"Poor Man's" Solder Iron Temperature Controller

This describes the poor man's solder iron temperature controller. Or rather a temperature controller for the DIY'er. Note this is a Work In Progress for 120V, if you want to finish it, sure, else it should be complete for 220V. This was stolen from Laser Pointer Forums but really from Erwo, a Yugoslavian magazine. This is a recreation of the schematic and layout for this controller and an attempt to convert this for 120V use.

I have an Ungar 4624 rework station that is already temperature controlled but also have a Craftsman 45W uncontrollable iron. This iron gets HOT, maybe TOO hot. I don't want to get rid of it because I have an extra heating element for it as well, so I decided to look for some way to control the heat.

Due to the nature that the heating element has a positive temperature coefficient like most metals, the hotter it gets the more resistive it gets - which is good as it will self regulate temperature to an extent as it won't runaway like semiconductors. As it gets hotter the resistance goes up - less current flows - less heat is generated - cools down... resistance goes down - more current flows, etc... Great if you want to run it at one fixed temperature but not so great for soldering heat sensitive components!

Note you should use an iron that gets "too hot" - because this can only regulate downwards. This will limit the heat the "too hot" iron can get. Ideally you want to use an iron at least 30-40 watts. Do NOT use this for a solder gun - as this has semiconductors, use with an inductive load can have unexpected consequences. Besides, solder guns, with a bit of experience, you don't need a temperature controller.


Note that the screen shot may not be the latest rev of the xpcb source. I am still making final edits on it to improve safety and fabricatability.

Files

source schematic in gEDA gschem format and pdf format. Appears to be UNIX only for now but schematic is same as on the other site except annotated in English.
SCR Symbol that's used but missing from gschem libraries
printed circuit layout in xPCB GNU PCB layout editor format. There is a Windows binary!

Theory of operation and circuit notes

This circuit tries to compare the resistance of the heating element versus a fixed value created by the voltage divider made from R1 and R2 (Ignore D1 - it's small enough we won't worry about it. It's used to make sure the op amp only make the comparison with the part of the wave when the heater is always on and not get negative voltages). Also note... isn't it weird, the voltage at R3 depends on line voltage? Well not to fear, it's being compared to the "reference" voltage that is also affected by line voltage! So it won't have as much effect as if we were comparing against a constant voltage source. (If you stare at it enough... Mr. Wheatstone will pop out!)

Now I don't have a good way to measure temperature and resistance so I search on the web and find a table:

Nichrome properties (copied from Wiretron):
°F68200400600800
°C2093204315427
NiCrA %increase00.82.03.34.8
NiCrC %increase01.73.55.26.9

As can be seen the coefficient of resistance related to temperature is very small. I measured the cold resistance of my 45W 120VAC/DC iron to be around 305Ω at room temperature, so that will be our baseline. It and R3 will make a voltage divider and the voltage will be Vmain*R3/(R3+305) when cold. Assuming the heater uses nichrome-A, using the above table, at 600 °F, the voltage will be Vmain*R3/(R3+305*3.3). Using a 15Ω resistor for R3 (seems most appropriate for a 45W iron), 120*15/(15+305)=5.625V vs 120*15/(15+(305+3.3%))=5.453V. This is less than 200 mV difference for over 500 degrees difference! The potentiometers will cut this down even further. Since the gain of the LM339 is very high, a few millivolts will make a difference - then again we will have to watch out - the offset voltage will play a big factor in this. Looking at National Semiconductor's data sheet on their LM339, we can see there's an offset voltage of ±3mV. Fortunately we're dealing with hundreds of millivolts so this won't be too bad (plus this is poor man's - it's close enough :). However perhaps a better op amp may produce more accurate control... Alas some circuit modification would be needed to use op amps instead of a comparator-driver like the LM339.

The "reference" of R1 and R2 will be Vmain*10K/(10K+330K) or around 3.53V. Comparing this with the voltage we got from the other divider will determine whether or not to turn the heater on for the other half of the cycle. A 47KΩ resistor and the 1µF capacitor low pass filters it and holds the voltage through the second phase of the AC signal. The heater is powered by one half of the AC signal due to D2, so it will always be generating at least half the wattage of the iron without control. The circuit will turn the other half of the AC cycle through D4 and the SCR if the resistance is too low - meaning, if it's too cold. The LM339 will let go of the output, allowing the 100KΩ resistor to supply current into the transistor turning it on. The transistor then pulls the gate to the cathode of the SCR, turning it on and locking it on for that phase when the temperature too cold. A neon pilot light is on when the second half of the wave is enabled - use much like an oven - when the light goes out, it has reached the right temperature. Once the phase ends, the SCR will shut off and measurement starts again.

On the other hand, if the element is too hot, the resistance will higher and the comparator will have lower voltage. It will make the comparator output pull the output down. This will starve current flowing into the base of Q1, shutting it off. R11 will then hold the SCR gate at the same potential as the anode, keeping it from conducting.

Since this circuit was originally designed for 220VAC, operation on 120V that's typical in the US and Japan will require modifications. While I believe that most of the components can be used as-is, this is still being analyzed. The notable exceptions: R3 must be reduced considerably. Because we're going from 220V to 120V, we need to divide by four to keep the ratios correct - so this resistor needs to be around 30 ohms in the example to keep the ratios right. This is because power is proportional to the square of voltage, and to keep the voltages in the divider at the same ratio as the reference divider while keeping the same power draw, we need to reduce the resistance by not two, but four (actually, (220/120)2 but close enough). Note that many of the resistors need this correction but the circuit may still work properly with the 220V values and dissipate less power as it does not drop as much voltage with 120V mains.

Now with R3 modified, we still need to bias the variable resistor tree which I think is fairly poor but there's not really any way around it due to soldering iron tolerances. We need to calibrate to the resistance of the resistance wire in the iron which differ from iron to iron, and thus need the potentiometers.

R7 and R8 also need to be modified for 120V use, likely along with variable resistors. As of right now I think R7 and R8 need to be halved but never below four times R3 or so, else you'd be dissipating a lot of heat in that resistor or the pot. The pot itself may need to be halved too but you may be able to just use the 500Ω anyway. The divider of R1 and R2 will still divide properly at 120V so it may not need attention. The hope is that the voltage remaining with the 33:1 divider (at 120V it's about 3.53V) is within range of the LM339 amplifier. Both the reference and the sense legs need to be fairly close else the amplifier will have no effect. R11 will likely also need to be changed to around 22KΩ but only needs to be ¼W as far as I can tell. R6 also may need to be reduced, probably to around 20KΩ (though it seems at 220V it would be dissipating more than 1W...)

I changed the circuit to use a TO92 2N3904/PN2222/2N4401 because these are common in the USA where I live, instead of the BC548 in Europe. Just reverse the flat side to use the BC548. Most TO-220 SCRs should be the same pinout so that was left as is. Crunching the numbers it's possible to use a TO-92 but be safe and use at least a 400V TO-202 SCR.

Building the controller and DANGER

Now this is the disclaimer: I AM NOT RESPONSIBLE FOR ANY DAMAGE YOU CAUSE TO YOURSELF OR YOUR BELONGINGS FROM BUILDING/USING THIS CIRCUIT!!!

This is a very dangerous project as it deals with 120V directly. The main reason why I don't like perfboard is reemphasized in this: since we're dealing with lethal 120V we can't really play games with loose wires. So an etched board it is. Please use the stronger fiberglass 1/16" FR4 to build this. Also keep in mind this circuit is LIVE. You may assemble this into a GROUNDED metal case with 3-prong plug, else this needs to be specially built to ensure safety. I don't know what it would take to get this UL listed but the potentiometers are VERY dangerous as they are NOT isolated from line voltage. You'll likely need plastic shaft, plastic potentiometers so the user has at least two forms of insulation from the resistive element. You'd definitely need a huge knob that will have one additional insulation layer along with the special potentiometers, and the circuit needs to be in a plastic box as the second insulation layer - to be double insulated. Better yet, this should be run from an isolation transformer like this one from MCM/Tenma. People should always be working on electronics that are connected to live wall power through an isolation transformer.

Note carefully that this layout is not quite the same as the original. Besides the fact that there's component annotation on mine (along with the fact mine might be slightly smaller), I changed it slightly to remove the two jumpers that are hard to see in the original photos, and have no traces that goes between components that have less than 0.2 inch clearance between leads - it should be easily prepared via toner transfer fabrication. Also the link through the IC is no longer needed, I found another way to get to the emitter of the transistor without cheating through the IC, and WATCH OUT, the 470µF capacitor is inserted the OPPOSITE direction as the original!!! Do NOT put the capacitor in backwards and have it explode if you're using this layout with the original assembled image!!!

Since I redrew this so I could fabricate the PCB, the circuit is available. The schematic capture is in gEDA schematic format and the PCB is drawn in xPCB (both Free Open Source Software). As source files are available you are free to generate Gerber files if so desired and get a third party house to fabricate the board.

Note that the theory of operation of this is very similar to most thermocouple-less temperature controlled soldering stations, except they have "integral" isolation transformers - they tend to run at 24VAC at a few amperes - to reduce the risk of electrocution since all the circuitry is isolated from the mains. Because the heating element is the same as the sensing element, the tip itself may not get to the desired temperature before the heater turns off, so that's another drawback of this circuit. Note that most of the "real" integrated sensor/heating element solder irons also are designed differently - the tip and "thermal mass" of the iron itself is minimized so that the tip stays as close to the same temperature as the heating coil, and thus the sensor, as possible. The tip tends to be well within the heating element. This will not be the case with "cheap" irons and thus you will see heavy deviation in thermal control. The best type, of course, is with a discrete sensor embedded within the tip and the "T12" Hakko is one such.

Decent cheap solder tip You'll find two basic kinds of cheap noncontrollable irons out there, one has this fat sharpened short piece of metal wire that's male threaded and sticks less than a half inch into the hole of the iron, with a set screw to tighten. AVOID THIS TYPE for temperature control, it's a horrible iron to begin with. This type has absolutely horrid temperature equalization and the tip can have many tens of degrees delta from the heating element. The other cheap type has a "fat" female end on the replaceable tip. The iron has a threadded male end that goes into the tip with no set screw. This type is better for temperature equalization and thus more ideal for this cheap controller though still not great - at least it has more thermal mass.

There's a reason why this is called "Poor Man's" solder iron temperature controller!!!


Updated 22 Oct 2015 - blurb on solder irons
Updated 07 Jan 2018 - typos and corrections