Teardown: "HOTAIR GUN Desoldering Soldering REWORK SMD STATION 3 Nozzles RHOS Yihua 858D OZ"

Normally I wouldn't bother with a "teardown" post. They're popular, yeah, but I don't really see the point myself; it's like watching someone else eat ice-cream. But since I pointed out the potential & actual dangers of 'eBay special' hot air rework tools in my earlier post, I figured it'd be worth going through mine to uncover all the faults & explain them in a little more depth.

Opening it up, at first glance it doesn't look too bad:

Exploded view. (Note: comments and captions may not be literal…)

The fuseholder in the pic above is the dangerous temporary replacement one from my earlier post and yes, ideally should have heatshrink insulation on the tags. I'll certainly do that when I put a safe permanent replacement in.

Speaking of heatshrink…

More than meets the eye…

While it can often cover a multitude of sins, in this case it's not managing to even cover one. This little gem of a joint on the neutral wire was originally bundled up tightly with the rest & hidden under a cable tie. It was barely noticeable & would easily have escaped casual inspection.

By the way, as reported by others, this station has its neutral wire switched rather than the active:

Unlike others I'm not so sure this is against the regs, at least in Oz. Fixed wiring, hardwired appliances, & 'accessories' are all required to switch the active (or both A & N), but I don't recall that the same is required in portable appliances. I'll change it over later regardless; it's easier to do that than dig into the sometimes-contradictory regulations…

While we're down there, let's have a look at the earthing:

Crimp terminal, secured to the transformer mounting screw with its own separate nut. Better than other people have found, but still not compliant (or even great). The regs require protection "against mechanical and chemical deterioration and electro-dynamic forces" e.g. a shakeproof star washer to at least bite through the anodising/paint & ensure a good connection while resisting loosening.

(It also only has the earth wire from the mains lead connected to it, yet the wand has an exposed bare metal barrel. More on that later…)

The board itself (labelled "YH858D V6") is tidy enough, although it shows the lack of attention to component dress that's common in cheap gear. That said, it's not by any stretch the worst I've seen:

Excuse the Lindstroms; they keep muscling in to every shot…

Based on the unpopulated switches above the main chip (& other components elsewhere), it seems the board is designed for use in several variants. From the position of the switches and the fact that one of the unpopulated spots is for a TL431 reference, my first guess would be the 8786D iron / hot air combo unit.

The main chip is a Zilog / Samsung S3F94C4; a typical 8-bit MCU with 4kB flash and the usual complement of GPIO, PWM, and ADCs that we've come to expect from small MCUs. It's not dissimilar in concept & specs to a Microchip PIC or Atmel AVR chip - in fact, it seems that some manufacturers (e.g. Atten, Youyue) have designed their boards around an Atmel ATmega8, while Yihua and others have based theirs around this MCU. Unfortunately, they're not interchangeable.

(Although: EEVblog user madworm has Arduino-ised his Youyue with an ATmega168 by replacing the ATmega8 with an ATmega168 & custom firmware (EEVblog thread), and EEVblog user wguibas is working on an adapter board to replace the S3F94C4 chip with an ATmega168/328 & madworm firmware.)

On the left below the MCU is an Atmel-labelled 24C02B EEPROM (2kbits), while a bit further down and to the right is a bog-standard LM358 dual opamp. The former is presumably for retaining the last-used temperature setting &/or calibration data, while the latter is probably used to amplify/condition the signal from the temp sensor in the wand before feeding it to the MCU's A/D converter.

Rounding out the semiconductors is:

• An MOC3041 optocoupler (the white chip on the upper-right) for LV micro / HV triac & heater isolation. Allegedly Fairchild, though I'd take that with a Ghandi-load of salt.
• A BT16-600B triac (just above & to the right of the opto) for controlling the heater.
• An LM7805 (top-right) regulator to generate a 5v rail, and
• a TIP122 NPN darlington transistor (centre-right) for controlling the fan.

The triac, regulator, and transistor all carry what looks like a legitimate ST logo, although the packages & finish don't look quite right for ST. I won't say they're fakes, but…

Looking a little closer, the wiring on the left of the picture is the cable to the wand. While it may seem a bit lightweight, in reality it's heavier than it looks & I'm actually OK with it. The earthing of the wand, though, is an issue. The problem is that the earth wire from the wand is only connected to the front panel - and that through a threaded stud that looks like it's painted with primer:

Earth wire from the wand is connected to a front panel standoff.

Remember last time I measured the resistance between the earth pin and the wand barrel as ~6Ω? Well, from the wand barrel to that terminal measures 0.6Ω. The other 5Ω or so is entirely due to the poor connection between the two points through the painted threaded post and dipped/powdercoated/painted front panel to the chassis, via self-tapping screws through clearance holes in the front panel. I reckon a proper earth wire between the two will sort that out - the earth at the wand still won't be great, but it'll be an order of magnitude better than it is now…

Speaking of which, others have found the earth connection in the wand itself to be dodgy. Let's have a look at that:

Wand wiring

Wow, that's much better than most have reported! The board between the cable & fan/heater/sensor is clean & tidy and, most importantly, the earth wire is crimped to the barrel rather than being held there by luck and the sort of good intentions that would bring tears of joy to Hell's Department of Main Roads. The mains separation & creepage distances on the board are best described as 'adequate' - without actually measuring, they look to be standards-acceptable (though I suspect a lot of hobbyists who've never seen the AS/NZ or IEC standards would find it "totally unacceptable!!").

Strangely, there doesn't seem to be the usual reed switch glued inside the wand to detect when it's in the cradle. The wires connected to the "handle" track on the board simply disappear into the heatshrink at the end of the heater element. I know it does run through a cool-down process when returned to the cradle, so I suspect it uses a mercury or ball tilt switch to detect orientation rather than actually detecting the cradle. That could make it hard to use in some situations (i.e. upside-down), so I'll have to double-check.

(Post-modification update: nope, it works in all orientations, so it's not a mercury/ball switch. The cradle has magnets mounted on either side - as I discovered when it picked up all sorts of crap off the bench while I was working on it - so it must be a reed switch under the heatshrink.)

My overall impression? While it's not top-notch quality, for <AU$60 I didn't expect it to be. It is, however, much better constructed than I expected it to be (give or take the shitty fuseholder, which at least was safe until it broke). I'd class it as 'more or less safe', rather than one of the 'dangerously unsafe' examples posted at EEVblog and elsewhere.

Issues I … errr, found issue with:

• The fragile fuseholder I keep mentioning - I've already replaced this with a temporary fuseholder, and a proper standards-compliant safety type will be fitted as soon as I buy one.
• The partially-exposed joint in the neutral wire - easily fixed by remaking & re-heatshrinking the joint.
• Earthing - although much better than others have found in similar units, it's still far from perfect. Biggest issue is the lack of direct connection between the main safety earth & wand earth, while there is still an issue with the lack of clean metal-to-metal connections between earth & chassis points. These can be fixed by running a separate earth wire from the board to the safety earth point, cleaning the paint from around the chassis earth connections, and the use of external star washers on chassis earth points.
• Switching of the neutral - though I don't believe this is against regs, it's nice and 'proper' and generally good practice to switch the active. Since I'll be fixing the neutral joint, it's only a little more work to swap the switch into the active wire.

Pics of the fix 

To improve the earthing, I started by removing the transformer & sanding away some of the paint underneath one of the transformer mounting tab:

I also sanded away the anodising/lacquer from the mounting tab. I re-installed the transformer with a shakeproof star washer between case & mounting tab, and a small flat washer under the nut.

The existing earth wire from the mains cable and a new earth wire for the wand were then secured using a large flat washer & shakeproof washer underneath, and a small flat washer & nut above. A dab of external thread lock was then used to secure the earth retaining nut.

You can also see the joint in the neutral wire, which now runs straight from the mains cable to the board. The joint was remade using a small uninsulated crimp butt splice and insulated with heatshrink.

The new earth wire was then run to the board, where it meets up with the wand earth:

Originally I was going to solder the earth wire to the board, but in the end I used another ring terminal here. After removing the board, I added a star washer between the standoff and the bottom of the board. I then re-fitted the board with the wand earth terminal & safety earth terminal secured by another star washer & nut. Again, all the mounting nuts were secured with thread lock.

You can also see the rewired switch, which is now in the active lead between the fuse and the board.

This has made a big improvement to the earthing. Previously, the resistance from the mains plug earth pin to the chassis was <0.2Ω, and to the wand barrel was ~6Ω. After the modifications above, the earth resistance is now <0.1Ω to the chassis and ~0.7Ω to the wand. While it's probably not to spec, it's a definite improvement…