UM2+ Power Splitter: An Attempt at Improving Wall Consistency

[Porkpie 24]

Many moons ago, forum member [torgeir] reported a phenomenon on his UM2+, where he noticed wall inconsistencies on his prints that were related to effects from the hysteretic ("bang-bang") regulator of the bed heater.
at that time , i was looking into that issue as well and came to the conclusion that,opposed to other's believes, the wall incosistencies did not arise from temperature fluctuations of the heated bed, but were actually caused by voltage fluctualtions of the main 24V power supply. to prove that, i altered the bed temp regulator in the firmware from hysteretic to PWM PID control. this made the wall incosistencies immediately disappear, but at the cost of (and again, pointing at a power supply issue) an ever so slightly and annoyingly flickering LED illumination. i tried to ignore the flickering, but alas, after a couple of months, i could no longer bear it and decided to do something about it.
the idea was to split the main 24VDC bus from the main power supply into two branches: one branch would feed three MOSFET switches that would actuate BED, EXT1 and EXT2 by using the respective ultimaker main board's outputs as mere control signals, rather than as power-delivering outputs.
the other branch would go into a  beefy step-up regulator that would provide a super-tight regulated 24VDC power output to supply just the guts of the ultimaker main board itself, i.e., cpu, display, stepper drivers and all other paraphernalia.

you can see the result of my efforts in the pictures below. the power supply is a fanless UHP-350-24, connected to an AC switch box that is mounted on the side of the printer, so that i have the main power switch accessible from the front side of the printer.
the green-yellow PE connection is also connected to the 24VDC minus, to ensure that 24V-side electronics are not building up charge that could cause a destructive discharge when, say, connect the USB cable to it.
the 24VDC output is connected to the "power splitter" board through the main entry fuse (2 polyfuses in parallel). to the left of the 24VDC entry point is the regulated 24VDC power output for the ultimaker board, also fused with a single polyfuse. on the right hand half of the power splitter you can spot three identical power trains BED, EXT1, EXT2 (EXT2 partially populated, not used), each consisting of a "real" gate-driving opto-coupler, a power MOSFET, a free-wheeling diode, an indicator LED and a polyfuse. BED doesn't have an extra fuse, it is using the 24VDC entry fuse, while EXT1, EXT2 do have their own fuses.
the opto-couplers serve two purposes: they allow for noise isolation (not used here, but can be configured via jumpers), and drive the MOSFETs with sufficiently high gate voltages. in this case 12 VDC, which is generated on-board via the chunky device on the top left of the board. this in turn allows the MOSFETs to operate at 10 A continuously, while getting just warm to the touch.

the two test cubes side-by-side tell the whole story. on the left cube, the center area is printed while bed heater is on, operating in the original hysteretic mode. top area is printed with the bed heater turned off. the right cube was printed with the power splitter installed and bed heater turned on with PID PWM control.

both cubes were printed with Ultimaker Silver PLA, 60 °C bed, 210 °C nozzle 0.4mm, 0.15 mm Z, spiral mode, 40 mm/sec.

 

bottom line and TL;DR:  mind your power supply!

 

p.s.:

- you have seen a similar post from me before, discussing a different main board installation on the UM2+. well, it took me a while to figure out that the board would never live up to my expectations, so it had to leave the building,  r.i.p..

- no, i'm not going to make this commercially available, as the boat for all UM2+ mods has set sail long time ago already. this mod was meant to be a proof of concept only.

 

[CarloK 205]

These are impressive improvements! Do you have an idea why the bed has this impact?

Is it the bed cooling down / heating up, which creates thermal deformation, and so tiny layer height differences?

Or, is the power supply not keeping up? Extra noise induced into the system which causes nozzle temperature to vary?

Or, is it a poor designed pcb where the 5V power supply shares the ground with the bed power ground, and as such gets an induced offset?

Or, other reason?

[Porkpie 24]

i never suspected temp fluctuations on the print bed for one simple reason:

the hysteretic control of the heatbed (i.e., in the original UM2+ FW) switches the bed on and off at deliberate times, but still much faster than the glass bed could follow those switching events. also, at higher build heights, one would expect the influence of the bed's  thermal fluctuations to fully diminish, simply because the printing happens farther and farther away from the oszillating heat source. however, in reality, wall inconsistencies as shown in above picture do appear at any build height, not just close to the bed. that clearly contradicts the "thermal influence of the print bed" theory.

you can prove this by using a front door on the UM2+, which you would close for, say, 20 print layers and then open for the next 20 print layers. that is a much mor drastic temp change to the printing process than the hysteretic heater control of the bed would cause. and still, the effect on the layers will be barely visible.

so, yes, i suspected either a fluctuating power supply or voltage fluctuations on the GND plane of the Ultimainboard (rev. 2.14), which somehow seem to upset the circuitry inside the motor driver chips. the power splitter does away with both:

1. the Ultimainboard and its drivers now gets rock stable 24 VDC, which was previously fluctuating by approx. 100 mV, and

2. both extruder and bed heater have their own GND return paths that no longer interfere with the GND plane of the Ultimainboard.

as to whether there was / is any effect on the stability of the 5 V supply, i can't tell for sure. i have never experienced any (power supply related) hangs on the digital part of the board, i would assume that the 5V was just fine from the beginning.

 

 

 

 

Edited March 10, 2022 by Porkpie

[GregValiant 1,359]

I am electronically handicapped but I'm intrigued.

Searching around I see this has been addressed with generic add-on boards from TH3D and other suppliers.  I think I'll bite and get one for the bed and one for the hot end and then do a couple of tests.

 

Ordering the boards has brought up a rhetorical question...How come it costs $0.20 to ship something from China but $15.00 to ship something from California?  The Chinese package has to go through/over/around California to get to my house.  FTL travel?  The shortest distance between two places is not a straight line?  Einstinian space is a myth?

 

[gr5 2,239]

People have been asking this for about 10 years now.  It's cheaper to ship something from china than to ship something to my neighbor across the street.

 

For each country's national postal system, each one sets the cost of outgoing international shipping.  The receiving country doesn't get any money.  Once per year the countries meet and decide which country owes which country how much.  Always the decision was just to not worry about it.  For about 100 years this was an advantage to USA as we shipped more than we received.  But now it is a huge advantage to China.  International outgoing shipping from USA costs about $14 for a PCB that weighs less than 8 ounces.  From China to other countries I'm not sure what it is but it is less than $1.

[GregValiant 1,359]

News.  I ordered the boards on March 13 and with $0.20 shipping they showed up today.  Installation may be tomorrows project.  The printer is due for a regular maintenance anyway.

[GregValiant 1,359]

They are installed and are working.  The bed and hot end are heating up an maintaining temps.  I can't see any difference in prints yet but I haven't done any real testing.  I am pleased to get the current load off the mainboard though.  I have already burned up one main connector on the board and I'm hoping that I will avoid a second.