I meant to share this with the community… Not just my printer circles.
Originally shared by Eric Lien
PSA: I recently ran into issues with my heated bed which required replacement of the thermistor inside my alirubber silicone pad. After replacement I was having trouble running the Marlin PID Autotune feature because of failure of the bed to reach 110c quick enough. So PID tune would timeout. So I modified the firmware back to bang bang but forgot to enable hysteresis. The part on the left is with bang bang.
After looking at the part I re-enabled the heated bed PID in firmware, changed PID tuning target to 90c, and saved the resulting PID values. The print on the right is the identical gcode, on the same printer, only hours after the left print, but with PID tuned.
Moral of the story. Not all z ribbing is from mechanical moving components. It is possible to get a similar effect by heated bed overshoot/undershoot cycles. Since the bang bang ossilations vary the thermal expansion it causes changes the bed to nozzle distance, if you only look at the mechanical parts you may miss the root cause of your z-ribbing.
That is a very noticeable difference. Also the PID is typically set at a 10 degree swing on either side of the chosen temp. That can be changed to what ever value you want say a 20 degree variance. This will give more room for the PID to settle down.
That’s really smart. It should be possible to do a bed design that doesn’t have that much vertical travel with temperature, but if you can fix it in software why bother.
@Mark_Rehorst it is almost certainly the bed. My bed is almost 17"x17", so it is likely the large size of the bed, hence large span that amplifies it so much. Nothing else changed. They are uniform because without hysteresis enabled swings are large and cyclical. Also I print very fast. Both prints are at 150mm/s (and with acceleration at 10,000 that is a sustained speed). At high speeds there are multiple layers between swings. Hence ribs across many layers in a repeating pattern.
It’s not the plastic expanding. It’s the aluminum that is expanding, possibly even crowning on the span when the bang bang hits. Aluminum has a large expansion per degree C in comparison to plastic. It would crown since the edge of the plate will be cooler than the center until some thermal equilibrium happens. But equilibrium never happens due to the cycling heater temp.
@paul_wallich I am thinking of going back to my 1/4" thick heat spreader when my 800W 120v heater arrives. I currently have a 1/8" spreader now because the thermal mass of my 1/4" took too long with a 24v/400w heater. I ordered it on Monday from alirubber, as well as some high quality Omron SSR’s I found on eBay.
Just for clarity on the setup of my heated bed, here are some notes.
From top to bottom.
Top layer = 4mm glass.
2nd layer = 1/8" heat spreader.
3rd layer = 350mm x 350mm 400w / 24v silicon heater.
4th layer = 1/2" cork insulation (encloses all 4 sides of heater pad and the bottom).
5th layer = 1/4" Masonite hardboard to sandwich cork to heater.
Total bed size is 17" x17". It is mounted with 3 point leveling using long bolts that act to hold the sandwich together and retains the leveling springs.
The PDF instructions that were online for my printer said to calibrate the Z height after the heated bed and nozzle were heated to temperature.
A piezoelectric speaker heats up the center quickly and it cools quickly distorting material and causing the center to move due to the heating and shrinking.
What does this mean? Our heated beds have too much in common with a type of speaker. Also, precise heated bed temperature control is important.
@Maxim_Melcher Yes the hot end is rock steady regardless of the bed varying or not. This could happed if my PSU was undersized and the bed pulled too hard and effected the hot ends ability to maintain temp. But this is not the case.
@Eclsnowman have You tested a power supply with more power? Additional , make the 0.1uF condensator parallel to termistors, this should to give more precision for temperature measurement.
@Eclsnowman the power supply can be defect after one year. And I see, this is a trouble with filament temp, and not bed temp. Please, make the hot-end +10c for test
@Maxim_Melcher I will test later. I have several work prints to complete, so since PID is yielding good results I need to keep the settings to get through the backlog.
I wanted to add that I’ve noticed this effect is varied with the properties of the filament you’re using too. I’ve tested quite a few different filament vendors and some of the less viscous ABS I’ve used is much more prone to the ribbing from temperature differences.
