EEVblog #1171 - Yihau WEP SMD Rework Station Meltdown!

[EEVblog]

What caused David2's Yihau/WEP 898D+ SMD rework station to literally melt down?
Autopsy time.

[Zbig]

I have a similar thing, only mine's branded "Zhaoxin", the enclosure is plastic and the soldering iron's DIN-style connector is also made from plastic (so perhaps a knockoff after all?). Funny thing, mine went thermonuclear on the soldering iron part. Turns out, when the thermocouple in the soldering iron goes open circuit, the control circuitry isn't smart enough to detect it and just applies the power full time instead of the usual pulsing PID action. (Un?)fortunately, the fault in mine wasn't nearly as spectacular as David's: the soldering iron's tip merely went red hot and smelling bad until I noticed the fault.

[johnlsenchak]

That squirrel   cage  fan  looks like  something  out of  a   cheap  laptop  from around   10 years ago


Good to hear  Dave  #2  in this video , I though  Dave kept him  locked up in the bunker  LOL

[ataradov]

That thumbnail smoke and fire overlay pays for itself

[richnormand]

In the portion where you look at the pcb plugs with the mains line coming in, next to the heather plug, transformer and then next to the triac... around 21:00 to 21:41 ...hmmm

Is it one of those units always on standby without a real mains switch?

If so, you might want to look at this thread:
https://www.eevblog.com/forum/chat/electronic-workbench-fire/

My solution is posted at #22. A real mains switch with a neon light and fuse.

I know this not related to the initial issue presented here but everyone that has a similar hot air unit should examine it and check it.  Should it be only a triac holding the mains from the heather (or via a transformer)  you might want to consider what could happen due to a line transient in the middle of the night gets to the triac full on or shorted. In mine such a failure would not turn on the fan motor in the base as it is on a different circuit.

So check to make sure the "power off" on the front panel really powers down the puppy for real.

[mariush]

I don't see how you can like the internal build of such a product.

There's connectors glued with loctite (so they won't break) on one end, but there's wires soldered directly to the circuit board on the other end in the hand piece.
There's also wires soldered directly to the main board, so you have both connectors AND wires soldered directly

There's several kinds of connectors, i think i counted at least 3 different ones. I get that they need one with bigger separation between contacts for the high voltage stuff, but they could have used the same one for all the other connectors, with the same gauge wire ... there's red and black wires, there's the white wires with dashes of red, there's the connector with thin black wires 
I would have used just one 3 pin connector

Then you have DIP socket for the main mcu but instead they could have used a double sided circuit board so they could solder the actual LED digits properly ... look how the pins go through the pcb and how they soldered those digits from the other side.

They went through all the effort to make everything through hole, only to have that single chip there surface mounted - guess that added an extra manufacturing process, an extra pass through a machine, and programming the weird orientation (not 45 degrees) ... unless they soldered them by hand (wouldn't surprise me)
If you give up and decide to use a surface mount chip, then maybe spend a few more minutes and put some pads for surface mounted resistors or maybe use surface mounted opamps or whatever and clean the board a bit...

There's a couple of 1w or whatever (bigger resistors) practically touching that to220 chip on the side, which is probably the regulator for the microcontroller.. and those resistors also touch or are very close to that capacitor cooking it over the life of the product... not sure what those resistors are used, i guess either to drop some voltage before the regulator does its job, or maybe as current sense (if it's regulator i guess they wanted to save money on a heatsink?)

There's also the triac heatsink touching that film capacitor, but I'm not sure that film capacitor is that much affected by heat

I also think it's kinda small area to run 700w+ through, where those mains connectors and where the triac is located.

And what's with zip tie squeezing all cables together, is it even allowed to bundle mains voltage and low voltage like that?
 

[tautech]

Jeez, you lot aren't having much luck with soldering gear lately. 

Tell David to get a K style (knife/skew) Hakko tip for prototyping with passive SMD's as they reach/bridge both pads so you can sweep passives off easily and use the same tip to tack another one in.
No need at all for hot air.

[Dundarave]

Since the thermocouple relies on the fan to actually detect the output temperature, i.e. it normally detects the temperature of the hot air passing over it, the failure of the fan pretty much guarantees that the heating element will melt down before the thermocouple actually gets hot enough to regulate anything.

Without any air passing through, the element would have to get way, way hotter than normal before the thermocouple picked it up, due to the lack of the heat "coupling" via the air stream.

[orion242]

Without any air passing through, the element would have to get way, way hotter than normal before the thermocouple picked it up, due to the lack of the heat "coupling" via the air stream.

Nonsense.  Its close enough to the element that if it got red hot, it would have read off the charts.  Maybe the wires melted and it went open or shorted, but it should have seen a problem.  And that hand piece is in the holder upright...heat rises to this sensing element...

Not buying it.  It should have read thermal nuclear meltdown temps without any fan.

Wondering if its a sensor or software bug, where it went over range and failed to a low value or something retarded like that.  Any issue with that sensor in close contact as it was, should be a shutdown.

[texaspyro]

Did you check for a shorted/always on triac?   The fan on these plays very little part in temperature control.  The heater is PWM modulated via a PID loop to set the temperature.   If the triac circuit goes bad, the heater can always be on.   

The temp sensor is close to the heating coil.  A bad fan should not cause thermal runaway.  The thing will heat up and the PID loop should still control the temperature.  A bad temp sensor circuit could cause thermal runaway, particularly if they don't have a open/shorted sensor detector.

Dave, check out Madworm's thread on open source firmware for these systems.  It started out as a replacement firmware for the ATMEGA328 that was used on most of the original units.   Later ones used the Samsung chip and then all sorts of other processors...  there are now adapter boards to stick a MEGA328 in most of the clones.

[amyk]

Triac short is my guess too. It doesn't necessarily need to be a full short detectable with a multimeter, but just enough to have it turn on when mains is applied to it. A tiny anode-gate leak would be sufficient.

There's also the fact that sometimes semiconductors which shorted due to overload will appear open once they've cooled down.

Testing the triac with a suitable load, at mains voltage, will definitively tell whether it was stuck on.

[Dundarave]

Nonsense.  Its close enough to the element that if it got red hot, it would have read off the charts.  Maybe the wires melted and it went open or shorted, but it should have seen a problem.  And that hand piece is in the holder upright...heat rises to this sensing element...

The thermocouple looks about 1/4" away from the front coil.  The handle was horizontal with no air flow.  Only radiant heat from the bit of glowing-red front coil would be beaming onto the low-mass thermocouple (i.e. not conductive, and not convective), while the rest of the coil is also red hot, and with the heat accumulating in the steel tube.

I speculate that the thermocouple was possibly regulating that front coil at (what appeared to it to be) 350 degrees or whatever from 1/4" away, just from the radiation heat of the small bit of coil, while the rest of the iron soaked up the heat from the other 5 heating coils and got red hot as a result.

[floobydust]

TEST THE TRIAC. It's got brown marks and is a high failure rate part on chinese hot air stations, due to either its tiny heatsink or a low quality semi.
Heat can also get stuck on due to a ground-fault with the heating element touching (ground) the nozzle metal sleeve, as Dave mentioned.

I found fan fail is detected in the Youyue 858D+ firmware in MK1840D3 MCU.
Since the thermocouple is close to the heating element, it does read reasonably, even with no airflow.
The firmware looks at the derivative of temperature.
If temperature is climbing too fast when the heating element is on, the station shuts down.

I suggest testing the fan with 24VDC power, not ohms.

Never use these hot air stations unattended, and never leave them plugged in. Never unplug the wand with the station plugged in to mains.

[orion242]

Nonsense.  Its close enough to the element that if it got red hot, it would have read off the charts.  Maybe the wires melted and it went open or shorted, but it should have seen a problem.  And that hand piece is in the holder upright...heat rises to this sensing element...

The thermocouple looks about 1/4" away from the front coil.  The handle was horizontal with no air flow.  Only radiant heat from the bit of glowing-red front coil would be beaming onto the low-mass thermocouple (i.e. not conductive, and not convective), while the rest of the coil is also red hot, and with the heat accumulating in the steel tube.

I speculate that the thermocouple was possibly regulating that front coil at (what appeared to it to be) 350 degrees or whatever from 1/4" away, just from the radiation heat of the small bit of coil, while the rest of the iron soaked up the heat from the other 5 heating coils and got red hot as a result.

Never mind the ceramic material that holds everything together and conducts the heat.  Still don't see it with that separation between element and sensor. Not talking about being 100F over temp, more like 500F.

[SeanB]

Transformer looks like it will fit the Weller, and probably be an upgrade, as it likely has an internal thermal fuse as well.

[jesuscf]

My W.E.P 858D had the mains fuse connected to the neutral wire.

http://www.williamosman.com/2015/12/dangerous-wiring-fault-858d-hot-air.html

[EEVblog]

TEST THE TRIAC.

I did, it seems fine.

[Psi]

I had a very similar hot air station fail in this same way. Heat with no fan power.
I saw it happen and quickly hit the power switch.
Luckily i got it powered off in time and the element didn't blow.

On my unit the fault was caused by an intermittently faulty rocker switch on the front panel (the one for hot air on/off).
Failure mode was having only one side of this DPST switch go open.

Looking at your one it also has a DPST switch in the same place, so this is the first thing i would check!
From memory one side of the DTST powers the fan motor.

[ilporcupine]

My first thought was as Dundarave. Especially if that tcouple was on the bottom side of the element when it was in the holder.
It already is set to quite a high temp, when under normal operation.  Even if it was still trying to regulate, would be MUCH hotter than with fan on.
TRY IT!

[SeoulBigChris]

David #2 made an interesting comment. When he turns off the unit by resting in its cradle, the fan turns off. All the hot air rework stations I’ve used have a delay. When you turn off the unit, the fan keeps running for some time until the element cools down. But my Hakko 474 unit is over fifteen years old, so maybe more modern designs do this cool-down phase differently.

[Zbig]

David #2 made an interesting comment. When he turns off the unit by resting in its cradle, the fan turns off. All the hot air rework stations I’ve used have a delay. When you turn off the unit, the fan keeps running for some time until the element cools down. But my Hakko 474 unit is over fifteen years old, so maybe more modern designs do this cool-down phase differently.

Mine, from the post #2, is < 8 years old and does the same. When you place the unit down on its rest, first the power to the heater is cut but the fan keeps running until the temperature falls below 100 degrees C or so.

[Fungus]

My theory:

If air isn't flowing then the sensor can't measure the temperature of the heating element correctly - it's a couple of centimeters away from it.

If the gun was oriented with the sensor at the bottom then it might be a lot cooler than the element there - heat is going upwards away from it.


nb. There's a fuse on the transformer primary!

[GreggD]

You can't just "ohms" check the fan. It likely needs higher voltage before it connects a coil. Test it with 24V.

[Urs42]

At least the WEP Station has a good earthing connection on the handpiece, attached is a picture of my handpiece.

[IanMacdonald]

I have the Atten 858 and I was under the impression that they had a thermal cutout, but looks like this one does not. Now I'm worried. I think I'll check mine and fit a cutout if it has none.

Only thing is it's difficult to get cutouts rated at more than 240C, but if it's fitted behind the heating element in the cold air flow that should be OK.

[free_electron]

fan is a bldc type..

[eNBeWe]

I would chime in with everybody suggesting that the fan failed (due to mechanical stress, maybe).

Last year I started tracing a similar 858D rework station (without the soldering iron part). In my station (which looks very similar) the transformer creates 10V and 30V (both AC). The 30V are then used to power the fan. In the end I measured about 28V at the terminal for the fan that is rated for 24V. I guess constant overloading will take its toll.

If you are interested have a look at the pictures from my teardown and my reverse engineered schematics: https://blog.enbewe.de/tags/858d/

[Cnoob]

I have a WEP 8508D which has an air pump inside the case. Which I didn't remove the transit screws before I first, fired it up.
The pump is rubber mounted. Anyway I use cable tidy around the hose, (a bit like spiral band but thinner walled.)
which is ideal in preventing the hose from kinking.

Below is the ebay address for the cable tidy.

https://www.ebay.co.uk/itm/2-Metre-Cable-Tidy-PC-TV-Wire-Organising-Wrap-Spiral-Office-Home-s/122714804493?epid=2254727580&hash=item1c925f510d:m:mOPVoN_ro_qtGNK1XGXR4-A:rk:1:pf:1   

[Wolfgang]

What caused David2's Yihau/WEP 898D+ SMD rework station to literally melt down?
Autopsy time.

Dave please stay away from crappy soldering stations or your building insurance will cancel your contract.
No Wellers and other crap anymore. Play safe ! 

[SiliconWizard]

My theory:

If air isn't flowing then the sensor can't measure the temperature of the heating element correctly - it's a couple of centimeters away from it.

That was my first thought as well. Wondering what temperature it's able to measure when the fan is off compared to when it's on, would be interesting. Surely even if air was not flowing, the air temperature around the sensor must have gotten pretty high...

It's a very lousy design anyway. Doesn't seem to have been tested if the fan fails. No detection of fan failure. No additional protection from overheating.
I happen to have an Atten 858D+, don't know if it's any better. It looks very similar. I'll be cautious when using it now.

[floobydust]

My Youyue 858D+ can dish out up to 33V to the 24V fan... don't use it at full speed!

They're just hair dryer heating elements. I measure:
120V 6.1A 19.8 ohms; 727W.
240V 3.4A 71.6 ohms, 805W.

That generic heatsink I've measured 32°C/W and the usual rule of thumb for TRIAC/SSR dissipation is 1W per amp.
40% heater duty cycle gives over 100°C so I could see a heat issue for 120V systems where the TRIAC runs hotter.

I've added a GFI powerbar to my workbench, I wonder if Dave has one.

This is a repost of my schematic for the Youyue 858D+, including the wrong switched neutral, which was fixed.

[fsr]

Why not try to power the fan, to see if it works correctly?

[wraper]

It's 'Yihua', not 'Yihau'. many people likely won't see this video when searching before buying this station.

[free_electron]

These kind of system should check the current to the fan. If excessive current or no current is detected : shut down the heater.
That way if the fan fails short or open, the heater will shut down. all that is required is a small resistor and a sense amp. the micro can do the rest.

I am wondering why people still use triac's for this kind of stuff these days ... stick a mosfet in a bridge rectifier and you have much better control.

- you can switch on much closer to at zero cross.
- you can turn off at any point in time.
- you create less electrical noise due to phase chopping
- your load does not get hammered with 'hard' voltage spikes every time , it gets a nice clean ramp-up

[floobydust]

Here are some fan current waveforms. Easy to pick off with a data slicer and very expensive LM358  Another EEVblog member instead made an analog voltage for RPM. Typically 200-600Hz.  There is a 1/f relationship to RPM and I think it's a 4-pole motor.

But one problem I found adding a digital tach to the fan circuit, is the MCU needs another timer/counter to measure RPM. Arduino ATMega328P is already busy doing LED mux and PID so it would be a bit of work.
Easier to go bluepill or ATMega328PB, then you want a better display, and it never ends...

[Dundarave]

These kind of system should check the current to the fan. If excessive current or no current is detected : shut down the heater.
That way if the fan fails short or open, the heater will shut down. all that is required is a small resistor and a sense amp. the micro can do the rest.

Easier, cheaper, and ultimately more reliable would be a simple thermal cutout mechanically bonded to the metal tube part of the handle. 

Then if the tube gets hot enough to melt the plastic handle, for any reason, the device would positively shut down.

[wraper]

These kind of system should check the current to the fan. If excessive current or no current is detected : shut down the heater.
That way if the fan fails short or open, the heater will shut down. all that is required is a small resistor and a sense amp. the micro can do the rest.

Easier, cheaper, and ultimately more reliable would be a simple thermal cutout mechanically bonded to the metal tube part of the handle. 

Then if the tube gets hot enough to melt the plastic handle, for any reason, the device would positively shut down.

That's how decent stations are made. They have thermal fuse just before the heater. If heater is powered without airflow, it will be simply disconnected.

[amyk]

I am wondering why people still use triac's for this kind of stuff these days ... stick a mosfet in a bridge rectifier and you have much better control.

- you can switch on much closer to at zero cross.
- you can turn off at any point in time.
- you create less electrical noise due to phase chopping
- your load does not get hammered with 'hard' voltage spikes every time , it gets a nice clean ramp-up

One word: cost. It's switching a resistance heater in this application, no need for extra complexity... and if the MOSFET shorts out, you have the same problem.

[TechSpecB]

Not just soldering irons. Fans fail, sensors go bad, circuits short out and chips fail. Redundant safety devices are necessary!
I have seen fans fail to blow but still happily draw reasonable current. With a blocked inlet or stuck rotor, the fan will still draw power. Once found a cheap fan where the fan blades had become disconnected from the motor. A friction fit had failed.
The heater current drops during operation. Initial resistance of a cold heater will be highest, and the fuse will be sized accordingly to limit nuisance failures. The heater would not overload the fuse unless the ceramic support failed allowing the heater element to short.
To prevent a fire, a thermal fuse would be added in circuit with the heater. Best location would be in the power lead that goes through the center of the ceramic to the far end of the element. Value would be much higher than normal operating temperatures, but well below meltdown. Without this I doubt the device is actually CE rated.
Dave, please check: is there is a thermal fuse? It may have been ignored as it doesn't affect normal operation.

[Fellonium]

I was browsing Yihua's website for a bit after watching this video. It seems they were already aware of this failure mode (fan dead, hot air fries itself). Contrary to the older models, some (not all) of the more recent models actually have the following listed as a feature:

Protection function for air gun fan malfunction: at the process of using the air gun, the air gun features shutoff protection functionality that automatically turns off the heating element when the fan is stopped to protect handle and ensure safe operation.

Good to know they already figured out this was a potential issue and "fixed" it. YIIHUAAA!! (Sorry, couldn't resist.) Not so good is that this probably means quite a few people almost burned down the house.

Example: http://yihua-gz.com/Products_detail.asp?id=811