T12 Soldering Iron Controller and a Compact Soldering Station Built Around It

Some time ago, I reviewed a basic soldering station controller and had some ideas about how to use it. But after playing around with it for a while, I realized it wasn't quite right for my idea.
Then I got my hands on a controller board that was just what I needed for my project.


First, a quick overview. I’d like to thank Vladimir, who sponsored this review by sending me the controller. I decided to show how it can be put to use.
The controller was bought on Taobao, but it’s also available on AliExpress for about $20 — link.

The seller offers two package options. The only difference is the connector for the soldering iron.
The full kit includes:
1. Controller
2. Vibration sensor, or more precisely, a tilt sensor
3. 10 kΩ thermistor
4. Encoder knob
5. Five-pin “aviation” connector.

The board is really tiny, though that’s not the only thing I like about it.


The board has a color display with a visible area of about 25 × 15 mm, an encoder, and a spot for installing the connector.
The board measures 78 × 18 × 8 mm, not including the battery.


But the key feature isn’t just its size—or even mainly its size. The board can be split into three parts, which is exactly what I needed, since my project called for a shorter controller board.
The idea is that you can simply snap the board apart, but I’ll say in advance that it’s better to score it first.
The only model marking is “T12 colour” under the display. The display itself is attached with double-sided tape about 3–3.5 mm thick.


All the contacts are labeled, which is handy both when connecting everything and just for reference. The board is also marked version 2.01, though that doesn’t mean anything to me.


It’s controlled by an STM32F030C8T6 microcontroller. To the right is the power transistor that controls power to the soldering iron; at the bottom are the buzzer and voltage regulator.
An RTC chip is installed between the battery and the encoder board. Interestingly, there’s also a spot on the board for an external battery connector.


The board’s supply voltage depends on the soldering iron being used—or, more precisely, its heating element—and is 24 V. But the board works just fine at lower voltages too, for example, from a 10–12 V supply.
The board’s own current consumption is very low, around 7–8 mA.


The control interface is very simple and easy to understand, even in Chinese, which is the default language.
The top shows the operating mode, set temperature, and the temperature of the chip or external sensor (if connected).
The actual temperature is shown in the middle, but since the soldering iron isn’t connected, an error is displayed. The month, date, and day of the week are shown to the right.
The supply voltage and current time are shown at the bottom.

There are four operating modes:
1. Normal: the temperature matches the set value.
2. Standby: temperature is 150 degrees. It can be activated manually with a single press or automatically by a timer; the automatic function can be disabled.
3. Boost: activated with a double press. It raises the temperature for the duration set in the menu.
4. “Off” mode: the display shows the year, date, day of the week, time, and the temperatures of the soldering iron and sensor. You can enter this mode manually by holding the button in standby mode, or automatically via the timer.

You can switch from standby to normal mode either by pressing the button or automatically via the soldering iron sensor (this function can be disabled). To exit off mode, press the encoder. When power is applied, the controller starts in normal mode.

For some reason, the boost mode didn’t make it into this photo, but I think you get the idea.


The settings are a little more complicated. First, I had to figure out how to switch the language from Chinese to English. That’s menu item 11, which you access by holding the button in normal mode.
There are thirteen menu items. To exit, press and hold the button (unless Wake set, item 2, is enabled).


Settings overview:
1. Temp Adj. Temperature calibration, performed at 350 degrees. That is, set the temperature to 350 degrees and measure the actual temperature. If there’s a difference, apply a correction.
2. Wake set: 0 — wake-up disabled; 1 — enabled via the soldering iron sensor and button; 2 — enabled via the sensor only.
3. Standby: time before switching to standby mode, 0–30 minutes
4. Off time: automatic shutoff time, 0–60 minutes
5. Boost temp: how many degrees to raise the temperature in boost mode, 0–100 degrees
6. Boost time: how long to keep boost mode on, 0–255 seconds
7. Buzzer, turn the beeper on/off
8. Alarm.vol, supply voltage threshold at which a warning will be issued, for example when running from a battery
9. OFF.vol, supply voltage at which the controller will stop heating; essentially, this protects the battery from over-discharge
10, 11, 12. Contrast. As the name suggests, this adjusts the contrast, although it looks more like a brightness control. But I think you get the idea. The default is 80; it can be adjusted from 0 to 255, but be careful: at 0, the screen goes completely dark.
13. DateTime. Select 1 and press the encoder to open the date and time settings menu.
14. Init, restore factory settings.
15. And this is what the main screen interface looks like in English.


The screen brightness was set fairly low by default, so I turned it all the way up. It looks noticeably better now.


I'll run a few tests later, but for now I want to tell you what I was hoping to achieve.
In short, I have a small soldering station—or, to put it simply, a temperature-controlled soldering iron. The key thing is that it's all built into a very compact enclosure and has galvanic isolation.
Yes, this tiny enclosure contains not just the controller, but also a power supply rated at around 50–60 W. I bought it in 2008, but by then the people who designed it had already stopped making them. It served me reliably all this time, and it still works fine, really, but I've wanted something more modern for a long time.


What held me back was that the project needed a compact controller. The previous one might even have worked, but it was long, and mounting it vertically on the enclosure just wouldn't look right.
The controller I showed above is perfect for the job, and now there's a second challenge: building an equally compact power supply, rated at around 60 W.

I won't go into much detail about the circuit, since it's simple, or describe the whole construction, since it depends on a specific enclosure and won't be relevant to many people. But I'll explain the general idea and the problems I ran into; it may be useful to anyone thinking of building something similar.

So, the power supply.
I decided to use a controller I'm familiar with from Power Integrations' TOP series. The other option was a simpler power supply based on the IR2153D.

You can find a circuit, for example, in the controller datasheet, along with information on selecting the power rating. I have to give Power Integrations' engineers credit for their well-prepared datasheets, application notes, and examples.

The power supply can be very simple or relatively complex, like the 70 W, 19 V power supply circuit shown below.


But what's even more useful, especially for beginners, is their design software. Based on the parameters you enter, it provides not only a circuit but also a complete power supply design, including component values, a sample PCB layout, transformer winding data, and instructions for building it. You can even change component parameters on the fly: the software recalculates everything automatically and provides recommendations.

As an example, I put together a 24 V, 60 W power supply design. It looks noticeably simpler than the example above.


Unfortunately, newer versions of the software don't work with older GX-series chips. But that's not really a problem; in most cases, you can use the design data for the HX series.

I decided to do it the old-fashioned way anyway, and even use the chip in a simplified configuration. In fact, I didn't even draw a schematic; I went straight to laying out the PCB.

I salvaged components from several devices, starting with a Mean Well power supply. It didn't last long and “died young.”


I usually praise this company's products, but not this time—I had plenty to complain about. Sure, it was a budget-series power supply, but it was loaded to no more than 50% and failed after about a year, if not sooner.


I salvaged the bridge rectifier from it, but for the switching transistor I needed its case, which is quite unusual. Mean Well's designers made the case with a built-in heatsink, so to speak. There's aluminum inside the plastic, which carries away heat through tabs that extend out and attach to the high-voltage transistor and output diode assembly. It is also soldered to the PCB at two points, where it acts as a shield to reduce EMI.


A few words about the power supply.
If I didn't know it was a Mean Well, I'd have guessed it was from a less well-known manufacturer.
It has everything you need, but the overall look is somehow rather unattractive. By the way, this power supply is based on a self-oscillating converter, which is extremely rare in devices in this segment.


The input and output capacitors are made by CapXon. The output capacitors are rated for 16 V, which was partly why they failed; the other reason was the manufacturer of this particular capacitor series.
For some reason, the output diode assembly uses ordinary diodes rather than Schottky diodes, and they're rated for 200 V. They're MOSPEC parts, which are more often found in cheap Chinese power supplies.
The board itself looks pretty rough too, especially the crookedly soldered LED and SMD resistor. It looks as if they forgot about the resistor and soldered it on later.


One of the key components in any power supply is the transformer. At first I wanted to use a transformer from an LCD monitor power supply board. They have enough power, but are designed for 12–15 V. I honestly couldn't be bothered to rewind one, and boosting it to 24 V would be unsafe for the controller.

In the end, I remembered that I had a disassembled power supply left over from one of my reviews, rated for 60 W and 24 V.
For a size comparison:
1. Transformer from some no-name 250 W ATX power supply.
2, 3. From LCD monitor power supply boards
4. From the LRC power supply.

The first transformer has no gap, so it won't work for my power supply and would also need rewinding. I think I'll use it in another project, though.


I originally laid out the board for one of the transformers from an LCD monitor, and I even have a nearly finished layout somewhere. But as I wrote, I ended up using a transformer from another power supply.
That's why there's no point in publishing the layout. While working on it, though, I thought about asking the manufacturer whether I could buy some of these transformers from them. They're universal, after all, for 12 V and 24 V power supplies; the two windings can be connected in parallel or in series.
If they're available to buy, then it makes sense to build this design.

After etching the board, I tinned the traces and then did a little "filing" so the board would fit properly in the case.
By the way, while assembling it I made a few changes to the layout just in case it might come in handy later.


Of course, you might ask what to do if building a power supply is too complicated, but you still want a compact station. You can use the MeanWell EPS-65S-24 power supply. In the photo, it's shown next to the board from the "donor" and my prototype.
Unfortunately, it won't fit in this case. The one that did fit was in an adapter-style form factor, but was noticeably larger—far too large, in fact. That's why I decided to build my own.


All in all, you could say it worked almost right away, but I messed up at least one thing from the start: I got the diode bridge connections mixed up. I noticed while installing it, so I had to cut the traces and reconnect them the other way around.
I used the calculator on the blog to work out the feedback network. The calculated voltage was 23.47 V, while the actual voltage turned out to be 23.35 V. I could have adjusted it upward; there's room on the board for an extra resistor, but for now I left it as is.

The main challenge when laying out and assembling the board was keeping its height down. The only tall component allowed was the transformer.
In the end, most of the parts came from various boards intended for parts harvesting.
1. The 10 µF, 400 V input capacitors were left over after modifying the light fixture drivers
2. The PWM controller and some other small parts came from a monitor power supply board
3. The transformer, fuse, and HTR20L150CT output diode assembly came from the LRC power supply
4. The diode bridge and output inductor came from the power supply whose case I used.
5. The output capacitors are new: Panasonic 470 µF, 35 V before the inductor Panasonic 470мкФ 35В, and Samwha RD 220 µF, 35 V after it

The TOP246Y PWM controller is configured in three-terminal mode, but runs at 66 kHz rather than 133 kHz, as I usually use it. That's because the transformer in the original power supply is designed for 65 kHz.


To test both the power supply and the controller, I modified the soldering iron handle connection from my review of a simple controller.
The seller provides a detailed description and wiring diagram, so there were no problems here.


I had to resolder both the connector wiring and the board, and I also mounted the thermistor on the board. I considered making a universal connection, but changed my mind.


First test: I set the power supply output to 23.5 V, and everything seemed fine. I also found out that at this voltage the soldering iron draws around 60–62 W, which is perfectly fine by me.


To test the power supply, I needed to install heatsinks: a large one on the PWM controller, cut from the power supply that I took the transformer from—in essence, a piece of aluminum angle. I found a small heatsink for the output diode assembly somewhere in my stash; it was probably from some switch.
You could say the second heatsink is tiny, but the diode assembly turned out to have very decent specs: the calculated voltage drop at 2–2.5 A is around 0.5–0.55 V.

I test-fitted the board in the case and realized I’d hit another snag, and this time a more serious one. The power supply case has snap tabs, and the top ones were obstructed by both one of the capacitors and the heatsink.


And of course, I checked that it worked and monitored the component temperatures.
For the test, I took another board from an old monitor and started desoldering its large components, heating up the copper pours in between—in other words, I simulated a lot of activity with a fairly heavy load.


At first I just heated the copper pours, but realized they barely warmed up (left photo). Then I desoldered various parts, including a couple of heatsinks. The hottest components were the snubber resistor and the output inductor, at around 50 degrees.
Of course, this wasn’t a complete test, since the power supply wasn’t in its case, but even so, it seemed to be behaving normally. I also found that the heatsink for the PWM controller was clearly overkill.


But there were problems too: it turned out that the power supply really doesn’t like repeated current surges. It was constantly switching between idle and maximum load, and beeping during the transitions, which got annoying.
I also found that the output voltage was fluctuating between 23.4 and 25.8 V. Naturally, that wasn’t acceptable, so I started investigating. It turned out that the standard 0.1 µF capacitor in the optocoupler’s power supply circuit was clearly too small. I added a 1 µF capacitor in parallel, and the problem disappeared completely. Incidentally, the software recommended a capacitance of 10 µF.
I didn’t have this problem when testing it with the Mean Well EPS-65S-24.


If we consider this board a test version rather than the final one, I immediately found three mistakes, which I corrected:
1. I routed the diode bridge connections incorrectly
2. I used too little capacitance in the optocoupler (and PWM controller) circuit
3. I placed the components too close to the top edge of the board.

I fixed all the problems except one: I had to desolder one of the input filter capacitors, so the input capacitance is now 40 µF instead of 50 µF. Just to be safe, I insulated the heatsink and the tops of the capacitors with special tape intended for insulating transformer windings, and coated the underside of the board with two layers of protective varnish.

I split the board into three parts. I cut a window for the display in the top of the case, and placed the connector and encoder at the bottom. Then I connected the boards so the wires wouldn’t get in the way of the transformer; there was about 3 mm of clearance between it and the cover.


The final schematic.


Unfortunately, while taking the power supply apart, I broke one of the four snap tabs. I thought the case was glued all the way around, so ideally it should be glued back together, but I decided to leave it as is for testing. The first thing I’m unsure about is the capacitance of the Y capacitor. I used a 1 nF one. I wanted a smaller value, but the ones I had at home were either 2.2 nF or 100 pF. The first was too much, the second too little; I think something around 330–470 pF would be ideal.
I put it together, plugged it in, and everything worked without a hitch :)


And the final assembly step: making the whole thing look reasonably presentable. I made the “front panel” with a self-adhesive label, covered with ordinary tape for protection—sort of a DIY lamination. To avoid breaking the delicate display, I recessed it slightly during assembly and fitted a piece of clear plastic from some blister pack into the opening. I was worried it would cause glare in photos, but it turned out great.
Incidentally, it’s interesting that the original power supply label was covered with a fairly thick (about 0.5 mm) piece of plastic, with adhesive on both sides.


Unfortunately, I couldn’t beat the previous station’s size record, but I still managed to pack everything into a volume comparable to a pack of cigarettes and get more functionality.
The only thing the new version loses on, in my opinion, is impact resistance. After all, an LED display is harder to break and easier to replace, but comfort comes at a price.


A few tests and comparisons.
1, 2. When switched on, the new station draws about 70 W until it reaches the set temperature, after which power drops to 9–11 W.
3, 4. The old one draws over 120 W immediately after switch-on, but within a few seconds power drops to 75 W, then to 50–55 W. Once it reaches temperature, power falls to about the same 10–12 W, perhaps a little more.


The new station reaches operating temperature incomparably faster.


In defense of the old one, I can say that although it heats up slowly, it accounts for thermal overshoot. That’s why it starts to “ease off” the heating about 20–25 degrees before the set temperature, and the subsequent temperature overshoot is only 5–6 degrees. It also does a pretty good job of maintaining a stable temperature during use.
The full version; I had to speed the video up 4.5 times.


I’d noticed for a long time that the old station’s controller gets noticeably warm even when idle, so I decided to compare it with the new one.
1. The two stations side by side after about an hour of operation. You can see that the old one gets hotter.
2. Most of the heat in the old one is concentrated around the power plug; the temperature easily reaches 60 degrees.
3. The new one is considerably cooler. For some reason, the highest temperature was on the display controller, but even there it was only 37–38 degrees.

Of course, someone will say, “What if you put the soldering iron under full load, say, by dipping it into a cup of water?” Friends, I don’t boil water for tea with a soldering iron. I have a gas stove and a kettle for that. The station can deliver around 60 W for a relatively short time (easily up to 10 minutes, I think). If that’s not enough, the problem isn’t the station; you probably need something more powerful or some additional heating. For normal use, the power supply has more than enough power.


The temperature can be set from 150 to 480 degrees by simply turning the encoder. Slow rotation adjusts it in 1-degree increments, while fast rotation adjusts it in 5-degree increments.
When the temperature reaches the set value, the controller beeps briefly. It gives the same beep if the temperature “drifts” by about 5–10 degrees during operation and then returns to normal.


Of course, I checked how accurately the tip temperature matched the setting, though without any additional heat loss.
I tested at 150, 200, 250, 300, 350, and 400 degrees.

Well, what can I say—not great. The difference reaches 20 degrees. I could blame measurement error, of course, but I generally trust the multimeter, and the thermal contact is fine too, since the temperature reads high rather than low.


But wait, the controller was supposed to support calibration.
We set it to 350 degrees, open the calibration menu, and see the same temperature for both the set and measured values, while the multimeter shows a noticeably higher reading.
I turn the encoder to increase the calibration value, while monitoring the temperature with the multimeter.
By making gradual adjustments, I got the temperature close to the set value, where it hovered between 350 and 351 degrees.


I run through the same range again. At 300–350 degrees, it’s fair to say the results are okay, but although the other settings have improved, I wouldn’t say by much—the nonlinearity is clearly noticeable. To calibrate properly, you need at least two points, say 150 and 400 degrees. Here, you have to settle for a compromise. I think the tip itself is partly to blame, since it’s unclear which controller it was originally calibrated with.


As I was finishing up the review, I realized that now I have a clock not only in my multimeter, but in my soldering station too :)
You can also see here how the handle’s thermistor temperature and the chip’s built-in sensor temperature are displayed. Unfortunately, you can't force the chip temperature to be displayed on screen. That would be more relevant for my use case, but as it is, the temperature is only shown when the soldering iron is switched off.



Conclusions.
For the most part, I liked the controller. For example, the "sleep" mode is convenient to use. I haven't gotten used to the forced mode yet, but I think it'll be just as convenient. I also like that these modes have additional settings, which makes the controls more flexible.
But as I wrote, I liked the design much more: the board can be split into sections and used to build a compact soldering station.

There are some drawbacks, too: the temperature setting isn't very accurate, and there's no memory for storing calibration data for different tips. To be honest, I set the temperature more "by feel" than by thermometer, and what matters most to me is that it stays stable. I have no complaints about that. Still, it would be nice if it were a bit better.

One unusual feature: the station has a clock. It's not really needed, especially since in my case it doesn't switch between standard and daylight saving time. But it turned out that the clock is actually handy when the station is switched off in software.

Now, about the device I ended up with.
First of all, it works, which is already a reason to be happy :)
Second, sure, you could ask what the point of this is and why not make it like everyone else does: a box, a cable to the wall outlet, and a cable to the soldering iron. It's simple: I've gotten used to this setup over many years, and for me, this turned out to be the most convenient option.
And third, I just wanted to try making a truly compact station, with a case about the size of a pack of cigarettes and its own power supply inside. Of course, I didn't achieve the same level of miniaturization as with my old one, but I still think I met the minimum goal.

Subjectively speaking.
I like everything about it. It solders easily, holds its temperature, doesn't get very hot, and now I have another clock :)

That's all for today, and thanks again to Vladimir for such a useful little gadget :)