An idea based on watching Grant Thompson’s description and demo video of his “Styro-Slicer” hobby hot wire cutter.
As an alternative to laser-cutting wood or MDF, you could cut it with a hot wire, like the one in a hot-wire styrofoam cutter. Some hot-wire styrofoam cutters are already capable of cutting soft woods with some difficulty.
I am envisioning a sort of bandsaw in which an endless loop of thin wire is run through the kerf, heating up on its way into the kerf and then being quenched as it plunges into the material, dumping its heat into the wood. Two conductive rollers, one above and one below, feed electricity through it; the distance between the rollers is so much shorter than the distance around the other part of the loop that the other part of the wire is not dangerously heated.
To be concrete, let’s consider 80-micron-thick (40 AWG is 79 microns) nichrome wire cutting 3mm MDF with 50 watts of power, leaving, say, a 100-micron-wide kerf, with the two electrical-feed rollers at a distance of 45mm between them and an additional 900mm in the loop. Nichrome’s resistivity is about a μΩm and apparently 800 mA is enough to keep an 80-micron wire radiating at 1100°, which is plenty hot enough. In this case the resistance of the 45mm should be about 1 μΩm * 45mm/((80μm/2)²π) or about 9 ohms, so you need about 2400 mA to get to 50 watts:
You have: micro ohm meter * 45 mm / ((80 microns/2)^2 pi)
You want: ohms
* 8.9524655
/ 0.11170107
You have: sqrt(50 watts / (micro ohm meter * 45 mm / ((80 microns/2)^2 pi)))
You want:
Definition: 2.3632718 A
This means your power supply needs about 22 volts, which is eminently feasible.
The resistance of the 900mm part is 20 times greater, or 180 ohms, so it will be carrying 120 mA and dissipating about two or three watts --- enough to require management but not an overwhelming problem.
This machine seems nearly feasible, but a problem with this is that the exposed 45mm of wire is going to be not only radiating like a motherfucker, perhaps losing most of its energy into the void, but also possibly getting rather hotter than the 1400° that it can in fact withstand before melting. And don’t forget that the MDF needs to have air blowing on it the whole time to keep it from catching on fire, which is also going to cool off the wire, although maybe this problem can be limited by careful management of the air blowing, so that it’s only blowing in a very thin layer.
So consider a revision: 127-micron-diameter (36-AWG) wire cutting a 150-micron-wide kerf (still at only 50 watts) with only a 12-mm gap between the conductive rollers. This drops the resistance per mm by 2.5 and the distance by 4x, getting us down to about 930mΩ, which means that we now need 7 amps running through the wire, powered by about 7 volts, and a power-supply output impedance of below an ohm or so to keep efficiencies reasonable. That’s still hot enough to melt the wire if there’s no thermal load on it, but I feel that it’s possible to keep it running through the rollers fast enough to keep it passively safe, even without rapid-response electronic control to keep it from burning. And it’s mechanically strong enough that it might survive encounters with things that unexpectedly fail to burn on contact with it.
(180-micron (33-AWG) nichrome wire is commonly used for styrofoam-cutting machines like the popular Argentine brand “Segelin”. Thicker gauges are common; thinner gauges are somewhat exotic. One Aldo Vignolo sells them on Mercado Libre here in Buenos Aires.)
What do we do about splicing the wire? One possibility is to simply not worry about it, instead merely reversing direction when the time comes, maybe turning off power to let the wire cool in between.
How much thermal energy can be stored in the gap before the wire melts? Nichrome has a specific heat of 450 J/kg/K and a density of 8.4 kg/liter, which works out to 3.8 kJ/liter/K. At 127 microns, that’s 48 microjoules per millimeter per kelvin, or 67 millijoules per millimeter, or about 800 millijoules in all, to melt the wire in the gap. The situation is slightly worse because nichrome’s resistance increases slightly with temperature, so it can experience a kind of thermal runaway where hotspots heat up faster, and also slightly better because the hottest spots are also radiating away energy most rapidly. Disregarding these, this means that keeping it passively safe at 50 watts involves feeding a gap’s worth of wire onto a quench roller about every 16 milliseconds, a speed of about 750 mm per second. But that’s just to barely keep the wire from melting.
I’m not sure that shortening the gap was a big improvement. It does lower the amount lost in the other part of the loop, but it also reduces the heat capacity of the wire in the gap, thus requiring faster response if we decide to use active feedback for safety, and so it doesn’t actually affect the necessary passive-safety wire speed. And, because it lowers the resistance, it requires a higher-quality power supply and heavier wiring. And, because it raises the required current, it also increases the possibility of melting the wire --- resilience is likely of more value than efficiency here.
A third revision, where we increase the gap size to the accursed 25mm, leave the wire thickness at 127 microns, and drop the power to 40 watts, requires less current, only 4.5 amps:
You have: sqrt(40 watts / (micro ohm meter * 25 mm / ((127 microns/2)^2 pi)))
You want:
Definition: 4.5020328 A
This is low enough that it is almost passively safe even for copper, which melts at 1084°, even without the quench rollers. The resistance in the gap is up to 2Ω, which means you need 9 volts, a very convenient number indeed. You no longer need quite such big hairy power cables or a high-quality power supply; Duracell’s spec for their MN1604 nine-volt battery lists a 1.7Ω impedance, so even that would be almost good enough; a dozen or so in parallel would do, though obviously a line-powered power supply would be more cost-effective. The gap can now hold 1.7 J of thermal energy before melting:
You have: 450 J/kg/K * 8.4 kg/liter * ((127 microns/2)^2 pi) * 25 mm * 1400 K
You want: J
* 1.675935
/ 0.59668186
Lowering the power to 40 watts means that it takes 42ms to reach melting temperature in the absence of any heat-loss mechanism, such as radiation, conduction into the workpiece, or the quenching roller. For the quenching roller alone to be sufficient, the wire needs to move at 600 mm/s:
You have: 25 mm * 40 watts / (450 J/kg/K * 8.4 kg/liter * ((127 microns/2)^2 pi) * 25 mm * 1400 K)
You want:
Definition: 0.59668186 m / s
The 150-micron kerf I expect from this is not quite as good as the 100-micron kerf I see from laser cutters. Contrast this with the Olson No. 2 spiral scroll saw blade, which nominally gives you a kerf of .035 inches in US folk units (890 microns), losing six times as much material and thus with six times as much opportunity for imprecision.
I don’t know if it will work as well as a 40-watt laser cutter at cutting wood, even if you blow air down the wire. I think the laser cutter is usually heating up a tiny point rather than a whole contact surface, so it might heat the wood or MDF to a higher temperature. But maybe this will instead work better, since it has some mechanical sliding action and won’t be impeded by smoke.
CNC foam cutters commonly use inconel rather than nichrome.
US patent 5,429,163 is relevant; it claims 600° to 800° is sufficient for cutting wood, and suggests using either a reciprocating or an endless wire. It says the wood itself need only heat up to 240° to 270°, and that irregularities in the wood pretty much require closed-loop temperature control, and claims 10 to 21 mm/sec of cutting speed. Unfortunately it doesn’t go into any detail about power usage, preheat gap size, wire speed, or wire diameter.
As an alternative to heating the wire by running electricity through it lengthwise, you could run it through a high-frequency induction heater and thus induce eddy currents in it, particularly if it’s ferromagnetic. Rudnev et al.’s Handbook of Induction Heating, Second Edition tells us on p. 38, in §2.2.4, that induction heating of wire is commonplace, using frequencies from 10 to 800 kHz and wire speeds through the inductor around 2 m/s. They cite a particular system they built for Radyne that heat-treats 1060 carbon-steel spring wire in the 4–8-mm diameter range with a 400 kW, 10 kHz inverter for heating up to its Curie temperature, then a 420 kW, 200 kHz for heating above its Curie temperature; but they don’t give a wire speed.