Kautschuk Platten Dämmmatten selbstklebend Isolierung XT | 32mm Dämmdicke (WHL #93)
Back to 3D printing for a bit – still lots of changes to the printer, but this one might be interesting to a wider audience: Thermal insulation of the print bed (hard data at the very end of this post!).
You see, the stock Ender 5 Plus printer bed comes pre-“insulated”:
Soo…that is a 3mm mat of black foam material that is glued down with some obnoxious acetone-soluble gunk, and garnished with a bit of aluminium foil on top. Yup, certainly better than nothing, but it doesn’t even cover the entire area, it’s just inside of the mounting frame. And the frame, obviously, is made from aluminium profiles with excellent thermal conductivity.
This is interesting since Creality, for whatever reason, designed three tracks onto that aluminium-backed circuit board: One heating track that has a nominal wattage of 430W at 24V, which is 1.34 Ohms, and two much smaller ones that lead to the center. These are clearly made for a thermistor or some sort of heat sensing device. Resistance is also okay, it’s in the order of one Ohm. With a 10k thermistor, that’s less than 0.01% typical skew at room temperature or 0.05% at 70°C, so completely negligible. The bed is however shipped with a fully wired thermistor. Someone put three heat shrink tubes around it and carefully taped it down near the two thermistor pads a at the center of the heating bed. I wonder why… (and no, the pads are unused, it’s clearly not a thermistor replacement or mod the previous owner did – also check the missing adhesive on the board where the wires snake along)
After quite a lot of acetone to clean up that goo, I did two things:
a) I tapped that strange indentation in front of the sensor with an M3 thread
b) I cut the thermistor from the cable and soldered it directly onto the PCB
As the glass bed is typically slightly raised towards the center, I figured I could potentially correct that in hardware (instead of software via Klipper + KAMP). M3 isn’t a massively capable thread, but the correction would be in the order of 100-200µm, so that should be sufficient. And if I don’t use it, well, now there’s a tiny hole in the center of my heater plate, whatever.
Soldering the thermistor is straight forward; soldering the cable onto the pads near the edge of the PCB is a bit more tricky since they carry away quite a bit of heat. But nothing the good ol’ 250W JBC cannot do
With those things prepared, now back to the original idea: Insulation.
Guess who only took one darn photo late at night and didn’t check if it was shaky? Yeah, me, of course.
But I guess the intention is clear – Outer frame cut to size, then the 2+2 mounting rods are added and insulated where possible, inner frame cut to size, and some chopping and hacking to get aluminium parts covered at a similar overall height. Afterwards, some cutouts for parts that otherwise not fit the printer, e.g. the compression/distance wheels (no more springs needed), the wiring section and bits near the mounting points.
Important thing to note here: Get a sharp knife. I did the initial cuts with a fresh blade on a box cutter – that works, but it’s ugly. After correcting for half a centimeter here and there the place was a total mess with all the bits flying around. I ended up using my sharpest knife from the kitchen – a Kiwi cooking knife, type 172 to be precise.
That 6€ non-stainless knife can be sharpened to a ridiculous level, and it slices through the entirety of the foam (or literally anything else) with ease. Having a more reasonable foam thickness to begin with should also help, if the glue on the insulation foam is sufficient. I used “INSUL ROLL XT” which does stick very well, and instead of the single 32mm layer I’d get a 9 or 13mm roll and do three layers. It’s available in 6, 9, 13, 16, 19, 25, 32, 40 and 50mm thickness. For thicker material, lay it flat for like a week, so it doesn’t warp after installation. There’s very little outgassing straight from the box, and it does not smell at all when under full 70°C load. YMMV with other vendors and products, though…
Funny photos aside, I also got data to back up my claims of this being a worthwile upgrade.
My full Ender 5 Plus setup in idle (at 19°C) has a wall plug (!) consumption of 32Wh over the course of 2 hours, 24 minutes. That’s 13 1/3W and sounds plausible – the 500W power supply is a quality MeanWell unit but way below any reasonable load (read: terrible efficiency in that state), and there’s a Raspberry Pi 3B running, the Octopus board with some small 12864 display, there’s two 120mm fans, an active BL touch and smaller bits and bobs everywhere. So 13.3W is the system baseline.
Heating up to and holding the whole thing at 70°C bed temperature for exactly one hour consumed 192Wh = 192W average. That’s 84Wh on the first 20 minutes (252W average) and 108Wh for the next 40 minutes (162W average). Of course the initial heating is done at full power, so load will drop after a while. Reaching 70°C takes 6 minutes sharp, unfortunately I don’t have a consumption figure on that one. Working back from the 162W over 14 minutes gives an estimate of 46Wh/6min = 460W, which is also plausible. I’d say the latter 40 minute readings is pretty stable and is indeed representative of long-term use. Note that all these figures obviously include the 13.3W baseline wattage of the system. Also, the LED panels were removed for those tests, so it’s just a stock open frame Ender 5 Plus.
After that insulation upgrade, and at a slightly colder 18.5°C (disadvantage!), it looks like this: One hour total consumption is 152Wh = 152W average power draw. First 20 minutes is 74Wh (222W average) and last 40 minutes are 78Wh (117W average). This time, I also got consumption readings after reaching 70°C, which is 43Wh at the very same 6 minutes sharp, equalling to an average power draw of 430W during that time. That matches pretty well with the nominal 430W of the heating bed, even if the 13W base power needs to be accounted for.
To sum that up, now with the base system draw removed from all figures (yes, might be less due to efficiency curves of the power supply, I know…)
Heating to 70°C: 447W (extrapolated/speculative) -> 417W (-7%)
First 20 minutes: 239W -> 209W (-13%)
First full hour: 180W -> 139W (-23%)
Any subsequent hour: 149W -> 104W (-30%)
Insulating the heated bed just from below saves up to 30% energy!
Of course the bed isn’t the only heated component in here, but the power consumption of the extruder is dependent on printing temperature, material and material flow, so that is hard to factor in without printing a part in exactly the same time for both tests. All the motion systems also consume power depending on the print job, and there might be additional power draw from RPi and processor board to compute the jobs, there’s additional fans coming on and this and that. Still, heating the bed on a large-scale printer like the Ender 5 Plus is clearly the major cost factor, with 430W of the 500W power supply capability dedicated to that job. And insulating this bit saves in the order of 20 to 30% of cost for the majority of prints! Additional power can be saved when adding blocks of the very same insulation material to the bare areas of the build plate for longer prints – covering let’s say 1/4 of the area should add another 10% of savings…
One more thing: “Tell me the savings in Euros”, some might say. Well, glad you asked.
Klipper/fluidd has this nice feature of logging print times. Now my Ender wasn’t set up with Klipper from the beginning and I printed quite a bit before that, but current stats from the last three months are:
*Total print jobs: 179
*Longest job: 26h 36m 46s
*Total time: 229h 19m 49s
*Avg per print: 1h 16m 52s
Consumption old vs. new was 192W to 152W avg over the course of an hour, let’s add 30W to cover extruder wattage and other stuff. At 229h, that would be 51kWh for the bare configuration vs. 42kWh with insulation. Delta 9kWh per 3 months. That’s annual savings of 36kWh if I continue to print that much stuff.
36kWh at my current tariff of 0.3014€/kWh is: 10,85€. Ten bucks a year saved by a layer of self-adhesive foam under the printing bed.
The smallest chunk (1m x 1m) of this Insul Roll XT stuff in the recommended thickness of 9mm costs about 18€, so it’ll pay for itself within two years. I bought the 32mm version which was 31.19€ including shipping. So it’ll take me three years – however, I got enough stuff to cover four printers with it (or to sell each remaining quarter for a tenner…)