Anyone running a 32-bit controller on their Mendel90?

Hi,

After periodically monitoring a print over an 8 hour period I turn my back and the printer stops with 30 minutes to go.

I have a stalled hotend with a heated mass of goo around it and the cooling fan still running. Pronterface cannot communicate with the printer - it says it is disconnected.

This is the first print I have attempted of this duration. It was running from SD-Card so USB communications should not be the problem.

Printer itself seems fine so I suspect operation hit a Marlin bug and printing stopped.

Has anyone any experience of 32-bit controllers with something other than Marlin? Smoothie seems to be gaining momentum.

Regards,
Neil Darlow

---

I try to write with consideration for all nationalities. Please let me know if something is unclear.
Printing with Mendel90 from fedora 25 using Cura, FreeCAD, MeshLab, OpenSCAD, Skeinforge and Slic3r tools.

Hi Neil

I have been using duets with our flavour of Mendel 90 for a couple of years now. The work that David Crocker (DC42 on these forums) has done on the RepRapFirmware has continued to expand the functionality and quality of prints. My blog post of using the Duet +Duex4, while a bit old now, is the first time I talked about using it for multi extruders:

[blog.think3dprint3d.com]

That said we routinely do long prints on the RAMPS based Mendel90s as well running Marlin, which we are yet to upgrade to Duets. Have you recently upgraded your Marlin becuase we are still using an old version that prints great.

Cheers

Tony

I have never had any build failures due to Marlin on Melzi running a couple of machines constantly for more than 3 years. I regularly print 25 wades small gears at 100% file and that takes more than 8 hours.

The latest firmware I have run is this one: [github.com]

The only problem I have noticed is that my machine that has a Panalolu2 drifts in speed. I.e. it sees encoder movement in the top level menu, when there isn't any. I have disabled the LCD on my machine without and that is the one that does the longest builds.

If you run from SD over the long P2 cable it may be prone to EMI and corrupt SD data might upset Marlin, or perhaps you might be using a too recent version. I.e. after the point it started getting massive re-factoring and became totally unstable.

---

[www.hydraraptor.blogspot.com]

Hi,

My current version of Marlin was prior to the move to the new repository so doesn't relate to the refactoring efforts.

I too have seen feedrate drift on the Panelolu2 and I too have rebuilt with Panelolu2 disabled to determine if it is a factor.

The SD card is inserted into the Melzi and, in fact, the Panelolu2 is disconnected from the adaptor card.

I am now experimenting with the Teacup firmware as a minimal implementation to isolate the hardware. So far I have done a 5.5 hour print over a USB connection without issue. I need to run this setup a little more just to be sure.

Regards,
Neil Darlow

---

I try to write with consideration for all nationalities. Please let me know if something is unclear.
Printing with Mendel90 from fedora 25 using Cura, FreeCAD, MeshLab, OpenSCAD, Skeinforge and Slic3r tools.

Hi,

I was using smoothieware on a azsmz with silentstepsticks for some month. I love it and want it back.

Before azsmz I had ramps with glcd. Was using the newest Marlin and did a lot of long prints (most of them lamp shades) with up to 16h.
My frame for the mende90 (you can see it on thingiverse) was only one plate and it took 24h.
Barely issues with long prints with ramps and Marlin.

Also with smootieware I did prints up to 19h.
Suddenly, two weeks ago, there were resets after around 1-2h of printing. Same like you had.

I had 5 plates to print. And all of them were between 6 and 10h. I could finish one of them and tried all of the rest without success.
Everytime the same - a reset after 1-2h. Normally I use octoprint on a rpi2 without any issue. But in this case I tried also to print from sd card with the same result - reset.
I removed the fuses, because there was once a similar issue that was related to one fuse. Again the same.
I was thinking the azsmz is dying.

So I put the ramps back and finished the prints.

But there are still issues. The extruder temperature is stable for hours and suddenly it jumps around for 10-20min in a range of about +-20°C.

So I changed my mind. Not sure if azsmz is the problem. It is only acting different. I am still searching. Either there is a unstable power supply or contact problem on ground or on any other cable.
The mendel is not a kit from nophead. I will replace the suspicios parts one after the other with the goal to get azsmz running again. The extruder connect PCB ist one and after that the power supply after rechecking all the power and ground cables.

The reason to leave ramps was the noise.
First step was to use drv8825 with 1/32 steps hoping it will help. But on the contrary it grew louder.
Periodically there were slight shifts (about 1-3mm) mostly on the x-axis, but sometimes also on the y-axis.
It is clear that there can be a lot of hardware related reasons. I think I am aware of most of them. Sure not all.
But turning the current of the stepper driver up was ending with even more noise and the whole machine shaking.
The silentstep drivers made the change. Even with ramps it is fine now.

But still I want azsmz back. It is again more silent. Very nice to configure ...

Lars

Edited 1 time(s). Last edit at 09/16/2015 12:56AM by danielis.

Yes power and ground connections of PSU would be my first suspicion. Also make sure you have auto reset disabled on all your boards as you don't want a USB disconnect to interrupt an SD print.

Look for burnt plastic on the thermistor connections. That can make the temperature inconsistent. If you also have burnt plastic on the heater it can feed 12V into the thermistor inputs and that might crash the CPU.

---

[www.hydraraptor.blogspot.com]

I received the PCB from you. Thank you for the fast delivery.

You were right, there was some melted something on the termistor wires. Although the nozzle was wrapped with several layers of Kapton and several Layers of PTFE (it was not the red silicon? self fusing one - but now it is there).
But I am wondering, because after removing the layers of PTFE and Kapton the head was clean except a small black something on the termistor wires directly on the whole where it enters the head. I could swear it was clean as I installed it and the isolation was there going far inside of the whole up to the glass of the sensor.
That cannot be, but it looks like the silicon-isolation was melted on the corner of the hole where the wires bends where they leave the head. No idea how that could happen.

The resistor wires were clean and safe isolated.

But sill, I did some measurements after cleaning the wires. With a oszilloscope I found peaks on the head up to 6V. The resistor induces that voltage.
By design it behaves like a coil as it is a wire-wound ceramic. And every time when the controller switches the current on or off it induces those peaks in the brass of the head.
Seems that a 5V CPU can tolerate those more than a 3,3V CPU.

Then I reinstalled the azsmz, replaced my PCB with yours and checked all wires again. I added a ground wire directly to the sensor terminals.
Since then the printer is working fine again and did successfully a 12h print.

Thank You for your help.