  Aside from upgrades, some items were replaced on suspicion and others to test, and others because they were faulty).

• New hot end using a mix of E3D V6 and clone parts (due to issues with the original hot end clogging due to the Bowden setup).
• New Bowden tube and fittings (did not help the original hot end problems but I kept them on there)
• New Themistor (original did not seem to be accurate).  The first replacement was accurate but the new one is a cartridge which I prefer.
• New heater cartridge (I bought a few previously and picked the one closest to 14.4 ohms).
• New firmware (tried both Marlin and TH3D), currently running TH3D v1.8J (running that v1.8 since I could find instructions to get the touch probe up and running quick, but may try v1.9 or Marlin again)
  
• New build tac - using PEI now, though nothing wrong with the original build tac except for the texture.
• 3D Touch Probe, which seems like it will be a time saver.
• Added a shim to the Z-motor mount which seems to have reduced binding on the Z-Axis leadscrew.
• Moved the Y-Axis limit forward, which seems like a common problem as it has tons of room in the back but goes right to the edge when moving the plate all the way back.
  
• Moved the Z-Axis limit up, and then removed it completely when I added the 3D Touch.
• Swapped the extruder gear, this was cheap and probably a worthwhile upgrade.
• Swapped the extruder with modified parts allowing spring adjustment (then swapped back to stock when I found it was too tight with the modified extruder).
• Adjusted Vref of the Extruder and verified the rest (values are in a prior post somewhere down there).
• Shimmed the vertical frames where they meet the baseframe and straightened the pre-assembled base frame since it would not sit flat when I got it.

Seems like a short list considering the time spent on this, and am not even done yet.  Admittedly many of the mods did not help and made things worse possibly (like the extruder mod, which I later found was causing excess load on the stepper and causing it to skip steps).  Other things may have helped or not, but I suspect the one mod that will hopefully benefit the most will be the hot end and the polished E3D heat brake. 

The firmware upgrades are a mixed bag, I am not sure if there will be a big difference in print quality to the stock firmware but they do make things simpler.  However the Ender 3, like the original CR10 has a pretty minimalist mainboard with only 128KB of space for Marlin, which requires some compromises.  I found that I could not compile Marlin 2.0 with all the bells and whistles, the most memory intensive optional features were the LCD, SDSUPPORT and UBL  (or Universal Bed Leveling).  It's one of those puzzles where you can pick 2 but can't have it all, so I lost SD Support in favor of some basic onboard leveling and the LCD (I was able to get all 3 at one point but it used almost all the space and warned of likely stability problems so I ditched it).  So without SD Support I just need to print from OctoPrint which normally works fine.  If I can get this working with a degree of reliability and am happy with the print quality then I will consider upgrading it to something with more memory or possibly a 32 bit controller. 

Still yet to be added are:

• TL-Smoothers (reviews look promising and they are cheap)
• Titan extruder (may even try the Aero, though it will be all clone parts due to the expense).
• Upgraded blower fan (on the slow boat from China right now)

All in all, I think I can finally see the light at the end of the tunnel.

So the V6 clone hot end was not the fix I was hoping for.  In fact it was just another time sucking detour, since I had to try (and fail) to fix a problem with the clone.  The trouble was with the heatbrake, which has some burs or roughness inside that is stopping the filament and clogging the works.  I tried to polish the feed ramp inside the tube but it only helped a small amount.  I never did get a good print from the new hot end, and found that what it did print had poor layer adhesion, even after pushing temps to 270 deg C on ABS (bed at 110).  I think that should be in the upper range for ABS, but it was clearly not heating properly.  I took out the thermistor (which was just recently swapped), and tested it using some boiling water, and then compared it's readings to my cooking thermometer.  The values were within 2-3 degrees so the thermistor seems to be good.  Next I pulled the heating cartridge (which I retained from the stock Ender 3) and found that it was measuring 13.5 Ohms (ideal is 14.4 Ohms for a 40W cartridge on a 24v system).  I am not sure if that could be related to the layer adhesion but I am swapping it with one that measures at 14.5 ohm.  The heater cartridge that shipped with the V6 clone hot end measured at 15.4 Ohms, so there is probably a good deal of variation in these cartridges.  None of this fixes the problem with the clone V6 heatbrake though, and I have no spares.

In the end I have ordered a (hopefully) genuine E3D heatbrake which cost as much as this whole hot end.  I also went ahead and ordered the E3D upgrade kit for the V6 which includes a new block and cartridge thermistor.  Although I am sure the thermistor is working, I don't know if it is making good contact with the block which may be skewing the results.  The clone V6 blocks just have a small hole to push the thermistor bead into.  On the V6 clone, I added some thermal paste to help with the "connection" between the thermistor and the block but that did not help with the problem.  So, with all the problems I have been having, it is one less variable if I can be sure there is good contact, which the cartridge type on the genuine E3D V6 block should have.

More Ender 3 troubles, and as usual it is the damn extruder again.  I went to print some brackets (2nd print on this printer of a non-test item) and it failed miserably.  I could see that it was going to have problems since the extruder was slipping so I tried adjusting the tension on the extruder up and down.  There was no happy medium, if it didn't slip, it was popping.  This is after flashing the firmware with Marlin 2.0 and then TH3D, adjusting the vref of the extruder (now at 0.815v which seems too hot for it but with it lower it was lacking the torque needed).  I have also printed a different extruder with an adjustable spring tension (which is backed all the way off now since the stepper is unable to keep up), and swapped the extruder gear with something more aggressive (36 tooth).  The extruder was re-calibrated after the mods, and was about 14mm off which is quite a bit - though I accounted for that in the firmware and when I re-checked it was about spot on.  I am going to measure the stepper however I doubt it is faulty.

Compounding things, the Kossel is back to it's old tricks also with bad level shifts so I cannot rely on that either.  I think I need to swap the extruder on the Ender 3 with a Titan since the stock stepper lacks the torque to push the filament, but damn if I don't also need to swap the hot end too since it is garbage to begin with, and I suspect part of the trouble with the under extrusion due to the design of the Boden setup and the hot end.  I have been unable to get the stock hot end adjusted tightly enough to say put, so it ends up leaking which leads into more problems.  It's been weeks since I have been able to get a decent print for an actual project out of my printers so I need to start solving some of these problems, and that cannot be done in this case with the stock parts I think.

I have the parts to do these upgrades on the Ender 3, but lack a reliable printer to make the mounts for them.  So looks like I need to go back and give the Kossel some attention and a brain transplant to get it back in operation reliably * so I can then get the POS Ender 3 upgraded to the point of usability.  Amazing that I Ender'ed up with two lemons, but now I have to double down on making some damn lemonade...

* I may have lucked out though, I'm trying to print a simple mount for the new hot end on the Kossel.  It is a quick print so hopefully it can do it without shifting layers.

After attempting to run a print with the new hot end, I found that the stock Ender 3 temp probe which I moved over to the new hot end was not working accurately and seemed to be reading low, it was probably the source of much of the problems.  I swapped it with he probe from the new hot end and found temps were much closer to reality now.  Hopefully I am getting closer to a functional printer. 

M304

Well technically the firmware on the Ender 3 is Marlin. just a customized version of it, but I wanted the full Marlin firmware on the Ender 3.  And because I can't seem to stop getting into deeper problems with this, I am trying Marlin 2.0 which is not quite released yet.

To load up Marlin on the Ender 3, I had to do two things (really only one thing if I wanted to use a pre-compiled image):

1. Burn the bootloader since the Ender 3 does not have a bootloader from the factory.
2. compile and load an image of Marlin 2.0

The bootloader is just needed to load other firmware on the motherboard, and the Arduino IDE can take care of it with some work.  Some jumper wires however are required.  The following are good guides to burning the bootloader:

https://matterhackers.dozuki.com/Guide/Flashing+the+Bootloader+on+a+Creality+CR-10/62

https://www.youtube.com/watch?v=27RDXtKjrCI

Part 1 is pretty straightforward, and I could have stopped there and just used a pre-compiled binary for the Ender3, but I wanted to compile a version from Marlin2.0 using the Arduino IDE.

To get started I used the configuration.h from Blinkii (https://www.thingiverse.com/thing:2981891) who was kind enough to upload a working image of Marlin 2.0 for the Ender 3, and his configuration files for Marlin 2.0. 


==========================================================================================
Update - I removed some bogus info due to my lack of understanding of PID - the following is correct to the best of my current knowledge:

To tune PID for the heated bed, I used the info here:

https://www.youtube.com/watch?v=6gr-XOJ_1Fo
 
That is basically running this GCode from the Octoprint terminal:

M303 E-1 C8 S90

E-1 specifies the heated bed
C8 = run the calibration for 8 cycles
S90 = use 90deg C as the target bed temp

That will then run for a bit and if you have Octoprint, it will show the temp going up.  On the terminal it helps to suppress the temp messages, but I noticed that I had to uncheck that to see the results.  When it was done I had this:

Recv: Kp: 866.45 Ki: 170.59 Kd: 1100.17

Which I then set using this:

M304 P866.45 I170.59 D1100.17

Then M500 to save the settings to EEPROM, and M501 to confirm they are correctly saved.  This calibration routine is custom to each printer so don't just pop these in, does not work that way.  A similar calibration routine can be run on the hot-end, but the parameters are different.  In the linked video, E303 E0 S200 C8, he used to calibrate and M301 was used to save them, with M500 to save the settings and M501 to view are the same.

The same values can be entered directly into the Marlin configuration.h by searching for these and changing the values, below are mine:

    //Default Bed PID
    #define  DEFAULT_bedKp 866.45
    #define  DEFAULT_bedKi 170.59
    #define  DEFAULT_bedKd 1100.17

As an aside, I found that in TH3D's Marlin flavor which I am now trying out, these are in configuration_backend.h which it says not to edit so probably, though I did it and then re-ran the calibration utility anyway.

Getting back to the Blinkii Marlin 2.0 configuration.h files, I also changed this line in configuration.h to English (en) since the file from thingiverse was set for German (de):

#define LCD_LANGUAGE en
 

Once I had the configuration.h and configuration_adv.h settings the way I wanted them for my printer, I had to compile them with the Arduino IDE which is easier than it sounds.  I did have a few errors however.  

This error popped up when attempting to compile Marlin 2.0 using Arduino IDE 1.8.5:

Arduino: 1.8.5 (Windows 7), Board: "Sanguino, ATmega1284 or ATmega1284P (16 MHz)"

sketch\src\lcd\dogm\u8g_dev_ssd1306_sh1106_128x64_I2C.cpp:72:20: fatal error: U8glib.h: No such file or directory

 #include <U8glib.h>

                    ^

compilation terminated.

exit status 1
Error compiling for board Sanguino.

This report would have more information with
"Show verbose output during compilation"
option enabled in File -> Preferences.


I did two things to get around this:

1. Followed the instructions here to change my build.path=c:\temp (or something short like that)

https://www.v1engineering.com/forum/topic/avr-gcc-exe-the-filename-or-extension-is-too-long/#post-56045

2. And then I just installed the missing library from the Arudino Library manager ( Sketch > Include Library > Manage Libraries > search for "U8glib" and I think it was the 1.19.1 version bu Oliver that I used.

Then Marlin 2.0 was able to compile.  I had the Arduino IDE configured to complie for a Sanguino /  "ATmega 1284 or ATmega 1284P (16 MHz)" and then plugged in the board for the Ender 3 and uploaded the sketch.  I also exported the binary in case I need it later.  I almost forgot that the Sanguino board profile is not available by default - it needs to be installed maually in the Arduino IDE - just follow step 6 here or heck just follow that guide, it is really good.

Thanks to Marlin I can now get my extruder calibrated and it is starting to look better, I am just a few steps from hopefully getting it dialed in.  Unfortunately the damn think bucked a bit when I was printing a test cube and tore up the build tac.  This is one stubborn machine.

My purchase of the Ender 3 has been a bit regretful due to poor quality prints and hours lost in troubleshooting.  Fortunately my Kossel has really stepped up and is printing pretty well, though I cannot trust it for anything over an inch tall (when it will probably start layer shifting). 

Aside from the set up issues which I don't really mind, the problems with the Ender 3 began on the first print out of the gate.  I did some research first to find a good Cura profile (which seems to require a Facebook account), but then I found this video which seems to over complicate things (but oh God the music sux so be fore warned).  I followed the video and set mine up to match.  The first print was garbage, it looked like a drunken spider ran all over it.  Then I tried calibrating the extruder (which I should have done first), and found it was only putting out 88% of what it should.  I made adjustments in Cura (flow 114%) and tried again - crap, and again with different retraction settings - crap, and again with other retraction settings and lower acceleration and speeds - crap, then I swapped the Bowden tube and couplers (thanks to a suggestion), tried again - crap, then I printed a test piece, forgetting to compensate for the under-extrusion - that looked ok but it was clearly under-extruding, but when I put the flow back up to 114% to compensate it was crap again.  So I am running out of weekend to mess with this thing and realizing once again, that cheap is expensive. 

This printer is going to end up costing me about half as much again before I can hope to get a decent print from it.  A lemon as it may be, I am not giving up on the Ender 3 yet, the frame is at least decently straight (no thanks to the factory), and the electronics and steppers work.  What doesn't work is the extruder and hot end.  The under extrusion is quite bad and even after swapping the Bowden tube and couplers there was no change.  The hot end is a hot mess.  I found that it had leaked PLA around the nozzle which seized it up requiring me to remove the hot end and clamp it in a vice, remove the heat blanket, clean the cooked off PLA from around the nozzle and then with a good deal of effort remove the nozzle (which at that point was ruined).  I got all that cleaned up and re-assembled with new parts, but it did not help with the stringing or the under-extrusion.

Unlike the Anycubic which uses a E3D V6 clone hot end, the Ender 3 uses a style that requires the PTFE Bowden tube to sit flush against the nozzle.  I was not really familiar of this type of hot end until now (I had seen them but never used one).  The weakness is that the coupler on the Bowden tube is all that keeps the clamping pressure of the tube against the nozzle, and if there is any slip, there will be a gap between the end of the tube and the nozzle - which is bad.  PLA will pile up in the gap and cook off, eventually clogging up the works which would have happened to mine in short order.  I have found that I don't like this style of extruder, since it is too delicate and really just an outdated design. 

The E3D V6 hot end on the other hand, has a threaded insert that forms the transition between the melted plastic and the unmelted plastic (the heat break).  The nozzle then seats up against this threaded insert so there is not much chance for a gap, and the clamping pressures between the insert and nozzle are much higher than the setup on the Ender 3's hot end - so it is unlikely there will be a gap for plastic to fill and burn off.  Good description of this here.  The E3D clone hot ends are pretty cheap, however the Ender 3 is a 24V system, which means most of the hot end kits made for 12V will not work (the fan and heater need to be 24V).  So a bit more expense and work, and waiting for more parts. 

Pretty clear there were corners cut in this design (had to for a $200 printer), and to that point, here are the most glaring design problems with the Ender 3 (in no particular order), as well as any fixes for them by the community of makers at thingiverse:

• Filament enters the extruder mere mm from the greasy Z-axis lead screw (fixes here, here and here)
• There is a cut in the extrusion which cables are run through from the control board, problem is it has very sharp edges (fix here)
  
• The fan opening for the electronics is positioned below the home position for the hot end, and is facing up, so any garbage that falls down there can go into the fan (fix here)
• Cable management is a mess, and they can get caught during operation (fixes here and here)
• The back side of the display is flapping in the breeze (cover it up here)
• The extrusions are not flat on top, though they treat the design as though they were (requires shims).
• Mine did not include any manual, only the Ikea style assembly instructions (not a big deal but would be nice to have a setup guide for the menus and board pins).
  
• The Z-axis end stop had an indexing tab which put it near the limit for the adjustment range of the bed (had to modify the end stop bracket to lower it) - this may be something I am missing since I have not heard of this problem before.
  
• No bootloader on the main board which means there are a few extra steps to get Marlin installed.
  
• The location and design of the extruder makes it difficult to get the filament loaded.
  
• Mine is under extruding by a good margin (14%), possible fix with an extruder mod here.
  

Some of the good points are:

• It is relatively inexpensive at $200.
• 24V should help with heating things up faster.
• There is a large community for the CR10 and the Ender 3 is really just a small CR10.
• the carriages are metal and there did not seem to be any alignment issues with the wheels.
• All the holes for the parts were spot on.
• It was 60-70% assembled - though the base did need adjustment to sit flat.
• The build tac suface works well and is not glued on, but rather stuck to some hard plexi material so it can be easily removed or replaced.
• Cables and wires are well labeled, even at the motherboard end.
• They seem to have fixed several issues with earlier models of the Ender 3 so Creality is improving things.

I am probably biased against this printer after all the time I have wasted to get it working but I think the above is still pretty fair and I still have hope that I with some more work this printer can make some good prints.

My plan now is, instead of trying to fix something that is likely to never work right, I am going to try and swap the hot end with an E3D V6 clone and the extruder with a titan clone which I am going to mount as a direct drive.  I am also going to install TL-Smoothers and install Marlin.  I'm not going to spend much more time trying to fix on the Ender 3 until I get these upgrades installed, but may try some simple prints on it.  Maybe if I am lucky I will get a good print out of this after that is all done.  I do have a job though, and this printer has stolen most of my weekend so I am going to step away from it for a while (need to wait a couple weeks for parts anyway).

Update - I found this video which describes a better way to adjust the X-Axis:

Although in the above video he did not adjust the outer wheels (which do not have eccentric nuts, it seemed like there was some small slop in them on mine which allowed me to push them a bit when tightening and it did seem to work for me in the end - though it may not be the easiest way - I would try the method from the video first next time by just adjusting the ilders.


And this is a great video on how to deal with a really out of alignment X-gantry (after seeing it, I would do this on my printer if I was building it again) - in fact just stop reading the stuff below and check out the video below:

Will leave the original post in case it may help - but please see the above videos (or at least the EdgeOfTech video) which are better and clearer guides:

After getting the frame square, I noticed that the X-Axis 2020 extrusion was not parallel to the top 2020 extrusion on the frame.  When I measured it was about 3-4mm lower on the right side.  I just considered it to be an arm on a pivot point and adjusted it accordingly.  (update 6/16/2022) In the process below, the idea is to use the eccentric nuts, and also the bit of slop in the holes for the other wheels, to first bring the x-axis parallel with the top bar and then snug up the outside wheels first by pushing them in while tightening.  It can be a bit tricky to push the outside wheels in while tightening them, but if the slack can be removed from the non-adjustable wheels, the eccentric nuts will have more adjustment range, and won't need to push excessively from the inside of the frame to tighten all the wheels up (which would seem to also affect the geometry possibly).  Below is the process I used to fix that - from memory so I may have left something out, but this is the basic process:

1. Unplug the printer and then disconnected the Z-Axis stepper (this will keep any back EMF from damaging the board or stepper drivers).
2. Rotated the Z-Axis coupler until there was 10CM between X-Axis bar and top of the printer on the left side (didn't matter which side i set to 10CM, it is just for reference).
3. Noted the measurement on both the left and right side, measured from the 2020 extrusion on the X-Axis bar to the 2020 extrusion on the top frame (not between mounting brackets).
4. Write the measurements down and noted which way the X-Axis bar is tilted.  In my case the right side is sagging (low).
5. Loosened the yellow wheels by turning the eccentric nuts, only loosened them enough so they turned freely.  The eccentric nuts can be turned either way, so just need to adjust a bit and feel if the wheels are getting lose or tight, it should not take more than 1/2 turn to see some change.  See pics below for more info on the eccentric nuts.
6. Loosened the wheels in green in the pic below, and if I could not then adjust the X-Axis bar, I would have loosened the bottom purple wheel also (but I did not need to).
7. lifted the right side of the X-Axis bar so it was level, if possible clamp it here or just hold it - both sides should now measure the same 10CM distance from the top 2020 extrusion (X-Axis bar is now parallel to the top 2020 extrusion).
8. Tighten the bottom green wheel on the right side, then the bottom purple wheel on the left side (if loosened), and then the top green wheel on the right side.  These outside wheels need to be pushed snug to the frame while tightening since there is just a bit of slack in the mounting holes that will allow that.  It may be necessary to "split the difference" when taking up this slack between the left and right side outside wheels.
9. Re-adjust the yellow eccentric wheels to snug them down, I adjusted them so I could turn them with my fingers but not easily.
10. Verify all wheels are tightened and I was able to turn them using my fingers (but again they were not easily turned).
11. Re-measure the parallelness of the X-Axis bar and the top 2020 extrusion by repeating #2, then make any adjustments again.
12. Once done, re-connect the Z-Axis stepper and verify that all the steppers and end stops remain connected.

A note on the eccentric nuts. they can be loosened without loosening the locknut, they should turn a bit with the included wrench. The eccentric nuts have a hole that is not centered, so when they are rotated, they move the bolt that goes through them, this enabled the adjustment of the wheels in or out by turning the eccentric nuts.

I bought a new printer - the Creality Ender 3, and for a $200 printer it looks impressive.  I just assembled it today and ran into a few problems but none as difficult as the assembly issues with the Kossel.  The Ender 3 is a 222x220x250 Cartesian style printer which uses belts to drive the X and Y with a single screw for the Z.  Probably the biggest plus is that it runs on 24V which should mean faster heating.  I watched a couple review videos and read a bit on forums before buying it, and it has a pretty good reputation so when I found it on sale I bought it.  

It is about 70% assembled when it arrives but there is still a good bit to do to get the printer working out of the box.  It is the standard 2020 extrusion frame and every hole was spot on and the threads looked good where they were needed.  The only issues I had were with the extrusions themselves and the Z endstop mount. 

The 2020 extrusions are not perfectly flat on top, appears that with both the base frame parts, the center of the extrusion is just about a perfect 40mm, but the sides are 0.16-0.3mm shorter.  I ended up shimming these but made a bit of an error which seems to have not mattered in the end.  I had seen a video where somebody ran into the same issue and used some paper to shim, and others where people had to widen holes on the top cross frame due to the same problem.  Widening the holes though seemed like an extreme way to handle it, and would not correct the geometry that is off (the verticals tilt inward - or possibly outward, due to the higher center on the 2020 base rails).  I found that when I got it shimmed so the distance between the inside edges of the vertical frames was 25cm at the top and bottom (that is, they were parallel), the screw holes lined up on the top frame which indicates that it was manufactured correctly.  The problem is that the vertical rails won't be parallel without shims, or a lot of luck.  I also ended up shimming more on the inside edges of the frames since there was still a tilt to the inside after the initial round of shimming.  Below is a pic of the problem:

I found that the amount of shims needed was different between the sides so here is what I ended up up:

Now here is where I messed up, I should have divided the amount of difference measured between the heights of the sides and middles of the base 2020 extrusions by 2 and then used that as the starting point, but instead I used the whole value.  As it ended up however it seems to be fine and the screw holes in the top cross frame line up (I actually over shimmed a bit on the inside edges since I had to put some inward pressure to get the screws to line up).  However I see a slight (under 1mm) skew to the right as would be expected since the left side is now shimmed higher than the right side - I will probably need to go back and fix this by removing some shims from the left side, likely around 0.1mm from each side of the left side.

Below are some pics of the shims and how I measured the 2020 base rails to get the sizes I needed (to start with):

The second problem was that the z-endstop is too high. I'm not sure why I couldn't find anyone else reporting this problem but I acknowledge that I may just be missing something stupid.  There is a small indexing catch on the z-endstop mount plate that does not allow it to be moved lower (without cutting the tab off).  The manual indicates that there should be about 32mm from the bottom of the base 2020 extrusions to the bottom of the z-endstop mount but mine is closer to 36mm.  This means the z-endstop will trigger before the extruder nozzle is still a few mm from touching the bed - and that is with the bed screws all the way out.  I am printing a new mount now, but this just seems like a strange oversight (again could be me doing something stupid too).

And below are some other pics I took during the assembly, for reference primarily: