Successful calibration with 626Pilot Smoothieware + FSRs!

[kiiwii]

Since I first built my RMAXv2 last July, I was unable to ever achieve a usable calibration that allowed a decent first layer on the full extent of the bed. There would always be areas between the towers that were so close to the bed that no plastic would extrude, and other areas that the plastic would lift off. Adjusting the endstop screws was a huge pain. Basically, every print was hit or miss. I was able to print small objects in confined areas of the bed, and I often had to try the first layer over and over, sometimes tolerating messed-up first layers hoping that further layers would cover up the errors. (Bad idea--often, the errors actually accumulate higher up.)

I hadn't touched my printer in a while since it was so frustrating and was even considering getting rid of it. But after seeing the great results that 626Pilot's software has been giving people, I decided to give it a shot. I upgraded to the Smoothieboard, got the FSRs and endstop controller kit, and printed these FSR mounts by rpress/Nylocke, and installed 626Pilot's firmware. After much experimentation and tweaking I was able to get "average" to "good" repeatability on the probe and ran calibration.

Like magic, I was able to print out the Onyx Bed Leveling Aid for the first time. There's still one area between the X and Y towers is a bit thin. But no more plastic lifting off! And now I longer have to worry about the first layer not sticking; as long as I put down glue stick I'm fairly confident the first layer will come out OK.

If anyone else is having chronic difficulty with calibration and first layer evenness, I highly encourage upgrading to the Smoothie and trying out 626Pilot's firmware. It's the only thing that's been able to cure this problem for me.

--------------------

Some notes I took along the way:

I initially built the magnetic Z-probe that 626pilot designed. Mounting it on the side of my effector was completely useless for calibration, though; the sampled points were all shifted on the bed and it was not able to converge on an improved calibration. I didn't want to swap out my hot end to calibrate.

Instead, I went with the FSRs, which replace the nylon spacers that hold up the snowflake. Printing out the mounts without a well-tuned printer was tricky. First, I was having trouble printing dimensionally accurate parts. The fix was to 1) calibrate E-steps (mark filament with Sharpie at precisely 100mm, remove Bowden, scale the E-steps so that 100mm extrusion command moves from mark to mark), and 2) reduce perimeter width in Cura (I used 0.43mm shell with my stock 0.5mm nozzle). The reduced perimeter caused holes in the outer walls of the mounts, but it was the only way I could get the pieces to be close to fitting. Later I found that KISSlicer produced FAR more accurate parts than all the other slicers I've tried (Cura, Slic3r, Craftware).

I couldn't find #4-40 tee nuts for the upper mounts at the hardware store so simply drilled out the upper mounts and press-fit regular #4-40 nuts. They don't have to be super tight anyway. In the end I needed 3x #4-40 x 3/8" flat head screws to attach the lower mounts to the base and used the original #4-40 x 3/4" screws through the Onyx, snowflake, the nylon spacers, and the upper mount. (#4-40 x 5/8" screws would avoid the need for the nylon spacers, but I forgot about the snowflake when I made my hardware store shopping list, so had to improvise. The spacers don't hurt though.)

[img]http://i.imgur.com/O1zZbCJ.jpg[/img]
[img]http://i.imgur.com/6RfvU8U.jpg[/img]

After the FSRs were installed, I noticed that it was taking A LOT of pressure to trip the FSR switch. Enough to displace the effector platform, flex the bed, and slightly move the bed when pressing in the corners of the bed between the FSR mounts. I couldn't get decent repeatability with any combination of probe smoothing, speeds, and FSR endstop controller sensitivity.

I debugged this by removing the hot bed and pressing on the FSR mount's plastic "puck" with my finger: a lot of pressure required. But when I removed the puck and pressed on the FSR directly with my finger, it tripped with minimal pressure. Looking at the FSR sensor, only the CENTER portion has traces printed! The puck evenly distributes pressure, including to the edges, where the FSR is not sensitive.

The fix: I put a dab of hot glue on the center of the puck to concentrate pressure in the middle of the sensor. Now, pushing slightly on the bed with my finger causes it to trip. No more bed movement during calibration.

Here are my final settings:

FSR endstop controller: jumper set to 0.92 (max sensitivity).

626Pilot firmware:
smoothing: 3 samples
priming: 1
fast speed: 70mm/s
slow speed: 3mm/s <-- very slow! but gives me better repeatability
probe height: 5mm
no runout deceleration (since the hot end is already touching the bed when probe triggers)

Repeatability (tested with G29) is often within 1-2 steps, "average" to "good".

I auto-calibrated with G32 (geneb-style calibration), followed by annealing (G31 O P Q R S), and depth mapping (G31 A).


Test print: (flipped because my fan mount is on the right)
No adhesion problems anywhere.
[img]http://i.imgur.com/HJJlDES.jpg[/img]

[forrie]

Awesome..will have to give this a go on my new smoothie build. I was planning to mount a zprobe off to the side of the platform but after lots of reading I'm now not so sure.

[forrie]

Had a thought....

Would you mind posting your final config file for the masses?

[kiiwii]

Here you go: https://github.com/iceboundflame/3dprin ... hie/config

My KISSlicer settings are also in that repo.

[forrie]

Awesome....thanks

[forrie]

A question...

I have an older version of the V2 and the top plate doesn't have screw holes for the snow flake in front of the towers.
Would it matter if the FSR's were mounted offset from the towers? I suppose I could drill some holes if need be but at this stage would prefer not to.

[kiiwii]

It should be fine; the FSRs will sense slight pressure anywhere on the plate.

[jamesm]

I just finished setting this up on my smoothieboard using the same FSRs... the results are spectacular. every delta machine should have this capability. thanks!

[gestalt73]

One of the things I've been curious about for this type of FSR mount, is how "floaty" the heated be ends up being.

Does this type of solution constrain lateral movement? I'm worried that my print quality would suffer as a result of the bed not being affixed to the printer.

It looks like the bed is mounted on three points, and only by a puck that sits in a recess.

Also, how long does the calibration process take once it's kicked off?

[Nylocke]

He's using my "old" design (there isn't yet a new design, but I still consider it old), and I still have this on my printer. I haven't experienced any movement of that sort while printing, it pretty much just sits and prints.

[gestalt73]

Heh, sounds like fun. Just ordered two kits, one for each printer.

Thanks!

[miglo]

Got mine working with Smoothieware + 626pilot firmware + FSR + nylocke's FSR mounts. Thanks guys!