Driving a DC PCB heated bed with AC mains?

Hello,

I'm in the early stages of desiging my own kossel-style printer. I would like to drive the heated bed with mains voltage in order to reduce the power requirement on the power supply.
Even if the following is an insane idea, I would appreciate an explenation along with the name-calling!

So, the basic idea is to order a board like this, and power with mains voltage.
BUT! since just connecting it to mains will certainly fry it within a second or two, the idea is to use a solid state relay like this one in order to PWM the mains power.

The above bed is 1.2 ohm, so with 240vac, it would produce about 48Kw (right?), assuming I want the bed at ~100w, I'd need the PWM to be about 0.2% (right?), or on for 2ms out of every sec...

• Is my math right?
• would this work, theoretically?
• what's the minimum PWM pulse time with major boards? i'm guessing 2ms isn't possible, but maybe 20ms is?

Thanks!
-Leav

Even with PWM you have to design for full on conditions

so 1.2 ohm, with 240vac

so thats 171 amps!

Its never going to work, your just going to have lots of instantaneous copper evaporating and fire!

In the long run, you would probably be much better off looking on YouTube at how Werner Berry made the mains-powered heated bed on the BerryBot and duplicating that. I believe that there are a few threads on the SeeMeCNC forums (they have a section for heated beds) that discuss mains power and other options, and IIRC the most common solution is to run a specific 24V DC power supply with the output trimmed to 19-20 volts. My Onyx bed will not hold 110C wired straight to my beefy 12V ATX power supply, and takes over half an hour to reach 100C unless I place a magazine over the glass for insulation.....

Quote
Leav
So, the basic idea is to order a board like this, and power with mains voltage.
BUT! since just connecting it to mains will certainly fry it within a second or two, the idea is to use a solid state relay like this one in order to PWM the mains power.

Your mains fuses will blow within half a second (hopefully). AC isn't really PWM-able, especially no to cut down power to less than 10%.

Seeing such a calculation I wonder wether I can recommend you to play with mains power. Perhaps an industry made heating mat would be the better choice.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

He has got me wondering about some some sort of sine wave following pulsed power supply though

that grabs voltage from 0 - 15v and 0 -15v part of the cycle for use. It wouldn’t need to be continuous.. as long as the power produced was enough to keep it at temp.

Just a idea though... no idea how you would do this, if its even possible.!

Quote
Dust
He has got me wondering about some some sort of sine wave following pulsed power supply though

AC solid state relay typically switch when voltage crosses the zero line. There are semiconductors which do this nicely. As such your theoretical max. PWM frequency is 100 Hz. PWM delivered by a controller isn't synchronized with these 100 Hz, though, and also doesn't respect the 1/100 s minimum period, so you get a messy timing and chaotic output.

Those using mains power use heat mats made for mains power and drive this bang-bang. This works. But not, if you have just fractions of a second to blow the assembly.

Another one: these 12/24 V heated beds are not sufficiently insulated for 220 V ... trying to use it anyways is a pretty dangerous adventure.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

You missed the central point...

generating ~24v ac from the mains by extracting the first and last part of the sin wave.

ie a type of power supply.

I am not advocating putting the HBP anywhere near real mains voltage.

Edited 1 time(s). Last edit at 08/04/2014 05:02AM by Dust.

If you're going to cut down the mains 240VAC down to 24VAC, you may as well just use a toroid transformer suited to the job (with a 10:1 winding ratio) to do it and avoid all this really unnecessary (and probably a hell of a lot more expensive than the toroid) switching gear to get it to switch with the precision you want.

Also: Trying to force something to switch part through a cycle will lead to a very ugly waveform, which would not be a nice thing to try and calculate the actual power for.

One other thing of note: There is a reason zero-cross switching devices exist - You probably don't want to switch something on mid-cycle every cycle.

While I applaud your thoughts Dust, I seriously doubt the practicality.

This is a very frequent question, and very interesting subject to think about.

If we have a current flow of y amps and 20 volts, there is no known method to coniniously just let 10 volts through while not without affecting the difference in some way. If we would of had something like that, we would not have efficiency issues with different power sources. We do tend to think at amps and volts being different things, but its just one phenomena behind it even if it has two sides of its sort of same coin, so things are tied to each other in certain way. A linear regulator or a resistor divider would just let 10volts y amps through, and just dissipate the difference in heat for y amps x remainder of volts. The switched mode power supply employs pwm-ing the input by principle of volts*sec balance or amps*sec balance to determine the output from the input and pwm filling factor. But still smps is complicated (e.g. many failure modes) and while has better efficiency its far from being a great. This is why this idea of controling the sine wave to accomplish something similar is very appealing.

A light dimmer contains a triac that is used to start the conduction at an angle. E.g. from that entire input sinusoid only lets a part of it through. It doesnt change the output at a time, it just starts to conduct at a certain point, after which it fully conducts. But as for the light bulb example, the systems runs in good operating point / parameters while its fully on and the dimmer just goes down from that e.g. it "eases" operating points. This is why you cant use a dimmer for the heated bed like in the OP issue. It still gets same voltage peak as the initial sine, what it drops is the initial lowest parts while the voltage is low. Sort of speaking the ones using a diac to trigger would drop at least like 20-30 volts from zero, and conduct in the rest, including peaks of 308volts. Same thing the optocouplers for triacs like mocxxx types also dont switch perfectly on zero, but more or less around that, again some +/- 15v or something. Also its only 50hz so if one would use it like so to for a voltage conversion, then would need a huuugeee bank of capacitors and still have significant ripple. And those capacitors would need to be like 400v rated, which means extremely big and expensive. And few other things. But on short, the solution is not feasable for voltage conversion.

What one could do practically, is to use a type of dimmer circuit with a triac that is only 3 quadrant (because the 3quadrant ones can work ok with inductive loads), and put it at the input of a transformer and you could get that trafo to lower its output down from its normal value, because only parts of sine gets into the trafo. A normal light dimmer probably does not contain a 3 quadrant triac so its better not to try directly just like that, but use a dedicated circuit for the job. Also can do something similar on the output of a transformer, there are also some solutions that employ triacs in the middle of rectification bridge to control the exact form of the output, more to cut down the peaks instead, and again that would be non-standard, and for other purposes a proper syncron rectification would be generally more accomodating.

Its a nice idea to be able to use only parts of the sine and to get a different output, it works for incandescent bulb (light dimmer) or other cases, but it has its limitations. Sort of speaking, if the heated bed would have like at least ~60-90++ ohms resistance, then one could put a light dimmer on it, sort to control it down from its normal operating point. But it needs a high resistance, enough that it would work nice on full sine first. So wont work with that pcb heated bed because the peaks would be way too huge. Very similarly to what the op asked, there are integrated circuits that do the exact thing to control temperature, e.g. a circuit that "fires" the triac at different angles (proportional control), e.g. takes different parts of the sine depending on the feedback coming from a temp sensor. Most of us have one circuit like that in our homes, its quite commonly used in fridges, where it does exactly this thing with the feedback from a thermistor, but it just controls it the other way like we would in a heated bed, e.g. makes the temperature go down. Otherwise something like that could be easily done for a heated bed running on mains, and have the temperature on proportional control. It would be good like that, with a uC compatible on-off input. But ... But i dont think its generally advisable to promote heated bed running on mains voltage for our users. My bed runs from mains with a traic and a moc_xx, and its a aluminium plate with 8 power resistors of 50w each and runs fine. But i do not want to promote that for the regular user, which is less aware, to have something like that to be used in homes with children, with exposed wires, on the desktop, etc, on short i dont see good things coming from that, and as long as there are atx supplies those should be safer.

Edited 1 time(s). Last edit at 08/05/2014 12:23PM by NoobMan.

I might be wrong but I think that you are missing a point .

If you want to get 100 Watt from a 48 Watt resistance you will overheat the resistance and in a short time it will burn.

If you want to go directly with 230 Volts in a large heating plate with the size and pattern you prefer you should go with this solution :
[forums.reprap.org]

Quote
lucamax
If you want to get 100 Watt from a 48 Watt resistance you will overheat the resistance and in a short time it will burn.

Not if you apply better cooling than still air or stop heating when some temperature is reached. Look at extruder heating resistors: they're spec'd for 3W and get heated with 25 Watts. They don't burn, because heating stops when temperature is reached.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

IMHO you are wrong, extruder resistors do not burn because there is a dissipator and the temperature do not reach high values for the resistor, but in a bed heating plate you cannot put a dissipator.
More there is a big difference in the amount of heat that a cartridge resistor can hold and the heat that a pcb resistor can , the second is much , much less.