Questions about cartridge heaters

[rthorntn]

Hi,

Thanks to everyone who has helped in the past.

I have some Dernord 12v 100w cartridge heaters, I want to power them with a PC ATX PSU, I'm planning a tiny oven.

So I will embed the heaters in some aluminium plates and surround those plates with refractory material. A thermocouple will be measuring the oven temperature, some fans perhaps and an MCU running a PID controller.

What is going to be the best way to go about controlling the temperature?

Be gentle, my understanding is the heater is a resistor, creating heat from that, the heater is 12v so each one will need  8.3 amps to get to 100w.

This is where my AC and DC are a bit fuzzy, if it was AC I could use a relay to switch the heaters off to cool it down or maybe a variac to control voltage for more precise control.  Is DC different, I could use a relay but would PWM be used?

Thanks.

Richard

[kosine]

Hi Richard, you don't say what you intend to do with it, but I'm guessing maybe a surface-mount reflow system...?

If you need double-sided loads, then an oven may be required, but for single-sided boards a hot plate is much easier. I put a quick 1500W rig together a few months ago and it'll reflow a board in about 40sec. (Takes around 6mins to reach 300C - then just place the PCB on top and watch the paste melt! It's pretty much effortless.)

It was made from two 100x150x12mm aluminium plates (6082/HE30), with ball nose slots each side to recieve the cartidge heaters. Then I just bolted it together and added some 8mm bolts added as standoffs. (See photo.) I've run it for hours at a time with no problem at all. (Was expecting the standoffs to scorch the table, but they don't actually get that hot. I does take a long time to cool down, though.)

So long as you can drill or machine some alumimium plate (which I assume you can from your post), then it's actually really cheap and easy. Check ebay for PID kits. You can get the controller, thermocouple and solid state relay for about $15 delivered. Mains cartridge heaters are also dirt cheap (I used two 750W, 10x150mm in size). The wiring is trivial, and there's no need to hack a PSU or write any MC code.

If you absolutely need an oven, then the same approach in a suitable box would probably work. However, your 100W heaters will likely be far too weak. (Little more than light bulbs!) I originally tried 500W but it wasn't anywhere near powerful enough. 1.5kW did the trick.

[rthorntn]

Thanks kosine, nice job.

It's a small roasting oven, maybe 10 x 10 x 2cm internal, nothing to do with reflow.

I was thinking of using four 100w heaters together, four 6mm (should I consider copper?) plates (two bottom/two top), two heaters sandwiched between two plates with ball nose slots.

I want it to take at least 12 minutes to get to around 230C and I want to be able to slow the temperature ramp at 2 minutes in.

I want to use DC to keep it beginner friendly.

If 12v is a flop, could I put 24v or 48v through these Dernords to speed things up:

https://www.aliexpress.com/item/12v-100w-8x150mm-Electric-Heater-Cartridge-Tubular-Heater-Element/32713061852.html

Or would I be better buying specific 24v & 48v ones (I have seen a 300w 48v)?

[kosine]

Hi Richard, that sounds quite doable. 400W in total over that volume should be adequate. (My 500W rig was quite slow, but that's what you need in this case.)

You can probably run it from something like an Arduino with a DC solid state relay. (Which have worked quite well for me on other projects.) They're pretty cheap on ebay, though the DC variants are a little less common than the AC versions (which won't work for DC).

I think you'd only need to implement the proportional control part of PID, which makes it easy - just roll off the duty cycle as the temp/time reaches your set points. (The integral part ultimately just adds a bias to compensate for cooling, and most systems don't respond quickly enough to make differential control necessary. A lot of industrial PID systems only uses PI in practice, and you can often get by with just P if your temperatures aren't too critical.)

For temperature sensing with a microcontroller I've always preferred PT100s rather than thermocouples. They're not too expensive these days (few $$s) and for simplicity you can just use them in a voltage divider like a thermistor. The change in resistance is linear with a PT100, so it's very easy to work out the temperature. If you use a precision resistor (0.1% is OK) and add a precision voltage reference (I use a 5V LP2950), then you can get quite linear measurements to about 0.1C. No need for a fancy bridge circuit or special amplifier chips.

Quick edit - I should probably also have mentioned another pitfall I encountered. The PWM output from a typical micro controller is way too fast for an AC sold state relay, and possibly also for DC. (The Arduino PWM fequency is about 500Hz by default.) For AC the cycle time has to be slower than mains frequency (say 10Hz), so bit-banging an output pin is the way to go. I think this would be a better solution for DC as well if you go that route.

[rthorntn]

Thanks, could I just use a nice electromechanical relay?

Apologies I'm confused about your quick edit, how does PWM work with an SSR?

Bit banging (digitalwrite, gpio) is just setting a pin high as fast as you need or up to the limit of the mcu?

[kosine]

Yes, a standard coil relay would work fine, although you might get sick of the constant clicking!

PWM just sets the amount of on-off within each cycle. If your cycle was 1s, then when cold, your heaters would be on for 1000ms each cycle (100% PWM duty cycle). That would gradually drop to 0ms (0%) as you get closer to - or exceed - your required temperature set point. (So for 200C you could just subtract 5ms per degree of temperature increase: i.e., "on time" = 1000 - (temp * 5).)

You just need a 1s loop (or thereabouts): Measure the temp and calculate the on time, then set a pin high, wait for the required ms, then switch it back off. Wait for the remaining time to elapse and repeat. (It's essentially the same as analogWrite(), except you have complete control over the timings by manually "bit-banging" the pin on and off. As you say, as fast as you want up to the limit of the MCU clock frequency.)

Because you have two set points, you'd need to pass to a second loop after the first (with a different "on time" formula), and to get the required rise times you'd either have to write some more sophisticated code, or just fine tune your formula by some experimentation. If your temps and times aren't super critical, then a simple approach like this should get you in the ball park.

[beanflying]

You trying to build a Coffee Roaster? I own several Electric Drum Roasters BTW 

PWM from the Arduino to a MOSFET would do the Cartrige heaters nicely. There is already an Arduino plug in Board available that will integrate into either RoastLogger or Artisan for Temp Logging too and or controlling the Arduino Ramp profiles.  http://www.mlgp-llc.com/arduino/public/arduino-pcb.html

Even though Jim's boards are designed for AC elements and an SSR for Power DC is easy too.