A5S hybrid

Earthbound

Background:

I bought a damaged A5S on ebay and have been slowly getting it back to working condition. It had some shipping damage to the X/Z gantry tower which was easily fixed. But then it became apparent that there were electronic issues as well. The USB connection was dead - I found burned traces on the controller PCB, and a cooked IC. The touchscreen worked and I could move/adjust, but it would not print from files on the SD card. I tried updating firmware first via SD card and later by ST-link, but board would always refuse to boot if anything other than the original firmware was loaded. Tried several more steps to reprogram and eventually bricked it.

Decision:

I chose to replace the controller with something other than another JGAURORA Main Board. Based on price and features, I went with a BIGTREETECH SKR 1.3 board and TMC2130 drivers. It is obviously not a direct replacement and I have had a few challenges but no show stoppers. Since the machine would be substantially dismantled, I am doing other upgrades at the same time.

Progress:

I have the SKR board mounted and most of the wiring connected - everything except the hot bed and its thermistor since the top of the base is still off.

I've researched the possibility of reusing the stock touchscreen but my conclusion is that it would require some support hardware to be used with most aftermarket controllers. One of the unusual features of the stock A5S controller is that it serves as the memory resources of the screen. For now I am using a RepRap Discount Full Graphic Smart Controller because I already had it in a drawer.

I've replaced the linear bearings in the Y-axis (X will get the same treatment) and the new ones are much smoother (cheap ones found on Amazon).

I hope to have it operational within the next couple of days. Pics to follow.

Earthbound

My first go at a mounting bracket. Retained most of the vents on the bottom of the case. Gave plenty of space to bring wires into the screw terminal blocks. SKR board would sit as low as possible in chassis.

But it was not to be. The main 24v power supply leads would not reach the board and wires for the filament sensor were a bit short as well.

Earthbound

The next try crowds the SKR board quite close to the rear-most chassis post. To allow clear access to terminal blocks, I had to raise position of the new board a few millimeters.

Earthbound

I didn't get a pic of the 2nd tray in place before I mounted the SKR. Interim graphic display is lurking in the corner temporarily. All wiring to the SKR board is complete in this pic except for the bed's heater and thermistor.

Earthbound

Front of tray will hold 40x10 fan for airflow across driver heatsinks. Fan due to arrive later today.

Earthbound

I fitted TMC2130 drivers because I wanted to experiment with sensorless homing. I don't know if it can be dialed in well enough to use for automatic bed leveling, but I plan to try it. If it works, then in the long run, only Z axis would need to have a TMC2130 and the others could use cheaper alternatives.

I read something that suggested there was a performance issue when running an SPI controlled touchscreen and TMC2130 drivers (which use SPI for communication). Will need to play with that when the TFT35 arrives in January.

eyes

I had that board in my printer when I thought my JG board was dead. I figured out what was wrong and have the original board back in just because I didn't like the RepRap Discount Full Graphic Smart Controller hanging out of the front. I was using TMC2208 drivers set in UART mode. I like that I could adjust the current via G Code without having to open the printer up. It was also definitely quieter. I think you can adjust your driver currents the same way with the steppers you have in SPI mode. I also like that it has a dedicated servo plug so BLTouch is an easy upgrade if you can't get automatic bed leveling to work. If you do please post how you do it! I would recommend adding an SD card extension cable so you will be able to flash new firmware without having to take it apart. I used this:

https://www.amazon.com/gp/product/B01D9JIUU0/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

If you haven't already figured it out, you will need to use different steps/mm on the drivers if they work the same as mine. I have mine set at:

#define DEFAULT_AXIS_STEPS_PER_UNIT   { 80, 80, 800, 405 }

Your extruder will be different. 405 is for a BMG extruder.

If you are interested in doing the extruder cooling fan mod I posted (fan off until extruder reaches 50C) let me know because it is very easy on this board and no additional parts are needed. 

I really like your mounting bracket. I can only design in tinkercad and never attempted to create anything for it. Would you mind sharing it?

I don't know if the fan on the end will cool all of the steppers well enough but I like it. I would watch the axis furthest from the fan for problems. Before I knew cooling was an issue with TMC's on another printer it printed crazy when they got too hot so I think you will be able to tell if temps are an issue.

I have the TFT35 v2.0. It needs a bootloader fix that bricked the display when I followed Bigtreetrechs instructions for uploading it on their github. What version did you order? Or is it a different brand?

It is larger than the original so the front plastic would have to be cut away some and a new mount would need to be made. From the looks of your mounting board, it looks like that would be something you would be able to do. The stock display mount is just screwed in place so it looks like it would be fairly straightforward for someone with graphic design knowledge. 

Earthbound

I've attached an STL for the revised tray design.

I enabled hot end fan control as part of the initial configuration. Will be exploring options for a quieter fan up top. Man, that thing is loud.

I will be monitoring driver temps to see if the tray fan is adequate. Next stage would be a duct/shroud to keep airflow directed over the heat sink fins. The little fan(s) that I ordered showed up yesterday but were not what I intended. Reordered immediately (thanks Prime!) for the correct size and as added benefit this time found a 24v version with long leads - useful since the 24v terminals are on the opposite end of the board. The original plan had been to use 5v power available at an empty end-stop connector.

Thank you for the link to the SD extension. Since the SKR board's SD card appears as a drive over the USB port, I don't need physical access to the card slot.

I wrote TFT35, but that was on my mind from another printer project. For this machine I ordered a BIGTREETECH TFT28 which should be here in about a week. The RepRap Discount Full Graphic Smart Controller is only a means to keep moving ahead until the touchscreen get here.

I have it all assembled again and have started running through some tests. Heaters and thermistors are responding. X/Y/Z all move - Y only after I went back inside and plugged in the motor. X and Y steps look good. Still using the end stop switches for those. Will start dialing in the Z axis next.

eyes

Thanks for the link. I have 2 new cooling fans I just got. I'll let you know if there is a difference. Both ball bearing and one is the "quiet" version. 

Earthbound

Curious. The part cooling fan is nearly silent at full speed. Can hear some PWM noise from it at lower speeds. The hot end fan must be a different model.

eyes

The hot end fan is a regular 30mm fan like on an E3D V6 hotend. The part cooling fan is the kind where the air comes out the side. Its called a blower fan. And that type is usually louder. I just bought both fans. This is the new part cooling fan. Mine now sounds horrible. And jgaurora already replaced both of them. 

I'll let you know how this one does too. 

Earthbound

Today's project was creating a clean way to mount the RepRapDiscount Full Graphic Smart Controller. It may be temporary, but it was a useful exercise. I decided on a 3-piece solution. First, a back plate that would use the original display mounting points. Then a front plate that would attach to the rear plate. Lastly, a housing for the new display that would clip onto the front plate.

Here is the back plate, installed. Reused the factory screws.

The opening in the plastic overlay is not centered between the standoffs!

Here is the front plate. Couldn't get a shot that clearly showed the beveled edges at top and bottom, but trust me they're there. This mounts to the back plate with 4 M3x10 screws

Rear view of the housing. Not planning to use the SD card slot, but could add a opening for it.

It's not the prettiest thing ever, and it sticks way out. Needed the clearance for the cable ends, since they don't line up within the opening of the metalwork.

 But hey, that bed sticks way out at high Y values, too.

I'll probably make some small changes to the inside edges of the housing and print it again in black. There is a lot of light bleed-through from the sides of the display. Or, could probably just mask the bleed with some tape or paint.

Earthbound

The files for the plates and housing, if anyone cares to play with this.

Earthbound

And of course, the money shot.

eyes

Very cool thank you. I will definitely use this if I change back to the skr v1.3 board

Earthbound

Won't be using the sensorless homing feature of the TMC2130 drivers. It would work for homing X and Y, but my desire was to use it as a bed leveling sensor on Z. I learned some things while trying to set it up. The "accuracy" of the method varies with travel speed, and at its best is only so-so. The slower the travel, the worse it gets.

I will try the IR reflector sensor instead. Hopefully my last trip inside the chassis for a while.

Install the cooling fan.

Unplug the Z limit switch, install IR harness in its place.

Swap all of the TMC2130 drivers with more economical TMC2208. Check/move jumpers if needed.

Earthbound

One other to-do item while it is open - check the Y axis.

I swapped out the Y-axis linear bearings and the new ones felt much smoother. But after everything was reassembled and I ran the axis by hand, it felt chunky. Full travel under motor power is ugly. It might be something silly like the wire harness catching on a belt tooth. Must be resolved before any kind of print test is possible.

I had planned to replace the X axis bearings but after beginning to disassemble it, I found the bearings to run quite smoothly once the belt was disconnected. The roughness might be poor fit between belt teeth and plastic pulley. More to investigate.

Earthbound

Test print underway!

Tightened the Y axis belt a bit which improved its smoothness.

Getting the IR sensor installed and working was simple enough. Getting it adjusted to work well is less simple. Getting the automated bed leveling to a point where it is useful is a work in progress.

Need to play with the motion settings for the Y axis. The table mass really shakes the machine with its current values.

Earthbound

Test print completed. Looks awful. It's not the worst Benchy I've ever seen, but it is really, really bad.

Some of this is because the spool of filament I used for the test is terrible. It was cheap stuff from ebay purchased because I needed a particularly color for something 2+ years ago. The leftover has been sitting unwrapped in my basement since then. I threw it away today because it kept breaking every time I tried to unspool enough to load the machine. The moisture content got too high and it became extremely brittle. I have supply of the Hefty 2.5 gallon bags now for storing partial spools.

The bed adhesion was a little too good. After cooling to room temperature, the bed still had a firm grip on the Benchy (printed without any extra skirt). I wish I had tried warming it slightly in hindsight. I used the brute force method and managed to tear off some of the coating from the bed. The Benchy is magenta, but the bottom surface is gray. Bed has visible scar, but still feels smooth when I run a finger over it.

Part cooling was a factor. The overhangs and bridges needed more cooling than I had set in the slicer. Hot end temp was probably a little high for the old filament, too.

But the biggest concern is the distinct mechanical flaws in the print. The defects vary from one side to the other. This is from either roughness or play in axis travel. X feels smooth and sounds smooth. Y has been giving me fits. It seems tight enough, but sometimes stumbles like the belt is slapping. It doesn't sound smooth. There are things I sure like about the A5S design, but access to the controller and Y-axis is not on that list.

The Y axis trouble proved to be of my own making. In the process of swapping the linear bearings I failed to tighten two screws. Specifically, the ones which anchor the Y axis belt to the carriage. I must have just put them in finger tight. One had fallen out - I spotted it sitting in the wiring bundle a couple of days ago. I did not realize its origin and thought it was left behind from work done on other parts. The anchor bracket is also the part that trips the Y limit switch, so when the bracket got loose enough, it pivoted and missed the switch. When the Y axis tried to run off the back of the machine and stalled the motor, I investigated and fixed the issue. Now the Y axis is smooth and quiet again when moved by hand. Need to dial in the acceleration and jerk settings before I'm done.

Earthbound

2nd Benchy (with recently opened high quality filament) looked really good, right up to the tops of the doorways and windows. The print broke loose from the bed. Not enough adhesion (no skirting) to withstand the hotend striking an overhang that had curled upwards and hardened. But it looked darn good up to that point!

Re-sliced, slightly slower print speed, some skirting to improve bed grip, and placed it off-center away from the bed area that was stripped of coating. Upped bed temp 5C to 60C, and giving it another try.

Earthbound

Adding the IR sensor and abandoning the stock Z limit switch has been more work than expected. The reward should be worth it, though. Getting the IR sensor at just the right height in relation to the nozzle is critical. Maybe there is a way to compensate within Marlin, but I struggled to find something that worked using only the settings. I ended up making a mounting bracket with snug circular holes. The design posted in the IR sensor thread has slotted holes for adjustment, but that wasn't working for me. I made a test bracket first, and then measured the hot end gap to the bed after homing with the test bracket. 2nd/final bracket was made shorter, by the measurement achieved from using the test bracket.

Learned a bunch about SD card support. The SKR board has a microSD card slot onboard, as well as a USB jack. The microSD card was visible as a drive letter when connecting to the controller over USB. This made firmware updates really simple during all of my testing. Even though the card was buried inside the chassis, I had easy access over USB. Notes within Marlin warn you that when you enable SD support, the ability to access the card over USB will go away, so firmware changes have to then be done by physically handling the card. I read all those notes and thought I would be fine... I hadn't yet noticed there were no card related functions on the LCD menu.

You have to enable SD support to have those menu items populate. OK, edit config, make a new build, send to printer over USB, reset and I'm a rockstar with fancy SD menu. Ta-da! Let's print something from SD. Hmm, I don't have any gcode files on the SD card, yet. This is where you realize that the card is deep inside the printer and no longer accessible over USB, so you have no way to put gcode files on it. You need to make further changes to the firmware, except that it too needs to go onto that card deep inside... Frustration grows.

Amazon brings you the microSD slot to SD slot adapter/extenion that user 'eyes' mentioned above.

I used the extension cable a few times, but eventually made more changes and enabled the SD slot of the RepRap Discount Full Graphic Controller. I forget the exact parameter you have to edit, but the value needs to be changed from ONBOARD to LCD to make the magic happen. I printed a new housing for the LCD screen, this time with an opening for the SD slot. I elected to make the wall a little thicker for stiffness and opacity. The new one has no light bleed through the housing, but this difference could be attributed to the different color material. I need to verify that the slot on the LCD can be used for firmware uploads to the SKR board before I remove the extension.

Earthbound

Happy to report the TMC2208 drivers are super quiet. The hot end fan is by far the loudest component. The part cooling fan is 2nd. While printing you can hear a slight growl while Y axis is moving. There is more noise during the travel moves. I may be able to mitigate travel noise with lower acceleration values.

Also just noticed that the drive pulley on the X-axis is out of round. As it turn, you can see the belt rise and fall in relation to the rods.

Earthbound

Strike that. Power supply might be the loudest. Closely followed by the hot end. Can't hear the fan I added for cooling the drivers.

Earthbound

The 3rd Benchy looks good. All of the mechanical flaws are gone. The surfaces look really good. The mild overhangs are fine. The severe overhangs and bridges are terrible. No question that the part cooling fan is nearly useless. I'm off to the Thingiverse to find something new.

Earthbound

Have determined that firmware for the SKR must still be loaded via the onboard microSD slot. The SD support for the slot on the LCD is only for printing of gcode files. 

Earthbound

Samuel Pinches

Niiiiice

Earthbound

My conclusion is that the IR sensor may be fine for probing/leveling, but using it as a limit for homing as well is problematic. All UBL data is referenced to a machine's index point. If that point is on the bed and measured by the IR sensor that has a fixed position in relation to the nozzle, then it doesn't provide a solid reference.

Today I reattached the Z-limit switch and moved the IR sensor to the unused X+ (max) limit input on the SKR board. On my machine, the X- limit closes at about X= -7, which provides enough room for the nozzle to travel below the bed surface.

With Z Safe Homing enabled, I adjusted the Z- limit switch to trip just before the tapered side of the nozzle reaches the edge of the glass. Now I can get a consistent Z zero before building/using the UBL mesh.

So far, this has given very good results.

Samuel Pinches

I didn't have good results with IR sensor and the coated glass bed, even when using the Z-endstop as my reference. YMMV, and glad it's working for your printer, but its not something I recommend.

Earthbound

Things have taken a turn.

The dead space on the original screen is on the left side as installed, so the factory overlay is cut to reveal only the screen area used. On the BIQU/Bigtreetech screen, the dead space is on the right. I'll need to do a little trimming to see the left edge.

Earthbound

I wanted a screen that would fit the factory front panel without major rework. The BIQU 2.8TFT fit the bill. It has a wider and shorter mounting pattern than the original, but the SD card is positioned to line up correctly.

https://www.biqu.equipment/products/bigtreetech-tft28-touch-screen

I've already started to devise a mounting bracket when I took these pictures.