Wanhao i3, Part 2

Lots of updates on this...

I finished refurbishing the MK10 heater block. This required sanding all the crud off of it.

Mostly sanded and cleaned

 I went a bit crazy with the insulation. I covered all sides with 3mm of ceramic fiber insulation and "Koptan" (knock off Kapton, i.e. polyimide) tape. I also soldered and installed a new thermistor. I was having a lot of weird heating issues before. I'd have to run PLA at ~250C to prevent delaminating layers, but measuring with a thermocouple multimeter showed that the nozzle wasn't getting anywhere near that hot. Turning the cooling fan on would also lower the temperature significantly. So I decided to fix all possible reasons for this: new insulation, new thermistor, new cooler (DiiiCooler) aimed at extruded filament and not the nozzle/heat block.

Insulation, new thermistor

More insulation

While I'm talking about this, I later discovered that the fan cooling off the nozzle problem had gone away, but I still had to run PLA at 250C. This meant two things: 1. Insulation helps, 2. the old thermistor was problem fine. I replaced the nozzle and PTFE liner next (the old PTFE tube looked fine actually). The old nozzle was either nickle coated brass or SS, the new nozzle was definitely brass. This made a big difference. I could now print PLA at 215C and get similar layer adhesion strength. I think the old nozzle was SS, which has a much lower thermal conductivity, resulting in a higher temperature gradient from thermistor to nozzle. I also added some more insulation around the nozzle since this one was longer. Remember to re-zero your Z axis when you change nozzles! This heating mess took many hours to resolve. What a pain.

Brief review of Koptan tape: Koptan is the chinese knock off of Kapton polyimide tape. Kapton has a fairly high heat resistance and is electrically insulative, making it suitable for heating and electronic applications. I've used it a lot in the past. Surprisingly, Koptan seems to work ok, though I don't think it has as high of a temperature rating. Well, it's probably rated the same, but won't actually work at the higher temperatures. 

I printed a DiiiCooler before I took everything apart. I had an ABS CiiiCooler that came with the printer, but it was pretty warped and was too high in Z, which caused the air to mostly blow on the nozzle. I figured a PLA cooler would be fine as long I insulated the hot end well. I wouldn't recommend printing it with external supports (definitely not internal)...the supports spaghetti'd on me and I was barely able to save the print with a bunch of superglue. Unfortunately, the DiiiCooler was also too high in Z (I did not print the "short" version, which is even shorter in Z). I used washers as spacers. Update: currently using 4 washers and some paper as shims, which results in the cooler being ~1mm above the part.

I ended up using fewer washers than this. The cooler was not sitting flat, so I had to crack the supports and reglue them slightly angled.

You should barely be able to see the nozzle peaking through. This means the air flow is aimed at the part and not the nozzle.

I still ended up needing to shim the cooler with paper to get it level to the bed.

While I had the hot end apart, I decided to do the rotate the extruder stepper motor mod. This lets the wires clear the top of the frame at the max Z height. Unfortunately, I discovered this:

Stripped thread...ugh

The previous owner must have stripped the threads out of that hole while trying to get the extruder lever screwed in. Turns out a M3 helicoil kit is more expensive than replacing the stepper. But since buying the exact same model is impossible, I'd have to go through stepper tuning (steps and current) again. Ugh. I decided to just switch out the Y-axis motor for the extruder motor. Turns out the screws holding the Y-axis motor are long enough to engage the threads at the back of the stripped hole, so it worked out.

I replaced the belt and added a 3D printed y axis tensioner. This ended up working ok, but not amazing. The little nubs that the belt ends wrap around kept breaking off and falling out...poor layer adhesion or something. I managed to superglue them in, and they held long enough to get the belts in, but then broke off again. I added zip ties to the belt ends to help keep them from pulling through. While I was doing this, I also aligned the y-axis belt (adjusted the idler and belt gear. I also added a zip tie around the frame and end of the bolt holding the y-axis idler pulley so it wouldn't be cantilevered. I also printed the X version of this tensioner (comes with the download), as well as another X axis tensioner. I couldn't get the second X axis tensioner to print well, but I was able to manually tighten the X axis belt enough that I decided I didn't need it or the first X axis tensioner. I did replace the screws holding the belt ends with longer ones and add a zip tie between them so they screws wouldn't be taking the whole belt tension load as cantilevered beams. I wasn't able to align the belt perfectly, though, so I'm going to wait to replace that belt until I can figure out a better solution.

I also replaced the y-axis carriage plate. I purchased it from Tehnologika in Slovakia for about 25 Euro with shipping. It's about 145g lighter than the stock plate (230.7 vs 375.5g). You need longer screws because it's 6mm thick (18x M4 12mm button caps, and 4x M3 25 or 30mm button caps); order these while you wait on the carriage plate to come. The longer screws only add about 5-6g, so you definitely come out ahead in terms of weight. It's also way stiffer than the original stamped steel build plate. It's made of two 3mm thick plastic core aluminum sandwich panels glued together. It has a coat of white paint on one side and gray on the other, though the paint job sucks and the paint flakes off easily. They're faster "airmail" shipping option is horribly slow, even to the UK. It took over 2 weeks to arrive in London, then it only took a day to get to me. Ironically, I had gotten fed up waiting and had just finished reinstalling the original carriage plate when it got here -_- ...ugh. So I took that back off and installed the new one. I tried using the spring cups I printed, but they kind of fell apart...poor layer adhesion. I then put the new glass+PEI build plate on, using 3-4 pieces of Koptan tape on each edge to hold it down. Then an endless string of problems began.

I could not level the bed at all with the glass build plate. I used the paper under the nozzle method. No matter what tightening/loosening pattern I used, the screw/springs on one diagonal would be full tightened and the two on the other would be fully loose. This basically was pringle-ing the heated bed and trying to warp the glass.  The glass was very flat...I felt suction trying to lift it off of flat surfaces, so it had nothing to do with that. I also checked this by laying a thick edge ruler on edge on it and looking for gaps. Using these methods, I also determined that the new carriage plate and heated bed were level within 0.5mm, so I don't think they were the problem either. Geometrically, this doesn't make any sense...why would a perfectly flat piece of glass need to be heavily warped to be level?  The most frustrating thing was that I was able to level the old bed (a 3mm thick plate of PEI plastic that was definitely not flat) better!!!  It took ~5 incredibly frustrating hours for me to figure out what the cause was. It can't be the glass: that's flat. It can't be the carriage plate (I tried both the old and new one). It can't be the X axis level-ness because I leveled it by placing an object between the lower x-axis rod and the Z-axis stepper mounts on both sides. Besides, that could only be the problem if the left or right side were higher than the other, not the diagonals as was the problem here. This left only one thing: the y-axis rods. If one rod is angled (rotated about X) relative to the other, then you will get this diagonal mismatch behavior. This isn't because the rods are warping the bed. It's because the bed moves on the y-axis. If the y axis rods are tilted, then the bed will tilt as it moves. You have to think about it some, but it finally clicked with me. The worst part is that the bed can only be as level as these rods are co-planar, IF the bed is perfectly stiff...like a 3mm thick piece of glass, but very much unlike a 3mm thick piece of plastic. THAT is why I was able to "level" the old bed better. I was actually warping the plastic build plate (and heated bed) to compensate for the non-coplanar rods. I used some digital caliper's depth post to try to measure the y-axis rods' Z distances to the frame. I would then loosen the metal bracket holding one end of a rod, adjust it up or down, then retighten the bracket. Then remeasure and repeat. I was able to get them within +/-0.3mm this way, but that means that the glass build plate can only be as level as that...which sucks. There really wasn't an easy way to use the calipers to do this. What I ended up doing was sliding the thick metal ruler under the rods but over the frame, then using pieces of paper as shims between the ruler and the rods. I'd then take the stack of paper pieces out and measure them with the calipers. Then I would adjust the rod heights and repeat. I was able to get them within 0.1mm this way. The pringle problem remained though. I then tried taking one of the screws out (back right), leaving 3 points defining the build plate's plane. This helped marginally, but the front left and back right corners were still about 0.05-0.1mm higher than the front right and back left. This makes sense because that's about as level as I could make the y-axis rods. I just can't figure out a better way to level them. What a pain in the ass. Update here.

I decided to just go with that. For small parts in the center of the bed, it shouldn't matter that much. The new glass+thin PEI will be worth it, right? I immediately had first layer adhesion problems. I had no adhesion problems with the old build surface. What the ****. The print will start out seemingly fine, but then if the nozzle bumps a slightly raised portion of the part, the part pops right off. I can manually pop them off of the heated build surface without much effort at all. Tried heated bed settings from 60-70 C (PLA). I always clean the surface thoroughly with 99% rubbing alcohol.  I tried sanding the PEI with 3000, 1500, and 1000 grit sand paper...didn't seem to help much. The whole point of using PEI on glass was to avoid using adhesives, so I didn't want to go that route. I still haven't solved this problem. I'll just try using rafts for now. Maybe the PEI I bought isn't actually PEI...no good way to check that. I may abandon it and either try adhesives on the other side of the glass or ditch it all for an ultrabase plate.

Then I started having extruder problems, specifically problems with random but infrequency under-extrusion, and no retraction. 

Ignore tower that popped off...bad bed adhesion as mentioned above. The angled blob towers indicate poor retraction and overheating. At least I know the cooling is working well since the blobs solidify before they can droop.

The retraction command would be sent, and I could hear the motor move, but the filament wouldn't move. I tested this with just the extruder position menu option, as well. It would start to retract after after about ~5mm, which is odd. Why was there suddenly a 5mm deadband? I took the extruder apart. Turns out that the extruder gear's grub screw had come out enough to cause the gear to rotate, but not all the way around the shaft because it would catch on either side of the flat spot on the shaft. This was causing the filament retraction (and probably random extrusion) problems. I put it back in with loctite.

During one of the many test prints, I tried pausing the print (under SD card menu) to clear a blob of plastic. With the current firmware, "pause" means move the X axis to 0 and y axis to 200mm (forward). Normally, this would be fine, but I had replaced the y-axis belt and left the end sticking out past the zip tie about 5mm too long. This causes it to jam in the y-axis idler pulley and skip steps. Resuming the print results in a y-layer shift. UGH. I had to flip the printer over, cut the zip tie, cut the belt shorter, and replace the zip tie. Take away: make sure your belts aren't too long.

I noticed that whenever I use the SD card menu (mount, print file, etc), that the extruder fan spins a little faster. I think this is related to the ground/power issues with the control board that ships with these printers. While no one has reported the fan thing, people have reported temperature fluctuation and reset issues. I may look into that some.

Near term to do: 

1. Fix first layer adhesion problem. Update: Actually, it seems this is ok. Anything with a very small contact area needs brims/raft, but most things seem to be sticking adequately.
2. Fix bed level problem

3. Reprint spring cups
4. Figure out weird power/grounding SD card thing.

I'm probably up to about 100 hours of work spent on this printer. It's all stubbornness now...definitely NOT monetarily worth it if factor in time spent.

My impressions of 3D printing so far: 

• Most 3D printers are NOT precision machines. 
• They aren't designed to be squared and leveled easily, they aren't made from precision machined parts, most of them have some sort of fundamental design flaw (or more than one).
• Stepper motors are not great substitutes for real servos
• The cheap ones are just barely usable. 
• It takes a fuck ton of work to get them to print anything, let alone print well. 
• Don't buy a cheap 3D printer, especially used, unless your time is worth nothing, you really like fixing things, and/or you're a masochist
• About half the things on thingiverse don't work, and about 75% have some sort of flaw. 

These things are what you pay the extra $1000 for a high end hobby printer and $10000 more for a professional printer.