Any opinions on the AiXun T3A?

[floobydust]

I'd expect push-pull. With the two 300Ω base resistors, about 8mA. Not sure what BJT they are using, it might have a problem. Newer T12 V3.3 use a mosfet 2N7002 instead.

You can troubleshoot or verify port firmware issues by scoping the MCU heater output pin and then add a 1kΩ pullup or pull down and look at the waveform.
That can reveal if the port config is getting corrupted or it goes open-drain etc.

[c0d3z3r0]

I'd expect push-pull. With the two 300Ω base resistors, about 8mA. Not sure what BJT they are using, it might have a problem. Newer T12 V3.3 use a mosfet 2N7002 instead.

You can troubleshoot or verify port firmware issues by scoping the MCU heater output pin and then add a 1kΩ pullup or pull down and look at the waveform.
That can reveal if the port config is getting corrupted or it goes open-drain etc.

Hm, GD32F303 datasheet says +/- 25mA for IOs.

Well, the push signal (1) works at least. If that would be broken completely, it wouldn't work at all, right? So, either the pull signal (0) is broken or there is a stuck PU.

I had a similiar problem on a Intel i7 SoC where one I2C line (hm, or I2S) for the audio codec broke for unknown reasons (maybe ESD, maybe manufacturing fault).
It had a stuck pull-up breaking any communication because the pull/PD was not able to pull down to 0V. I fixed it with a 10k external pull-down.
The signal still looks a bit weird but audio on my laptop works fine again \o/

I guess I will to desolder that transistor Q4 or better R14 and test that GPIO through SWD first...

[c0d3z3r0]

Ouch! R14, R15 are 3,3k, not 330! Fixed in the schematics. https://github.com/c0d3z3r0/aixun_t3a_rev/blob/master/schematics/aixun_t3a.pdf

[floobydust]

Oh. R14, R15 3k3 gives 0.78mA MCU load and 0.56mA base current. Q4 sees around 22mA spike switching on, due to the mosfet capacitance. Q4 seen as S8050 "J3Y" or "G1" MMBT5551.
Still no idea of why the Q4 NPN base waveform looks so bad. Even if the pin tri-states the transistor turns off fine due to R15 across the E-B.

edit: I would scope the MCU output pin or at the resistor R14 ahead of Q4-B. Maybe they are trying to turn off the mosfet slow and doing a bit of PWM there.

[drksy]

Heater GPIO init in firmware v1.26:

Code: [Select]

  rcu_periph_clock_enable(RCU_GPIOA);
  gpio_init((uint32_t)&GPIOA_CTL0, GPIO_MODE_OUT_PP, GPIO_OSPEED_50MHZ, PA8_HEATER);
Heater function:

Code: [Select]

void __fastcall set_heater_timer(int cnt_of_50us)
{
  __int16 cnt_50us_0; // r4

  cnt_50us_0 = cnt_of_50us;
  if ( cnt_of_50us )
  {
    gpio_bit_set((uint32_t)&GPIOA_CTL0, PA8_HEATER);
    timer_autoreload_value_config((uint32_t)&TIMER0_CTL0, 50 * cnt_50us_0);
    timer_enable((uint32_t)&TIMER0_CTL0);
  }
  else
  {
    gpio_bit_reset((uint32_t)&GPIOA_CTL0, PA8_HEATER);
    timer_disable(&TIMER0_CTL0);
  }
}


Where are you getting this source code? Is the firmware code for let's say v1.33 available somewhere?

[floobydust]

Heater GPIO init in firmware v1.26:

Code: [Select]

  rcu_periph_clock_enable(RCU_GPIOA);
  gpio_init((uint32_t)&GPIOA_CTL0, GPIO_MODE_OUT_PP, GPIO_OSPEED_50MHZ, PA8_HEATER);
Heater function:

Code: [Select]

void __fastcall set_heater_timer(int cnt_of_50us)
{
  __int16 cnt_50us_0; // r4

  cnt_50us_0 = cnt_of_50us;
  if ( cnt_of_50us )
  {
    gpio_bit_set((uint32_t)&GPIOA_CTL0, PA8_HEATER);
    timer_autoreload_value_config((uint32_t)&TIMER0_CTL0, 50 * cnt_50us_0);
    timer_enable((uint32_t)&TIMER0_CTL0);
  }
  else
  {
    gpio_bit_reset((uint32_t)&GPIOA_CTL0, PA8_HEATER);
    timer_disable(&TIMER0_CTL0);
  }
}


Casual glance and I'm unfamiliar with the MCU - Why is the port control register passed into the bit set/reset routine, not the actual port?
Giving this (*3 lines changed):

Code: [Select]

  rcu_periph_clock_enable(RCU_GPIOA);
  gpio_init(GPIOA, GPIO_MODE_OUT_PP, GPIO_OSPEED_50MHZ, PA8_HEATER); //*
Heater function:

Code: [Select]

void __fastcall set_heater_timer(int cnt_of_50us)
{
  __int16 cnt_50us_0; // r4

  cnt_50us_0 = cnt_of_50us;
  if ( cnt_of_50us )
  {
    gpio_bit_set(GPIOA, PA8_HEATER);  //*
    timer_autoreload_value_config((uint32_t)&TIMER0_CTL0, 50 * cnt_50us_0);
    timer_enable((uint32_t)&TIMER0_CTL0);
  }
  else
  {
    gpio_bit_reset(GPIOA, PA8_HEATER); //*
    timer_disable(&TIMER0_CTL0);
  }
}


[c0d3z3r0]

Where are you getting this source code? Is the firmware code for let's say v1.33 available somewhere?

Reverse engineered (IDA Pro incl. decompiler)

[c0d3z3r0]

Casual glance and I'm unfamiliar with the MCU - Why is the port control register passed into the bit set/reset routine, not the actual port?

It's exactly the same, bc GPIOA is just a macro that results in GPIOA_CTL0 register:

Code: [Select]

GD32F30x_standard_peripheral/Include/gd32f30x_gpio.h:#define GPIOA                      (GPIO_BASE + 0x00000000U)


[tony359]

Soldering tweezers for the T3A!

@3DBeerGoggles

Shall I assume you received something a bit less wonky than I have from Handskit?

[c0d3z3r0]

Oh. R14, R15 3k3 gives 0.78mA MCU load and 0.56mA base current. Q4 sees around 22mA spike switching on, due to the mosfet capacitance. Q4 seen as S8050 "J3Y" or "G1" MMBT5551.
Still no idea of why the Q4 NPN base waveform looks so bad. Even if the pin tri-states the transistor turns off fine due to R15 across the E-B.

edit: I would scope the MCU output pin or at the resistor R14 ahead of Q4-B. Maybe they are trying to turn off the mosfet slow and doing a bit of PWM there.

Doesn't look like intentional PWM there in the code. It's just turning off the heater pin. I will do more measurements at the weekend

[3DBeerGoggles]

Yeah mine were pretty straight - if you haven't already, maybe try reseating the heating elements in case one is just in the holder crooked but yeah... aside from that it'd be time to disassemble the tweezers and see if you can't shim it straight.

[tony359]

Yeah mine were pretty straight - if you haven't already, maybe try reseating the heating elements in case one is just in the holder crooked

Thanks!
How is that element attached to the holder? I can twist them but they don't seem to move in and out.

[3DBeerGoggles]

Yeah mine were pretty straight - if you haven't already, maybe try reseating the heating elements in case one is just in the holder crooked

Thanks!
How is that element attached to the holder? I can twist them but they don't seem to move in and out.

The movable "jaws" just have a built-in set of "teeth" that hold onto the ridge on the heating element. Removal is by gently twisting and pulling at the same time until it pops past it

[tony359]

The movable "jaws" just have a built-in set of "teeth" that hold onto the ridge on the heating element. Removal is by gently twisting and pulling at the same time until it pops past it

Thanks. Mine are kept in place by friction if I see that right. Anyways yes, I can fiddle them around and "align" them.

Can I please check with you what needs to be done about wiring? I've read your post and the one you refer to but I am not 100% sure. My cable from the station has 3 wires, they go the the contacts of the heating elements.

I guess this is why I bought a spare plug, do I need to chop off the moulded one and add the resistor in there between pin 3 and 4? Then I leave everything as is - but get rid of the mercury switch?

Mercury in the handle. Mindblowing.

Thank you!

[3DBeerGoggles]

Hi - reposting from my reddit reply so it's here for others too:

> I've heard the cable varies on these tweezers. In my case, there was an extra wire hiding inside the cable (it was trimmed flush with the outer jacket) that I recall making use of - so I was able to get away with stuffing the 10K resistor inside the tweezer housing instead of replacing the molded connector BUT I do know other people had to fit a separate connector to add the resistor (The 10K Identifies to the station that it's a T12 iron)

That said, it sounds like you're on the right track. The only thing I would double check is exactly what left/middle/right connectors actually connect to on the cartridge. I can't remember if it was identical to the T12 holder Aixun uses or if the order of the connections is different. (IE make sure the "left hand" connection in the diagram goes to the lower contact).

You can see where on the cartridge each contact grabs so you should be able to double check that without it being a big problem.
So (with mind that the color code in the Hanskit cable is likely different, so double check that too):
pin 1 - NC
pin 2 and 5 - [Lower band on the cartridge]
pin 3 - to pin 4 via 10K resistor
pin 4 - [middle contact AND upper contact on the cartridge]

[drksy]

So after lots of experiments and trial and errors these past two days, I've found a solution that attenuates the buzzing to completely inaudible. Here's what happened.

Last time I cooked up the poor mosfet, so I ordered a few extra mosfets, including a replacement (SI4401DDY), and also the SD03C-7 ESD diode to protect the input for detecting when handle is in the cradle. And since this mosfet only tolerates 20V G-S, I changed R11 and R6 to both 3.3k. After that I added the diode as recommended by floobydust (https://www.eevblog.com/forum/reviews/aixun-t3a-misbehaving-on-grounded-pcbs/msg5103978/#msg5103978), and soldered 16 guage wires directly to the contacts of the main pcb:



Here's a summary of my initial findings:
- Trying a different mosfet (IRF5210) had no effect
- Slowing the switching edges heated up the mosfet, but didn't reduce the buzzing
- There was no weird artifact on the heater waveform, it was a normal looking pwm with no spikes.



- Buffering the output of Q4 has no effect.
-Replacing D8 with a 10 ohm resistor and adding a 100u cap from Vin to ground (to decouple the return paths) reduced the buzzing on the control board but the supply was still buzzing
- I measured significant droop on the power supply (3V peak-to-peak ac) in sync with the pulses



Then I thought maybe it makes sense to provide a low impedance energy reservoir for the big gulps of current in those fast PWM spikes. So I put a RC filter consisting of 0.1 ohm + 4.7mF between the power supply and the control board:



Here came the surprise. The buzzing disappeared! I put my ear really close to the supply, being careful not to get electrocuted by the live board, and could almost hear very close by, but it was no longer loud. I looked around in the parts bin and sadly didn't have many big caps to try. I found two more 1mF caps, and added those in parallel. It became pretty much inaudible at that point! So I soldered the caps to a piece of prototype board and taped it on the empty space on the power supply:



I still don't fully understand why this mitigated the buzzing. Maybe someone can explain to us. But it looks promising and may be just the solution to eliminate the buzz for good. I wonder if you can just wack a 10mF on the back of the Aixun using the DC jack without needing to even open it up or remove the smps. Just have to be careful because that big cap will essentially short the supply while it's trying to start up.

So I'm happy with this as it is. I can't hear it anymore at all. I'm thinking just to hot glue the protoboard and close this chapter. As luck would have it, the replacement for the glass screen has just arrived



though I won't glue it on just yet.

[dreamcat4]

I still don't fully understand why this mitigated the buzzing. Maybe someone can explain to us. But it looks promising and may be just the solution to eliminate the buzz for good. I wonder if you can just wack a 10mF on the back of the Aixun using the DC jack without needing to even open it up or remove the smps...

Very quick and simplez question regarding all this buzzing reports...

For you drsky (or other guys here, cz0...). Did you guys already do the SMPS full electrolytic caps replacement set? It's like the 5 main caps that you order them (from digikey etc). And replace with one of the 3-4 quality brands, and then it still was buzzing afterwards, after doing that replacement set?

Because I didn't actually even start using my aixun at all, until immediately replacing those things. So if you did replace them already. Then indeed I think you hit some good and quite sensible mod here. Which seems quite worth considering.

However if nobody who buzzes bothered replace the existing caps then... well kindda interesting to know or find out if buzzing got reduced with the quality caps. Or had nothing, no effect at all (and was still same buzz).

[c0d3z3r0]

- Trying a different mosfet (IRF5210) had no effect

Good to know!

- Slowing the switching edges heated up the mosfet, but didn't reduce the buzzing

I guess this is expected, bc the time with higher resistance is increased

- There was no weird artifact on the heater waveform, it was a normal looking pwm with no spikes.

Where exactly did you measure? I'll retry then

-Replacing D8 with a 10 ohm resistor and adding a 100u cap from Vin to ground (to decouple the return paths) reduced the buzzing on the control board but the supply was still buzzing

Be careful with D8! It protects your USB port from shorting by the PSU caps when you connect USB but not mains. Ok for testing, but D8 should be added again. Better cut the trace and add your resistor there.

- I measured significant droop on the power supply (3V peak-to-peak ac) in sync with the pulses

Yep, that's what we saw on my measurements as well. So at least that is consistent. That PSU is overloaded, which can be solved by either adding huge caps like you did or by modifying the PSU itself. The latter will be difficult since we don't really know the specs of the transformer. If the PSU really is limited to 200W and the heater draws more than that, the PSU PWM controller limits voltage. That is probably what we see here.

Then I thought maybe it makes sense to provide a low impedance energy reservoir for the big gulps of current in those fast PWM spikes. So I put a RC filter consisting of 0.1 ohm + 4.7mF between the power supply and the control board:

It would be good to know if that RC filter can be left out, when only adding your later changes (caps on the PSU).

Just have to be careful because that big cap will essentially short the supply while it's trying to start up.

Pretty sure. Something like this might do the job:

So I'm happy with this as it is. I can't hear it anymore at all. I'm thinking just to hot glue the protoboard and close this chapter. As luck would have it, the replacement for the glass screen has just arrived
though I won't glue it on just yet.

If that new glass has glue applied already, I recommend removing it (it's just double sided tape) and only use some small pieces of double sided tape just in case you need to open it again later.

[drksy]

Where exactly did you measure?

I measured it right across the diode behind the circular connector, with ground alligator on the anode and the probe clip attached to the cathode.

Be careful with D8! It protects your USB port from shorting by the PSU caps

Sure, I put it back anyway.

Something like this might do the job:

Hmm, if you put a diode on top of the cap like that, would that diode ever conduct? I suppose you want one with very low forward voltage. Have you tried it? Since you said your unit is also buzzing, I'm curious to know whether this mod works in your case.

[drksy]

Did you guys already do the SMPS full electrolytic caps replacement set? It's like the 5 main caps that you order them (from digikey etc). And replace with one of the 3-4 quality brands, and then it still was buzzing afterwards, after doing that replacement set?

I haven't, but really maybe I should have. But is it worth to spend more and more time and money on this? I'm 110 + 20 in, and probably 500 worth of my time! Hard to justify, really, other than being a learning experience, learning electronics, and learning to stick with genuine stuff 

We're recapping a supply whose designers were so cheap they took out basic heatsinks and glued the damn thing to the case. Isn't it smarter to just buy a meanwell supply? Big named brand caps aren't so cheap. Intuitively, I think recapping this supply is like putting lipstick on a pig. It's so low quality, it's not even worth it. And I remember when I was testing last night, even a big linear bench supply was buzzing. I just doubt it will do much of anything.

[dreamcat4]

ah yes because earlier versions were not glued into the case (as mine). but that aside doing the recapping was itself not actually very hard at all, and would have cost something low in parts, maybe only $5-7 (out of a digikey order for other stuff).

i don't disagree however the viewpoint it shouldn't happen, and i would be interested to hear if some guys finds entire replacement smps module. that can actually be fitted well enough (and perform well enough) into the device.

but i never regarded as high as meanwell needing such best possible quality, for this soldering job. rather not the worst quality either. but somewhere inbetween that is plenty good enough. and with the good caps,

actually come to think of it: the whole reasoning for why to replace these smps caps to begin with, it was because somebody here had identified what with the constant dumping of currents. that they would be getting worked pretty hard, and then some people took that to mean: well replace it sooner or later, and others could not be bothered and said they have general policy to replace caps only as needed. [edit] except for the fact that most other random equipment probably does not works caps anywhere as hard as this does. so perhaps worth a more careful consideration, such rules.

personally i already knew i definately wanted to do other types of mods.so it was like a 0 extra added effort, and not any bother whatsoever, to replace (5) caps. since having to open all up anyways.

and it also just seemed like with the amount of current to constantly dump, whatever the chinese factory caps they provide, i certainly am not trusting whatsoever theirs. not to throw out all the rest of the soldering iron. but surely the weakest link for what comes in the box... which is indeed those factory electrolytic caps.

so some guys also took my position (before me, actually not my idea, just to follow their recommendations). while other guys took the other position. that if it isn't broke, then don't fix it (until later, once it breaks).

but yours was indeed already broke, to force those dillemas between the choice of which hassles was least hassles. to either return it, or to try and fix it. after a newer hardware revisions where they already messed about (as china does) to make production changes. perhaps not even for reasons of bom cost purely alone. but to make them faster or other reasons to change parts or design. it is not entirely clear.

and basically yes i not longer wants to recommend this product, but most of all because handle is bad. but then you can buy without handle, but smps is glued. which is also bad things. that i don't wish at all onto others. such silent revisions changes.

[c0d3z3r0]

Where exactly did you measure?

I measured it right across the diode behind the circular connector, with ground alligator on the anode and the probe clip attached to the cathode.

Uhm, you get clean PWM on the heater output? Wat... o.O I'll check on mine.

Something like this might do the job:

Hmm, if you put a diode on top of the cap like that, would that diode ever conduct? I suppose you want one with very low forward voltage. Have you tried it? Since you said your unit is also buzzing, I'm curious to know whether this mod works in your case.

The diode is there for blocking unlimited inrush current and forces it to go through the resistor. When the voltage drops, current flows through the diode. You might be right with the low forward voltage, though.
I got this idea from there: https://www.powerctc.com/en/node/4564 They use SVC4200VB with 0.7-0.8V Vf. That should be fine but needs to be tested ofc.

I still didn't check if I have enough big caps to do any experiments Maybe I need to order a bunch.

[c0d3z3r0]

We're recapping a supply whose designers were so cheap they took out basic heatsinks and glued the damn thing to the case. Isn't it smarter to just buy a meanwell supply? Big named brand caps aren't so cheap. Intuitively, I think recapping this supply is like putting lipstick on a pig. It's so low quality, it's not even worth it. And I remember when I was testing last night, even a big linear bench supply was buzzing. I just doubt it will do much of anything.

This is exactly the reason why I refused to do this in the first place If that shitty thing dies, I'll either try to fix the caps later or just replace with MeanWell (if there is a model that fits in the case...). It might be worth checking what's so different in MeanWell SMPSs, just out of curiosity.

[c0d3z3r0]

actually come to think of it: the whole reasoning for why to replace these smps caps to begin with, it was because somebody here had identified what with the constant dumping of currents. that they would be getting worked pretty hard, and then some people took that to mean: well replace it sooner or later, and others could not be bothered and said they have general policy to replace caps only as needed. [edit] except for the fact that most other random equipment probably does not works caps anywhere as hard as this does. so perhaps worth a more careful consideration, such rules.

If these "bad" caps are the reason for that buzzing, I would agree that it's a good idea to swap them. Still unsure about this, though.
See, I'm not trying to say that you are wrong. I usually just don't replace stuff without clear reasoning. Otherwise I could also just buy a MW supply.

and basically yes i not longer wants to recommend this product, but most of all because handle is bad.

Well, you get what you pay for and if it really is a problem very much depends on how often you use it. I don't use it every day, so it might last pretty long. If you need it every day, better have a replacement at hand or buy a genuine handle. There is no one-size-fits-all recommendation. For me it (still) works.

... and still we have not seen how a genuine handle looks like inside

[vever001]

It looks like aixun just released a new station: https://www.aixuntech.com/newsinfo/aixun-t320-precision-soldering-station-is-coming-soon/
And it is now available on Ali.
Anyone tried it yet?