Heater Question for Electrical Engineers and Other Knowledgeable on the Subject

Some of you on here have read and/or commented on my topic I started a while back on a custom built printer project that I am now in the early stages of designing/sourcing parts for. https://3dprintingspace.com/t/custom-built-printer-ideas-and-questions/5424/56
Anyway, long story short, the hardest part to source is the bed heater. This is because I want a lower wattage per cm2 than the fairly standard 0.4-0.5 watt/cm2. The reason for this is that I want to reduce peak power requirements, as bed area will be in the ballpark of 3,360 cm2. Trouble is finding a bed heater that is low enough wattage, but not too low.

So my question is this, if I would supply 12v to a bed that is 0.4watts/cm2 at 24v, will it be more or less 0.1watts/cm at that lower voltage? My research indicates yes, but I figured I would ask here.

If the answer is yes, that would make my search for a bed heater positively easy.

P = V * i


i = V * R

Substituting in “V * R” for current in power:

P = V^2 * R

So, if you want to reduce your power by 25%, you’ll have to reduce your voltage by 50% (assuming that Resistance remains constant).

Buy why would you want to do this? It’s much easier (and cheaper) to let the controller board’s PWM maintain the heated board’s power with minimal draw than to have a variable output DC power supply or two DC power supplies and switch the active one with something like a relay. I don’t know which firmware you’re using but you’ll have to modify that as well.

Let the controller monitor the bed temperature and set the PWM for you.

1 Like

So in other words, you are saying that yes, halving voltage will quarter the wattage; a 240w heater at 24v would become a 60 watt heater at 12v.

Why I want to do it? Because a 3,360 cm2 bed preheating will draw 1344 watts when preheating if using a heater with a power density of 0.4w/cm2. That is not an option for me. Yes, I know that a 15 amp circuit could handle that load by itself with no trouble. Thing is, this printer won’t be by itself, all my other printers are on the same circuit(yes, I know I could shut them down, but I’d rather not have to shut them all down to preheat this one) besides, that isn’t all that is on the circuit. Everything else plugged in in this room and the neighboring room is also on that circuit(it’s the way this house was wired🤷‍♂️) So, yes, I know this way entails a little more expense and complexity, but if that is what it takes to be workable, so be it.

On the modifying firmware note, why? Will a mosfet not work with regular marlin firmware? Excuse my ignorance, but is PID control in a bed heater context really PWM? I mean, it is sort of in the sense that is turning the power on and off in long “pulses” if you will, but if that qualifies as PWM, then so does an old fashioned thermostat. Or do I have things completely wrong?

With my ender 3s for example, on the screen it shows the power to the bed being on for a couple seconds, then off for a few seconds, and so on. Or does that graphic display on the screen not actually indicate if the power to the bed is on or off? If I have that right, it means the printer is alternating between drawing 240watts plus whatever the hotend, fans, and steppers take, and just drawing whatever the hotend(if on), fans and steppers take. So this custom printer would be drawing 1400w(roughly?) , and 56w respectively if I went with a 0.4w/cm2 heater. Contrast that to a cyclic load of 392w(roughly) to 56w. Granted, with a lower wattage bed heater the higher draw pulses will be roughly 4 times as long(or longer?), but the peak load will still be lower, making this printer workable in my situation.

Please correct me if something I said above is wrong.

Yes, it’s a true PWM on your printer - the period is surprisingly long (I believe Marlin has an 8s period for the bed) but it is constant so that the time when power is applied to the heated bed is an accurate fraction of the period. Personally, I prefer the much shorter period used with Klipper (0.1s for both bed and extruder) but I don’t think the processor in the Ender 3 can run with that.

A traditional house thermostat and furnace/air conditioning is not an example of PWM control. There is no defined period for on/off and no fraction or percentage applied to it. A thermostat brings the temperature into a “comfort zone” that is a couple of degrees or so above and below the desired temperature.

Honestly, as I understand your project, this is one of your lesser concerns and one that should probably be lower priority. You seem to be content to working with heating pads and providing them with DC power and that feels like an adequate plan for now.

You’re going to have a huge aluminum print surface which you’re going to have to mount in a way that thermal expansion isn’t going to be a problem. For a piece of aluminum one metre in length and heating it by 50C, you’re going to see it expand by over a mm (using the Coefficient of Thermal Expansion of 22.2*10^-6 mm/C). How are you going to hold the plate in place while allowing to expand and shrink without bending or causing unexpected forces on the rest of the structure?

Along with that, don’t forget that you have two radiating surfaces (the top and bottom of the aluminum plate). The modeling we discussed previously will probably be accurate to within 20% or so, but I wouldn’t count on anything more.

I don’t think you can adequately plan for reality and I honestly think your best course of action is to keep things as simple as possible for now, get the printer running (even if it means that for the first few trial runs you have to shut down all of your other printers when you’re testing) and then look for opportunities.

1 Like

I’m running skr mini e3 v3s in my Ender 3s, and will do the same on this printer. I don’t know if that makes a difference in the capability as far as heating time period length goes. The board is BTT supports Klipper for this board, for what that’s worth. I don’t plan on making this a klipper rig, at least not anytime in the foreseeable future, so this board’s klipper abilities aren’t really relevant for my current plans.

So each “pulse” is 8 sec with marlin, correct? Or is the period a time in which the board is sending short pulses? Sorry if I am being dense.

My plan for the bed mount is to do what @Jason said the Modix printers they sell have. That is have bolts from the bed to the mounting brackets, but instead of a spring between the bracket and and bed, there is another nut on top of the bracket. The nuts are snugged up enough to be secure, but still allow the bolt to slide in it’s slotted hole. The very center bolt in the bed would have the nuts tightened so there would be one fixed mount on the bed. That way the end bolts would only need to move around half a mm, and the bolts in between some fraction of that.

Yes, I realize that two surfaces will be radiating. I will deal with the underside to my best ability with foil backed insulation.

You are right that the best test will be building and testing it in real life. I still want to do due diligence though.

The site I use to compile my firmware for me(I’m gonna try it for this project and see what happens) has the option to not enable PID control on the bed. Presumably it would simply operate like a thermostat then. If PID won’t work with a mosfet, that would be an option to try. Yes, I know that would lead to more bed temp fluctuation, but that would still be something to try.

I think I would rather start with a 12v bed and see if it works, then do the reverse, and start with a 24v bed. That way I am starting with my first priority and not my last resort. Preheating will still be drawing full power regardless of how the printer is set up, and I don’t want 1400 watts of preheating draw if at all possible, that much I know.

No, each pulse takes place within the 8s (or whatever the Marlin) PWM period and is the period multiplied by the power level. 10% power would have a 0.8s pulse, 50% power would have a 4s pulse, etc.

Fastening down at one point, I use the lower left (0, 0) corner, seems to be the way that people are doing it now.

There’s “due diligence” and there’s appropriately prioritizing which I’m trying to push you towards. As I said in my last response, you have a basic plan for the heated bed (big aluminum plate that is heated by adhesive pads powered by a DC source). You have a huge and somewhat unusual build area that I would think you should be concentrating more on things like how to build the frame and jig it so that it is accurate and your rails are straight and parallel, both to each other as well as the print bed.

The heated bed is meaningless unless you have the moving parts running down very accurately placed rails with no chance of binding or cables being snagged as the gantry goes from one end of the printer to the other.


Gotcha, that makes sense now about the pwm period.

Yes, I realize that normally the fixed point would be in the corner, but as you say, this is not an entirely normal printer design. Having a center fixed point reduces the max amount of travel needed from a bolt. I don’t really see any technical difficulties that could arise from a center fixed point.

I have decided to use exoslides as opposed to linear rails, it will hopefully be a little easier to setup, as opposed to linear rails. I have a few ideas for how to square things up. As the design I would be going with would essentially be a Cartesian “gantry slinger” as opposed to a hbot or corexy, it should be somewhat easier to square up.

As far as snagging cables, all moving cables are going to be inside openbuilds cable chain, that is a no brainer I think.

Anyway, the rest of this discussion can be migrated back to the other topic when I need to ask more questions, which I am sure I will. Thank you for taking the time to answer my questions, you have definitely helped me out.

With regards to the Exoslides, how are you planning on implementing them? I’ve played around with them a bit and they’re probably well suited to something long like yours but I don’t love them. They’re 80mm long, which is a potential problem as they add considerably to the size of a traditional printer (not an issue with yours) and they’re more expensive than other approaches for linear movement (but would be cheaper than linear rails in an application like yours). The big thing I didn’t like about them is that they work best when four are used to a rail and that complicates things like cable management (can’t have drag chains with them).

I’m curious, how are you keeping sag in the middle (I think you have a fair amount of mass on your gantry, if I remember correctly, you’re putting your spool on it) to a minimum? Are you using something like 20x40 or 20x60 V-Slot for rigidity because I know you’ll get a noticeable amount of sag with basic 20x20.