Aixun T3A misbehaving on grounded PCBs

[mahi]

thm_w: Indeed, the Weller WXR 3 switches DC as well, but one major difference is that the Weller switches a non-filtered full-wave rectified signal at the zero crossings whereas the Aixun T3A switches a regulated DC signal. So, one results into a smooth(ish) full-wave sine and the other into a harsh square wave.

And even the full-wave sine of the Weller WXR 3 seems to be a step back from the older generations that used triacs to switch the transformer AC directly (also at the zero crossings). When heating up I can hear a faint buzzing noise from the WXMT soldering tweezers. I guess the sharp "bottoms" of the full-wave sine are to blame for that.

Note that using linear transformers and switching AC at the zero crossings is not guaranteed silent either. Transformers, especially non-toroidal, can hum and several Weller soldering stations show that.

[tony359]

other "real life" examples of when this issue happens.

- Equipment is unpowered but still connected to a serial port which is connected to the laptop which is connected to a device via USB which is connected to ground.
- Oscilloscope clip is still attached to the chassis.

With this I mean that there are a variety of cases where this issue happens.

[floobydust]

sorry I cross-posted your video in the other thread https://www.eevblog.com/forum/reviews/any-opinions-on-the-aixun-t3a/msg5094360/#msg5094360

What markings are on your thermocouple amp IC? "541E"? Many different op-amps are used in T12 clone controllers and I'll bet that's part of the problem. What ver controller board you got?
I would tack on a wire to observe the TC amp output with a scope and see it is probably oscillating or picking up HF noise due to a grounding or SMPS problem. Legendary firmware based on Boeing's MCAS seems to keep the displayed tip temp fake. Terrible.
I'll see if I can sketch a schematic of that controller board V1.1 based on pics to see if anything else in the hardware is wonky.

[dreamcat4]

hello i have recently checked my aixun t3a smps v2 and can say the ground is solid through the smps to the negative (ground floodplane) on the controller. which is then the gnd on the cartridge. so its a solid ground.

so like tim says, to do isolated work, consider the matter that i have tested this t3a station on dc power working (soldering things) from 24v down all the ways to about 15-16v. and it can still function and solder even that low. perhaps taking more current or less performance at lowest voltage level. then the controller board has some precise cut off like 15.25v or something particularly exact.

so for this isolation purpose: to direct power dc from a battery rather than through smps. for example can install some physical toggle switch, simply add wiring to switch and reuse existing dc power jack as instead input connector on the back (rather than output, which is not a recommended purpose). this can then isolate for this station. and then see if solves issue.

if you can open up station, then this is more easy to run a/b test case to grounded boards. to make sure that the theory works correct.

for my own purpose, i believe to mod xtc30 or similar dc power input to back, and then connect to 18v power tool battery. which is ideal because nominally the fully charged voltage on makita lxt is about 21 or 22v. then the most empty voltage coming from these battery is still 15 or 16 volts. so this is pretty good and simple. and convenient to recharge, or shared with other makita power tools. that you already have the batteries anyhow.

[c0d3z3r0]

I'll see if I can sketch a schematic of that controller board V1.1 based on pics to see if anything else in the hardware is wonky.

Well, there is a schematic for 1.2. The only difference in 1.1 was D7 or D8 being added and the Zener diode being part of the layout (not hackysoldered anymore).
https://github.com/c0d3z3r0/aixun_t3a_rev/blob/master/schematics/aixun_t3a.pdf

[c0d3z3r0]

Good teardown of a Weller station was just posted: https://www.eevblog.com/forum/reviews/weller-wxr-3-rework-station-teardown/
They are using DC and have a FET for each output (24V, 12V, and 12V). So clearly DC itself is fine. Although they've still gone with the linear transformer instead of a switch mode.

Well, what I meant is the combination of using DC + common ground for heater, tc versus AC + common ground. That makes a hughe difference. So, a relevant question is if the Weller tips have that setup (common connector).

[thm_w]

Good teardown of a Weller station was just posted: https://www.eevblog.com/forum/reviews/weller-wxr-3-rework-station-teardown/
They are using DC and have a FET for each output (24V, 12V, and 12V). So clearly DC itself is fine. Although they've still gone with the linear transformer instead of a switch mode.

Well, what I meant is the combination of using DC + common ground for heater, tc versus AC + common ground. That makes a hughe difference. So, a relevant question is if the Weller tips have that setup (common connector).

People used to debate here that DC would ruin a tip "designed" for AC which is clearly BS.

Anyway, the poster in that thread noted: "With the equipotential bonding jack (bottom right corner of the PCB), you can choose different bondings like potential free (floating tip) or a different earth reference. Some hand pieces have TIP and GND connected, others keep them apart. I don't know why that is."

The weller RT tips themselves have a common gnd, heater, and TC pin (3 pins total).

[floobydust]

I'll see if I can sketch a schematic of that controller board V1.1 based on pics to see if anything else in the hardware is wonky.

Well, there is a schematic for 1.2. The only difference in 1.1 was D7 or D8 being added and the Zener diode being part of the layout (not hackysoldered anymore).
https://github.com/c0d3z3r0/aixun_t3a_rev/blob/master/schematics/aixun_t3a.pdf

Capacitor C23 is a very bad idea, and C27 is in the wrong spot should be other side of R33. I hope neither is big like 0.1uF. The TC op-amp "541E" looks to be a TC1541 but the markings are different. No idea if this op-amp is well behaved with HF noise coming in or phase-reversal with -ve input voltages which I suspect is aggravating things here as far as the ground problem. The IC datasheet is all unicorns and rainbows which of course is la bull.

Does that make sense? HF noise coming into the op-amp (from the SMPS and Y-cap) at the tip with C23 are possibly causing it to oscillate, or the DC ground loop making it phase-invert. Then the firmware freaks out.

[dreamcat4]

it does sound like some set of possible scenarios. and especially as you noted what with how these chinese designs and components are not such a bedrock to ignore. to get to the bottom it seem affected users needs to connect up oscilloscope to there etc. and that introduction (scope probes) will not interfere to observe or ignore/dissapear such oscillations?

[c0d3z3r0]

The TC op-amp "541E" looks to be a TC1541 but the markings are different.

No. The op-amp is a AO4413. https://aosmd.com/sites/default/files/res/data_sheets/AO4413.pdf

EDIT: Sorry, brainfart....... AO4413 is the mosfet

The closest match I could find was TP1541A https://datasheet.lcsc.com/szlcsc/3PEAK-TP1542A-SR_C99005.pdf

[floobydust]

I was trying a LTSpice sim of the mosfet switch circuit and changing the resistor values helps a tiny bit for switching times- at the expense of them heating up.
More important is the heater and handle inductance causes the mosfet to ring/oscillate at RF extending turnoff. This could explain OP's observations of oddball sensitivity when the tip is being grounded in a loop. I imagine it also causes issues reading the thermocouple voltage right at the edge.

Like I've said a few times, adding a Schottky or fast recovery diode across the heater (backside of GX connector) stops all that. Try a 1N5819, SS24 or ES1B etc. what you have lying around in the junkbox.

There was drama with using cheaper TC op-amps in the T12 clones, so the 541E or 541LE mystery part could be part of the problem as well.

edit: added pic

[tony359]

meanwhile after many days of silence Aixun has replied to my message saying that their technical team are looking into the problem and trying to find a solution.

Finger crossed.

[c0d3z3r0]

Like I've said a few times, adding a Schottky or fast recovery diode across the heater (backside of GX connector) stops all that. Try a 1N5819, SS24 or ES1B etc. what you have lying around in the junkbox.

I always thought "Naaah, no problem. I don't solder live circuits anyway". Now, how shall I solder that damn diode to the positive connector to the T3A itself while it's tip is negative?! 

[floobydust]

meanwhile after many days of silence Aixun has replied to my message saying that their technical team are looking into the problem and trying to find a solution.

Finger crossed.

The "Golden Week" holiday has ended so they'll be getting back to work now...

I always thought "Naaah, no problem. I don't solder live circuits anyway". Now, how shall I solder that damn diode to the positive connector to the T3A itself while it's tip is negative?! 

You don't own enough soldering irons lol

[c0d3z3r0]

meanwhile after many days of silence Aixun has replied to my message saying that their technical team are looking into the problem and trying to find a solution.

Finger crossed.

The "Golden Week" holiday has ended so they'll be getting back to work now...

I always thought "Naaah, no problem. I don't solder live circuits anyway". Now, how shall I solder that damn diode to the positive connector to the T3A itself while it's tip is negative?! 

You don't own enough soldering irons lol

Yeah, obviously. Looks like I need to hook up my McGyver soldering station again  https://www.eevblog.com/forum/reviews/any-opinions-on-the-aixun-t3a/msg4412704/#msg4412704

[c0d3z3r0]

Like I've said a few times, adding a Schottky or fast recovery diode across the heater (backside of GX connector) stops all that. Try a 1N5819, SS24 or ES1B etc. what you have lying around in the junkbox.

Can you give more insights on how one would select the "right" diode for that job? What values are important, besides recovery time? Max. Voltage, current, peak reverse voltage?

[thm_w]

There is not a ton of value in analyzing it IMO, flooby's given recommendations that are safe: iron is 24V, so 40V is a reasonable and common max voltage rating.
1A is overkill for the amount of energy, but, that is a standard diode rating. You won't find much below 1A unless you get a tiny package.

Not quite the same as its for motors but:
https://electronics.stackexchange.com/questions/357392/selecting-flyback-diode-for-inductive-load

[tony359]

another "real world" scenario where the iron misbehaves:

- removing SMD capacitors using two irons, one being a traditional GROUNDED one. On some components the issue happens and the tip reaches 500C.

Still no answer from Aixun.

[eliocor]

Please take a look to the following post:
https://www.eevblog.com/forum/reviews/aixun-t3b-strange-behavior-(must-invert-liveneutral)/msg5135604/#msg5135604
and (if possible) check if even the T3A behave at the same manner, thanks

[floobydust]

another "real world" scenario where the iron misbehaves:

- removing SMD capacitors using two irons, one being a traditional GROUNDED one. On some components the issue happens and the tip reaches 500C.

Still no answer from Aixun.

If I owned a Aixun T3A, I would do a test if it's HF noise upsetting the temp readings. Capacitance from the tip to PE (such as when soldering a board with scenarios OP mentions), not necessarily a direct ground loop. Prob. it's own SMPS noise gets into the op-amp.
Just try it - take a 0.1uF cap and touching the tip to one lead, the other lead is grounded - should have no effect at all - unless the TC op-amp flips out, and then the software propaganda campaign cover up fails, which is what I think is going on. It ignores TC temperatures wildly out of range - causing the usual head in the sand (ostrich) problem... tip temp is skyrocketing... that crappy algo is a total fail and needs a rewrite.

Like I've said a few times, adding a Schottky or fast recovery diode across the heater (backside of GX connector) stops all that. Try a 1N5819, SS24 or ES1B etc. what you have lying around in the junkbox.

Can you give more insights on how one would select the "right" diode for that job? What values are important, besides recovery time? Max. Voltage, current, peak reverse voltage?

The back-EMF diode should be fast-recovery and take peak current of what the heater gets (for usec though) and PIV of well over 24V. 1N5819 40V 1A 25A peak 1/2 sine is plenty overkill. For SMT consider B140 or SS14 (SMA package) or bigger SMB B360, SS24 etc.
It does two things: clamps -ve spikes from ESD and the heater/cable inductance from damaging things like MCU inputs, as well as blowing the power mosfet. As well it will help with the op-amp input recovery time but it's a mystery part they are using with no noise filtering either.
Another T12 clone controller just put a 0.1uF? capacitor across the heater, as a cheap sort of snubber. We need to save micropennies here.

I say it's at least two problems - noisy hardware and the Eversolder software algo.

[tony359]

I went to make a video review today and when I tried to re-create this issue, it would not happen anymore.

the tip still overshoots, I see the spike in power but it then stops (it's likely being detected now) but if I measure with a thermometer, it's +100C than the set temp.

I re-upgraded from 1.33 to 1.34. It looks like Aixun secretly changed 1.34 to make it better!!

[tony359]

I made a video on this issue (and more generally to compare the T3A to a more traditional soldering iron). Let me know what you think!

[thm_w]

Good demo. Showing the overshoot on the display is of course better than hiding it.

[tony359]

Thank you!

Indeed. At least I know it happened and maybe give the tip a few extra seconds to cool down.

[Icchan]

I tried it on my unit. Grounded some board to mains earth and touched the tip of the iron to some ground point. I couldn't reproduce the effect. Nothing happened.

Can you check the hardware version of your unit? I have 1.0 (at least if the aixun software is to be believed) and I'm wondering if that matters..
If PCB was easier to take out (you need to heat up the front panel and tear the double sided tape, could crack the glass if not using suction cup method) I'd ask you to do that and take pictures of the PCB so we coudl check if this truly is hardware and not software issue...

Thanks