Any opinions on the AiXun T3A?

[SteveyG]

i do not understand why these GPIO input protections have to be so elaborate.

A simple small TVS diode should be all that's required. If you're designing from scratch you may wish to add additional protection but you would need some specific circumstances to make it worthwhile.

[dreamcat4]

wonder if these would be suitable. They are from the same series as the ones meant for the firewire device. Of course in this application there isn't any such need for high switching frequencies. And as was noted 'it bounces around a lot' with the MCU firmware doing its on software debouncing.

Anyhow here is the TVS protection device i was interested in from before (but as a single channel, instead of needing the bigger 4-channels version):

https://www.digikey.co.uk/en/products/detail/eaton-electronics-division/0603ESDA-MLP1/14645727

[al777]

@dreamcat4: careful with just this component on the GPIO pin, its datasheet lists 35V clamping - check it against the GD32F303 datasheet, where page 51 lists:

VIN:
Input voltage on 5V tolerant pin⁽³⁾     VSS - 0.3   ...  VDD + 3.6 V
Input voltage on other I/O              VSS - 0.3   ...   3.6 V

and footnote 3 lists:

⁽³⁾ VIN maximum value cannot exceed 5.5V.

This may not mix well enough in terms of protection. It will be better than nothing (35V is waaay better than several kilovolts!), but alone this particular component may not be sufficient... Just IMHO :-).

Also, if selecting a TVS with lower stand-off voltage - check its datasheet for leakage current, if I recall correctly, at least some TVS lineups show notable leakage for lower nominal stand-off voltages. If leakage current is tangible - it could pull the GPIO level down far enough to make MCU think it's a logic zero (or keep flapping in and out of zero randomly).

[yelkvi]

Then I think really Aixun should be more helpful in this regard and offer to send you a pre-flashed MCU. For a few bucks spare parts charge. If they want to preserve their firmwares.

I wrote to Aixun technical support. They didn't answer.
I described my problem here earlier.

[dreamcat4]

@dreamcat4: careful with just this component on the GPIO pin, its datasheet lists 35V clamping - check it against the GD32F303 datasheet, where page 51 lists:

VIN:
Input voltage on 5V tolerant pin⁽³⁾     VSS - 0.3   ...  VDD + 3.6 V
Input voltage on other I/O              VSS - 0.3   ...   3.6 V

and footnote 3 lists:

⁽³⁾ VIN maximum value cannot exceed 5.5V.

This may not mix well enough in terms of protection. It will be better than nothing (35V is waaay better than several kilovolts!), but alone this particular component may not be sufficient... Just IMHO :-).

Also, if selecting a TVS with lower stand-off voltage - check its datasheet for leakage current, if I recall correctly, at least some TVS lineups show notable leakage for lower nominal stand-off voltages. If leakage current is tangible - it could pull the GPIO level down far enough to make MCU think it's a logic zero (or keep flapping in and out of zero randomly).

yes thanks for bringing this up. i absolutely get what you are coming from here. and it is a legitimate concern. my own problem is that i really don't actually understand enough about this subject to be sure either way.

however it was my prior understanding that even something like a 3v3 input to a silicon die on a modern mcu will already have some modest levels of physical protection against something like a short lived transient. i.e. those do not exceed voltages in the data sheet are tested for continuus steady state situation. whereas a transient is very short lived. so in other words, so long as the tvs has a low enough pf effective capacitance. and can react fast enough. then it might still exceed 35v going up the spike before getting triggered sure... but then so long as it never reaches several kv of pd, and the transiend getts stopped  by the time it reaches lets say at 50v or 100v (as an example). then it no longer presents a high enough pd to the silicon die of the mcu to actually crack it. or break it whatever that catastrophic physical event is called. for example if the catastrophic reverse breakdown voltage of a diode (or a pn junction or whatever) is around 500v. that is going to then adequately protect the ic from the hv arcing type discharge. that jumps over and destroys that diode junction.


wheras i am assuming here (in my utter ignorance, please correct me) that those long term (steady state) volatage limit is probably more to do with a forward biased diodes or silicon junctions overheating. and then slowly burning up. like you would slowly burn and overheat a transistor.

what i am not accounting for however is the fast acting and electric field sensitivity of a fet / mosfet field gates at the gpio input of the mcu. because those are more sensitive and works differently. but presumably a fet gate must be yet another type of insulative gap to arc over. or any other such nano scale physical gaps between adjacent traces or pads at the input circuit of that mcu's gpio pin. which presumably should be able to stand more than only just a 5v over voltage (in terms of hv arcing only, and for that snubbed tansient of the modified or diverted esd discharge event spike).


so the reason that specific tvs i referenced was because it is supposed to have lower pf capacitance than the ti ones. however i remember now that is intended to improve the maximum speed of the slew rate and switching bandwidth of the valid data that can be passed though it and still remain valid signal. so it really is not matter as a desirable feature for this application. since the thing bounces around like crazy and there no point trying to preserve that.

i just am not as sure it actually matters (meaningfully compromises or reduces protections) during an actual short lived esd discharge event. maybe the datasheets for a tvs (in general) should state a minimum voltage somewhere. that then corresponds to these 3v3 cmos process technology. it just might be in the 100's of volts. haven't checked / don't know yet. not really my area of expertise unfortunately. but should really learn / look into it! thanks for bringing this up here

[dreamcat4]

ah ok... looking at the datasheet for the 0603ESDA-MLP1, it says the leakage current is less than 1nA... seems negligble. (these are supposed to be the highest possible quality TVS btw, so that does not surprise me).

* the trigger voltage being 300v (for an ESD surge). Seems in line with what you might reasonably expect
* and claims a 'less than 1nS' response time (presumably for that 300v trigger point)
* rated for at least 1000 ESD discharge events before failure

wheras the clamping voltage seems to be an independant feature. Like a 2-for-1 type deal. In this scenario having a clamping voltage... it probably does not matter?

You want a clamping voltage feature (for example). To protect connecting something that is feeding +5v signals into a 3v3 chip. So in other words to reduce component count in a design. And not need some seperate input network or other dedicated clamping circuit.

So I don't believe that would be applicable here (for the stand input). Since it is it's own self driven circuit designed. All we need is the ESD protection feature not the clamping feature here right?

However i take the point, because it could be useful for other projects when ordering a bunch of these. To have a proper 3v3 (or 3.6v whatever) clamping. For other stuff like ESP32s. Or whatever other 3v3 mcu based projects like arduino of black pill etc. Very useful.


I think the biggest thing in that datasheet is to be really careful when soldering them by hand. Not to go too hot with the iron and dump too much heat into them. Because they are such tiny SMDs and it can damage them. So maybe something else in a larger package would be better idea. Or otherwise doing some board preheating (according to their profile). And then a hot air gun. Since after soldering onto the target device it is difficult to 'test them safely'. Lets put it that way. Unless can do on some other separate test board first, and have equipment to measure them etc. like Joe smith does with his transients generator for multimeter testing.

[ramphands]

One thing I've noticed is that, when using an official JBC tip (C245-774), the station seems to cycle the tip power much more aggressively, causing the temperature to waver a lot more than the knock-off tips.  The tips that came with the station get a pulse of 1-3% power every few seconds to keep the temperature stable, but the JBC tip gets a 10-13% pulse, causing the temperature to go up and down over a 10 degree span.  Not really an issue at all, but all my other tips are rock solid temperature-wise.  Not sure if anyone else has experienced this.

On the knockoff tip front, I have the 1.6mm chisel and 3mm hoof tips from this vendor and they're really good so far, although still limited in terms of power delivery.  They max out at 25% on the power draw while even the skinny 1.2x0.3 JBC tip gets up to 40% when pushed.
https://www.aliexpress.com/item/1005002997292071.html

[al777]

@ramphands, it makes sense - looks like evidence of less thermal conduction in non-original tips. It takes only 2% of power to keep the **inside** of the tip hot with non-original tip, but with the original one because of better thermal conductivity between the heater core and outside air - it cools faster and that necessitates more power injection to keep the temperature the same... And it's supported by your second finding of max 25% vs max 40% - no matter how hard you try to cool it (by sticking the very tip into a glass of water, for example) - only a small amount of energy **can** be sucked away, so you only need 25% pulses to keep the **core** hot... not the tip... I bet if you x-ray the non-original - we would see a notable air gap between the core and the body of the tip... Sort of 900/936 heater+tip combo, sold for much more $$ :-). Still better than the true 936-tipped iron...

[al777]

ah ok... looking at the datasheet for the 0603ESDA-MLP1,

@dreamcat4: I'd try something like this:
https://www.ti.com/lit/ds/symlink/esd351.pdf
or similar from Bussman/Eaton if you prefer that brand. Don't worry about the capacitance value - super slow signal here . Do worry about the package size - this one is 0402, I believe .

[c0d3z3r0]

Hi all,

I got my T3A yesterday
Well, schematics in 3... 2... 1... https://github.com/c0d3z3r0/aixun_t3a_schematics

[c0d3z3r0]

Then I think really Aixun should be more helpful in this regard and offer to send you a pre-flashed MCU. For a few bucks spare parts charge. If they want to preserve their firmwares.

I wrote to Aixun technical support. They didn't answer.
I described my problem here earlier.

... working on that "problem"

[c0d3z3r0]

Then I think really Aixun should be more helpful in this regard and offer to send you a pre-flashed MCU. For a few bucks spare parts charge. If they want to preserve their firmwares.

I wrote to Aixun technical support. They didn't answer.
I described my problem here earlier.

... working on that "problem"

Readout protection is NOT enabled 
https://github.com/c0d3z3r0/aixun_t3a_rev

[yelkvi]

Excellent! Thank you.
Now you need to purchase a programmer for GD. I have one for STM, but it won't fit.
I'll try programming again. If it does not work out, then I will buy a new processor and flash it.
But it's not soon - I'm in no hurry. I also have other soldering irons. And also the main work that takes time.


P.S. I looked at Aliexpress programmer GD-Link. Very expensive (compared to STM programmer)

[c0d3z3r0]

Excellent! Thank you.
Now you need to purchase a programmer for GD. I have one for STM, but it won't fit.
I'll try programming again. If it does not work out, then I will buy a new processor and flash it.
But it's not soon - I'm in no hurry. I also have other soldering irons. And also the main work that takes time.


P.S. I looked at Aliexpress programmer GD-Link. Very expensive (compared to STM programmer)

The STM programmer *should* work well, too, but I guess you need to use one of the open-source tools (I saw stmflash or stm32flash or so...). Also, you can just use any FT232/FT2232H/FT4232H, or whatever you have around that is supported by OpenOCD.

[mastershake]

Hi all,

I got my T3A yesterday
Well, schematics in 3... 2... 1... https://github.com/c0d3z3r0/aixun_t3a_schematics

which board revision is this for? there are 2 different revisions. sadly all of mine are the original ones while i wish they were the newer one but im curious. ahh now i see it was rev 1.2. all of mine are on 1.1.

i wish i could get new boards for the front they seem to be okay for the most part but we did loose 2 of them now. i have not had time to go through them yet but its not the power supply on them. one will not recognize the tip inserted, tried brand new handle from another unit that has never been used before and works fine in another t3a. and the other no longer controls the temp properly. it shows proper temp on screen but turns the tip bright red. i do wonder if either of these would have been better protected with the 1.2 revision. i have been out of the office and no one is sure what happened to either. we did have some random power outages but nothing else has been damaged. i have been away from the office due to cancer treatments. so far out of the 20 we lost those 2. they did have a lot of time soldering on them though. kind of sucks if i waited like a month we would have 1.2 units not that it may have saved these 2 but i do wonder. so far no response from them when i asked if i could buy new board or even a whole new front panel for them. they only responded offering to sell me 2 units for normal price.

i have personally been mostly using the 420d at home over the t3a or t3b lately since bringing it home. i do have 3 brand new t3a i never used i may end up selling just because i have the 420d now. they do have the older stand though undecided yet.

[c0d3z3r0]

Hi all,

I got my T3A yesterday
Well, schematics in 3... 2... 1... https://github.com/c0d3z3r0/aixun_t3a_schematics

which board revision is this for? there are 2 different revisions. sadly all of mine are the original ones while i wish they were the newer one but im curious. ahh now i see it was rev 1.2. all of mine are on 1.1.

i wish i could get new boards for the front they seem to be okay for the most part but we did loose 2 of them now. i have not had time to go through them yet but its not the power supply on them. one will not recognize the tip inserted, tried brand new handle from another unit that has never been used before and works fine in another t3a. and the other no longer controls the temp properly. it shows proper temp on screen but turns the tip bright red.

Ouch. This really sounds like the ESD problems mentioned earlier in this thread... probably the GPIO/ADC got killed :/ Maybe also the op-amp... I'd suggest swapping the MCU.

i do wonder if either of these would have been better protected with the 1.2 revision.

Nope. From looking at the v1.1 pictures, it's is mostly the same:
- v1.2 has another additional reverse protection diode - for VBUS from USB-C IIRC
- v1.2 has proper pads for the Z-diode on the VR

i have been out of the office and no one is sure what happened to either. we did have some random power outages but nothing else has been damaged. i have been away from the office due to cancer treatments. so far out of the 20 we lost those 2. they did have a lot of time soldering on them though. kind of sucks if i waited like a month we would have 1.2 units not that it may have saved these 2 but i do wonder. so far no response from them when i asked if i could buy new board or even a whole new front panel for them. they only responded offering to sell me 2 units for normal price.

Yay, customer service at it's best. The front glass could probably easily be cloned with cnc milled acrylic glass, a bit of masking tape and spray paint 

i have personally been mostly using the 420d at home over the t3a or t3b lately since bringing it home. i do have 3 brand new t3a i never used i may end up selling just because i have the 420d now. they do have the older stand though undecided yet.

TBH I highly doubt the T420D is protected any better...

[dreamcat4]

ah ok... looking at the datasheet for the 0603ESDA-MLP1,

@dreamcat4: I'd try something like this: ti esd351

ok thanks for the recommendation. have now made a shortlist of 3 options for tvs diode. which are in stock at digikey...

1) the original 1st option for the eaton. (put forward by myself). it simply doesnt have a low clamping voltage. so that feature isnt going to work. but it is suitable for protecting against esd discharges. while being faster (for higher speed applications). and the package is larger. so should be a bit easier to solder

https://www.digikey.co.uk/en/products/detail/eaton-electronics-division/0603ESDA-MLP1/14645727

2) then your ti suggestion. is with a low-ish clamping voltage. and indeed it should be appropriate for most popular typical 3v3 or 5v mcus

https://www.digikey.co.uk/en/products/detail/texas-instruments/ESD351DPYR/9597828

3) then i also found this other alternative from diodes inc. which seems like it has roughly similar specs to the ti one with a low clamping voltage. but i didnt bother to check it. so it might be better or worse in some ways idk

https://www.digikey.co.uk/en/products/detail/diodes-incorporated/DESD3V3E1BL-7B/10295151

anyhow tvs diode is recommended here (for handle sense input). along with replacing those electrolytic caps in the psu. for anybody who was paying attention. and wants to get a maximum operating life from their unit

also wanted to know if we need anymore tvs elsewhere. for example on the thermocouple adc input. but it should already be buffered behind the op-amp right?

[al777]

also wanted to know if we need anymore tvs elsewhere. for example on the thermocouple adc input. but it should already be buffered behind the op-amp right?

When I check the schematics (thanks c0d3z3r0 !) - I see that there are total of three MCU-leading lines, one is the input I've put protection on (stand detect aka "Port1"), also there is "ID" and there is "Shake" . The remaining two are traveling "to outside" through the wiring towards the handle, so technically - also need to be protected. But... in order to add TVSs - one would need to access the front board, which is behind the glued front acrylic panel... Also, those inputs are in a more controlled environment, not as exposed as the back-panel stand sensing input. So, all in all I will probably pass on trying to add protection to these (and another one - temp sensing via opamp)... Too much hassle.

For myself I'm settling on just being careful ESD-wise when disconnecting (rarely if ever!) the handle connector from the base. I'm not in a high-speed manufacturing environment, so when I change the tips - I don't mind to turn off the base and do it in a controlled fashion, letting the tip to cool down before swapping it. Time for a cup of tea :-). This reduces the risk of mishaps during tip removal and insertion. And again, I rest my hands on a dissipating, grounded mat when I swap tips.

If I recall correctly, one of the users has menioned earlier in this trail that the "no dormancy" problem started once they have swapped the tips. I'm just thinking common sense here - hot tip, mechanical contacts surrounded by plastic, plus likely non-original quality of plastic and accuracy of handle manufacturing. Softened plastic moves under mechanical forces, contacts touch where they shouldn't, maybe arcing, maybe sending +24V down the signal lines... Lots of things could go wrong...

Will see how my T3A will behave under my light use, hopefully it will last. I'll keep checking this message trail as well :-).

[c0d3z3r0]

also wanted to know if we need anymore tvs elsewhere. for example on the thermocouple adc input. but it should already be buffered behind the op-amp right?

When I check the schematics (thanks c0d3z3r0 !) - I see that there are total of three MCU-leading lines, one is the input I've put protection on (stand detect aka "Port1"), also there is "ID" and there is "Shake" . The remaining two are traveling "to outside" through the wiring towards the handle, so technically - also need to be protected. But... in order to add TVSs - one would need to access the front board, which is behind the glued front acrylic panel... Also, those inputs are in a more controlled environment, not as exposed as the back-panel stand sensing input. So, all in all I will probably pass on trying to add protection to these (and another one - temp sensing via opamp)... Too much hassle.

The inputs on the connector probably are not that much affected by air discharge, as the spark would hopefully just go to the case. Also, one usually doesn't touch the pins (human body model), not even inadvertently I guess, and even then one would touch the case first anyway). But the handle might still be loaded and cause contact discharge (charged device model).

The ESD protection of the MCU can handle 4kV in human body model and 750V in charge device model. I wouldn't bet on it tbh.

The stand detect input (PORT1) is heavily at risk of both air and contact discharge in both models.

Now, what I did was this:
- slaughter a dead lenovo x260 mainboard to get two of the TVS packages (IP4223CZ6) and a common mode choke
- place one on the four inputs near the connector, before the diodes
- place the other on USB data lines; also cut the USB data lines and insert that common mode choke
- cut the traces between diodes/pull-up of ID, PORT1, SHAKE and the MCU and place 0201 series resistors:
  - 10k for PORT1 - max of what was possible without breaking the signal completely (100k was toooo much, 50k either)
  - 1k for SHAKE (probably 10k would work, too, but 1k should be more than enough here and I can't test it anyways due to using T245 only)
  - 200R for ID (to not influence the detection via voltage divider too much)
- for PORT1: move pull-up R26 (and rip off pad... R.I.P. pad) between diode D5 and series resistor to not build a voltage divider, which would break the active-low signal
- patch the firmware to *not* use an additional pull-up on the GPIO (again, this would result in a voltage divider and break the active-low signal)
- find out that the firmware does a crc16 check and patch that as well....

Maybe 100R would have been enough but since 10k/1k/200R worked I didn't want to try lower values.

For the TEMP input I only have connected the TVS, without adding a series resistor. I a new design one would want to protect the MCU from a failing OpAmp, too. However, this would need more firmware patching. Thus, I just hope that that the TVS is enough here...

Some pictures from this odyssey:

First, I had to solve a chicken-egg-problem  Yeeeeaaah.... I've chosen the cheap tip, not my shiny new originals 



This definitely did not become my best solder work, but what do you expect with that hacky "solder station"? 

For myself I'm settling on just being careful ESD-wise when disconnecting (rarely if ever!) the handle connector from the base. I'm not in a high-speed manufacturing environment, so when I change the tips - I don't mind to turn off the base and do it in a controlled fashion, letting the tip to cool down before swapping it. Time for a cup of tea :-). This reduces the risk of mishaps during tip removal and insertion. And again, I rest my hands on a dissipating, grounded mat when I swap tips.

Well, I don't want to disappoint you, but shit happens Also, see above (charged device model).

If I recall correctly, one of the users has menioned earlier in this trail that the "no dormancy" problem started once they have swapped the tips. I'm just thinking common sense here - hot tip, mechanical contacts surrounded by plastic, plus likely non-original quality of plastic and accuracy of handle manufacturing. Softened plastic moves under mechanical forces, contacts touch where they shouldn't, maybe arcing, maybe sending +24V down the signal lines... Lots of things could go wrong...

Will see how my T3A will behave under my light use, hopefully it will last. I'll keep checking this message trail as well :-).

[al777]

The stand detect input (PORT1) is heavily at risk of both air and contact discharge in both models.

...

Well, I don't want to disappoint you, but shit happens Also, see above (charged device model).

That's exactly why I did protect the stand input with optocoupler plus ESD protection of the +24V rail I had to use for the LED side of the optocoupler. From a risk management perspective - I'm satisfied with this balance.

BTW, the Charged Device Model implies transition of: "no contact with ground, no potential" --> "potential build-up (via tribocharging/induction)" --> "contact with a path to ground, discharge!". With the handle attached nearly permanently to the base, chances are there is always a path for potential to continuously flow towards ground, without much chance for high-level potential to build up... those "T4" reverse-protection diodes (likely equivalent of 1N4148) are not ideal, they do have reverse leakage current of at least few uA, so any slow electrostatic potential buildup (regardless of + or - ) will likely leak via that diode, 4K7 resistor, to power rail, through the power supply circuits, towards ground. At 5uA (that is at 75V reverse, at 20V reverse it leaks less, but still not zero) - it's equivalent of 15 megaohm.

The primary risk for those pins is when the handle gets connected / disconnected. Keep it connected and when you do disconnect - don't stick your charged pinky into the connector on the base (or touch its rim first) :-). With connect/disconnect transactions minimized - the risk is minimized as well. I also suspect that in T245 handle those detection wires either are not connected or looped to the ground right in the handle's connector (except the T sensor) - not much point in running long wires to the actual handle, our friends in China would rather save $1 on that extra copper :-). Even less chance of static induction.

NOW, you've provoked me to think outside of the box, literally. I got an idea now - we can introduce protection without ripping the front panel! I'm now thinking about finding two connectors (male & female) matching the ones used on the base & handle. Between those connectors we can place proper TVSs between each exposed MCU line and ground, maybe in a tiny metal enclosure (similar to RF filters/measurement equipment input dividers, etc.). Then the new "adapter" can be left permanently attached to the base and the exposed side will have "TVS-ed" lines only... Hmm...  :-)

I admire your determination to make AiXun's product better. Now that you've modified the firmware - you may want to think about how to prevent auto firmware update. If AiXun software (Windows) doesn't make any further checks (like refusing to update a firmware with unknown checksum/hash) - it may silently overwrite the modified firmware with a newer version, on connection - your modifications could be lost then... Or just remember to never install their software again :-). Good luck in your endeavours!

[dreamcat4]

thinking about finding two connectors (male & female) matching the ones used on the base & handle. Between those connectors we can place proper TVSs between each exposed MCU line and ground, maybe in a tiny metal enclosure (similar to RF filters/measurement equipment input dividers, etc.). Then the new "adapter" can be left permanently attached to the base and the exposed side will have "TVS-ed" lines only... Hmm...  :-)

see i like this thinking. however (for completely un-related reasons).... i did not want to put that extra little PCB in the back of the main base unit for myself. because (in my mind) I was hoping that perhaps there was enough room back there to somehow relocated the GH5 connector from the front of the unit. to the back of the unit. just to make the wiring more neater.

however also another idea I was hoping might be possible was making the cradle an extension for that same GH5. and pass thru the handle wire via the cradle.... except this t3a cradle is made of plastic and does not actually have a gh5 connector on it. But there is another version of the stand (perhaps for t3b?) that does?

[edit] and possibly changing the gh-5 aviation connector for a gh-6. then no longer need that extra seperate wire for the idle detection (labelled in port 1). because then you can remove that jack from the back of the base station and make some room to relocate the gh-5 (---> gh-6) connector itself. maybe also remove the other +24v dc output barrel jack. and or remove or relocate the esd jack on the read panel too

anyhow... so with either of those above 2 mods. then perhaps might also give the necessary opportunity to add those TVS diodes on the GH pins? to be a bit less clunky?

[c0d3z3r0]

The stand detect input (PORT1) is heavily at risk of both air and contact discharge in both models.

...

Well, I don't want to disappoint you, but shit happens Also, see above (charged device model).

That's exactly why I did protect the stand input with optocoupler plus ESD protection of the +24V rail I had to use for the LED side of the optocoupler. From a risk management perspective - I'm satisfied with this balance.

I see. Well, there are most often many ways :-)

BTW, the Charged Device Model implies transition of: "no contact with ground, no potential" --> "potential build-up (via tribocharging/induction)" --> "contact with a path to ground, discharge!".

Correct. This is exactly what applies to the unconnected handle aka charged device.

With the handle attached nearly permanently to the base, chances are there is always a path for potential to continuously flow towards ground, without much chance for high-level potential to build up...

"nearly"? That one time it isn't could kill the station. I agree, from a risk management perspective, though

those "T4" reverse-protection diodes (likely equivalent of 1N4148) are not ideal, they do have reverse leakage current of at least few uA, so any slow electrostatic potential buildup (regardless of + or - ) will likely leak via that diode, 4K7 resistor, to power rail, through the power supply circuits, towards ground. At 5uA (that is at 75V reverse, at 20V reverse it leaks less, but still not zero) - it's equivalent of 15 megaohm.

The primary risk for those pins is when the handle gets connected / disconnected. Keep it connected and when you do disconnect - don't stick your charged pinky into the connector on the base (or touch its rim first) :-). With connect/disconnect transactions minimized - the risk is minimized as well. I also suspect that in T245 handle those detection wires either are not connected or looped to the ground right in the handle's connector (except the T sensor) - not much point in running long wires to the actual handle, our friends in China would rather save $1 on that extra copper :-). Even less chance of static induction.

NOW, you've provoked me to think outside of the box, literally. I got an idea now - we can introduce protection without ripping the front panel!

Removing the panel was totally easy for me, actually. I heated it up to 180°C and used such thin metal thingy sold for smartphone repair to separate it. Afterwards, I glued it back with T-7000, making any subsequent removal substantially easier.

I'm now thinking about finding two connectors (male & female) matching the ones used on the base & handle. Between those connectors we can place proper TVSs between each exposed MCU line and ground, maybe in a tiny metal enclosure (similar to RF filters/measurement equipment input dividers, etc.). Then the new "adapter" can be left permanently attached to the base and the exposed side will have "TVS-ed" lines only... Hmm...  :-)

Nice approach!

I admire your determination to make AiXun's product better. Now that you've modified the firmware - you may want to think about how to prevent auto firmware update. If AiXun software (Windows) doesn't make any further checks (like refusing to update a firmware with unknown checksum/hash) - it may silently overwrite the modified firmware with a newer version, on connection - your modifications could be lost then... Or just remember to never install their software again :-). Good luck in your endeavours!

Well, that would imply using their software for firmware updates. I just dumped and flashed back via SWD. Heh! I forgot TVS'es for the SWD wires! 

[c0d3z3r0]

thinking about finding two connectors (male & female) matching the ones used on the base & handle. Between those connectors we can place proper TVSs between each exposed MCU line and ground, maybe in a tiny metal enclosure (similar to RF filters/measurement equipment input dividers, etc.). Then the new "adapter" can be left permanently attached to the base and the exposed side will have "TVS-ed" lines only... Hmm...  :-)

see i like this thinking. however (for completely un-related reasons).... i did not want to put that extra little PCB in the back of the main base unit for myself. because (in my mind) I was hoping that perhaps there was enough room back there to somehow relocated the GH5 connector from the front of the unit. to the back of the unit. just to make the wiring more neater.

however also another idea I was hoping might be possible was making the cradle an extension for that same GH5. and pass thru the handle wire via the cradle.... except this t3a cradle is made of plastic and does not actually have a gh5 connector on it. But there is another version of the stand (perhaps for t3b?) that does?

Nice idea! You could add the connectors to the stand. I saw someone doing that... can't find it right now.

[edit] and possibly changing the gh-5 aviation connector for a gh-6. then no longer need that extra seperate wire for the idle detection (labelled in port 1). because then you can remove that jack from the back of the base station and make some room to relocate the gh-5 (---> gh-6) connector itself. maybe also remove the other +24v dc output barrel jack. and or remove or relocate the esd jack on the read panel too

anyhow... so with either of those above 2 mods. then perhaps might also give the necessary opportunity to add those TVS diodes on the GH pins? to be a bit less clunky?

Hmm, you can still use GX12-5, when using the shield as GND and the original GND pin for the stand detect

[dreamcat4]

Thanks for these help.... this was the guy here. Who modified the stand:

Inspired by cfg5 user I made modification to new aixun t3a stand. I like the fact that I can change tips quickly same as with jbc stand.

Fantastic. Definately I want to try to do the same thing on mine.

However we still would prefer to need the tvs diodes inside the base unit, to be closer to the mcu and protect against hands touching of those pins on the gx12-5 connector. So I will take apart the back of my base station. And see how much room is there. Probably not very much. Because to retro fit a gz12-5 at the rear, it requires some depth to accommodate that panel mount connector.

[edit]

Yes, after removal of the +24v barrel jack (for aux / optional power out). Then this leaves a factory hole of diameter 12.0mm. The standard (same) gx12-5 pcb mount connector, will slide exactly though into this factory hole. It's almost like they designed it that way. To give an option. But then changed their minds and put the gx12-5 on the front panel instead. Where it can be relocated on the back... the depth will intrude 13.0mm into the chassis rear cavity. There is then a further 6.0mm of clearance before reaching the edge of the PSU pcb (at 19mm behind the rear panel).

If you solder the gx12-5 onto a little daughter board pcb. Or veroboard. Things like this. For then attaching TVS diodes and relay wires to take it to the front control board. Then you will still have the full 6.0mm of clearance remaining to insert some insulation sheet or plastic cards. To keep it isolated away from the 240v ac mains PSU.

Another mod i want to look into is push rod to back panel power switch. However that requires getting out the hot air station. To remove glue of the front plastic.

[c0d3z3r0]

Another mod i want to look into is push rod to back panel power switch. However that requires getting out the hot air station. To remove glue of the front plastic.

No plastic, it's actually real glass!