Use a transistor as a heater
Circuitsoft
Posts: 1,166
I've been thinking about making a 3D printer, and the typical "hot-end" has a 3 watt resistor (run at about 20W) inside an aluminum block, alongside a thermistor to maintain the tip temperature. Figuring that a TO-220 would be an easier package to mount, I found a TO-220 resistor, but it's ~$6.5 at Mouser, and ~$8 at Digi-Key.
It occurred to me that a TIP41 is ~$0.60, and that the B-E junction can be used as a diode-based temperature sensor, but then there's the temperature issue. It's not made to run that hot. So, I'm wondering what the failure mode of allowing the die to get that hot will be.
My own understanding is that thermal runaway is possible running at too-high a current, due to the -2.1mV/C temperature coefficient of the B-E junction, but I don't think I'll be running a high-enough power to have significant temperature differentials.
It occurred to me that a TIP41 is ~$0.60, and that the B-E junction can be used as a diode-based temperature sensor, but then there's the temperature issue. It's not made to run that hot. So, I'm wondering what the failure mode of allowing the die to get that hot will be.
My own understanding is that thermal runaway is possible running at too-high a current, due to the -2.1mV/C temperature coefficient of the B-E junction, but I don't think I'll be running a high-enough power to have significant temperature differentials.
Comments
Some silicon MOSFETs can run at 175°C.
There are now Gallium Arsenide and Silicon Carbide MOSFETs that can run at much higher temperatures. Thermal runaway is not really a factor here as, I assume, you will only have a single device.
Duane J
survives the temperature the chip will degrade as dopants start to diffuse and the device structure change. The Vbe v. temperature
curve is likely to change in unpredictable fashion as this happens.
A SiC (silicon carbide) MOSFET might be more feasible (operating temp 350 deg C or more) - but not cheap.
Mark, thanks for the explanation about diffusion of dopants - that's what I was looking for.
The outside of the case will generally have to run ~205-215C, which is outside any thermal range I've ever read about. I know SiCFETs can supposedly run a lot hotter, but I haven't seen that reflected in any datasheets that I've read. Also, a FET won't give me the B-E junction to use as a temperature sensor. Any idea if anyone makes a SiC BJT?
Thanks,
- Alex
A FET also has a tempco, but remember any tempco that reports, is Tj, not heatsink temp.
I would read both Vg-s (or Vb-e) and Current with a decent ADC, and then use curve-tracer maths to work back to a Tj
Use the biggest package you can, to get Tj elevation as low as possible, and I'd also err on the biggest-die side too.
(tho that may not be 60c anymore..)
Add in a separate sensor, for reading the plate temperature, and you can then even track long-term junction degrade.
Treat the heater as disposable, and then you do not care so much about long lifetimes.
Feel free to try, but all I can see in your future is melted electronics.
PLA plastic is typically melted at 185 degrees C, and ABS at somewhere around 250.
Your numbers doesn't match up, either.
A 3W Resistor would fail pretty soon if you sent 20W through it.
Normally, it's a 5W resistor, at 5.5 or 6.5Ohm.
Mounting the resistor is pretty easy.
If it's a snug fit in the hole, cover it in high-temp thermal paste and slide it into place.
If it' not a snug fit, take a strip of aluminium foil, cover it with a very thin layer of thermal paste and roll it onto the resistor, then slide it into place.
With thermal paste, less is definitely better.
One good reason for using the thermistor for temperature sensing is that t's well-understood, and all the Firmware versions ut there already comes with tables that fits most easily-available thermistor on the market.
Its nice that you're looking for alternative designs, but now you're trying to use a component that is either prohibitively expensive with little if no improvement on the finished product, or you end up with something that is going to self-destruct pretty soon.
A typical hot-end these days costs less than $50 as a kit of precision-machined parts.
These kits(possibly unless you buy one made in China) are pretty good, and most an be ordered 'tailored' for your preferences.
(Nozzle sizes from .25 to .5mm is common, and usually a choice between 1.75 and 3mm fiament. Some comes with different inlays to let you change filament size later)
I suggest that instead of trying to reinvent this, you get one and try it, and if you still think there's room for improvement, THEN try to reinvent it.
If you really want to help out the community, try making the rest of the HW simpler or cheaper.
A set of the NEMA 19 steppers commonly used these days isn't that cheap. A good way of reducing the number of them, or to switch to the smaller and cheaper NEMA 17 on part of the build, without compromising the mechanisms, now that would be nice.
A resistor - no thermal runaway situation
A transistor - potential thermal runaway situation.
I like being creative, but when the components are working against success, why bother?
That's 72Watts.
A Power transistor giving off that much heat is probably on fire...
74 watts would not be difficult. Even 5000 watts would not be difficult.
Where to acquire a small amount that is ready to use? Try your local Goodwill Store for a broken or working 2nd hand toaster or toaster oven.
The wire will work into the hundreds of voltage as long as you get the resistance right. So 12 volts shouldn't be too much of an engineering problem. But you may have to provide some parallel circuits to get enough heat.