Bonded Heat bed for 12"

[lakester]

Now THAT illustrates the point..., thx!

Is that just the standard 24mm cube?

Here is a picture, I am sure you can guess what one is with the heated bed on.

Attachment 7083

These two were printed with the exact same settings on both, except the one on the left is with the heated bed on the one on the right is with the heated bed off. As I mentioned earlier even with the bed off I still have some periodic banding at .8mm intervals.

[lakester]

Chadd: Another couple of questions:

Is the banding effect equal on the 4 vertical faces? Or is it equal on opposing faces? Just wondering if it's dominant on one axis or the other.

The other question: is your heat bed "pre" or "post" stiffener upgrade? (The ribs underneath the bed that Colin added after the warping problem became really apparent).

I ask because when I look at the 24mm cubes I first printed, they don't show anything like the effects you're showing. BUT..., not too long ago, I went around and did a general tightening of things, in particular the belts and x-carriage, and ended up with some artifacts I didn't have before. I haven't tried reprinting a cube since "the tightening"..., but I wonder if it would show something similar to what you're seeing now...

[Chadd]

Yes the banding is the same on all four sides.

I have replaced the wood heat bed mount with an aluminum one, my stock wood plate would warp so bad that the rollers would come in contact with the cross brace at the back of the printer.

Chadd: Another couple of questions:

Is the banding effect equal on the 4 vertical faces? Or is it equal on opposing faces? Just wondering if it's dominant on one axis or the other.

The other question: is your heat bed "pre" or "post" stiffener upgrade? (The ribs underneath the bed that Colin added after the warping problem became really apparent).

I ask because when I look at the 24mm cubes I first printed, they don't show anything like the effects you're showing. BUT..., not too long ago, I went around and did a general tightening of things, in particular the belts and x-carriage, and ended up with some artifacts I didn't have before. I haven't tried reprinting a cube since "the tightening"..., but I wonder if it would show something similar to what you're seeing now...

[lakester]

So your "y table sandwich" is the following, from bottom to top?

• aluminum
• cork(?)
• heat pad
• glass

I wonder if that "system" is just simply more responsive/compliant than with the wood. It fixes the big warping problem, but is more prone to the cycling error.

FWIW, I looked at the hysteresis management code in Marlin. I'd swear that my system operates exactly like the code would have it, if it were enabled..., which it supposedly isn't. Just the same..., it's a very dumb implementation, and not particularly tuneable..., doesn't seem worth fiddling with.

Enabling PID on the other hand seems like the way to go, maybe especially with an aluminum table.

Yes the banding is the same on all four sides.

I have replaced the wood heat bed mount with an aluminum one, my stock wood plate would warp so bad that the rollers would come in contact with the cross brace at the back of the printer.

[Chadd]

So your "y table sandwich" is the following, from bottom to top?

• aluminum
• cork(?)
• heat pad
• glass

I wonder if that "system" is just simply more responsive/compliant than with the wood. It fixes the big warping problem, but is more prone to the cycling error.

FWIW, I looked at the hysteresis management code in Marlin. I'd swear that my system operates exactly like the code would have it, if it were enabled..., which it supposedly isn't. Just the same..., it's a very dumb implementation, and not particularly tuneable..., doesn't seem worth fiddling with.

Enabling PID on the other hand seems like the way to go, maybe especially with an aluminum table.

Yes that is how mine is setup. I had the same banding issues with the original wood plate also, but if I had a long print the wood would warp enough to cause the Y axis to crash into the rear support.

[tsteever]

At the beginning of this discussion, I understood what was being discussed...but now I am loosing grasp of the concepts. I too have the Aluminum bed to combat the warping and I have also have the banding issues.

What is the hysteresis being referenced?


To start my understanding, would the current output on the Rumba that opens and closes the relay also be used to control the SSR? Or are other changes needed?

[BLKKROW]

I have also been printing PLA with the heatbed off and my banding issues have disappeared. After I finish a set of prints for a friend I will be switching to the aluminum bed to see if that fixes everything.

I am really interested in PID and will be watching this thread closely.

[Chadd]

I have also been printing PLA with the heatbed off and my banding issues have disappeared. After I finish a set of prints for a friend I will be switching to the aluminum bed to see if that fixes everything.

I am really interested in PID and will be watching this thread closely.

Switching to an AL bed alone did not fix the issue for me, it did help but didn't take care of the issue 100%. Again I think it is several issues stacking up causing the problem.

[Chadd]

To start my understanding, would the current output on the Rumba that opens and closes the relay also be used to control the SSR? Or are other changes needed?

To my knowledge yes the current output should be able to drive the SSR using PID. That's my plan at this point in time and I will find out for sure as soon as my SSR gets here.

I also think that the Aluminum Bonded MK3 heat bed that I mentioned at the start of this thread would fix the issue and I would like to swap to one but, the issue with that is that I can't find anyone carrying the MK3 bed in 12" and it would require ordering it directly from China and to order a single bed it would cost ~$100. I am going to have a $1500 makerfarm by the time I am done trying to fix the issues with it.

[printbus]

...would the current output on the Rumba that opens and closes the relay also be used to control the SSR?

As Chadd said, yes the existing output should be fine. The heater is still controlled by the controller output toggling in a digital, off and on way between 0V and 12V. All the discussion is in regards to the software algorithm behind how that output is being toggled.

"Bang Bang" is like your typical dumb thermostat. Heat stays on until an upper threshold has been met and heat will then shut off. Heat will stay off until a lower threshold is met, and heat will then turn on again. Hysteresis is the difference between the upper and lower thresholds. Simple and easy to implement. I don't know what the relationship of set point to hysteresis is in Marlin, but I assume the desired set point would be in the middle of the hysteresis window. One problem with bang-bang is that some amount of overshoot and undershoot is a given. For example, when the lower threshold is met and the heater is turned on, it'll take a bit for the heat to make a difference. In the meantime, the temperature continues to drop off, leading to undershoot.

PID adds smarts to this, using a model that "predicts" what it will take to maintain the temperature. The algorithm might turn off the heater a bit early as the threshold nears, knowing the temperature will still creep up a bit to reach the upper threshold. Likewise, it'll turn on the heater a bit early before the lower threshold is hit in order to keep the measured temperature from undershooting too low. It can even start to toggle the control output as threshold approach. "Tuning the PID" involves the software characterizing the system typically through a number of heating and cooling cycles. The tuning results in some coefficients that can be plugged into the predictive algorithm to optimize it for the specifics of your printer and environment - heat quickly but without a lot of overshoot.

Applying Pulse Width Modulation (PWM) will refine this a step further. Instead of crudely turning heat on and off early in advance of reaching a threshold, PWM can constantly toggle the heater on and off at a comparatively fast rate. "More heat" is obtained by increasing the ratio of on-time to off-time. Cooling is obtained by decreasing the ratio. This is no different than using PWM to control LED brightness. The LED is always either fully off or fully on. "Brighter" is obtained by more on-time. Dimmer is obtained by less on-time. If the PWM frequency is fast enough, the eye doesn't see the LED being constantly toggled on and off, and what you get is an "averaged" brightness level. Now leverage this to a printer. Toggle the heater on and off fast enough, and what you get is an averaged amount of heat.

I don't know what options Marlin has, but in a generic sense use of PID is not necessarily tied to using PWM. For example, I know the temperature controller on my green egg smoker is PID based but it does not use a PWM to control the speed of the airflow fan on the firebox. Marlin might have a similar way of applying the PID algorithm with a mechanical relay, but the SSR is going to be requirement if the PID algorithm inherently includes using PWM to toggle the output at a high rate.

DISCLAIMER: I may have some the specifics less than perfectly accurate, but this should help with some of the fundamentals involved.